G'day Guys,
This amp was finally fired back up again after the holidays.
I added some braided copper shield to the input wiring which is earthed at the same point as the PSU.
I also replaced C2 with a 1nF C0G ceramic cap.
I had an issue with the main fuse blowing immediately after startup. I traced this to the power supply wiring to the right channel.
A little tab of metal on the back side of a spade plug was poking through the heatshrink on the +ve rail and shorting straight to ground.
No further problems after that. Onto the scope it went.
The amp is now stable into a 4R + 220nF. Before it was stable into 4R + 100nF but would oscillate into 4R + 150nF.
To recap this is how it went into 4R + 150nF before:
Now:
4R + 220nF
4R + 330nF - Things get a bit unhappy again:
This amp was finally fired back up again after the holidays.
I added some braided copper shield to the input wiring which is earthed at the same point as the PSU.
I also replaced C2 with a 1nF C0G ceramic cap.
I had an issue with the main fuse blowing immediately after startup. I traced this to the power supply wiring to the right channel.
A little tab of metal on the back side of a spade plug was poking through the heatshrink on the +ve rail and shorting straight to ground.
No further problems after that. Onto the scope it went.
The amp is now stable into a 4R + 220nF. Before it was stable into 4R + 100nF but would oscillate into 4R + 150nF.
To recap this is how it went into 4R + 150nF before:
Now:
4R + 220nF
4R + 330nF - Things get a bit unhappy again:
I may yet take this step. However it is recommended against by the designer.
The big question I haven't answered sufficiently well for myself yet is:
How stable is 'stable enough'?
I have read somewhere (I can't remember where) that the most important range for capacitive loads is 47nF to 220nF.
This P3A is happy enough with square waves into 4R + 220nF. Happy with sine waves into 4R + 2.2uF.
I am inclined to think that this has reached the stage of 'Stable enough'.
I will sleep on it for a day or 2 however my inclination is to simply plug the amp into my main rig and start on the next amp (one with a thiel network at its output).
Why not try, could easily be done after speaker output, if no succes, remove again.
I had some issues with my utracer6, I could not get it to deliver its plus 15 volt, I checked and double checked all parts,
Traced pwm signal to mosfet, mosfet got pwm, but still no 15 volt, it turned out there was a broken trace under the 100uf capacitor, it took forever to find that fault, have you checked continuity between traces and parts? Those scope captures are not exactly pretty, for a 3pa that is.
I had some issues with my utracer6, I could not get it to deliver its plus 15 volt, I checked and double checked all parts,
Traced pwm signal to mosfet, mosfet got pwm, but still no 15 volt, it turned out there was a broken trace under the 100uf capacitor, it took forever to find that fault, have you checked continuity between traces and parts? Those scope captures are not exactly pretty, for a 3pa that is.
Check out section 3.4.1 of Rod Elliotts Power Amplifier Design Guidelines. He discusses the P3A being designed without an output inductor, but later mentions that with exotic loads, a small inductor is advised. This supports what PMA is advising.
Your test load with 10kHz square load is just that to an extreme level that may be needed for use with electrostatic speakers like Quad ELS62's. By all means run tests with parallel loads in the range of 47nF to 220nF but with a resistor load of 8 R. Lower value caps will have higher resonance frequencies.
In a zobel stability network the capacitor to earth will be at the higher end of the above test cap values i.e. 100nF or less commonly 220nF as in the Silicon Chip Magazine "Blameless" Hummingbird project in the December 2021 issue.
The series output coil value is 10uH damped by a 1Watt 4R7 resistor. The 220nF zobel cap is in series with a 1 Watt 15R resistor. This project is in the same output power zone as the P3A with power rails of +/- 32volts or 35 -40 volts maximum range. It is rated at 60 W into 8R for +/-32volt rails.
From the look of your square waves and in consideration of lower impedance loads you might have to take a more extreme solution than PMA has
proposed. You could test these by putting each in a plug in test box in series with your test load. For square waves the probe should be ahead of the stability network whichever you try.
I think I am going to leave it alone for now and just enjoy it.
I think this build is 'stable enough'.
My speakers are not exotic, my typical listening volume is low.
It is not going to be amplifying square waves into difficult loads in normal use.
My goal with this build was simply to build a well regarded 'nice' sounding amplifier, not develop a new variant.
I had some mishaps.
I learned a heck of a lot along the way.
My understanding of amplifier circuits has grown immensely.
I am now happy to say I have achieved my goal with this build and it is time to move onto the next one.
I thank everybody for their help along the way.
Special thanks to Mjona, Brian92fs, PMA, and Amplidude. Your patience in the face of my foolishness has been endless.
I will post some pics of the final build in the gallery probably after the weekend.
Next up will be a Mooly's MMADFM. https://www.diyaudio.com/community/threads/my-mosfet-amplifier-designed-for-music.119151/
I have a stash of 2SK1058/2SJ162 just itching to make noise.
PMA, I will be coming for your recently published design soon enough.
I think this build is 'stable enough'.
My speakers are not exotic, my typical listening volume is low.
It is not going to be amplifying square waves into difficult loads in normal use.
My goal with this build was simply to build a well regarded 'nice' sounding amplifier, not develop a new variant.
I had some mishaps.
I learned a heck of a lot along the way.
My understanding of amplifier circuits has grown immensely.
I am now happy to say I have achieved my goal with this build and it is time to move onto the next one.
I thank everybody for their help along the way.
Special thanks to Mjona, Brian92fs, PMA, and Amplidude. Your patience in the face of my foolishness has been endless.
I will post some pics of the final build in the gallery probably after the weekend.
Next up will be a Mooly's MMADFM. https://www.diyaudio.com/community/threads/my-mosfet-amplifier-designed-for-music.119151/
I have a stash of 2SK1058/2SJ162 just itching to make noise.
PMA, I will be coming for your recently published design soon enough.
Last edited:
Square wave testing does inform in a broad sense. If you want to be more precise a Tian Plot simulation will get you closer to the mark. For this the generally accepted standard for gain margin is 10 dB or more with phase margin in somewhat in excess of 45 degrees. For transient purposes some prefer a value about halfway between this and 90 degrees - around 67 degrees. At the lower end it pays not to get to close to the edge.
Unfortunately I cannot find the time to build much at present due to my wife's health needs having priority over hobby interests.
I
Don’t understand why but the second image looks like 3d on my phone. What did you do. 😊
IMHO the polarity of C5 is incorrect. Did it enter so the PCB design? Might explain the problems with the oscillations.
It is backwards yes.
I did print boards but never used them.
I got my boards from Rod stable to my satisfaction put the amp into service and moved on.
It's a good amp with a nice sound. Simple enough for a novice to build with enough fish hooks to learn some things.
Interested parties are welcome to the gerbers to try the boards.