If P3a is built competently with the ESP boards and legitimate components, it will work flawlessly for years. Measurement at 1W 8R dummy load attached. Couldn't measure below 100Hz because it's a midrange amp for an active system and P33 loudspeaker muting & protection activates the muting relays at low frequencies.
Please stop bagging Rod Elliott's designs. P3a is a brilliant and simple amplifier and mine (all 8 of them) are still in trouble-free operation after replacing amplifiers including very expensive Conrad-Johnson models many years ago. Within its power limitations P3a easily out-performs all of them.
The comments made by people here about circuit flaws, incorrect component value designations, colourations, instability etc. are baseless and false. You should typically expect anything from 1mV DC offset to around 20mV and clean output as per the attached measurement. I have NEVER seen instability on my scope from ANY ESP power amplifier!
P101 is also a great amp and so is P68. I've built these also and they also perform flawlessly. Even the little TDA7293 P127 works brilliantly, so I don't know what all the fuss is about.
Please stop bagging Rod Elliott's designs. P3a is a brilliant and simple amplifier and mine (all 8 of them) are still in trouble-free operation after replacing amplifiers including very expensive Conrad-Johnson models many years ago. Within its power limitations P3a easily out-performs all of them.
The comments made by people here about circuit flaws, incorrect component value designations, colourations, instability etc. are baseless and false. You should typically expect anything from 1mV DC offset to around 20mV and clean output as per the attached measurement. I have NEVER seen instability on my scope from ANY ESP power amplifier!
P101 is also a great amp and so is P68. I've built these also and they also perform flawlessly. Even the little TDA7293 P127 works brilliantly, so I don't know what all the fuss is about.
Attachments
I never purchase single layer boards. For personal use, I prefer to etch them myself.
There is enough P3A photographic information on Rod's site for a quick reverse engineering.
A vintage sportscar is likely less reliable and stable than a volkswagen golf, but does that make the latter a better car?
There are other criteria to judge amplifiers by, you don't mention for example 'how it sounds'. But of course then you could argue an amplifier should not sound like anything, just amplify. Endless discussion, lovely!
IMHO design complexity is a key parameter to consider for DIY amps.
Simple designs like P3A or Hitachi/Maplin provide a considerable performance with very few components.
Understand how they work and building them is fundamental in the learning process, especially for less experienced DIYers.
Further improvements can always make things better but sometimes I wonder if the present quest for perfection is reasonable.
I mean the escalation in amp complexity only gratifying for the very few thinking their ears can detect ppm distortion.
Simple designs like P3A or Hitachi/Maplin provide a considerable performance with very few components.
Understand how they work and building them is fundamental in the learning process, especially for less experienced DIYers.
Further improvements can always make things better but sometimes I wonder if the present quest for perfection is reasonable.
I mean the escalation in amp complexity only gratifying for the very few thinking their ears can detect ppm distortion.
Quest for perfection? You mean in terms of measurements?
Possibly related, here are some words from Dr. Earl Geddes:
"The bottom line here is that we know so little about how humans perceive the sound quality of an audio system, and in particular the loudspeaker, that one should question almost everything that we think we know about measuring it. From what we have found most of what is being done in this regard is naive."
https://www.gedlee.com/Papers/Comments on howard.pdf
Possibly related, here are some words from Dr. Earl Geddes:
"The bottom line here is that we know so little about how humans perceive the sound quality of an audio system, and in particular the loudspeaker, that one should question almost everything that we think we know about measuring it. From what we have found most of what is being done in this regard is naive."
https://www.gedlee.com/Papers/Comments on howard.pdf
But why not refine the circuit ? is that not a part of the learning process.
Why not add a couple of diodes in series with the bootstrap to make clipping symmetrical ?
Or , decouple the Vbe and VAS ? or add Re to the input pair to smooth over a wide range of beta ?
Flyback diodes and inductor/R at output ??
Unlike professor Leach's project , P3A takes many "shortcuts" .... like what you would quickly make on a Vero board.
My only point ...
I also don't believe in "complexity escalation" I won't go above 10-12 IPS designs , I also don't like overly complex compensation
schemes. Keep the cost $$ down , junkbox designs ( with fidelity). The " Wolverine" does not need to be as complicated as it is
presently ... they add a bunch of frivolous stuff , just to get a couple PPM.
OS
Why not add a couple of diodes in series with the bootstrap to make clipping symmetrical ?
Or , decouple the Vbe and VAS ? or add Re to the input pair to smooth over a wide range of beta ?
Flyback diodes and inductor/R at output ??
Unlike professor Leach's project , P3A takes many "shortcuts" .... like what you would quickly make on a Vero board.
My only point ...
I also don't believe in "complexity escalation" I won't go above 10-12 IPS designs , I also don't like overly complex compensation
schemes. Keep the cost $$ down , junkbox designs ( with fidelity). The " Wolverine" does not need to be as complicated as it is
presently ... they add a bunch of frivolous stuff , just to get a couple PPM.
OS
Last edited:
I have a simple question for you. What is the reason for using a 25 turn 2k Cermet trimmer in the Vbe where a normal single turn carbon track trimmer of suitable value would suffice in an EF2 circuit? The function is obvious so give me your analsis.
Last edited:
I think that would be a great idea for someone learning to build and design amplifiers. Add some parts, get it working and stable, and find out if they can hear (and like) the difference.
Rod designed the P3a for his purposes. He places great importance on minimizing parts count. This won't be important to everyone. I also understand he originally used this design as a PA amp and built them in great numbers.. hence a low part count kept the costs down. Also, as a PA amp, lower slew rate benefitted it (he states this somewhere on his site). He also mentions that the P3a doesn't like inductors and states that he doesn't fully understand why.
Just to throw my hat in the ring here, I have A P3A built using Rod’s boards. Like all the stuff I bought from him it worked first time and a quick listen on some old speakers sounded fine. I don’t actually have any passive speakers that are good enough to give me a real measure of how good it is.
It will however compete with TDA7293, LM3886, Doug Self compact Blameless, LM1875 and an MX50SE for a spot in my next active speaker!
My gut feeling is that the Blameless is the best of the bunch but is oddly the largest too!
I haven’t done any serious distortion measurement on any of these other than looking at the FFT on the scope at fairly high output and high frequency.
I was very hesitant about buying the P3A as I had read the Self and Cordell books both of whom were in complete agreement about the design flaws, but I read many complementary things on this forum so decided to give it a go.
Using Rod’s boards and recommended component values it worked first time and showed no sign of oscillation, even without an output inductor. My only variation in spec was that I used Toshiba output transistors as I already had them and felt that they were similar enough to the current recommended devices.
All that said, looking at the circuit does give you a strong urge to “improve” it 🙂, but it seems that might be slightly harder than it looks!
I like the look of OS’s Easy Amp above and if I didn’t already have enough amps, that would take its place in my projects list!
It will however compete with TDA7293, LM3886, Doug Self compact Blameless, LM1875 and an MX50SE for a spot in my next active speaker!
My gut feeling is that the Blameless is the best of the bunch but is oddly the largest too!
I haven’t done any serious distortion measurement on any of these other than looking at the FFT on the scope at fairly high output and high frequency.
I was very hesitant about buying the P3A as I had read the Self and Cordell books both of whom were in complete agreement about the design flaws, but I read many complementary things on this forum so decided to give it a go.
Using Rod’s boards and recommended component values it worked first time and showed no sign of oscillation, even without an output inductor. My only variation in spec was that I used Toshiba output transistors as I already had them and felt that they were similar enough to the current recommended devices.
All that said, looking at the circuit does give you a strong urge to “improve” it 🙂, but it seems that might be slightly harder than it looks!
I like the look of OS’s Easy Amp above and if I didn’t already have enough amps, that would take its place in my projects list!
Okay. However, there may be reasons to suspect it may not be completely blameless when it comes to SQ. Its as Geddes has explained, which IIUC is to the effect that we don't know exactly how humans hear so we don't know exactly what to measure, but at least we know what we typically measure does not correlate all that well with human perception.
https://www.gedlee.com/Papers/papers.aspx
First of all you must be the only one left of us to still use carbon trimmers
More seriously, according to D. Self, the CFP stage appears to be more demanding of Vbias compensation than EF and requires more accuracy.
@Markw4, I’m no expert on psychacoustics so can’t contribute much on the specifics there other than to say that the stuff we DO measure ought to give us some confidence that the system that measures well on all of these and isn’t suffering from oscillation or unpleasant clipping artifacts should give us some measure of confidence that the system is doing what it should.
Observations on Wolverine and Neurochrome amps would seem to support this even if it is far from being the full story.
I think we are wandering off P3A stability here somewhat!
Observations on Wolverine and Neurochrome amps would seem to support this even if it is far from being the full story.
I think we are wandering off P3A stability here somewhat!
@Jeffh01 I didn't bring up Blameless, nor Wolverine as possible alternate projects. Maybe you are complaining to the wrong person?
BTW, I am not against measurements. My only concern has to do with how some humans subjectively interpret FFTs and other measurements in ways that may not always be well supported by science. In that case I may feel compelled to comment in order to help inform newer-to-the-hobby readers.
BTW, I am not against measurements. My only concern has to do with how some humans subjectively interpret FFTs and other measurements in ways that may not always be well supported by science. In that case I may feel compelled to comment in order to help inform newer-to-the-hobby readers.
Well, my plan was to try the 100nF cap across q9 first and see if I could rebuild the blown channel to be stable.
However there isn't enough space to mount the 100nF film caps I have lying around under the board and still gave it sit flat on the heatsink.
So it looks like the first step for me will be to make a new board with space for the extra components:
-Degen resistors
-Extra decoupling
-Cap across q9
-Space for heatsinking on the drivers as Rod recommends this for 4R loads but doesn't easily allow for it.
-PP cap at the input.
As was stated above, Rod designed his board for his purposes, I will design one for mine. Pcb design is one of my favorite parts of this hobby. It's like a game of tetris.
However there isn't enough space to mount the 100nF film caps I have lying around under the board and still gave it sit flat on the heatsink.
So it looks like the first step for me will be to make a new board with space for the extra components:
-Degen resistors
-Extra decoupling
-Cap across q9
-Space for heatsinking on the drivers as Rod recommends this for 4R loads but doesn't easily allow for it.
-PP cap at the input.
As was stated above, Rod designed his board for his purposes, I will design one for mine. Pcb design is one of my favorite parts of this hobby. It's like a game of tetris.
@Markw4 It was not my intent to complain to or about anyone and I’m sorry if I came across that way.
I had rather hoped that I left room for everyone’s contributions without being drawn on areas that I have yet to gain much knowledge or experience.
I referenced those amplifiers as they both measure well and receive many subjective plaudits.
I have seen your contributions in many areas on this forum and found them interesting and informative especially wrt to DACs, an area I have a rather incomplete understanding of!
As far as I am concerned your opinions are interesting and I am happy to hear them. It’s part of what I really enjoy about this community.
I hope that puts you at ease with my previous post.
Jeff
I had rather hoped that I left room for everyone’s contributions without being drawn on areas that I have yet to gain much knowledge or experience.
I referenced those amplifiers as they both measure well and receive many subjective plaudits.
I have seen your contributions in many areas on this forum and found them interesting and informative especially wrt to DACs, an area I have a rather incomplete understanding of!
As far as I am concerned your opinions are interesting and I am happy to hear them. It’s part of what I really enjoy about this community.
I hope that puts you at ease with my previous post.
Jeff
Regarding driver heatsinks - take a look at Douglas Self's boards on the Signal Transfer Company's site. You'll see that he uses large heatsinks for the CFP drivers. Changes in the driver junction temp will cause the bias to drift. Keeping the drivers cool will reduce this.
While you are at it, add a current compensation resistor to the bias generator. This will also help stabilize the bias, potentially allowing you to reduce the size of of the output resistors. With these changes, you should be able to go to 0R22, maybe even 0R1.
While you are at it, add a current compensation resistor to the bias generator. This will also help stabilize the bias, potentially allowing you to reduce the size of of the output resistors. With these changes, you should be able to go to 0R22, maybe even 0R1.
@Sadface
There are ways to help figure out why an amplifier is unstable/oscillates. In some case it may be useful to open the feedback loop, although just doing that will change the system because impedances will change where the circuit has been broken apart. Its possible to approximately compensate for some of that, but even if not so-much compensated there may be something useful to be learned by opening the loop.
In general though, maybe better to get an amp stable, say, maybe with no load if that helps, then try to take a look at the feedback effects with the loop closed. A famous paper on the subject of feedback measurement is at: https://u.dianyuan.com/upload/space/2010/12/30/1293704376-389780.pdf
Some supporting material on what causes oscillation and one way to think about measurements at:
https://www.rohde-schwarz.com/us/products/test-and-measurement/essentials-test-equipment/digital-oscilloscopes/understanding-bode-plots_254514.html#:~:text=Bode plots show the frequency,phase, most commonly in degrees.
https://lpsa.swarthmore.edu/Bode/Bode.html
https://en.wikipedia.org/wiki/Bode_plot
https://en.wikipedia.org/wiki/Parasitic_oscillation
Don't know if any of that makes sense to you at this point? If not, maybe we could try to help you develop a little more intuition about the problem.
Reason I bring this stuff up is because I'm not so sure that its best to set you off on a course of random hacking to try to solve the problem.
There are ways to help figure out why an amplifier is unstable/oscillates. In some case it may be useful to open the feedback loop, although just doing that will change the system because impedances will change where the circuit has been broken apart. Its possible to approximately compensate for some of that, but even if not so-much compensated there may be something useful to be learned by opening the loop.
In general though, maybe better to get an amp stable, say, maybe with no load if that helps, then try to take a look at the feedback effects with the loop closed. A famous paper on the subject of feedback measurement is at: https://u.dianyuan.com/upload/space/2010/12/30/1293704376-389780.pdf
Some supporting material on what causes oscillation and one way to think about measurements at:
https://www.rohde-schwarz.com/us/products/test-and-measurement/essentials-test-equipment/digital-oscilloscopes/understanding-bode-plots_254514.html#:~:text=Bode plots show the frequency,phase, most commonly in degrees.
https://lpsa.swarthmore.edu/Bode/Bode.html
https://en.wikipedia.org/wiki/Bode_plot
https://en.wikipedia.org/wiki/Parasitic_oscillation
Don't know if any of that makes sense to you at this point? If not, maybe we could try to help you develop a little more intuition about the problem.
Reason I bring this stuff up is because I'm not so sure that its best to set you off on a course of random hacking to try to solve the problem.
Last edited:
For what reason? The last repair I did with blown driver and output transistors was in a standard Darlington. The trimmer for the Vbe bias set was a single turn trimmer. This was a Sansui amplifier classic? The same applies with the Leach lowTIM amplifier I built so you are going against the late professor.First of all you must be the only one left of us to still use carbon trimmers
More seriously, according to D. Self, the CFP stage appears to be more demanding of Vbias compensation than EF and requires more accuracy.
Last edited: