Thanks. 🙂 You may call me Henry, or anything else as long as it's not too insulting. I considered renaming my amps, but I fear that will just add to the confusion.
I wish I had more exciting developments to share, but I'm on a break for now. There is so much hype and mystery around the HPA-1, I think I will just document my efforts with the project in this thread rather than in the "official" one. That may be anti-social, but people seem to have strong feelings about the HPA-1 and I prefer to keep a lower profile in general. I know Mark will keep me honest. Bear with me and I should have some news eventually.
Here's a little OT to pass the time. Three years ago I built an Aikido line amp from Broskie boards. I built it with 6CG7 inputs and 6DJ8 outputs. I was playing with it today on the bench. It makes beautiful square waves, puts out about 70VRMS at clipping, and has a 350kHz bandwidth. It also sounds horrible. I hooked it up to my A30 Pro and listened again with headphones. Yup, still makes my ears bleed. I have no idea why it gets so screechy and congested on complex passages. Sounds like your worst preconceptions of a 60's transistor amp. The thing is a total loss, but I did such a nice job on the build, I can't bear to take it apart. I have no use for a line amp right now, and I don't feel like messing with it.
Final note: Two nights ago I sent Amir at ASR a PM and asked him to please expunge me from his system. I won't explain the thread that put me over the edge, but it was in the DIY sub forum, and I just couldn't take the rudeness and ignorance anymore. I'm really happy that diyaudio.com is a thing. Thanks for all the support with my projects, and happy holidays.
-Henry
I wish I had more exciting developments to share, but I'm on a break for now. There is so much hype and mystery around the HPA-1, I think I will just document my efforts with the project in this thread rather than in the "official" one. That may be anti-social, but people seem to have strong feelings about the HPA-1 and I prefer to keep a lower profile in general. I know Mark will keep me honest. Bear with me and I should have some news eventually.
Here's a little OT to pass the time. Three years ago I built an Aikido line amp from Broskie boards. I built it with 6CG7 inputs and 6DJ8 outputs. I was playing with it today on the bench. It makes beautiful square waves, puts out about 70VRMS at clipping, and has a 350kHz bandwidth. It also sounds horrible. I hooked it up to my A30 Pro and listened again with headphones. Yup, still makes my ears bleed. I have no idea why it gets so screechy and congested on complex passages. Sounds like your worst preconceptions of a 60's transistor amp. The thing is a total loss, but I did such a nice job on the build, I can't bear to take it apart. I have no use for a line amp right now, and I don't feel like messing with it.
Final note: Two nights ago I sent Amir at ASR a PM and asked him to please expunge me from his system. I won't explain the thread that put me over the edge, but it was in the DIY sub forum, and I just couldn't take the rudeness and ignorance anymore. I'm really happy that diyaudio.com is a thing. Thanks for all the support with my projects, and happy holidays.
-Henry
👍
I hope DIYaudio members / administrators will take care to preserve DIY spirit and share.
I really appreciate your approach and philosophy Henry, even if I'm not skilled enough to fully appreciate them.
Thank you for your wise contributions and hard works...
Congrats
Last edited:
Thanks for your kind words.
As promised, I've done some work the past two days on the Pass HPA-1 project. This is the very first cut and I need to do a lot of review, which includes the possibility that I'll decide it sucks and start over.
I included Molex connectors so I can take things apart for testing and making mods without having to desolder the wires. Since I have plenty of space, I plan to add my stock relay driver circuit to the board. This will solve a major nuisance for me (having to provide an external timer and power transformer). The output devices dissipate almost 3W each depending on the bias current and tend to run hot with board-mounted heatsinks. Instead, I've decided to mount them on the chassis floor, as I did with my other projects.
The only circuit modification I made was to include the 10 Ohm ground isolation resistor between input and output stages. This worked out well on my earlier projects.
The stock HPA-1 circuit board is laid out much more spaciously. I saw no reason not to make it more compact. Pass Labs uses a ground plane, but I went with a central ground bus, which may be six-to-one-half-a-dozen-to-another. This seems to work for me, and it's my project, so that's the way I'm doing it. The HPA-1 is a relatively low-feedback design so I expect it to work fine without a ground plane (not that I had any issues with my super-high gain HPA1/2). Here is the stock layout:
The presence of the ground bus forced me to put the bias spreader off to the side. The symmetry of the stock board is pretty, but I don't think the electrons care. I chose single-unit op-amps for the servos instead of the dual unit in the commercial amplifier because my initial plan was make separate boards per channel. I don't think this makes much of a difference.
Tomorrow, I will add the relay timer, then kick off many days of staring at the schematic and board layout looking for mistakes and opportunities for improvement. You will notice certain stylistic similarities to my HPA1 and HPA2 layouts. I could probably rotate the servo circuit 90 degrees and shorten up the board. We'll see. Like I said, this is all very preliminary.
As promised, I've done some work the past two days on the Pass HPA-1 project. This is the very first cut and I need to do a lot of review, which includes the possibility that I'll decide it sucks and start over.
I included Molex connectors so I can take things apart for testing and making mods without having to desolder the wires. Since I have plenty of space, I plan to add my stock relay driver circuit to the board. This will solve a major nuisance for me (having to provide an external timer and power transformer). The output devices dissipate almost 3W each depending on the bias current and tend to run hot with board-mounted heatsinks. Instead, I've decided to mount them on the chassis floor, as I did with my other projects.
The only circuit modification I made was to include the 10 Ohm ground isolation resistor between input and output stages. This worked out well on my earlier projects.
The stock HPA-1 circuit board is laid out much more spaciously. I saw no reason not to make it more compact. Pass Labs uses a ground plane, but I went with a central ground bus, which may be six-to-one-half-a-dozen-to-another. This seems to work for me, and it's my project, so that's the way I'm doing it. The HPA-1 is a relatively low-feedback design so I expect it to work fine without a ground plane (not that I had any issues with my super-high gain HPA1/2). Here is the stock layout:
The presence of the ground bus forced me to put the bias spreader off to the side. The symmetry of the stock board is pretty, but I don't think the electrons care. I chose single-unit op-amps for the servos instead of the dual unit in the commercial amplifier because my initial plan was make separate boards per channel. I don't think this makes much of a difference.
Tomorrow, I will add the relay timer, then kick off many days of staring at the schematic and board layout looking for mistakes and opportunities for improvement. You will notice certain stylistic similarities to my HPA1 and HPA2 layouts. I could probably rotate the servo circuit 90 degrees and shorten up the board. We'll see. Like I said, this is all very preliminary.
You might try shorting out the 10-ohm ground resistors to if they are doing anything you can hear. Or maybe short them out to begin with so that you can get used to the sound without them. Then maybe after a month of burn-in and listening, try cutting the shorting wires. Any difference? 
Last edited:
One of the most critical aspects of practical amplifier design is grounding and wiring layout. Like most beginners, when I got started I had problems with hum and noise. Over the years, I've adopted a formula based on known best practices that seems to work for me. My HPA1/2 amplifiers are dead quiet. On the other hand, I didn't follow the rules with the DCG3 builds and had problems.
The resistor in question is there to provide a local RF ground on the board from the input stage to the output stage. There will be a separate wire from the input stage to the central ground point, effectively shorting out the resistor at low and medium frequencies. This is well-known technique, for instance as documented in this paper by Cherry:
https://linearaudio.nl/sites/linearaudio.net/files/Cherry mutual inductance distortion_0.pdf
Cherry is mainly concerned with Class AB amplifiers and the resulting output stage harmonic currents, which is less of a problem with Class-A Amplifiers. On the other hand, the HPA-1 isn't a "pure" Class A amplifier, since it runs too low output stage bias current to stay in Class A up to its rated power.
We do not want the front end ground reference to be sitting on a non-zero impedance ground path that carries output stage distortion currents. But a separate input stage ground wire has stray inductance and capacitance, so we provide the 10 Ohm resistor as a short path to keep the amp stable. This forms a ground loop, but the resistance is high enough to limit the current that flows through it. These are second-order effects and in a relatively low-gain, high(er)-distortion design like the HPA-1, it's debatable whether the difference would be measurable.
As with my earlier layouts, I've tried to shorten up the ground paths and keep things tight in the high-current output stage. I don't like the huge loop areas of the power rails, but there doesn't seem to be a better way to do it on a two-layer board.
I think I said earlier, what I'm building isn't a Pass HPA-1. It's my version of Jeff Young's version of his best guess at the schematic of the HPA-1, built on a completely different PC board with different parts. We are in the realm where it took seven iterations of the power transformer to find one that worked well, and no doubt similar effort went into the selection of every other component and operating parameter in the amplifier. So this should have some similarities to the real HPA-1, and hopefully, given care in the execution, will be a nice-sounding amplifier. Beyond that, it's anyone's guess. For that reason, I'm not worried about trying to stay true to the original design, which we don't even have anyway.
Being me, I kinda have to do things my own way. So let's not lose sleep over whether this is true to the original.
I've decided it's time to change amplifier nomenclature. So, from now on, the amplifiers formerly known here as "HPA1" and "HPA2" will be called "A1" and "A2." And this new amp, which is not a Pass Labs HPA-1, will be known as "A3".
-Henry
The resistor in question is there to provide a local RF ground on the board from the input stage to the output stage. There will be a separate wire from the input stage to the central ground point, effectively shorting out the resistor at low and medium frequencies. This is well-known technique, for instance as documented in this paper by Cherry:
https://linearaudio.nl/sites/linearaudio.net/files/Cherry mutual inductance distortion_0.pdf
Cherry is mainly concerned with Class AB amplifiers and the resulting output stage harmonic currents, which is less of a problem with Class-A Amplifiers. On the other hand, the HPA-1 isn't a "pure" Class A amplifier, since it runs too low output stage bias current to stay in Class A up to its rated power.
We do not want the front end ground reference to be sitting on a non-zero impedance ground path that carries output stage distortion currents. But a separate input stage ground wire has stray inductance and capacitance, so we provide the 10 Ohm resistor as a short path to keep the amp stable. This forms a ground loop, but the resistance is high enough to limit the current that flows through it. These are second-order effects and in a relatively low-gain, high(er)-distortion design like the HPA-1, it's debatable whether the difference would be measurable.
As with my earlier layouts, I've tried to shorten up the ground paths and keep things tight in the high-current output stage. I don't like the huge loop areas of the power rails, but there doesn't seem to be a better way to do it on a two-layer board.
I think I said earlier, what I'm building isn't a Pass HPA-1. It's my version of Jeff Young's version of his best guess at the schematic of the HPA-1, built on a completely different PC board with different parts. We are in the realm where it took seven iterations of the power transformer to find one that worked well, and no doubt similar effort went into the selection of every other component and operating parameter in the amplifier. So this should have some similarities to the real HPA-1, and hopefully, given care in the execution, will be a nice-sounding amplifier. Beyond that, it's anyone's guess. For that reason, I'm not worried about trying to stay true to the original design, which we don't even have anyway.
Being me, I kinda have to do things my own way. So let's not lose sleep over whether this is true to the original.
I've decided it's time to change amplifier nomenclature. So, from now on, the amplifiers formerly known here as "HPA1" and "HPA2" will be called "A1" and "A2." And this new amp, which is not a Pass Labs HPA-1, will be known as "A3".
-Henry
Should I fix the silkscreen then? 😉
I ordered 10 boards, 4 of those I will definitely keep. The other three pairs are available if anyone from the EU is interested (2,50€/pair + shipping). They are standard JLCPCB leaded HASL, 1oz copper, red soldermask.
I still need to order some dual JFETs and a few resistors, but most of the other components are ready to go.
Yikes! I'm excited you're making progress, and nervous too. I hope it works out well for you.
I lose track of how many people are building these things, but I can't tell you how thrilled I will be to see some finished projects, and hopefully get some positive feedback on the sound quality. I felt kind of bad about the troubles with the A1 and the lukewarm review from the first person I lent it to. I've stayed in touch with my friend, Geoff, who has the A1 breadboard now and he continues to enjoy it a lot, says it's one of the best amps he's heard. So I haven't written this design off entirely and wouldn't discourage people from trying it. I think maybe all it needs is some attention to the compensation. Otherwise, it's just a mostly ordinary Blameless circuit, so don't, uh, blame me, haha.
Nobody but me has heard the A2. It's my daily driver and I'm very happy with it, except for the nagging feeling about the attenuator, which I intend to replace again eventually. As always, the details of execution matter so if it sounds bad, it must be that you built it wrong. 🙂
I lose track of how many people are building these things, but I can't tell you how thrilled I will be to see some finished projects, and hopefully get some positive feedback on the sound quality. I felt kind of bad about the troubles with the A1 and the lukewarm review from the first person I lent it to. I've stayed in touch with my friend, Geoff, who has the A1 breadboard now and he continues to enjoy it a lot, says it's one of the best amps he's heard. So I haven't written this design off entirely and wouldn't discourage people from trying it. I think maybe all it needs is some attention to the compensation. Otherwise, it's just a mostly ordinary Blameless circuit, so don't, uh, blame me, haha.
Nobody but me has heard the A2. It's my daily driver and I'm very happy with it, except for the nagging feeling about the attenuator, which I intend to replace again eventually. As always, the details of execution matter so if it sounds bad, it must be that you built it wrong. 🙂
I wanted to talk a little more about the amplifier grounding scheme, in response to a PM I received today. This is gonna be long (sorry) and I'm too lazy to make diagrams, so I'll try to paint pictures with words.
The specific concern is that the extra input ground wire can act as an antenna and you may end up picking up interference, especially RF interference, that may subtly degrade sonic performance even if the problem is not immediately obvious. I want to draw a distinction between ground loops and internal and external interference. Forgive me if I've written about this before; I'm constantly thinking about this stuff and have a hard time remembering what, if any, of it I've already posted.
The goal of any grounding scheme is to establish a common point of reference for zero volts, and make sure that every point in the circuit that needs a ground reference gets it, as closely as possible. Since currents flow through ground wires, and wires experience voltage drops and pick up interference, perfect ground distribution is impossible. But we can try to avoid screwing it up. The way to do that is to treat every wire as having a non-zero impedance and to organize things so the inevitable voltage drops don't get mixed up with one another. We also want to minimize circuit loop areas to avoid inductive pickup. And finally, we want to try to keep external interference from infiltrating into the chassis.
In my headphone amps, I've settled on making the output jack common terminal the central ground reference for the whole amplifier. What matters is the signal going to the headphones, so it makes sense to use this as our reference point.
The highest currents in the amplifier flow in the output stage. The power supply and power wiring have non-zero impedance, so the voltages will not be constant. As is common practice, I put a pair of medium-sized electrolytic capacitors on the circuit board, connected to the power rails, right at the output stage where the power comes onto the board. These capacitors cannot eliminate all signal and noise on the rails, but can divert some of it. Any odd-order nonlinearity in the output stage, especially if you drive it out of Class A mode, will cause rectified harmonic currents to flow in the output stage ground return. This is pure distortion. We do not want these currents (or the associated voltage drops) to leak into the rest of the circuit, so we have to keep them contained.
For starters, I place the two input capacitors right next to each other, with a short ground bus between their common terminals. I make the power traces as short as possible, and keep the overall circuit loops from the power inlet, around the output stage, and out to the load and back, as tight as I can. This should minimize inductance and stray electromagnetic coupling to other parts of the circuit. I also try to put the power supply physically right next to the output stage, for the same reasons.
With dual-mono design, the power supplies can float, i.e., their only ground connections are via the board power wiring. The supplies may be bouncing up and down on the voltage drops across their ground connections, but since they're not connected to anything else, no current can escape. All the damage is contained inside that one loop. The common connection of the two electrolytics becomes the "noise ground" and is the output stage reference. I run a wire from this point to the common terminal at the output jack, one wire per board. So now, my noise grounds are at the same potential as the headphone ground. Under zero-signal conditions, no current flows in these wires.
All is not perfect, however. With a signal playing, the ground wires carry the headphone output currents, so the noise grounds ride the voltage drops across them. Since we have two channels, the left and right channel noise ground voltages will also be different from each other, in real time. This is problematic. We could try other supply grounding schemes, but the fundamental problem seems impossible to avoid, so what we have to do is minimize its impact.
Turning to the amplifier front end, there are three ground circuits of interest. The first is the one for the attenuator and the input stage RC filter. The current loop here is from the input jack hot terminal, into the input stage, and back again, mostly. The second loop is for any "utility" currents, like bias voltages for current sources and cascodes and the like. These flow from the supply rails and are mostly DC, with minimal signal current, so we will ignore them. Real op-amps are designed to not have ground terminals at all, so these bias currents never make it into the ground system, but this is not a real op-amp. The third is the return current for the feedback network, which is critical.
We depend on feedback to reduce the distortion at the output of the amplifier, and the feedback needs a pure ground reference. If we connect the feedback network to the output stage noise ground, our amplifier will use that point as a reference, and we will end up amplifying whatever signal, noise, and distortion exists there. So I run my "clean" input stage ground via a separate wire to the designated reference ground at the output jack. Now the feedback loop can sense the actual signal at the output terminals, not some noisy, harmonic power supply junk.
This gets us to the crux of the problem. The input ground is a bare wire, and it's not twisted with anything. It has inductance and it forms a loop that can pick up interference floating around the chassis. Our amplifier potentially has open-loop gain out into the megahertz range. We don't want the feedback loop to go through a foot or more of wire, which could make all kinds of problems. To keep things stable, we want the input and output stages to be in sync. So we're in a bind. The solution is the 10 Ohm resistor linking the clean and noise grounds. It's a low enough impedance to close the loop at high frequencies, but high enough to force in-band signals to flow where we want them, through the dedicated input ground wires.
There is another problem. We have two channels. Their grounds are connected together at the output jack. Somewhere, the left and right input jack shells have to be connected together as well. This forms a loop, and it potentially extends out into the RCA cable shields and all the way back to the source component. We want this loop area to be as small as possible. Best practice says to put the input jacks close to one another and connect their grounds directly together. Now, if the two input cable shields pick up interference, these currents will cross over at the input jacks and not enter the amplifier. In theory, you should float the input jacks, and connect 0.01uF ceramic capacitors from the shells to the chassis. This provides additional immunity from external RF interference.
You still have an inductive ground loop inside the amplifier because the two channel PC board ground buses are physically separated. There isn't much you can do about this, other than to run the two input stage ground wires close together, and keep the power transformers as far away from the signal circuitry as possible.
I mentioned earlier that the noise grounds are at a non-zero potential due to voltage drops across the output ground wires. This could drive a signal current through the input stage grounds. The 10 Ohm resistors help to break that loop.
Finally, we run a wire from the output common to the chassis, so any voltage induced in the enclosure walls is tied to reference potential and can't capacitively couple to the circuitry. And don't forget the AC power safety ground connection to the chassis, either. And the fuse, of course.
In the end, evidently, everything is a compromise. I'm not an expert, so I keep trying things and slowly refine my technique as I run into problems and come up with new ideas. I don't promise to have all the answers.
I hear this is a very good book on this subject (among a lot of other things). I don't own it, but probably should:
https://www.amazon.com/Electromagnetic-Compatibility-Engineering-Henry-Ott
It costs a hundred twenty-five dollars and is eight hundred eighty pages long, so everyone, get out your credit cards, click Buy It Now, and start reading...
The specific concern is that the extra input ground wire can act as an antenna and you may end up picking up interference, especially RF interference, that may subtly degrade sonic performance even if the problem is not immediately obvious. I want to draw a distinction between ground loops and internal and external interference. Forgive me if I've written about this before; I'm constantly thinking about this stuff and have a hard time remembering what, if any, of it I've already posted.
The goal of any grounding scheme is to establish a common point of reference for zero volts, and make sure that every point in the circuit that needs a ground reference gets it, as closely as possible. Since currents flow through ground wires, and wires experience voltage drops and pick up interference, perfect ground distribution is impossible. But we can try to avoid screwing it up. The way to do that is to treat every wire as having a non-zero impedance and to organize things so the inevitable voltage drops don't get mixed up with one another. We also want to minimize circuit loop areas to avoid inductive pickup. And finally, we want to try to keep external interference from infiltrating into the chassis.
In my headphone amps, I've settled on making the output jack common terminal the central ground reference for the whole amplifier. What matters is the signal going to the headphones, so it makes sense to use this as our reference point.
The highest currents in the amplifier flow in the output stage. The power supply and power wiring have non-zero impedance, so the voltages will not be constant. As is common practice, I put a pair of medium-sized electrolytic capacitors on the circuit board, connected to the power rails, right at the output stage where the power comes onto the board. These capacitors cannot eliminate all signal and noise on the rails, but can divert some of it. Any odd-order nonlinearity in the output stage, especially if you drive it out of Class A mode, will cause rectified harmonic currents to flow in the output stage ground return. This is pure distortion. We do not want these currents (or the associated voltage drops) to leak into the rest of the circuit, so we have to keep them contained.
For starters, I place the two input capacitors right next to each other, with a short ground bus between their common terminals. I make the power traces as short as possible, and keep the overall circuit loops from the power inlet, around the output stage, and out to the load and back, as tight as I can. This should minimize inductance and stray electromagnetic coupling to other parts of the circuit. I also try to put the power supply physically right next to the output stage, for the same reasons.
With dual-mono design, the power supplies can float, i.e., their only ground connections are via the board power wiring. The supplies may be bouncing up and down on the voltage drops across their ground connections, but since they're not connected to anything else, no current can escape. All the damage is contained inside that one loop. The common connection of the two electrolytics becomes the "noise ground" and is the output stage reference. I run a wire from this point to the common terminal at the output jack, one wire per board. So now, my noise grounds are at the same potential as the headphone ground. Under zero-signal conditions, no current flows in these wires.
All is not perfect, however. With a signal playing, the ground wires carry the headphone output currents, so the noise grounds ride the voltage drops across them. Since we have two channels, the left and right channel noise ground voltages will also be different from each other, in real time. This is problematic. We could try other supply grounding schemes, but the fundamental problem seems impossible to avoid, so what we have to do is minimize its impact.
Turning to the amplifier front end, there are three ground circuits of interest. The first is the one for the attenuator and the input stage RC filter. The current loop here is from the input jack hot terminal, into the input stage, and back again, mostly. The second loop is for any "utility" currents, like bias voltages for current sources and cascodes and the like. These flow from the supply rails and are mostly DC, with minimal signal current, so we will ignore them. Real op-amps are designed to not have ground terminals at all, so these bias currents never make it into the ground system, but this is not a real op-amp. The third is the return current for the feedback network, which is critical.
We depend on feedback to reduce the distortion at the output of the amplifier, and the feedback needs a pure ground reference. If we connect the feedback network to the output stage noise ground, our amplifier will use that point as a reference, and we will end up amplifying whatever signal, noise, and distortion exists there. So I run my "clean" input stage ground via a separate wire to the designated reference ground at the output jack. Now the feedback loop can sense the actual signal at the output terminals, not some noisy, harmonic power supply junk.
This gets us to the crux of the problem. The input ground is a bare wire, and it's not twisted with anything. It has inductance and it forms a loop that can pick up interference floating around the chassis. Our amplifier potentially has open-loop gain out into the megahertz range. We don't want the feedback loop to go through a foot or more of wire, which could make all kinds of problems. To keep things stable, we want the input and output stages to be in sync. So we're in a bind. The solution is the 10 Ohm resistor linking the clean and noise grounds. It's a low enough impedance to close the loop at high frequencies, but high enough to force in-band signals to flow where we want them, through the dedicated input ground wires.
There is another problem. We have two channels. Their grounds are connected together at the output jack. Somewhere, the left and right input jack shells have to be connected together as well. This forms a loop, and it potentially extends out into the RCA cable shields and all the way back to the source component. We want this loop area to be as small as possible. Best practice says to put the input jacks close to one another and connect their grounds directly together. Now, if the two input cable shields pick up interference, these currents will cross over at the input jacks and not enter the amplifier. In theory, you should float the input jacks, and connect 0.01uF ceramic capacitors from the shells to the chassis. This provides additional immunity from external RF interference.
You still have an inductive ground loop inside the amplifier because the two channel PC board ground buses are physically separated. There isn't much you can do about this, other than to run the two input stage ground wires close together, and keep the power transformers as far away from the signal circuitry as possible.
I mentioned earlier that the noise grounds are at a non-zero potential due to voltage drops across the output ground wires. This could drive a signal current through the input stage grounds. The 10 Ohm resistors help to break that loop.
Finally, we run a wire from the output common to the chassis, so any voltage induced in the enclosure walls is tied to reference potential and can't capacitively couple to the circuitry. And don't forget the AC power safety ground connection to the chassis, either. And the fuse, of course.
In the end, evidently, everything is a compromise. I'm not an expert, so I keep trying things and slowly refine my technique as I run into problems and come up with new ideas. I don't promise to have all the answers.
I hear this is a very good book on this subject (among a lot of other things). I don't own it, but probably should:
https://www.amazon.com/Electromagnetic-Compatibility-Engineering-Henry-Ott
It costs a hundred twenty-five dollars and is eight hundred eighty pages long, so everyone, get out your credit cards, click Buy It Now, and start reading...
Yup.
Nice article from Bruno Putzeys are attached.
Also give a chance to a perfect Elya Joffe "Grounds for grounding" book.
https://www.amazon.com/Grounds-Grounding-Circuit-System-Handbook/dp/0471660086
Yes, great article. Putzeys is into balanced wiring and does a good job of justifying his position. All my gear is single-ended, but the core ideas still apply. It's all surprisingly complicated if you haven't thought much about it before, hence the length of my post. I always start out a project thinking about grounding. I encourage others to do the same.
True for the most part. OPA1622 is one exception. IIRC its internal compensation cap is referenced to ground via an extra pin, whereas in most opamps the reference pin is effectively the negative rail.
Here is the latest revision. I cleaned up the signal layout a little, and added the relay timer circuit. I decided not to run the relay off of the main amp power for two reasons: 1) There are no 24V relays in stock anywhere; 2) I didn't feel like putting an extra 20mA or so of load on one supply rail. So I will need an auxiliary 12V power supply, not a big deal.
I still need to fix up all the parts designators, and then continue checking and improving the layout.
FWIW and YMMV.
I cleaned up the relay circuit and the parts designators. Nothing offends me too badly about this layout, so I will move on to detailed checking.
Do you like using that particular TO-92 footprint? (And soldering that particular footprint?)
I just mention it since the KiCad TO-92_Inline_Wide can be easier to solder.
I just mention it since the KiCad TO-92_Inline_Wide can be easier to solder.
Thanks for the comments. Actually, the standard TO-92 footprint is a nuisance, not because it's hard to solder, but because the transistors don't stay in place in the holes. But I like that it's compact.
I wonder if the TO-92_HandSolder applies enough spring tension in the leads to hold them. It looks fairly compact too. I have not tried it so I don't know how easy it is to solder.
I use magnifying glasses and an SMD tip on the iron and it works ok for me. You've half tempted me to change the footprints, but I'm too lazy to rework the layout so I think I'll stick with it as it is. I imagine the hand solder layout is angled enough to keep the transistors in the board while soldering.
Edit: When are we going to get a listening report on your amp?
Edit: When are we going to get a listening report on your amp?