New Headphone Amplifier Design

Thanks. It's always satisfying when things work right from the start. Hopefully the second channel will be trouble-free as well.

The bias and offset settings adjust very smoothly and I didn't have any drama getting the board running. I wasn't sure how touchy the plus-and-minus balance would be, but it's very well-behaved and could probably run fine without the servo.

I confess, I like complicated stuff. This circuit is so simple, there's very little that could go wrong.

The board is really quite small (4" x 5.5"). It's kind of funny; this doesn't feel like the Mighty Pass Labs HPA-1, LOL. But, of course, we need to wait and see how it works with headphones.

One thing that surprised and slightly annoyed me: The TO-92 footprint I used is pretty tight and the folks at JLCPCB scribed the board between some of the pads, presumably to remove short circuits following the hot air solder plating. They didn't do this on any of my previous boards. It doesn't hurt anything, and is hard to notice, but it's not a super-duper feature.

So, this is all very gratifying. My HP power supply is rated for +-20V, but cranks to +-23V if I turn the knob all the way up. I will probably try to give the thing a listen this weekend if everything goes will with the second channel.
 
Thanks for the question. Sorry, what do you mean by "contribution?"

Edit: I don't have much opinion about the "sound" of resistors. I always use precision metal film resistors and choose Vishay RN/CMF55 types, typically, because they are reputed to be of good quality and neutral sounding.

Edit: I also just use 1% or 0.5% (whatever is in stock in the values I need) and don't worry about trying to save money by using 5% or 10% where I can get away with it. For a commercial product, you could save some money choosing lower precision resistors where the exact value isn't as critical.
 
Last edited:
Someone replied to me personally just now about the sound of components. To echo what I said a while back, my attempt to characterize the sound quality differences between the DCG3, A1, A2, and A30 Pro left me with more questions than answers. I'm pretty confident of my impressions of the DCG3 versus the rest. But the seemingly large differences I thought I heard among the others have diminished to point where I'm no longer sure about them. I don't know if it's listener fatigue, or maybe it's true that all properly-operating high-feedback amplifiers sound basically the same and I was just imagining things at first.

I still have the sense that the EIZZ attenuator has taken away some of the "magic" from the A2. It's not so easy to change that out, but I do plan to try something else in the not-too-distant future.

I also have the consistent feeling that the A30 Pro has stronger bass than the A2. My friend in SLC commented that the A1 is a little light in the bass. These amps measure flat down into the infrasonic range, so why the bass would sound different remains a mystery. I remember, decades ago, comparing a power amp I built to a commercial Threshold unit. I was "blown away" by how much more authority the Threshold had in the bottom end. It's not surprising given the relative numbers of output devices and the sizes of the power supplies. But you wouldn't know based on simplistic small-signal response measurements.

In my relatively limited subjective testing experience, I've been surprised how "huge" differences can surprise you when you thought what you were listening to before was already as good as it can get. It's very satisfying when an improvement is so obvious that you don't even need to try to hear it. When you have to strain to identify the difference, it's just a nuisance.

I have the feeling that it's easier to compare equipment using loudspeakers than with headphones, because the spatial qualities are more evident. But I don't have a good enough listening room or speaker-based system for that. I was going to buy some high-end cans (though more money doesn't always get you better performance) but that plan got sidelines by unexpected household expenses at the end of the year.

I will take this one step at a time. I'm not going to worry about whether Vishay resistors are a good choice before I've even had a chance to listen to the amplifier.
 
A3 PCB.jpg


Second channel is done and tested, same as the first.

I discovered this morning that one of the MOSFETs I installed yesterday had been damaged. The case had a chip in it and the mounting tab was bent. I replaced it. Bummer, but I ordered eight of each type, so I have plenty.

Now I have to figure out how to turn this into a working amplifier.
 
Breadboard1.jpg

Breadboard2.jpg


Well, well, well. I am listening to this thing right now.

Yes, I am running it off of a forty year-old HP bench power supply. Stay calm, Mark. I do understand serious evaluation is pointless until I give it a proper power source. I just wanted to see if it worked.

The sound is very promising. I don't want to get carried away with the subjective descriptions like I did with the A1/A2. It has a bit of a "tubey" quality. Organic, maybe. It's quite pleasant to listen to. Not hyper-detailed. It has plenty of bite on the top end, though. No doubt it will tighten up with the official power supply. It's a different presentation from what I'm used to.

I checked the bias and offset quickly but didn't try to trim anything. The panel meter on my power supply is holding steady at 260mA. The metal plate is warm, but not hot as expected since the dissipation is about the same as the A1/A2.

Unfortunately, it's going to be a while before I get the power supply going.

That's all I'll say for now.
 
I've been listening to this setup for about an hour now. Bias is still steady. No flames.

I just went in the other room and listened to my work system with the A30 Pro and my Gamma 2 DAC. I had some reservations about the A3 (HPA-1 clone, right?) at first, but doing the comparison, it's clear the A3 acquits itself very nicely. I'm playing Thomas Dolby "Aliens Ate My Buick" at the moment. A3 sounds quite excellent now, really.

I don't know how long my HP supply was sitting before I bought it, but this is the first time it's run for more than a minute or two at a time in my possession. I would not be surprised if both the amp and the power supply are settling in.

TBH, the A3 sounded pretty bad at first. I don't know if it's me, or the amp, but "the sound is opening up."

If a real power supply improves the sound, then this thing will be superb. I'm actually very satisfied listening to it as-is. It's a very engaging sound, and not at all fatiguing.
 
I listened to the A3 until 2:20 AM. I have a new cat and to keep it from destroying the house, I've been sleeping with it in my work room for the past month. It wakes me up at 6:30 in the morning every day and I am absolutely fried right now, so bear with me.

The problem with subjective reports is you either believe in them or not. Depending on your attitude, it's either all true, or all nonsense. So I can't prove anything about what I'm hearing, but only make objective statements about my unreliable subjective perceptions. In other words, I can state factually that I thought I heard something, but have no idea if what I heard is "real" in any factual sense.

To repeat my earlier comments, what I thought I heard at first was that the A3 sounded awful. It had that "organic" quality but seemed very colored. Everything was there, but there was a warm haze over the music and I had to work, mentally, to sort out the pieces. After running it in for a couple of hours, it started to sound really good, and it just kept improving. For whatever reason, by the time I went to bed, everything had snapped into focus. It sounded so nice, I didn't want to stop listening. It all made sense and I didn't feel there was a barrier between me and the music at all by that point.

This could just be my ears adapting. The brief listen I did in the other system reset my frame of reference. I find I am very susceptible to expectations. I think in my mind I was imagining the sonic effect of my crusty old HP power supply. On the other hand, it's not at all inconceivable that both the amp and the power supply were limbering up over the course of the listening session.

I'm more than a little surprised how good it was with the bench supply. I'm very motivated to get working on the official power supply now.

I have often experienced the effect of "settling into the music," where a system seems to improve dramatically after listening to it for a while. I believe this is a psychological effect, and I think it explains a lot of reports of "sound opening up" over the course of a listening session. On the other hand, I can't disprove that there is something technical going on. All I can say is how dramatically my impression of the sound quality of the A3 improved over the course of my listening to it last night.

I was concerned from the very start of this project about bias stability. The diamond buffers in the A1 and A2 stabilize very quickly and the bias barely varies. The A3 bias slowly drifts down over time, but as I mentioned, settles at a constant value in the test rig and doesn't get too high as long as I stay away from the spreader transistor. It has a negative temperature coefficient, i.e., if I heat the spreader the bias drops. So it should not be prone to running away.

I understand now that the proximity of the spreader in the commercial design, and the unvented cabinet, are part of the thermal design. If I have problems, I can experiment with replacing the spreader with a TO-126 device mounted on the chassis floor and connected to the board with short wires. The temperature slope of the whole circuit will depend on the thermal resistance between the output and spreader transistors, evidently. I think it will probably work ok as-is. The aluminum plate gets pretty warm and is in close proximity to the board from underneath.

Overall, I'm impressed. The unknown (presumed negative) impact of the current power supply arrangement is frustrating. This has been a lot of work on the project so far, and I need to regroup and move onto the next phase.
 
I'm listening to the A2 now. I'm going to take the A3 apart shortly before the cat chews it all up.

The A2 is a very nice amp. It has lots of body and detail and ambience, as I said during my initial evaluation. It's doing that high-feedback "straight wire with gain" thing, and can really rip you a new earhole depending on the source material. It has a touch of warmth and I like it a lot. After listening to the A3 breadboard, I notice a little congestion in complex passages on the A2. What I don't know is if that's inherent to the amp, or a property of the recording, the other equipment, and my ears. The A3 sounded thick, but never congested or painful.

I won't make any more comments on the sound of the A3 until I finish building the amp.

I think people who are building versions of my A2 will be very happy with how it sounds. It's too early to say, but I also think the A3 is going to be a winner. When the project is done, I will publish my design files and give away my spare boards for anyone who wants to build it.

Thank you for enduring my commentary on my latest A3 developments.
 
LOL, I have a couple of those supplies that I use for the same reason as you.
They're actually not bad - designed & built during the time of "thorough and robust" analog work, complete with nicely spec'd parts for what was available in those days. Back then, you got what you paid for ...

Good to see you, Michael. It's hard to believe it took me all these years to get a bench supply. I specifically wanted one of these because it's what they had at the TV station where I used to hang out as a teenager.
 
Quick note for the night. The A3 PCB is in a plastic bag on my desk. I don't plan to fire it up again until the power transformer arrives and the filter/regulator board is done.

I've been thinking about packaging. The power transformer is going to be 4 1/2" in diameter and 2 3/8" tall, not counting the mounting plate it will sit on, which will add even more height. This makes for a challenge if the goal is a compact chassis. My constraints are:

1. Regulator board faces the output transistor side of the main amp board, so I can have the shortest power connections.
2. The transformer, rectifiers, and AC socket are as far from the signal circuitry and connectors as possible.
3. The input wiring is as direct as possible.

In my previous three amps, I had a logical left-to-right progression of the transformers, power board, and signal board. I would like to stick with this. The problem is, this will make the amplifier damned wide. I think I can hold it down to 16", which is about the maximum size to fit on the shelf on my bedside table. With a big old attenuator, depth will be 10-11".

I would love to make a tidy desktop form factor, but that's not going to happen, not with a 100VA transformer and 40,000uF of filter capacitance. Other layouts would make the amp narrower, but greatly increase the depth and the amount of space wasted in the chassis. I've thought through several options, but couldn't come up with anything better than the left-to-right plan. This is important to nail down, because it affects the power supply board design.

Alternatively, I may just throw away the A3 board and start over with an LME49600 design. Anyone want it?

JK.
 

donovas

Member
2014-06-22 6:45 am
Someone replied to me personally just now about the sound of components. To echo what I said a while back, my attempt to characterize the sound quality differences between the DCG3, A1, A2, and A30 Pro left me with more questions than answers. I'm pretty confident of my impressions of the DCG3 versus the rest. But the seemingly large differences I thought I heard among the others have diminished to point where I'm no longer sure about them. I don't know if it's listener fatigue, or maybe it's true that all properly-operating high-feedback amplifiers sound basically the same and I was just imagining things at first.

I still have the sense that the EIZZ attenuator has taken away some of the "magic" from the A2. It's not so easy to change that out, but I do plan to try something else in the not-too-distant future.

I also have the consistent feeling that the A30 Pro has stronger bass than the A2. My friend in SLC commented that the A1 is a little light in the bass. These amps measure flat down into the infrasonic range, so why the bass would sound different remains a mystery. I remember, decades ago, comparing a power amp I built to a commercial Threshold unit. I was "blown away" by how much more authority the Threshold had in the bottom end. It's not surprising given the relative numbers of output devices and the sizes of the power supplies. But you wouldn't know based on simplistic small-signal response measurements.

In my relatively limited subjective testing experience, I've been surprised how "huge" differences can surprise you when you thought what you were listening to before was already as good as it can get. It's very satisfying when an improvement is so obvious that you don't even need to try to hear it. When you have to strain to identify the difference, it's just a nuisance.

I have the feeling that it's easier to compare equipment using loudspeakers than with headphones, because the spatial qualities are more evident. But I don't have a good enough listening room or speaker-based system for that. I was going to buy some high-end cans (though more money doesn't always get you better performance) but that plan got sidelines by unexpected household expenses at the end of the year.

I will take this one step at a time. I'm not going to worry about whether Vishay resistors are a good choice before I've even had a chance to listen to the amplifier.
Youre right about speaker systems being better tools for monitoring imho. Whenever i try to AB something with headphknes the psycho acoustics set in quicker and i behin to doubt my objectivity. Not so much so if i use the same circuit as a preamp to my ss amp. I think this happens due to thw simple fact that cables are easier load than transducers and that whatever difference needed to sucessfully AB test gets amplified 20 30 times by the amplifier. Plus there can be other factors such as stereo imaging easier to distinguish on speakers if the song was mastered on speakers (which is most of the time) ect.