Have I just been living under a rock or why the hell haven't I heard about this before now!?
http://www.njr.com/MUSES/index.html
http://www.njr.com/MUSES/index.html
↧
MUSES premium audio semiconductors from NJR
↧
Refining the Open Loop Diamond Buffer Headphone Driver
A couple of years ago I built a standard op amp + diamond buffer headphone amplifier, called the Sapphire.
It's a fairly common circuit, from eBay on up its not hard to find similar designs. The Sapphire is distinguished by its full dual mono design, Zener voltage regulators (Z-regs), and by keeping the buffer circuit outside of the op amp feedback loop.
Regarding the feedback loop, there are two schools of thought:
1. Put the buffer inside the feedback loop, so the op amp does active error correction on the output signal, and the buffer provides the high speed currents required to make that correction.
2. Put the buffer outside the feedback loop to isolate the load from the op amp. The load currents are provided by the buffer, but the output signal relies entirely on the low output impedance of the buffer for accuracy.
The explanation is too long to bother with here, but I am strongly in favor of using a buffer for isolation, outside the feedback loop. This means, however, that the buffer has to do all the work by itself, without the help of global feedback. It is important that it have a low output impedance, low distortion, low noise, and high PSRR.
My original circuit was the simple four transistor four resistor diamond buffer of the LH0002. Later small resistors were added to the emitters of the driver transistors to boost the output bias current.
In this next go-round, I've replaced the emitter resistors with current sources. This provides a significant improvement in PSRR, over 20 dB in simulation. The output pair has been reinforced with a Sziklai pairing for lower distortion, and the primary output transistors five-way paralleled. The output impedance is 1~2 ohms, limited essentially by R25.
It simulates to <-100 dB harmonics for 0 dB (1 V rms) output into 60 ohms. The total circuit standing current is less than 50 mA per channel.
LTSpice files below. R5,R6 (12 ohms shown) adjust the output bias currents, and may have to be trimmed up or down slightly in practice as the simulation does not take thermal effects into account.
It's a fairly common circuit, from eBay on up its not hard to find similar designs. The Sapphire is distinguished by its full dual mono design, Zener voltage regulators (Z-regs), and by keeping the buffer circuit outside of the op amp feedback loop.
Regarding the feedback loop, there are two schools of thought:
1. Put the buffer inside the feedback loop, so the op amp does active error correction on the output signal, and the buffer provides the high speed currents required to make that correction.
2. Put the buffer outside the feedback loop to isolate the load from the op amp. The load currents are provided by the buffer, but the output signal relies entirely on the low output impedance of the buffer for accuracy.
The explanation is too long to bother with here, but I am strongly in favor of using a buffer for isolation, outside the feedback loop. This means, however, that the buffer has to do all the work by itself, without the help of global feedback. It is important that it have a low output impedance, low distortion, low noise, and high PSRR.
My original circuit was the simple four transistor four resistor diamond buffer of the LH0002. Later small resistors were added to the emitters of the driver transistors to boost the output bias current.
In this next go-round, I've replaced the emitter resistors with current sources. This provides a significant improvement in PSRR, over 20 dB in simulation. The output pair has been reinforced with a Sziklai pairing for lower distortion, and the primary output transistors five-way paralleled. The output impedance is 1~2 ohms, limited essentially by R25.
It simulates to <-100 dB harmonics for 0 dB (1 V rms) output into 60 ohms. The total circuit standing current is less than 50 mA per channel.
LTSpice files below. R5,R6 (12 ohms shown) adjust the output bias currents, and may have to be trimmed up or down slightly in practice as the simulation does not take thermal effects into account.
↧
↧
Headphone Sensitivity: AKG K702 vs. Sennheiser HD 600
I recently obtained a pair of AKG K702 headphones to complement my long-standing reference Sennheiser HD 600s. I figured since I'm building headphone amplifiers it would be a good idea to have a reference grade low impedance model as well as the high impedance HD 600s to use for evaluation.
First though, a note on sensitivity:
K702: 62 ohms, 105 SPL/V
HD600: 300 ohms, 97 dB/mW
Different units. Grrr!
At the same volume position I quickly discovered the HD 600s play slightly louder than the K702s. The datasheet values predict the K702s should be about 3 dB louder, so it seems the sensitivity is off by as much as 6 dB.
In numbers,
K702: 62 ohms, 105 SPL/V ... 93 dB/mW from datasheet, 87~89 dB/mW (99~101 SPL/V) in practice.
HD 600: 300 ohms, 97 dB/mW ... 102 SPL/V.
The K702 requires as much as ten times more power to drive than the HD 600s. The voltage sensitivity is about 3 dB lower than the HD 600, which means that a headphone amp with even more gain is needed and the HD 600 already works best with amps in the 20~26 dB range...
Take home: true to the AKG heritage, the K702 are essentially "ear speakers". Unless your headphone amp has a high gain setting and loads of clean power on tap, my advice is to stay clear. My Nexus 7, for example, is incapable of driving them to a reasonable volume level. To put it in perspective, to achieve 110 dB on the AKG K702s requires 200 mW: 3.5 V and 56 mA rms ... 80 mA peak! :eek: The same output level from the HD 600 needs 21 mW: 2.5 V and just 8 mA rms.
(At my usual listening level, I estimate the K702s require a maximum peak power of 20 mW, 25 mA peak current. That's just within the class A envelope of the Sapphire headphone amp I'm using, but well beyond the datasheet output power of most mobile devices.)
First though, a note on sensitivity:
K702: 62 ohms, 105 SPL/V
HD600: 300 ohms, 97 dB/mW
Different units. Grrr!
At the same volume position I quickly discovered the HD 600s play slightly louder than the K702s. The datasheet values predict the K702s should be about 3 dB louder, so it seems the sensitivity is off by as much as 6 dB.
In numbers,
K702: 62 ohms, 105 SPL/V ... 93 dB/mW from datasheet, 87~89 dB/mW (99~101 SPL/V) in practice.
HD 600: 300 ohms, 97 dB/mW ... 102 SPL/V.
The K702 requires as much as ten times more power to drive than the HD 600s. The voltage sensitivity is about 3 dB lower than the HD 600, which means that a headphone amp with even more gain is needed and the HD 600 already works best with amps in the 20~26 dB range...
Take home: true to the AKG heritage, the K702 are essentially "ear speakers". Unless your headphone amp has a high gain setting and loads of clean power on tap, my advice is to stay clear. My Nexus 7, for example, is incapable of driving them to a reasonable volume level. To put it in perspective, to achieve 110 dB on the AKG K702s requires 200 mW: 3.5 V and 56 mA rms ... 80 mA peak! :eek: The same output level from the HD 600 needs 21 mW: 2.5 V and just 8 mA rms.
(At my usual listening level, I estimate the K702s require a maximum peak power of 20 mW, 25 mA peak current. That's just within the class A envelope of the Sapphire headphone amp I'm using, but well beyond the datasheet output power of most mobile devices.)
↧
Gustard H10 Headphone Amplifier
Rhymes with "custard"...
Each channel has a BB OPA134PA - socketed - for voltage amplification and an eight transistor discrete buffer with 2 ea. 2SA1837. 2SC4793, C546B, C556B. Dual mono layout (more or less ... the circuit board itself is shared and not completely symmetric).
There's a pair of NE5532s at back for balanced-unbalanced input conversion, and a pair of transistors (TIP122 TIP127) with trim pots and circuitry between the input and the main amplifier which looks a bit like voltage regulators to me but I can't see where they fit in to the scheme of things.
Pretty decent quality for the price. (I paid $350/delivered)
First impressions of the sound are quite positive, plenty of depth and decent smoothness. Letting it cook overnight now and will put it in the main system tomorrow.
****
Comparison notes vs. Sapphire (with Sennheiser HD600s)
H10 immediately impresses with huge soundstage, deep and controlled bass, and low noise floor. The midrange is smooth, with just a hint of extra warmth which gives a pleasant burr to vocals. Lyrical, with a sweet touch for melody. The high frequencies are extended, again smooth, but a bit mild. I would say the amp only falls down in one area: it starts to come apart as the going gets tough. On big, loud, complex tracks the details start to smear and run into each other, and the dynamics stall out. All around excellent performance though.
In contrast the Sapphire (which I still prefer, I'm addicted to its visceral thrill) is the literal embodiment of the saying "when the going gets tough, the tough get going". There is no muddiness or congestion or strain no matter how loud its playing or how complex the source. Indeed it plays better loud. I admit though that the H10 has blacker backgrounds and generally better resolution, both spacial and tonal.
****
The final word for now: The H10 is a velvet glove, the Sapphire is a steel fist. I find there are things to like about both. Given the level of polish on the H10 (selectable gain switch, soft-start relay, overall fit'n'finish) and the low price I'm inclined to recommend it as a point-of-entry reference. I mean, you could fairly call this your baseline definition of "good" and judge other headphone amps in terms either of how close they get to the H10 (for cheaper models), or if the extra cost is worth it compared to the H10 (for more expensive gear).
Each channel has a BB OPA134PA - socketed - for voltage amplification and an eight transistor discrete buffer with 2 ea. 2SA1837. 2SC4793, C546B, C556B. Dual mono layout (more or less ... the circuit board itself is shared and not completely symmetric).
There's a pair of NE5532s at back for balanced-unbalanced input conversion, and a pair of transistors (TIP122 TIP127) with trim pots and circuitry between the input and the main amplifier which looks a bit like voltage regulators to me but I can't see where they fit in to the scheme of things.
Pretty decent quality for the price. (I paid $350/delivered)
First impressions of the sound are quite positive, plenty of depth and decent smoothness. Letting it cook overnight now and will put it in the main system tomorrow.
****
Comparison notes vs. Sapphire (with Sennheiser HD600s)
H10 immediately impresses with huge soundstage, deep and controlled bass, and low noise floor. The midrange is smooth, with just a hint of extra warmth which gives a pleasant burr to vocals. Lyrical, with a sweet touch for melody. The high frequencies are extended, again smooth, but a bit mild. I would say the amp only falls down in one area: it starts to come apart as the going gets tough. On big, loud, complex tracks the details start to smear and run into each other, and the dynamics stall out. All around excellent performance though.
In contrast the Sapphire (which I still prefer, I'm addicted to its visceral thrill) is the literal embodiment of the saying "when the going gets tough, the tough get going". There is no muddiness or congestion or strain no matter how loud its playing or how complex the source. Indeed it plays better loud. I admit though that the H10 has blacker backgrounds and generally better resolution, both spacial and tonal.
****
The final word for now: The H10 is a velvet glove, the Sapphire is a steel fist. I find there are things to like about both. Given the level of polish on the H10 (selectable gain switch, soft-start relay, overall fit'n'finish) and the low price I'm inclined to recommend it as a point-of-entry reference. I mean, you could fairly call this your baseline definition of "good" and judge other headphone amps in terms either of how close they get to the H10 (for cheaper models), or if the extra cost is worth it compared to the H10 (for more expensive gear).
↧
Z-reg II improved simple Zener voltage regulator
I've added an additional RC filter stage (R3, C4 in the schematic below) before the Zener diode, substantially reducing the amount or ripple on the transistor base by cleaning up the voltage applied to the Zener reference. (The original Z-reg is described here.)
Circuit shows C2 with a value of 300 uF. Typically much larger values are used. I kept the filter capacitance to a minimum here to show circuit working with a reasonably high ripple (1 V p-p) on the input. The rectifier diodes used here are of no particular consequence, I just wanted the simulation to generate a realistic sawtooth for the input.
***
OK, this doesn't do as much as I originally thought. The improvement is mostly below 100 Hz, whereas the ripple is mostly in the 100Hz-1kHz band. There's perhaps 3 dB less output ripple, but that's about it. You can verify this yourself in LTSpice, just cut the wire between C4 and the junction or R1-R3 and rerun the sim.
Circuit shows C2 with a value of 300 uF. Typically much larger values are used. I kept the filter capacitance to a minimum here to show circuit working with a reasonably high ripple (1 V p-p) on the input. The rectifier diodes used here are of no particular consequence, I just wanted the simulation to generate a realistic sawtooth for the input.
***
OK, this doesn't do as much as I originally thought. The improvement is mostly below 100 Hz, whereas the ripple is mostly in the 100Hz-1kHz band. There's perhaps 3 dB less output ripple, but that's about it. You can verify this yourself in LTSpice, just cut the wire between C4 and the junction or R1-R3 and rerun the sim.
↧
↧
Which is better, Sennheiser HD600 or AKG K702?
The HD600s.
Ok, so why dont you like the K702s?
I didnt say I didnt like them. Just that I think the HD600s are better.
Its pretty simple really:
The K702s have a strident, hard upper-midrange "bump" that I find disagreeable. Yes, it makes tracks sound more live, but its also fatiguing and a bit clinical, and - as many others before have noted - makes the sound overall somewhat thin. In direct comparison the HD600s seem full the point of boominess, but I'm willing to accept that midbass plumpness for the Sennheiser's warmer, luxurious midrange. In imaging, the K702s trend to a wide, distant, airy soundstage while the HD600s run towards a closed in, intimate presentation. In that sense the K702 are more like listening to speakers, and I can certainly see people being attracted to that.
These are both top-shelf headphones at the top of their game, I don't mean to imply that the AKGs are bad. The two headphones do however sound very very different, so if you like one it's unlikely you'll be satisfied with the other.
(I used both headphones with both the Sapphire and Gustard H10 amps, with essentially similar result. The Gustard narrowed the difference between the two headphones, while the Sapphire accentuated it. If I was playing matchmaker I'd probably use the K702s with a transformer coupled tube amp, or perhaps as single-ended MOSFET design. Something which would add some extra warmth and smoothness.)
Ok, so why dont you like the K702s?
I didnt say I didnt like them. Just that I think the HD600s are better.
Its pretty simple really:
The K702s have a strident, hard upper-midrange "bump" that I find disagreeable. Yes, it makes tracks sound more live, but its also fatiguing and a bit clinical, and - as many others before have noted - makes the sound overall somewhat thin. In direct comparison the HD600s seem full the point of boominess, but I'm willing to accept that midbass plumpness for the Sennheiser's warmer, luxurious midrange. In imaging, the K702s trend to a wide, distant, airy soundstage while the HD600s run towards a closed in, intimate presentation. In that sense the K702 are more like listening to speakers, and I can certainly see people being attracted to that.
These are both top-shelf headphones at the top of their game, I don't mean to imply that the AKGs are bad. The two headphones do however sound very very different, so if you like one it's unlikely you'll be satisfied with the other.
(I used both headphones with both the Sapphire and Gustard H10 amps, with essentially similar result. The Gustard narrowed the difference between the two headphones, while the Sapphire accentuated it. If I was playing matchmaker I'd probably use the K702s with a transformer coupled tube amp, or perhaps as single-ended MOSFET design. Something which would add some extra warmth and smoothness.)
↧
Doin' a "Gilmore" : a discrete transistor headphone amplifier
Recently I spent some time updating the diamond buffer of Sapphire headphone amp circuit.
And then later I stumbled on Kevin Gilmore's headphone amp circuit. Well, I mean, I'd read it before, but it had completely slipped my mind.
On seeing the Gilmore circuit again the thought process re. Sapphire went something as follows,
"Toss out op amp, convert the Gilmore dual-LTP front end to bipolar, bolt the Sapphire3 buffer stage to the back, and substitute in the Sapphire3 current sources. Wrap in a mild feedback loop."
The result is shown attached. The Vbe multiplier is still a simple resistor (R33) ... that may need to be refined to add thermal throttling.
The offset servo is not shown, but the action is shown as Vadj. Alternatively a trim pot would be placed between R30 and R32 to provide a small measure of offset adjustment.
Most of the open loop gain is controlled by R14,R15 ... it seems to me that some work could still be done in that area.
Despite going to BJT the input impedance is still high, and there's no obvious performance hits. However, you would normally add a coupling capacitor and input resistor between the input and the volume control to avoid the DC offset changing with volume position.
The PSRR is still terrible, just like the original...
And then later I stumbled on Kevin Gilmore's headphone amp circuit. Well, I mean, I'd read it before, but it had completely slipped my mind.
On seeing the Gilmore circuit again the thought process re. Sapphire went something as follows,
"Toss out op amp, convert the Gilmore dual-LTP front end to bipolar, bolt the Sapphire3 buffer stage to the back, and substitute in the Sapphire3 current sources. Wrap in a mild feedback loop."
The result is shown attached. The Vbe multiplier is still a simple resistor (R33) ... that may need to be refined to add thermal throttling.
The offset servo is not shown, but the action is shown as Vadj. Alternatively a trim pot would be placed between R30 and R32 to provide a small measure of offset adjustment.
Most of the open loop gain is controlled by R14,R15 ... it seems to me that some work could still be done in that area.
Despite going to BJT the input impedance is still high, and there's no obvious performance hits. However, you would normally add a coupling capacitor and input resistor between the input and the volume control to avoid the DC offset changing with volume position.
The PSRR is still terrible, just like the original...
↧
Refining the Open Loop Diamond Buffer Headphone Driver (RJM Audio Sapphire 3.0)
A couple of years ago I built a standard op amp + diamond buffer headphone amplifier, called the Sapphire.
My original circuit (Sapphire 1.x) was the simple four transistor four resistor diamond buffer of the LH0002. Later small resistors (Sapphire 2.0) were added to the emitters of the driver transistors to boost the output bias current.
In this next go-round (Sapphire 3.0), I've replaced the emitter resistors with current sources. This provides a significant improvement in PSRR, over 20 dB in simulation. The output pair has been reinforced in a Sziklai configuration for lower distortion, and the primary output transistors five-way paralleled for improved thermal stability. The output impedance is 1~2 ohms, limited primarily by the output resistor.
It simulates to <-100 dB harmonics for 0 dB (1 V rms) output into 60 ohms. The total circuit standing current is less than 50 mA per channel.
LTSpice files below. R5,R6 on LTSpice circuit adjust the output bias currents and are set for 40 mA total through R13,R14. The operating points of the actual circuit ended up extremely close to the simulation, with minimal thermal drift.
The Sapphire 3.0 (photo, schematic shown) fronts the buffer with a noninverting OPA134.
My original circuit (Sapphire 1.x) was the simple four transistor four resistor diamond buffer of the LH0002. Later small resistors (Sapphire 2.0) were added to the emitters of the driver transistors to boost the output bias current.
In this next go-round (Sapphire 3.0), I've replaced the emitter resistors with current sources. This provides a significant improvement in PSRR, over 20 dB in simulation. The output pair has been reinforced in a Sziklai configuration for lower distortion, and the primary output transistors five-way paralleled for improved thermal stability. The output impedance is 1~2 ohms, limited primarily by the output resistor.
It simulates to <-100 dB harmonics for 0 dB (1 V rms) output into 60 ohms. The total circuit standing current is less than 50 mA per channel.
LTSpice files below. R5,R6 on LTSpice circuit adjust the output bias currents and are set for 40 mA total through R13,R14. The operating points of the actual circuit ended up extremely close to the simulation, with minimal thermal drift.
The Sapphire 3.0 (photo, schematic shown) fronts the buffer with a noninverting OPA134.
↧
I've been cloned!
Back at the dawn of time one of the first audio circuits I worked on was the Gainclone, followed closely by The Dac of the Klones (Oh my, the nostalgia!) and of course the Phonoclone.
The VSPS was a side-project that grew out of the Phonoclone, and actually ended up first out of the gate as a working circuit.
Apart from the general design philosophy (low parts count, simplicity, careful layout and grounding) it has no particular link to 47 Labs. While the concept of a non-inverting op amp active phonostage is nothing original the circuit is mine, particularly the configuration and values of the RIAA filter which I calculated and simulated myself. The rest is an amalgam from a dozen or so different sources, textbooks, datasheets and application notes &c. All the values are quite carefully chosen.
That said I've always put the circuits and everything else on the internet, with source attribution as I felt necessary. The boards and kits came much later, and indeed I've always encouraged people to go out and make their own.
I just made two rules re. the circuits and related materials, as stated on my web page,
Don't pass them off as your own work, and don't use them for commercial gain.
So anyhow, it turns that this phono stage on eBay, the ElectronicsSalon MD-A310 (information attached below) is 100% the VSPS circuit and parts values, lifted from the VSPS web page (probably, from the looks of it, a very early version). Yep, I've been cloned., They've done their own layout though, removing the on-board voltage regulation and adding switches for the gain and cartridge loading. The assembled board sells for $12.99 including shipping. (Shipping, ebay fees, Paypal fees, the cost of the board, the OPA2134, capacitors, resistors and screw terminal blocks... never mind labor for assembly. That it can be worth someone's while to sell the module for that price .. is utterly depressing, to be frank.)
Pots and kettles, glass houses and stones: yeah, I fully realize I'm in no position to take the moral high ground on this. Fine. I don't even care. My own layout is better, and the components in the kit I sell are better. I'm fine with them offering a cut-rate alternative with the convenience of loading options. Flattered even.
It's the "don't pass them off as your own work" part that bothers me. Attribution or lack thereof. In my day job as a scientist proper attribution of other people's work and ideas is a huge deal, and I guess I just come to intuitively expect the same standards to apply to my hobbies (photography, audio) as well. ElectronicsSalon know as well as I do that even if I objected they could go about as they please so why not have the courtesy to contact me and ask my permission, or at the very least keep "RJM Audio VSPS" in the name? It seems like a small thing to ask.
The VSPS was a side-project that grew out of the Phonoclone, and actually ended up first out of the gate as a working circuit.
Apart from the general design philosophy (low parts count, simplicity, careful layout and grounding) it has no particular link to 47 Labs. While the concept of a non-inverting op amp active phonostage is nothing original the circuit is mine, particularly the configuration and values of the RIAA filter which I calculated and simulated myself. The rest is an amalgam from a dozen or so different sources, textbooks, datasheets and application notes &c. All the values are quite carefully chosen.
That said I've always put the circuits and everything else on the internet, with source attribution as I felt necessary. The boards and kits came much later, and indeed I've always encouraged people to go out and make their own.
I just made two rules re. the circuits and related materials, as stated on my web page,
Don't pass them off as your own work, and don't use them for commercial gain.
So anyhow, it turns that this phono stage on eBay, the ElectronicsSalon MD-A310 (information attached below) is 100% the VSPS circuit and parts values, lifted from the VSPS web page (probably, from the looks of it, a very early version). Yep, I've been cloned., They've done their own layout though, removing the on-board voltage regulation and adding switches for the gain and cartridge loading. The assembled board sells for $12.99 including shipping. (Shipping, ebay fees, Paypal fees, the cost of the board, the OPA2134, capacitors, resistors and screw terminal blocks... never mind labor for assembly. That it can be worth someone's while to sell the module for that price .. is utterly depressing, to be frank.)
Pots and kettles, glass houses and stones: yeah, I fully realize I'm in no position to take the moral high ground on this. Fine. I don't even care. My own layout is better, and the components in the kit I sell are better. I'm fine with them offering a cut-rate alternative with the convenience of loading options. Flattered even.
It's the "don't pass them off as your own work" part that bothers me. Attribution or lack thereof. In my day job as a scientist proper attribution of other people's work and ideas is a huge deal, and I guess I just come to intuitively expect the same standards to apply to my hobbies (photography, audio) as well. ElectronicsSalon know as well as I do that even if I objected they could go about as they please so why not have the courtesy to contact me and ask my permission, or at the very least keep "RJM Audio VSPS" in the name? It seems like a small thing to ask.
↧
↧
A headphone amplifier gain calculator
You input the headphone sensitivity and impedance, and it spits out what I think is the ideal amplifier gain.
Even if you disagree (personal preference, difference input levels, etc.), the difference will be consistent regardless of headphone as long as the specified parameters are correct.
The gain value setting is tailored to normal line level input and listening fairly loud with the volume control at 9~10 o'clock. The output series resistance is assumed to be zero ohms.
Adjust as desired, and note that 3~6 dB either way will still be a usable. If your amp has a large output series resistance the gain Av should be scaled up as,
(Routput + headphone Z)/(headphone Z)
Even if you disagree (personal preference, difference input levels, etc.), the difference will be consistent regardless of headphone as long as the specified parameters are correct.
The gain value setting is tailored to normal line level input and listening fairly loud with the volume control at 9~10 o'clock. The output series resistance is assumed to be zero ohms.
Adjust as desired, and note that 3~6 dB either way will still be a usable. If your amp has a large output series resistance the gain Av should be scaled up as,
(Routput + headphone Z)/(headphone Z)
↧
Commentary on the TDK Life on Record A33 Wireless Weatherproof Speaker
There is something freakish about a brick-sized block that sits there and plays room-filling music ... with no wires attached whatsoever and no obvious moving parts. It gave me the same "I'm living in the future!" sense of wonder I got buying my first 1 TB hard drive.
It doesn't take too much searching the internet to discover that among wireless portable (bluetooth) speakers, the TDK A33 is highly recommended for its exceptionally good sound quality. That comes with a massive caveat, however: Most of the people writing these reviews only have Bose, Beats, and the internal speaker of their iPhone as references for comparison.
So does the A33 sound good in a hifi context?
Read on to find out...
No, okay don't bother. The answer is "no".
But it doesn't sound bad. I'm listening to it now as I type this, hooked up via the AUX input to my Onkyo PCI-200SE sound card, with the A33 tucked under the desktop monitor. Perfectly clear, solid, pleasant background music. No hiss, no rattle, no boominess. Set against, say my Sennheiser HD600s driven by the Sapphire3 amp there's no stereo, no treble, no air, and no bass either. Sins of omission, all. The worst I can say against it is there is some midrange roughness as soon as the music volume goes above even a fairly low threshold.
Back off a second: this is a sub $100, tiny speaker running off a 6 AA NiMH battery pack!
This page at TDK explains some of the design philosophy, particularly the digital equalization. As you can see (diagram reproduced below) in addition to correcting for the response of the speaker itself, the overall frequency response is drastically re-shaped into something they call "TDK Signature Sound", with a sloping response over the midrange, and early cuts to both the treble and bass.
To which I say: "Give the engineers at TDK who came up with this a raise and a vacation!" Good work guys. If the sound coming out is pleasant and clear, and stands out from the competition, mission accomplished. It doesn't sound grossly manipulated, either. The treble cut gives it weight and smoothness (at the loss of "air") and the slope/bass cut keeps the boominess in check. (I suspect reflections from nearby surfaces compensate against the sloping response anyway.)
The best way I can contextualize the A33 is to frame it against what people would be using for the same purpose 10-20 years ago: ghastly plastic "PC speakers" from Creative, CD radio cassette players, or big, shiny, floppy plastic boomboxes. The A33 is so far ahead in both sound quality and convenience its like its from a different universe.
Impressive.
It doesn't take too much searching the internet to discover that among wireless portable (bluetooth) speakers, the TDK A33 is highly recommended for its exceptionally good sound quality. That comes with a massive caveat, however: Most of the people writing these reviews only have Bose, Beats, and the internal speaker of their iPhone as references for comparison.
So does the A33 sound good in a hifi context?
Read on to find out...
No, okay don't bother. The answer is "no".
But it doesn't sound bad. I'm listening to it now as I type this, hooked up via the AUX input to my Onkyo PCI-200SE sound card, with the A33 tucked under the desktop monitor. Perfectly clear, solid, pleasant background music. No hiss, no rattle, no boominess. Set against, say my Sennheiser HD600s driven by the Sapphire3 amp there's no stereo, no treble, no air, and no bass either. Sins of omission, all. The worst I can say against it is there is some midrange roughness as soon as the music volume goes above even a fairly low threshold.
Back off a second: this is a sub $100, tiny speaker running off a 6 AA NiMH battery pack!
This page at TDK explains some of the design philosophy, particularly the digital equalization. As you can see (diagram reproduced below) in addition to correcting for the response of the speaker itself, the overall frequency response is drastically re-shaped into something they call "TDK Signature Sound", with a sloping response over the midrange, and early cuts to both the treble and bass.
To which I say: "Give the engineers at TDK who came up with this a raise and a vacation!" Good work guys. If the sound coming out is pleasant and clear, and stands out from the competition, mission accomplished. It doesn't sound grossly manipulated, either. The treble cut gives it weight and smoothness (at the loss of "air") and the slope/bass cut keeps the boominess in check. (I suspect reflections from nearby surfaces compensate against the sloping response anyway.)
The best way I can contextualize the A33 is to frame it against what people would be using for the same purpose 10-20 years ago: ghastly plastic "PC speakers" from Creative, CD radio cassette players, or big, shiny, floppy plastic boomboxes. The A33 is so far ahead in both sound quality and convenience its like its from a different universe.
Impressive.
↧
Standard Resistor Values for RJM Audio Projects
From now on every effort will be made to consolidate to the following values, 1/4W metal film:
1, 4.75, 10, 47.5, 100, 150, 221, 475, 1000, 1500, 2210, 4750, 10000, 22100, 47500, and 100000.
Also, the 1/4W 47 ohm and 68k carbon comp. resistor is widely used as damping and bleeder functions, respectively.
Exceptions will be made for the RIAA eq. of the Phonoclone and VSPS circuits, and the business end of the X-reg, where specific, non-standard resistance values are required.
*****
On resistor selection-
I honestly don't know if one resistor sounds better than another. I do know, however, that one resistor can be more expensive than another. At Mouser your basic 1/4W metal film resistor can run between 5 cents and 2 dollars in 100 unit quantities. Setting aside sound quality, it's not at all clear that you are getting any kind of material benefit at all by getting the expensive parts. Instead, it really just seems to be the whim of Mouser's inventory management. The standard line items are going to be relatively cheap. Non-standard values or low volume part numbers are going to cost you.
Here at RJM Audio I've always used Vishay CMF55, but lately the prices - and the inconsistency of the pricing - has been getting out of hand. If you want to use these throughout, you'll perhaps find 70% of the resistors you need at a reasonable price, and then find the remaining ones cost 10x as much. Also, there's about a half dozen obscure variations of each part, tolerance, tempco, and packaging ... sometimes 100 ppm/C bulk KHEK will be cheaper, sometimes 50 ppm/C tape FKEB will. I was having to re-optimize my BOM line by line each time I ordered, which is hugely time consuming.
Cut to the chase, I'm throwing CMF55 under the bus and moving to KOA Speer MF 1% 100 ppm/C metal film resistors. Updated BOMs for all projects will be propagated shortly. It's cheaper, but more importantly its consistent - every value I need is available within the model range, and apart from tape vs. bulk packaging, there are no variations.
I briefly considered Vishay RN60, which Mouser also seems to keep a large stock of. Most of the values are available and the prices are reasonable. The problem is these resistors are actually CMF60 in military uniform, it's a larger 1/2W body de-rated for military use and won't fit comfortably in my boards, which are 0207 packages for the most part.
Finally if you just want to stick with standard values the CMF55 KHEK are still reasonably priced if you really want to go Vishay. Just be aware that the available values are somewhat limited and if you need anything off that range the price goes up exponentially.
1, 4.75, 10, 47.5, 100, 150, 221, 475, 1000, 1500, 2210, 4750, 10000, 22100, 47500, and 100000.
Also, the 1/4W 47 ohm and 68k carbon comp. resistor is widely used as damping and bleeder functions, respectively.
Exceptions will be made for the RIAA eq. of the Phonoclone and VSPS circuits, and the business end of the X-reg, where specific, non-standard resistance values are required.
*****
On resistor selection-
I honestly don't know if one resistor sounds better than another. I do know, however, that one resistor can be more expensive than another. At Mouser your basic 1/4W metal film resistor can run between 5 cents and 2 dollars in 100 unit quantities. Setting aside sound quality, it's not at all clear that you are getting any kind of material benefit at all by getting the expensive parts. Instead, it really just seems to be the whim of Mouser's inventory management. The standard line items are going to be relatively cheap. Non-standard values or low volume part numbers are going to cost you.
Here at RJM Audio I've always used Vishay CMF55, but lately the prices - and the inconsistency of the pricing - has been getting out of hand. If you want to use these throughout, you'll perhaps find 70% of the resistors you need at a reasonable price, and then find the remaining ones cost 10x as much. Also, there's about a half dozen obscure variations of each part, tolerance, tempco, and packaging ... sometimes 100 ppm/C bulk KHEK will be cheaper, sometimes 50 ppm/C tape FKEB will. I was having to re-optimize my BOM line by line each time I ordered, which is hugely time consuming.
Cut to the chase, I'm throwing CMF55 under the bus and moving to KOA Speer MF 1% 100 ppm/C metal film resistors. Updated BOMs for all projects will be propagated shortly. It's cheaper, but more importantly its consistent - every value I need is available within the model range, and apart from tape vs. bulk packaging, there are no variations.
I briefly considered Vishay RN60, which Mouser also seems to keep a large stock of. Most of the values are available and the prices are reasonable. The problem is these resistors are actually CMF60 in military uniform, it's a larger 1/2W body de-rated for military use and won't fit comfortably in my boards, which are 0207 packages for the most part.
Finally if you just want to stick with standard values the CMF55 KHEK are still reasonably priced if you really want to go Vishay. Just be aware that the available values are somewhat limited and if you need anything off that range the price goes up exponentially.
↧
JLH-2005 headphone amplifier
I was perusing this thread earlier today. Which led me to what I think is the original source, at least as a modern, relatively clean headphone amp version of the original original (by way of ESP).
Some comments from our own Nelson Pass are seemingly relevant.
The rough sim, below (LTSpice .zip attached), shows that the circuit performs well enough. It strikes me as something from another era though. It's using pnp transistors as some sort of precious commodity, instead of the dirt-cheap part they are today. As a result, all the voltage gain is basically achieved with just three transistors, plus the current sources. As I said, the linearity of the circuit is commendable, but the bias currents and DC offset stability (before the coupling cap, which I added to the sim - the original JLH-2005 circuit is DC coupled!) are most definitely not.
It's the nature of the beast. And while there are ways to deal with it maybe some circuits are better off retired and replaced with better, modern alternatives? As a headphone amp output stage the class A / current source configuration is fine, but trying to DC couple the output and insisting on a high gain, asymmetric BJT voltage stage just seems like more trouble than merit.
In my humble opinion single-ended output stages should be AC (transformer or capacitor) coupled and the biasing circuit should be designed to have reasonable operating point stability.
Okay ... come to think of it do my own single-ended efforts (here and here) meet my own standards? Well, at least they are AC coupled, so your headphones are safe in my hands. But also the MOSFET follower + source resistor is inherently a self-stabilizing design element. So the answer is "yes".
Some comments from our own Nelson Pass are seemingly relevant.
The rough sim, below (LTSpice .zip attached), shows that the circuit performs well enough. It strikes me as something from another era though. It's using pnp transistors as some sort of precious commodity, instead of the dirt-cheap part they are today. As a result, all the voltage gain is basically achieved with just three transistors, plus the current sources. As I said, the linearity of the circuit is commendable, but the bias currents and DC offset stability (before the coupling cap, which I added to the sim - the original JLH-2005 circuit is DC coupled!) are most definitely not.
It's the nature of the beast. And while there are ways to deal with it maybe some circuits are better off retired and replaced with better, modern alternatives? As a headphone amp output stage the class A / current source configuration is fine, but trying to DC couple the output and insisting on a high gain, asymmetric BJT voltage stage just seems like more trouble than merit.
In my humble opinion single-ended output stages should be AC (transformer or capacitor) coupled and the biasing circuit should be designed to have reasonable operating point stability.
Okay ... come to think of it do my own single-ended efforts (here and here) meet my own standards? Well, at least they are AC coupled, so your headphones are safe in my hands. But also the MOSFET follower + source resistor is inherently a self-stabilizing design element. So the answer is "yes".
↧
↧
Szekeres 2015
The circuit was originally hosted on Headwize, but the site seems to have gone offline.
It was a single stage resistively-loaded MOSFET follower, a unity gain current buffer used to drive headphones.
If I was going to built it today, I would build it as the attached schematic. It's little changed really, just an active current source replacing the source resistor. It runs about 2 watts per channel. Distortion figures aren't great, but the operating point is reasonably optimized for 16-300 ohm loads.
Mostly though the LTSpice sim is just to serve as a reminder of how poor this circuit topology is, and how well it was designed originally. Even on a "best effort" basis there's little more to eke out from it than was already in the original.
It was a single stage resistively-loaded MOSFET follower, a unity gain current buffer used to drive headphones.
If I was going to built it today, I would build it as the attached schematic. It's little changed really, just an active current source replacing the source resistor. It runs about 2 watts per channel. Distortion figures aren't great, but the operating point is reasonably optimized for 16-300 ohm loads.
Mostly though the LTSpice sim is just to serve as a reminder of how poor this circuit topology is, and how well it was designed originally. Even on a "best effort" basis there's little more to eke out from it than was already in the original.
↧
AT-HA5000
This post, about a push-pull MOSFET output stage for a headphone amp, got me thinking again about the Audio Technica AT-HA5000.
I think with any circuit like this, the differences are less about the MOSFETs and the operating points and more about the front end and what tricks you do with power supply. And, yes, making sure it doesn't go up in a puff of vaporized silicon, taking your headphones with it.
The Audio Technica schematic has nice old-school Zener regulators, a discrete JFET front end, a long tailed pair + current mirror for voltage gain and "proper" BJT driver stage. Q7 is presumably in thermal contact with Q10,11 providing overtemp protection, and the output has a protection relay for overcurrent (overvoltage?) protection.
I like it a lot, though I've not heard it: conceptually it looks rock-solid. Though 200 mA through 36 V (7.2 W heat per channel, probably ~20W idle power consumption for the unit) is perhaps a tad excessive...
I think with any circuit like this, the differences are less about the MOSFETs and the operating points and more about the front end and what tricks you do with power supply. And, yes, making sure it doesn't go up in a puff of vaporized silicon, taking your headphones with it.
The Audio Technica schematic has nice old-school Zener regulators, a discrete JFET front end, a long tailed pair + current mirror for voltage gain and "proper" BJT driver stage. Q7 is presumably in thermal contact with Q10,11 providing overtemp protection, and the output has a protection relay for overcurrent (overvoltage?) protection.
I like it a lot, though I've not heard it: conceptually it looks rock-solid. Though 200 mA through 36 V (7.2 W heat per channel, probably ~20W idle power consumption for the unit) is perhaps a tad excessive...
↧
Jenson JE-990 discrete op amp
LTSpice copy (protection diodes removed) of the original JE-990 circuit. Even with BC327/337 subbed in for all the original transistors the simulation works without further modification.
C1 seems to be critical for stability. C2 and C3 damp overshoot seen on the simulated square wave response, hinted at by the high frequency rise in frequency response shown in the screen grab below.
My impression is that this circuit is of the heavily optimized, no-stone-left-untouched variety.
Sourced from m.nats page and The John Hardy Company.
C1 seems to be critical for stability. C2 and C3 damp overshoot seen on the simulated square wave response, hinted at by the high frequency rise in frequency response shown in the screen grab below.
My impression is that this circuit is of the heavily optimized, no-stone-left-untouched variety.
Sourced from m.nats page and The John Hardy Company.
↧
capacitance measurements with the hp 4192A
Work stuff. I was writing Labview vis for an hp 4192A LF impedance analyzer and needed something to measure to check the data acquisition program. So I stuck some of my audio capacitors I happened to have into the 16047A test fixture "just to see".
I have no idea what these measurements are telling me other than yes, the 0.47 uF capacitors are indeed 0.47 uF ... up to about 0.5 MHz anyway. Maybe someone can do some technical analysis. I was struck though by just how quickly the inductance of these big film caps kicks in. As audio coupling caps they are fine, but if you are silly enough to use them as power supply bypass for example...
There are some reproducibility issues I'm still coming to grips with, but the differences shown in the plots is definitely from the capacitors themselves and not the leads or random variations. I've measured them several times over with similiar result.
I have no idea what these measurements are telling me other than yes, the 0.47 uF capacitors are indeed 0.47 uF ... up to about 0.5 MHz anyway. Maybe someone can do some technical analysis. I was struck though by just how quickly the inductance of these big film caps kicks in. As audio coupling caps they are fine, but if you are silly enough to use them as power supply bypass for example...
There are some reproducibility issues I'm still coming to grips with, but the differences shown in the plots is definitely from the capacitors themselves and not the leads or random variations. I've measured them several times over with similiar result.
↧
↧
Audio Op Amps
I'm often asked "which op amp sounds better".
The reply is usually a scowl and muttered "does it look like I care!?" Which is something of a lie... I do care about getting op amps to sound good. It's the phrasing of the question I dislike.
Op amps do not come in "good, better, best" flavors. All it is - and this is pretty obvious I would have thought but apparently not - all this is about is matching an op amp to the job it's going to do; the circuit it's going to be sitting in.
The op amp you'd choose to use as a DAC IV converter is different from the one you'd choose to back a 100k volume potentiometer in a preamp is different from the one you'd choose for an MC phono preamp input stage...
Why do you think there are like a gazillion different op amps to choose from in the first place? It's because there are about a gazillion combinations of op amp characteristics and properties ... not because companies like Nat Semi want to finely grade op amps by sound quality.
Even ignoring all the industrial, test, and measurement applications and just sticking to audio, there are still dozens of options.
Indeed many op amps are marketed as being "for audio". To the cynics among you: no, that doesn't mean the prices are inflated to market them to gullible audiophiles (well, may Muses...). In fact audio op amps are usually on the cheaper end of the price scale. These ICs are really designed for, or at least well-suited to, audio applications. And finally, no, contrary to what you might imagine, it doesn't just mean low distortion. That is just one relevant aspect of the performance, and there are many others.
Since I'm not up to writing a textbook on this, let me just state that probably most important thing is not distortion but the optimal, designed for, circuit impedances seen at the op amp inputs which for audio tend to run between hundreds of ohms to tens of kohms.
A good general-purpose audio op amp is designed to be able to handle up to, worse case, an unbalanced DC resistance of 100 kohms on one of its inputs and impedances in the tens of kohms, since a fairly typical application is a preamplifier where you have a 100k volume control, a coupling capacitor, then a 100k resistor to ground, followed by a non-inverting op amp gain stage. See the attached Bryston preamp schematic as a generic example.
While certain audio applications like IV conversion or phono preamps can use op amps designed for lower impedances, if you market an op amp as "for audio" it's going to have to have low distortion and low offsets when dumped into a circuit like the Bryston.
Audio lies just about on the borderline where both bipolar and jFET input types are in play. Both types are workable and both are common actually, though in the particular case of the Bryston, being on the upper range of impedances, FET would be most likely be a better match from a noise, distortion, and offset perspective.
And I haven't even touched on the loading, drive, PSRR, or gain-bandwidth considerations yet.
And you want to just know which op amp sounds better? Of course you do. I just hope you have a better appreciation after reading this why I don't answer such a question.
The reply is usually a scowl and muttered "does it look like I care!?" Which is something of a lie... I do care about getting op amps to sound good. It's the phrasing of the question I dislike.
Op amps do not come in "good, better, best" flavors. All it is - and this is pretty obvious I would have thought but apparently not - all this is about is matching an op amp to the job it's going to do; the circuit it's going to be sitting in.
The op amp you'd choose to use as a DAC IV converter is different from the one you'd choose to back a 100k volume potentiometer in a preamp is different from the one you'd choose for an MC phono preamp input stage...
Why do you think there are like a gazillion different op amps to choose from in the first place? It's because there are about a gazillion combinations of op amp characteristics and properties ... not because companies like Nat Semi want to finely grade op amps by sound quality.
Even ignoring all the industrial, test, and measurement applications and just sticking to audio, there are still dozens of options.
Indeed many op amps are marketed as being "for audio". To the cynics among you: no, that doesn't mean the prices are inflated to market them to gullible audiophiles (well, may Muses...). In fact audio op amps are usually on the cheaper end of the price scale. These ICs are really designed for, or at least well-suited to, audio applications. And finally, no, contrary to what you might imagine, it doesn't just mean low distortion. That is just one relevant aspect of the performance, and there are many others.
Since I'm not up to writing a textbook on this, let me just state that probably most important thing is not distortion but the optimal, designed for, circuit impedances seen at the op amp inputs which for audio tend to run between hundreds of ohms to tens of kohms.
A good general-purpose audio op amp is designed to be able to handle up to, worse case, an unbalanced DC resistance of 100 kohms on one of its inputs and impedances in the tens of kohms, since a fairly typical application is a preamplifier where you have a 100k volume control, a coupling capacitor, then a 100k resistor to ground, followed by a non-inverting op amp gain stage. See the attached Bryston preamp schematic as a generic example.
While certain audio applications like IV conversion or phono preamps can use op amps designed for lower impedances, if you market an op amp as "for audio" it's going to have to have low distortion and low offsets when dumped into a circuit like the Bryston.
Audio lies just about on the borderline where both bipolar and jFET input types are in play. Both types are workable and both are common actually, though in the particular case of the Bryston, being on the upper range of impedances, FET would be most likely be a better match from a noise, distortion, and offset perspective.
And I haven't even touched on the loading, drive, PSRR, or gain-bandwidth considerations yet.
And you want to just know which op amp sounds better? Of course you do. I just hope you have a better appreciation after reading this why I don't answer such a question.
↧
Choosing a USB DAC
My Onkyo soundcard drivers stopped working when I upgraded to Windows 10. Onkyo says they have no plans to release a patch, so I'm left with no high quality audio solution for my computer. Since I already have a good headphone amplifier, what I'm mainly looking for is a high quality line level analog output.
One options is another soundcard, the ASUS Xonar STX being the obvious choice. I dunno, it doesn't grab me.
I was thinking with going with an external box this time, connected via USB. As this opens up about a zillion options, I'm going to limit things to,
My current thinking is that while the Denon has the nicest form factor by miles, with a plastic case and external 15V wall wart power supply it isn't particular good value compared to the Lux or the TEAC 501. Unfortunately the Lux - and the Onkyo - are still stylin' like it's 1985. The TEAC units seems to be borrowing some Tascam DNA with a pro-audio vibe I personally find attractive.
*****
Update 1. The Luxman seems to get so-so reviews. It uses the PCB5102 DAC which does not inspire confidence. The TEAC UD-501 is highly rated and ticks all the right buttons with me - full marks for true dual mono circuitry. PCM1795 DACs - much better. It is available for little more than the cost of a UD-301, making it my "object of primary interest" currently. Pity it is relatively large.
Update 2. Almost got a UD-501 but declined to fight a Sunday night auction. Thinking more carefully about the UD-301. It shares most of the circuitry of the UD-501 but with a simpler power supply and cheaper casework. The main advantage is the smaller chassis.
Update 3. The ORB Jade Casa DSD just made my shortlist ... as well as the now discontinued JADE-1.
Update 4. Ordered a used JADE-1 for ~$350US (41,500 yen).
Update 5. JADE-1 delivered. Bigger and heavier than I imagined. Very high quality construction, it certainly feels like high-end audio!
Update 6. The JADE unit is good, but the DAC section sounds to my ears too much like digital. I tend to agree with some of the Japanese language reviews I've read: resolution (signal to noise, i.e. low noise floor) is exemplary but it lacks musicality. The upper midrange has a hard, overwound, shouty quality I do not care for.
Update 7. Bought a Xonar STX. I have very low expectations, but we'll give it a shot. Failing that it will have to be the TEAC unless I through caution to the wind and decide to spend a lot more money. The Meridian Prime for example.
Update 8. Xonar STX arrived today. Installed without a problem on Windows 10 64 bit. It's ... pretty decent! It's a little rough and boisterous maybe, slightly cutting while not the last word in resolution but it's makes a good first impression with lively and enjoyable sound. My Onkyo card was smoother and reserved, but probably even less resolving.
One options is another soundcard, the ASUS Xonar STX being the obvious choice. I dunno, it doesn't grab me.
I was thinking with going with an external box this time, connected via USB. As this opens up about a zillion options, I'm going to limit things to,
- Respected audio brands with a solid reputation for digital audio.
- Small enough to be placed on top of my computer case.
- $500-ish, used or new.
My current thinking is that while the Denon has the nicest form factor by miles, with a plastic case and external 15V wall wart power supply it isn't particular good value compared to the Lux or the TEAC 501. Unfortunately the Lux - and the Onkyo - are still stylin' like it's 1985. The TEAC units seems to be borrowing some Tascam DNA with a pro-audio vibe I personally find attractive.
*****
Update 1. The Luxman seems to get so-so reviews. It uses the PCB5102 DAC which does not inspire confidence. The TEAC UD-501 is highly rated and ticks all the right buttons with me - full marks for true dual mono circuitry. PCM1795 DACs - much better. It is available for little more than the cost of a UD-301, making it my "object of primary interest" currently. Pity it is relatively large.
Update 2. Almost got a UD-501 but declined to fight a Sunday night auction. Thinking more carefully about the UD-301. It shares most of the circuitry of the UD-501 but with a simpler power supply and cheaper casework. The main advantage is the smaller chassis.
Update 3. The ORB Jade Casa DSD just made my shortlist ... as well as the now discontinued JADE-1.
Update 4. Ordered a used JADE-1 for ~$350US (41,500 yen).
Update 5. JADE-1 delivered. Bigger and heavier than I imagined. Very high quality construction, it certainly feels like high-end audio!
Update 6. The JADE unit is good, but the DAC section sounds to my ears too much like digital. I tend to agree with some of the Japanese language reviews I've read: resolution (signal to noise, i.e. low noise floor) is exemplary but it lacks musicality. The upper midrange has a hard, overwound, shouty quality I do not care for.
Update 7. Bought a Xonar STX. I have very low expectations, but we'll give it a shot. Failing that it will have to be the TEAC unless I through caution to the wind and decide to spend a lot more money. The Meridian Prime for example.
Update 8. Xonar STX arrived today. Installed without a problem on Windows 10 64 bit. It's ... pretty decent! It's a little rough and boisterous maybe, slightly cutting while not the last word in resolution but it's makes a good first impression with lively and enjoyable sound. My Onkyo card was smoother and reserved, but probably even less resolving.
↧
Orb JADE-1 headphone amplifier and DAC : review
This is a headphone amplifier with digital inputs, not a DAC with a headphone jack. Though technically given equal board space, the headphone amp, with hot-running single-ended class-A output stage, is surely the centerpiece of the design. (The Asahi Kasei DAC, with MUSES01 for the I-V, is no slouch mind you.)
First impressions. It is large, solid, and very nicely made, but - after seeing the inside - rather simple, spartan even. From the DAC output to the headphone jack is just two op amps and two transistors, the op amps being shared between channels. A third op amp most likely just buffers the analog line output. Apart from the headliner MUSES01 op amp none of the parts are especially expensive, though many were clearly carefully chosen for sound quality - the 2SC5196 for example. The TE7022 USB receiver is a disappointment, as is, to be honest, the single set of power rails and the use of dual op amps shared between channels.
On a positive note, the front panel is gorgeous, if understated. The volume control knob is a work of precision machining. The controls and layout are excellent (I especially like the position of the headphone jack bottom-right, well away from the volume knob.) Too bad there is no way to adjust gain - I would have taken that over the direct/comfort switch function any day. About that, though: the "comfort mode" is not, as I had feared, a treble cut filter but appears instead to be a crossfeed. It narrows the sound stage and puts in more in front rather than inside your head. As these things go it is restrained but more of a curiosity than something I'd feel a need to use.
After some testing I've settled on listening to this connected by TosLink to the computer motherboard, with the analog outputs connected to my Sapphire3 headphone amp. I'm switching back and forth between connecting the headphones to the JADE unit and the Sapphire3.
The sound of the JADE-1 headphone output is much as expected for a single-ended output coupled through 1000 uF, i.e. Szekeres type circuit: smooth, plenty of mid-bass, subdued highs. It is nicely tuned to balance the DAC section, which has some hardness or glare in the upper midrange but is otherwise blameless apart from perhaps the slightly folded-in soundstage. The Sapphire3 has an unfair advantage of having a high gain optimized for my headphones. Still, it sounds more lively and clear. Unfortunately it also highlights the synthetic quality of the JADE-1 DAC.
I think I prefer the subtractive combination of the native Orb headphone amp, but it is a very close thing. Perhaps the most surprising thing to come out of this is how similar the JADE and Sapphire headphone amps are tonally. Both are basically transparent and allow me to hear the DAC for what it is. It's just that the Sapphire does this more rigorously.
chassis
2.5 mm aluminum top and bottom sheets, steel back plate, front panel is actually milled from a thick block of solid aluminum.
Heavy steel partitions between power supply, audio, and front panel controls.
The circuit boards are all mounted on a steel baseplate.
Milled, satin finish aluminum volume knob and buttons.
Volume control is the standard Alps blue velvet potentiometer.
power supply (analog +12 V, -12 V, and digital)
Custom R-core transformer from Kitamura Kiden (Single secondary, probably 12-0-12 VAC and 40-50 VA) Actually manufactured by Pheonix, R40 size.
JRC 7812A 7912A regulators with heatsinks.
Four very large filter capacitors, about 10,000 uF ea.
A separate small switching module powers the digital and control circuits.
digital
Tenor TE7022L USB receiver - bleh
Asahi Kasei AK4113VF receiver
[note complete board isolation between receiver and DAC]
Asahi Kasei 4382AT DAC (2 channel, 24/192 delta-sigma, 112 dB)
MUSES01 JFET dual op amp - DAC IV conversion and LP filter as per DAC datasheet, a good part of the BOM got spent on this part alone.
analog
MUSES 8820 bipolar dual op amp (NE5532A equiv.) - I'm guessing this is the buffer for the analog line output, following MUSES01.
LF412CN JFET dual op amp - the input / gain stage of the headphone amplifier, following the volume control.
2SC1815 NPN transistor, TO-99 - this is probably the driver stage of the headphone amplifier
2SC5196 NPN Toshina "triple diffused type" 40W 6A 80V - this is the output device for the headphone amplifier
68 ohm power resistor (5W?) - emitter resistor for the output stage.
Nichicon KW 1000 uF - the output coupling capacitor.
The rated output power (1100 mW/16ohms, 30mW/600 ohms) defines an envelope of 262 mA and 4.24 V ac rms output, which for a single-ended follower, sets the supply voltage to 12 V (2.8*4.24) and the bias current to 370 mA (262*1.4) The resistor is connected from the transistor base (+5.4 V) to the negative supply (-12 V). 68 ohms gives a bias current of 264 mA ... not 370 mA. So the person doing the calculations (incorrectly) set the bias current equal to the AC output current amplitude rather than the peak maximum. A common enough mistake.
Good. I think I have the output circuit worked out now: two NPN followers with simple resistive current sources, connected to the positive and negative rails. Some bias network to put the input up to around +6 V so the output coupling capacitor does not reverse polarity.
Product homepage: http://www.orb.co.jp/audio/jade.html
First impressions. It is large, solid, and very nicely made, but - after seeing the inside - rather simple, spartan even. From the DAC output to the headphone jack is just two op amps and two transistors, the op amps being shared between channels. A third op amp most likely just buffers the analog line output. Apart from the headliner MUSES01 op amp none of the parts are especially expensive, though many were clearly carefully chosen for sound quality - the 2SC5196 for example. The TE7022 USB receiver is a disappointment, as is, to be honest, the single set of power rails and the use of dual op amps shared between channels.
On a positive note, the front panel is gorgeous, if understated. The volume control knob is a work of precision machining. The controls and layout are excellent (I especially like the position of the headphone jack bottom-right, well away from the volume knob.) Too bad there is no way to adjust gain - I would have taken that over the direct/comfort switch function any day. About that, though: the "comfort mode" is not, as I had feared, a treble cut filter but appears instead to be a crossfeed. It narrows the sound stage and puts in more in front rather than inside your head. As these things go it is restrained but more of a curiosity than something I'd feel a need to use.
After some testing I've settled on listening to this connected by TosLink to the computer motherboard, with the analog outputs connected to my Sapphire3 headphone amp. I'm switching back and forth between connecting the headphones to the JADE unit and the Sapphire3.
The sound of the JADE-1 headphone output is much as expected for a single-ended output coupled through 1000 uF, i.e. Szekeres type circuit: smooth, plenty of mid-bass, subdued highs. It is nicely tuned to balance the DAC section, which has some hardness or glare in the upper midrange but is otherwise blameless apart from perhaps the slightly folded-in soundstage. The Sapphire3 has an unfair advantage of having a high gain optimized for my headphones. Still, it sounds more lively and clear. Unfortunately it also highlights the synthetic quality of the JADE-1 DAC.
I think I prefer the subtractive combination of the native Orb headphone amp, but it is a very close thing. Perhaps the most surprising thing to come out of this is how similar the JADE and Sapphire headphone amps are tonally. Both are basically transparent and allow me to hear the DAC for what it is. It's just that the Sapphire does this more rigorously.
chassis
2.5 mm aluminum top and bottom sheets, steel back plate, front panel is actually milled from a thick block of solid aluminum.
Heavy steel partitions between power supply, audio, and front panel controls.
The circuit boards are all mounted on a steel baseplate.
Milled, satin finish aluminum volume knob and buttons.
Volume control is the standard Alps blue velvet potentiometer.
power supply (analog +12 V, -12 V, and digital)
Custom R-core transformer from Kitamura Kiden (Single secondary, probably 12-0-12 VAC and 40-50 VA) Actually manufactured by Pheonix, R40 size.
JRC 7812A 7912A regulators with heatsinks.
Four very large filter capacitors, about 10,000 uF ea.
A separate small switching module powers the digital and control circuits.
digital
Tenor TE7022L USB receiver - bleh
Asahi Kasei AK4113VF receiver
[note complete board isolation between receiver and DAC]
Asahi Kasei 4382AT DAC (2 channel, 24/192 delta-sigma, 112 dB)
MUSES01 JFET dual op amp - DAC IV conversion and LP filter as per DAC datasheet, a good part of the BOM got spent on this part alone.
analog
MUSES 8820 bipolar dual op amp (NE5532A equiv.) - I'm guessing this is the buffer for the analog line output, following MUSES01.
LF412CN JFET dual op amp - the input / gain stage of the headphone amplifier, following the volume control.
2SC1815 NPN transistor, TO-99 - this is probably the driver stage of the headphone amplifier
2SC5196 NPN Toshina "triple diffused type" 40W 6A 80V - this is the output device for the headphone amplifier
68 ohm power resistor (5W?) - emitter resistor for the output stage.
Nichicon KW 1000 uF - the output coupling capacitor.
The rated output power (1100 mW/16ohms, 30mW/600 ohms) defines an envelope of 262 mA and 4.24 V ac rms output, which for a single-ended follower, sets the supply voltage to 12 V (2.8*4.24) and the bias current to 370 mA (262*1.4) The resistor is connected from the transistor base (+5.4 V) to the negative supply (-12 V). 68 ohms gives a bias current of 264 mA ... not 370 mA. So the person doing the calculations (incorrectly) set the bias current equal to the AC output current amplitude rather than the peak maximum. A common enough mistake.
Good. I think I have the output circuit worked out now: two NPN followers with simple resistive current sources, connected to the positive and negative rails. Some bias network to put the input up to around +6 V so the output coupling capacitor does not reverse polarity.
Product homepage: http://www.orb.co.jp/audio/jade.html
↧