Plugged it in this evening.
Sounds very good so far. Only 2 minutes of listening but I would say it tends towards "warm".
So if I am understanding you correctly:
on the 1kHz square wave the time from the 0v cross to the peak of the square wave appears to be:
200us scale divided by 5 subdivisions = 40us per subdivision.
I count 33 pixels per subdivision: 33 pixels / 40us = 1.2 us per pixel.
From this I see the rise time on the positive going half of the 1khz square wave to be 2.4us from 0v to the 1.774v peak.
I don't quite follow what can be inferred from this? Slew rate?
The settling time is a bit hard to decipher due to the noise on the square wave trace. I've stared at it for a few minutes and its not obvious what is the settling of the overshoot versus the general noise.
Definitely not high volume listening under normal circumstances. My sleeping toddler would not approve of this.
Tomorrow night I will hook up my scope while at a typical listening volume and see what I actually listen at.
Rise time is measured from 0 volts to the peak of the square wave on the horizontal time base. Slew rate is the voltage measured on vertical axis at 1us on the foregoing time basis.
Perfect. And most critical several nF range is fine as well.
https://www.diyaudio.com/community/...-class-a-amplifier-my-way.404147/post-7474526
It is true that is is spec'd without a coil. However, I recall reading somewhere on the site that Rod suggests it may benefit from a small coil if long speaker leads are used. Maybe try adding something like a 1u inductor to see if it helps?
8 to 10 turns of 18 gauge wire around something like a AA sized battery should give you around 1u.
8 to 10 turns of 18 gauge wire around something like a AA sized battery should give you around 1u.
Even 0.5uH may often help, in parallel with a resistor of several Ohms. In the past, I was also often designing without output coil, and it was a BIG mistake. Exactly what you have said, long speaker cables, especially coaxial. It is not about LF capacitance in audio band, but about transmission line input capacitance in MHz range (reflections), that may collide with UGF loopgain and kill the stability.
G'day Guys,
I did a bit of testing on my old LM3886 power amp that this build is supposed to replace.
The square waves are much cleaner looking on this amplifier, I only went as far as a 4R load and established that the messy square waves I am getting from my P3A are due to the P3A and not the test equipment:
The sine waves look the same. Square waves are much nicer looking.
Here is the schematic for the LM3886 board:
I notice that the input filter is more aggressive.
I think the next step for me is to increase the value of C2.
I did a bit of testing on my old LM3886 power amp that this build is supposed to replace.
The square waves are much cleaner looking on this amplifier, I only went as far as a 4R load and established that the messy square waves I am getting from my P3A are due to the P3A and not the test equipment:
The sine waves look the same. Square waves are much nicer looking.
Here is the schematic for the LM3886 board:
I notice that the input filter is more aggressive.
I think the next step for me is to increase the value of C2.
My unit gives nice clean square waves upto 100 Khz. No coil on the output. The Zobel resistor did get hot.
I need to repeat the tests without the Zobel. (100nf and 8 ohm resistor). Its good that you compared another amp. I have had issues with the scope, my probes and with my body introducing signal measurement issues.
I need to repeat the tests without the Zobel. (100nf and 8 ohm resistor). Its good that you compared another amp. I have had issues with the scope, my probes and with my body introducing signal measurement issues.
The filter is a multi stage one with a higher roll off slope for frequency input. Normal single pole filters roll off at 6dB/Octave = a slope of 0.707. The one in your old LM3886 will have a higher rate of roll off that allows higher frequencies to pass without compromising stability. The square waves are better than you could expect with normal filtering because of the sophistication of the chip and the structure built around it. I like what I see in this as a model to work with - of course not possible with out re-designing your P3A.
Your tests have been done in the raw without consideration of such items as a volume control and speaker cable. The former can add series resistance unless a low stage buffer stage is used at the input. I think Rod Elliott expected a higher impedance in series with the P3A input and some inductance in the speaker cable.
I have simulated the P3A with multiple parallel cap values and while this was stable in previous simulations as I believed to the sufficient given experience of others who have built this amplifier, I did not go so far as to test lower parallel capacitor values which I know will have higher resonance frequencies which will therefore be more challenging of circuit stability. I was prompted to check lower values by comments in post 465 by PMA which has substance.
There is no scope to modify your input low pass filter to emulate the LM3886 and I believe you are using a low output impedance buffer ahead of your input. In that case your only option is to increase the value of C2. You can solder some test values to tracks under your pcb however I think you will end up with a value of around 1nF. The other thing you could do is put your test load at the end of the length of speaker cable you intend to use and put your scope on the load attached to the far end. As there will be some inductance in this some ripple will be observed which is normal. If you then put the probe at the output of your amplifier you should see a clean square wave. Don't fret if this has some rounding.
I have previously suggested the inclusion of a 0.22 Ohm resistor instead of an output coil. I you try that you should see the same shape square wave at the near and far load end and the output of your amplifier. As for a coil I looked at some Supra low inductance cable to see what inductance that has, What you have will be different for inductance per metre and you should check the specification to see if you have enough.
I'm not at home until Friday if you let me know the result I will include it in my simulation and post the .asc file
Yes, that's him
He giveaway some PCBs, I was among those who got them, I sent you them because I already build my own handmade pcb at that point, so I thought you might find them useful,
Anyway there are other people who also selling this design like vasp,
I thought you find other designs too, so I asked
I have a P3A that was oscillating on the rise around 0V and fall just before the negative peak when I left the power transistor leads long. After clipping them, the (rather large) parasitic oscillations went away, but I shelved the amp for a while thinking it must be pretty marginal..
My Adcom amplifier started making a buzzing sound (the amp itself, not heard through the speakers) yesterday, so I took that out and put my P3A on the bench.
I applied your changes except for the 1.8pF lead compensation cap. Ran the amp through some simple tests with an oscilloscope and my phone as an audio source. No signs of oscillation. Installed it in the living room. Sounds great.
Wish I could do some THD measurements but I don't have the equipment. Thanks for the tips.