That is a worry actually and I need to give that some more thought. The KSA1220/KSC2690 do have pretty big Cob. I was thinking that Q9/Q10 would be protected by the cascode, but I’ve probably misunderstood how this part of the design works and will need to think about this again. I could go back to TO92 devices, but Q9 in particular dissipates ~250mW and that feels a little high to me in a TO92 package. The other alternative is KSA1381/KSC3503, which are very low Cob, but I’ll not be able to get good hfe matches in the complementary pair with the devices I have and they are lower gain.
Thanks for the link to Bonsai’s slides on diodes. I have a lot of respect for his designs and have built a couple of his amps, which I’m very pleased with. I have a bag of through-hole 1N4148s, who doesn’t, but I will probably go for SMT here too, so the BAS21J look like they might be the ones.
By the way, I don’t want to pollute your thread. Would you prefer me to post any progress here, to keep this version together with your design work, or start a new thread?
Q9 and Q10 minimize the effect of output capacitance in Q7 and Q8. But the collector of Q10 is a high-impedance node with the full signal swing on it. So its output capacitance will in principle produce some distortion.
I agree the T0-92 VAS transistors dissipate a little power, and they do get warm, but not too hot, to the touch so I haven't worried about it.
Everything I read says you gotta just build and listen. My reason for doing these projects was just to find out what would happen. I can only build so many amplifiers, so the more variations people try, the more information we have about what works and what doesn't. I think maybe one day a reference project might come out of all this. For now, I absolutely don't mind people reporting their own work in this thread. The forum belongs to everyone, so have at it.
Just don't post anything about First Cycle Distortion. ;-)
I agree the T0-92 VAS transistors dissipate a little power, and they do get warm, but not too hot, to the touch so I haven't worried about it.
Everything I read says you gotta just build and listen. My reason for doing these projects was just to find out what would happen. I can only build so many amplifiers, so the more variations people try, the more information we have about what works and what doesn't. I think maybe one day a reference project might come out of all this. For now, I absolutely don't mind people reporting their own work in this thread. The forum belongs to everyone, so have at it.
Just don't post anything about First Cycle Distortion. ;-)
I thought your 250mW dissipation figure for Q9 seemed a little high so I just pulled up my SPICE simulation. It shows static current of 4.5mA per side on the VAS diff amp, and a dissipation of about 125mA in that transistor, which makes sense. Dissipation will go up with signal level, but average power should remain low unless you plan to drive the amp continuously to near-clipping with a signal generator.
When two amplifiers sound different but both measure very well, or when a less well measuring amplifier sounds better than the two well measuring ones already mentioned, we are probably missing something.
Ron Quan still believes FM distortion is a problem.
I suspect that some kind of noise modulation is a common problem. To further explain, there are two kinds of symptoms of audio-signal-modulated noise to look for: (1) a change in the FFT noise floor level when an audio signal is present as opposed to when no signal is present, and (2) noise that appears as a broadening of an FFT spectral line for fixed frequency tone.
Wouldn't be too surprising if there are some other things we could measure but usually don't.
Aaargh, don't feed the trolls in your tread!
Just try some different capacitors in your HPA.
I've understand different designs, different compensation, different wiring.
But why under at least 60-80 dBs of feedback this can be an issue?
Exactly the right question. Maybe very high loop gains can produce side effects in some cases that are not being fully characterized by typical measurements? Either that or any audible differences are purely imaginary. Choose your poison. For myself, I will choose some mix of both possibilities.
What's missing is the context in your post. If someone is searching for a better sounding amp to him personally, then the measurement of whichever amp sounds better to him is better in that context. If he is looking for better sounding amp for sound reproducing purpose, then higher output fidelity to the input source is better amp. Whether the measurement of that amp translates to audible difference or not will need to be evaluated via objective listening test, not the usual subjective auditioning that you do.
Hi Ashley. I bought some from Micross at the beginning of the year for a phono preamp and got a spare pair at the same time - just in case of any mishaps! They have changed their webstore since then though so you can't simply order direct, only by requesting a quote. I'm guessing their MOQ may well have gone up. DigiKey now sell Linear Systems stuff through their marketplace, but only on the US site it seems. It is frustrating that there are no UK distributors for modest quantities. I read somewhere that somebody had some luck with the Linear Systems distributor in Germany, but I'm not sure if they will sell to the UK. Good luck!
NAC Semi sells the Linear Systems parts in unit quantities. I have no idea what their policy is shipping overseas.
I tried looking up information on Frequency Modulation Distortion but mostly turned up links to articles on repairing tuners, LOL. It makes sense that noise will intermodulate with the signal, producing signal-dependent sidebands. One of my other interests, currently on the back burner, is RF receiver design and this is the kind of thing that radio designers worry about (e.g., phase noise and reciprocal mixing). You don't see much in the usual literature about this kind of thing. Cherry published a paper on "Anti-Distortion" and in his introduction has a good and not-too-complicated argument for why the usual linear analysis doesn't tell the whole story about the relationship between feedback and distortion.
My math skills, which were marginal at best when I was an EE student, have declined badly over the years. I didn't study control theory in college and don't have the math tools to say anything intelligent on the subject.
The output we're trying to optimize is the hidden internal state of mind of the listener, which is only observable through unreliable subjective self-reporting. The most complicated signal processing in the entire chain happens inside the brain. It's unfathomably complex and nobody really understands how it works (though I understand a lot of good work has been done on the subject, which I don't claim to know that much about).
Many people confidently declare that simpler circuits sound better. My original goal for building HPA1/2 was to see whether or not adding a lot of transistors compared to the baseline DCG3 would yield an improvement or a reduction (or no difference at all) in sound quality. I don't claim the work I did is optimal in any way. The results are satisfactory to me, but I encourage people to keep trying different things, as I keep saying.
I tried looking up information on Frequency Modulation Distortion but mostly turned up links to articles on repairing tuners, LOL. It makes sense that noise will intermodulate with the signal, producing signal-dependent sidebands. One of my other interests, currently on the back burner, is RF receiver design and this is the kind of thing that radio designers worry about (e.g., phase noise and reciprocal mixing). You don't see much in the usual literature about this kind of thing. Cherry published a paper on "Anti-Distortion" and in his introduction has a good and not-too-complicated argument for why the usual linear analysis doesn't tell the whole story about the relationship between feedback and distortion.
My math skills, which were marginal at best when I was an EE student, have declined badly over the years. I didn't study control theory in college and don't have the math tools to say anything intelligent on the subject.
The output we're trying to optimize is the hidden internal state of mind of the listener, which is only observable through unreliable subjective self-reporting. The most complicated signal processing in the entire chain happens inside the brain. It's unfathomably complex and nobody really understands how it works (though I understand a lot of good work has been done on the subject, which I don't claim to know that much about).
Many people confidently declare that simpler circuits sound better. My original goal for building HPA1/2 was to see whether or not adding a lot of transistors compared to the baseline DCG3 would yield an improvement or a reduction (or no difference at all) in sound quality. I don't claim the work I did is optimal in any way. The results are satisfactory to me, but I encourage people to keep trying different things, as I keep saying.
I've been reading Audio Science Review a lot lately. I like the idea of ASR, but I think in practice the site is a sad joke, borderline cargo cult, and super-spreader of Dunning Kruger syndrome. On the other hand, there are a few very talented people over there (I don't consider site owner Amir M. to be one of them) so the site isn't completely worthless. I'm on the fence with respect to the ultimate value of a measurements-only approach to equipment evaluation. It seems fatally inconsistent to rank components according to their measured performance, while at the same time declaring all components that exceed an arbitrary threshold of SINAD to sound the same. I would go over there and tell them so, but it would just make heads explode and I'm not in the mood for arguing with closed-minded idiots.
I think it's plausible that all amplifiers sound the same. I also think it's plausible that they don't. When I think about it, it makes my head hurt. The question isn't easy. I will claim that anyone who says it's straightforward immediate disqualifies him or herself from serious consideration of their viewpoint.
This is a rant (sorry) but IMHO, the main purpose of ASR is to build tribal following that Amir will one day monetize for several tens of millions of dollars. Also, to feed the owner's narcissism. There, I said it.
Edit: Also, to serve as free advertising for Chi-Fi, to the detriment of domestic equipment manufacturers.
I think it's plausible that all amplifiers sound the same. I also think it's plausible that they don't. When I think about it, it makes my head hurt. The question isn't easy. I will claim that anyone who says it's straightforward immediate disqualifies him or herself from serious consideration of their viewpoint.
This is a rant (sorry) but IMHO, the main purpose of ASR is to build tribal following that Amir will one day monetize for several tens of millions of dollars. Also, to feed the owner's narcissism. There, I said it.
Edit: Also, to serve as free advertising for Chi-Fi, to the detriment of domestic equipment manufacturers.
LOL. I just posted something mildly inflammatory over there in response to a question about the Mjolnir "Pure Bipolar" headphone amplifier. I'm gonna regret doing that, I'm sure.
You probably figured out I know a fair amount about audio design, but also know how much more I still have to learn. Whenever I see people who claim to have all the tidy answers, it just makes me want to take them down. I acknowledge this is a personality flaw on my part.
Did I mentioned I posted a lengthy description of my HPA2 over there? Predictably, someone said I should "send it to Amir for measurement." My response was, "I'm not gonna do that because it would serve no purpose." I would love to have an Audio Precision analyzer because it would give me more insight into how my circuits work. But I don't think it would tell me a single thing about how good they are at playing music.
I'm doing this because, TBH, the world is going to hell and it makes me feel a little better to have something relatively harmless to think about and give me a tiny sense of control.
You probably figured out I know a fair amount about audio design, but also know how much more I still have to learn. Whenever I see people who claim to have all the tidy answers, it just makes me want to take them down. I acknowledge this is a personality flaw on my part.
Did I mentioned I posted a lengthy description of my HPA2 over there? Predictably, someone said I should "send it to Amir for measurement." My response was, "I'm not gonna do that because it would serve no purpose." I would love to have an Audio Precision analyzer because it would give me more insight into how my circuits work. But I don't think it would tell me a single thing about how good they are at playing music.
I'm doing this because, TBH, the world is going to hell and it makes me feel a little better to have something relatively harmless to think about and give me a tiny sense of control.
Maybe its more likely that they will sound the same to a listener who is strongly expecting them the sound the same based on measurements held as a matter of faith to be infallible. After all, humans can and do fool themselves sometimes. High intelligence does not render one immune, rather it can have the opposite effect: Highly intelligent people are better at thinking up long lists of all the reasons why they are right, even when they are wrong. So, care has to be be taken no matter what, whether relying on measurements, listening tests, or some combination of both.
The OP has shown a design and implementation of a high quality headphone amplifier so I have no trouble accepting his statements of the sound quality. However when somebody known only to butchered chinese abominations with stuff hanging from wires makes a statement of sound quality I have my doubts.
Thanks for the comments, folks. I'm gratified by the interest, and humbled by the depth of knowledge and experience here. I lie awake in a sweat at night (JK, not really, but maybe a little) worried people will build my amp and then get angry at me for wasting their time and money, LOL. This is why I keep warning people not to trust my subjective findings. I figure with enough disclaimers I can claim plausible deniability.
I watched a movie last night ("Logan." Such a sad ending.) I run a Ropieee endpoint into a Topping DAC and then into the HPA2, casting via Shairplay from my MacBook. Not hi-fi, but when I was done I thought to myself how awesome it was that I didn't need to use my fire extinguisher at any point during the performance.
I watched a movie last night ("Logan." Such a sad ending.) I run a Ropieee endpoint into a Topping DAC and then into the HPA2, casting via Shairplay from my MacBook. Not hi-fi, but when I was done I thought to myself how awesome it was that I didn't need to use my fire extinguisher at any point during the performance.
What kind of listening test? You've been avoiding the answer to that question but maybe that can change this time around...
