Speaking of the zener string.. how much current do you have flowing through it? What is the actual measured voltage across the string? There is some component tolerance variation on those things. And some current starvation (if that's happening) won't drive them fully into the breakdown region.
Here is how I work this out:
Optimistic voltage after rectification:
64V*sqrt(2)=90.5V
But rectifiers drop voltage, so:
90.5V-3.0V=87.5V
If we assume 80 mA are drawn...then
1.48V of ripple...
87.5V-1.48V=86.02V
Drop across balancing resistors...
80mA*33R=2.64V
2.64V*4=10.56V
Voltage left before MOSFET...
79.62V-10.56V=69.06V
Drop due to MOSFET...
80mA*2.75R (Rds On)=0.22V
Voltage left after MOSFET...
69.06V-0.22V=68.84V
WHICH is pretty much what I get at the output!
What an interesting exercise!
So, the Zener string doesn't clamp because there is no excess.
Is a Zener string noisy in these conditions?
Last edited:
Zener string is fed by the 6k8 resistor (relying on memory here). Should have > 1 mA flowing through it. If not, there is not enough clearance in the supply, i.e., the supply to the reg fet needs to be higher. I can explain more later.
In your case it looks like you have enough clearance.
In your case it looks like you have enough clearance.
I'll have to put that board back in circuit to investigate further.
But based on my previous calculations, only leakage current would flow through the Zener string...and I since learned that the Zener string should not be a source of noise when no current flows through it.
I put that board in circuit.
My math went wrong somewhere, probably because I overestimated the current draw.
In any case, reality is real and all that matters...here are some numbers:
64.5VAC of input
79.2VDC at pin 2 of the MOSFET
74V across the Zener string (18V+18V+18V+20V = 74V), so that's quite excellent and the string is in fact clamping.
But only 68.8VDC at the output, so the MOSFET appears to be dropping more voltage than expected.
PS: I must apologize for this distraction. I am trying to explain why one PSU board introduces noise while the other does not, and I ended up dissecting the voltages of the board that introduces noise. For all I know, the board does NOT introduce noise and the noise is a side effect of higher supply voltage to the preamp board when the output JFETs are removed. Alas, I can't think of a way to verify that.
My math went wrong somewhere, probably because I overestimated the current draw.
In any case, reality is real and all that matters...here are some numbers:
64.5VAC of input
79.2VDC at pin 2 of the MOSFET
74V across the Zener string (18V+18V+18V+20V = 74V), so that's quite excellent and the string is in fact clamping.
But only 68.8VDC at the output, so the MOSFET appears to be dropping more voltage than expected.
PS: I must apologize for this distraction. I am trying to explain why one PSU board introduces noise while the other does not, and I ended up dissecting the voltages of the board that introduces noise. For all I know, the board does NOT introduce noise and the noise is a side effect of higher supply voltage to the preamp board when the output JFETs are removed. Alas, I can't think of a way to verify that.
That paints a better picture of what's going on... With only 5.2V across the feed resistor, the current through the string looks really low to me. I would remove the 6.8K resistor and swap it out for a 1.5K to get a more reasonable amount of current through the zeners (5.2 divided by 1500 = 3.5mA). Some people shoot higher... i.e. 5mA to 10mA through them. To my knowledge, zeners are not less noisy when starved.
Source pin of mosfet sits about 4V lower than what’s on the gate. So you should get about 70V if there is 74V on the gate.
Pin 2 is the drain, isn’t it? That sounds like enough clearance. A few volts here and there is nothing to worry about. 68V out is good.
Also, if you increase current through the zeners then check the dissipation. 10 mA with 18V is 0.180W, that means at least a 0.25W zener. Should be okay.
Also, if you increase current through the zeners then check the dissipation. 10 mA with 18V is 0.180W, that means at least a 0.25W zener. Should be okay.
Sharing this in case it's of help to @ElArte.
I'm using a 50VCT transformer, a Hammond 166G50, and an STP45N10F7M MOSFET.
I initially had it setup with a 20 + 20 + 20 + 15 Zener string for 69.4V out. I was getting noise on the scope, 30mV or so if I recall.
So I reduced the string to 20 + 20 + 20 + 6.2 for 60.7V out and the noise went away.
The notes on the schematic tell the story.
Would be interested to learn whether I could manage to eek more supply voltage out with the same transformer by making some adjustments to the FET or R6 value.
I'm using a 50VCT transformer, a Hammond 166G50, and an STP45N10F7M MOSFET.
I initially had it setup with a 20 + 20 + 20 + 15 Zener string for 69.4V out. I was getting noise on the scope, 30mV or so if I recall.
So I reduced the string to 20 + 20 + 20 + 6.2 for 60.7V out and the noise went away.
The notes on the schematic tell the story.
Would be interested to learn whether I could manage to eek more supply voltage out with the same transformer by making some adjustments to the FET or R6 value.
Attachments
Again, I started this autopsy of the PSU board because removing it and replacing it with another PSU board that gives ~58V of regulated DC instead of 68V eliminated the noise issue I had with the preamp build.
From there, I concluded it could be 1 of 2 things:
1) There is a problem with my build of the PSU board.
2) Feeding 68V DC to my build of the preamp board when the output buffer is desoldered causes noise.
The autopsy of the PSU board didn't reveal any obvious problem as far as "1)" is concerned.
I had AI summarize the autopsy:
AI is not ready for electronics yet, it makes too many sloppy mistakes, even more sloppy than my mistakes, but this summary seems about right.
So, if it's not "1)", it must be "2)" and I do not imply it is "2)" in the general case, since some of you feed > 68V to the preamp board and have no problem. It must be my build of the preamp board.
All in all, it was a matter of curiosity, as I already know that using the PSU that regulates 58V fixes the problem.
I was just hoping there was something wrong on the supply side, so that I could fix it and enjoy the benefits higher supply voltage brings.
From there, I concluded it could be 1 of 2 things:
1) There is a problem with my build of the PSU board.
2) Feeding 68V DC to my build of the preamp board when the output buffer is desoldered causes noise.
The autopsy of the PSU board didn't reveal any obvious problem as far as "1)" is concerned.
I had AI summarize the autopsy:
AI is not ready for electronics yet, it makes too many sloppy mistakes, even more sloppy than my mistakes, but this summary seems about right.
So, if it's not "1)", it must be "2)" and I do not imply it is "2)" in the general case, since some of you feed > 68V to the preamp board and have no problem. It must be my build of the preamp board.
All in all, it was a matter of curiosity, as I already know that using the PSU that regulates 58V fixes the problem.
I was just hoping there was something wrong on the supply side, so that I could fix it and enjoy the benefits higher supply voltage brings.
Ah, now I get it. It can take me a while sometimes to catch up. This is about the noise. Try increasing the the current through the zeners as William suggested. Maybe start with 3k3 instead of 6k7. That may cure it. If not, would be interesting to find out what is causing it. I will check my build also. Is the noise always present? Or is it linked to pot position?
Last edited:
Yes, always present when the source and amplifier are both connected. When no signal is going, it’s easy to hear it. When music is playing, it gets drowned, but you can verify it’s still there during quiet passages.
As far as the nature of the noise, it’s unlike any I heard before. Not a hum, not a hiss, not static…more like some kind of puttering. Similar to the noise you hear when big MIFLEX KPCM caps are new and forming.
It’s audible at all positions of the volume knob, but a bit louder in a certain range.
Anyway, with the 58V supply, I was able to confirm the preamp makes class D sound glorious! The 58V supply with output buffer desoldered gives 0.21% THD, 100% H2 at 12’o’clock on the knob.
As far as the nature of the noise, it’s unlike any I heard before. Not a hum, not a hiss, not static…more like some kind of puttering. Similar to the noise you hear when big MIFLEX KPCM caps are new and forming.
It’s audible at all positions of the volume knob, but a bit louder in a certain range.
Anyway, with the 58V supply, I was able to confirm the preamp makes class D sound glorious! The 58V supply with output buffer desoldered gives 0.21% THD, 100% H2 at 12’o’clock on the knob.
With the "58V" supply, the zener string was 18V+18V+18V+10V = 64V. So if you run the napkin math, 79.2 minus 64 = 15.2V. Now you can calculate the amount of current through the string using ohm's law. 15.2 volts divided by 6800 ohms = 2.2mA. A much more reasonable number.
Thanks!
I missed your argument before.
Your point is that “starving”, i.e. having too little current go through the Zener string, could be the source of the noise.
I checked The Art of Electronics and they specifically refer to that Zener noise as “popcorn noise”, which matches quite well what I heard!
The datasheet of the Zener diodes states their characteristics at 5 mA of test current, so I guess they should get this much.
I will change R6 to 470R or 560R and test again.
Hold on a minute. If you're targeting "5mA" on the noisy power supply, you would want a 1K resistor at R6. Again, with the math... 5.2V divided by 1000 ohms = 5.2mA. R6 acts as a current limiter here..
With all due respect to the classic textbook, the latest LEDs reach full brightness at lower current. I recommend about 1mA, or a resistor value of 4.7k.