If I put my notes here, I might be able to find them again later!
This vs. that.
47 Labs 0247 vs. RJM Audio Sapphire
I wasn't even aware that the folks at 47 Labs were developing a headphone amplifier until after the Sapphire was finished, and I'm sure they didn't know what I was working on, so it's all the more remarkable just how similar the two designs are. Two separate answers from two separate people sharing the same general design philosophy.
Similarities:
- Solid state op amp voltage stage front end and solid state push-pull buffer running open loop for the output stage.
- Gain of 5-6x (14-16 dB)
- 20-30 mA bias current in the output stage.
- All BJT circuit** (see footnote)
- Use of "diamond buffer" circuit element. (Albeit in very different ways)
- 10 V voltage rails, split supply.
Differences:
- My voltage stage is an OPA134, or any 8 pin DIP op amp IC. 47 Labs voltage stage is a fully discrete current feedback op amp with a diamond buffer derived input stage. Point to 47 labs.
- My output stage is a classic diamond buffer, with thermal feedback to keep the bias current stable. 47 Labs output stage is a Sziklai pair. Both are 4 transistor, unity gain designs. I'll give myself half a point here.
- My power supply is dual mono, with simple regulators consisting of a filtered Zener reference voltage connected to the base of a pass transistor. 47 Labs power supply is shared and unregulated, with separate two-stage CRCRC filters for each channel. The output stage is tapped off the first RC filter, which the voltage stage is connected to the second filter stage for lower noise and additional isolation. Evaluation here is subjective, depends on what you think is important. Both are perfectly sufficient and both are well optimized. While some people might prefer RC filtering as being "cleaner and quieter" others might prefer the "directness" of the low impedance, regulated supply. In any event this is probably the biggest diverging influence on the sound of each.
- I'm using standard BD135,136 BJTs throughout, 47 Labs provide hand-matched pairs of Toshiba and Hitachi audio transistors. Point to 47 Labs.
Any difference is sound is subtle, a matter of presentation rather than quality level.
Based on an initial impression:
The Sapphire is more powerful, more extended.
The Treasure is more rounded, more restrained.
In graphical form,
/~~~~~~~~~~~~\ is the Sapphire
and /~~~~~~~~~~\ is the Treasure
** footnote : actually with the OPA134 the Sapphire has a differential FET input stage. To get an "all BJT" I'd have to swap out the op amp. Which I did, replacing with OPA134 with the OPA27. This change pretty much neutralized any audible difference between the two amps. I need to confirm this by trying a couple more different op amps, both FET and BJT types, but so far it seems that the type of transistor used in the input stage after the volume control makes a big difference, I'd go so far as to say defining difference, to how the circuit sounds.
I wasn't even aware that the folks at 47 Labs were developing a headphone amplifier until after the Sapphire was finished, and I'm sure they didn't know what I was working on, so it's all the more remarkable just how similar the two designs are. Two separate answers from two separate people sharing the same general design philosophy.
Similarities:
- Solid state op amp voltage stage front end and solid state push-pull buffer running open loop for the output stage.
- Gain of 5-6x (14-16 dB)
- 20-30 mA bias current in the output stage.
- All BJT circuit** (see footnote)
- Use of "diamond buffer" circuit element. (Albeit in very different ways)
- 10 V voltage rails, split supply.
Differences:
- My voltage stage is an OPA134, or any 8 pin DIP op amp IC. 47 Labs voltage stage is a fully discrete current feedback op amp with a diamond buffer derived input stage. Point to 47 labs.
- My output stage is a classic diamond buffer, with thermal feedback to keep the bias current stable. 47 Labs output stage is a Sziklai pair. Both are 4 transistor, unity gain designs. I'll give myself half a point here.
- My power supply is dual mono, with simple regulators consisting of a filtered Zener reference voltage connected to the base of a pass transistor. 47 Labs power supply is shared and unregulated, with separate two-stage CRCRC filters for each channel. The output stage is tapped off the first RC filter, which the voltage stage is connected to the second filter stage for lower noise and additional isolation. Evaluation here is subjective, depends on what you think is important. Both are perfectly sufficient and both are well optimized. While some people might prefer RC filtering as being "cleaner and quieter" others might prefer the "directness" of the low impedance, regulated supply. In any event this is probably the biggest diverging influence on the sound of each.
- I'm using standard BD135,136 BJTs throughout, 47 Labs provide hand-matched pairs of Toshiba and Hitachi audio transistors. Point to 47 Labs.
Any difference is sound is subtle, a matter of presentation rather than quality level.
Based on an initial impression:
The Sapphire is more powerful, more extended.
The Treasure is more rounded, more restrained.
In graphical form,
/~~~~~~~~~~~~\ is the Sapphire
and /~~~~~~~~~~\ is the Treasure
** footnote : actually with the OPA134 the Sapphire has a differential FET input stage. To get an "all BJT" I'd have to swap out the op amp. Which I did, replacing with OPA134 with the OPA27. This change pretty much neutralized any audible difference between the two amps. I need to confirm this by trying a couple more different op amps, both FET and BJT types, but so far it seems that the type of transistor used in the input stage after the volume control makes a big difference, I'd go so far as to say defining difference, to how the circuit sounds.
Total Comments 1
Comments
-
So I've spend some time over the weekend, comparing both. If I had to score it, and I am reluctant to, I'd say Sapphire 92, Treasure 87. That's not so much an accurate assessment as it is a starting point for the discussion.
Both amps are very good, and the subtle difference in sound is within spitting distance of the differences induced by changes in the volume, so level matching is crucial ... or to put it another way, at the end of the day the differences are probably not all that important. But since it interests me as a designer, I am insisting on drilling down on this as far as possible.
Ok, so the main difference, and the one that is immediately obvious I think to anyone even on casual listening, is the relative distance to the sound-stage. With the Treasure, it's a fairly traditional you-are-here-the-music-is-over-there affair, everything is arranged out in front of you. The Sapphire moves you right up to the edge, surrounding you with the musical events in every direction.
There are arguments in favor of both, but I generally preferred the more close up, enveloping presentation of the Sapphire.
Everything else I could discern in the way of differences were much further towards the limits of detect-ability.. and I am proportionately less confident in the claims.
Depending on how one decides to spin it, the Sapphire is either more extended in the bass and treble, or has some audible distortion in both extremes that can artificially accent transients. Is that extra shimmer and slam I hear in the Sapphire really on the recording or is it added glitter and rumble produced inside the amp itself? I'm honestly still not sure, and I'm not sure whether I'd call it a good thing or a bad thing either. Overall though I's say the Sapphire is more percussive, while the Treasure is clearly smoother.
Finally, there's a matter of what you might call "grain" or "haze". That last word in clarity that obstructs the perfect realization of the music. Here both the Treasure and Sapphire amps miss out by similar measures but in rather different ways. The Treasure has a slight cloudiness in the upper midrange, sounds seem to press together and details are lost or obscured. The Sapphire is resolutely three-dimensional, with wonderful blackness between the sounds, but perversely has a residual graininess over everything that means that sounds, while more realistically delineated, are not as cleanly reproduced as by the Treasure.
I've always thought that the Sapphire amp was still needing a couple of final tweaks, I haven't had the time to do the kind of sea-trials I did with the Phonoclone for example, trying out and comparing different versions - it was a "build it once, build it twice, done!" affair.
The events of the weekend has convinced me I need to go back and try out a couple of things with the Sapphire. First and foremost a quick op amp substitution, but there are a couple of other things I could try to reign in perceived instability or interference that might be adding grain.Posted 17th January 2012 at 12:09 AM by rjm
