[KimLaroux]

Great information right here. Thanks.
 
In my build, I have each MOSFET decoupled with a 330 µf capacitor. I'm thinking about adding a low value resistor before them, to create a CRC network at each FET. Would that be a sensible solution?
 
I just crunched some numbers real quick. I think a 20 R resistor would be good enough. It should drop 3 V and dissipate a little under half a watt. I don't know though how to calculate how many dB such a filter would attenuate. I'm also concerned 330 µf per FET may be too low. I have a 4700 µf cap on the output of the power supply, and isolating the output MOSFET from it just seems like a silly idea.
 
Now that I know how the source follower works, I feel safe about removing 3 V from the MOSFET Drain. It should not hurt it's operation.
 
Another solution would be to replace this 4700 cap with two 2200 and place a resistor between them. Since I can adjust my power supply up to 65 V by changing a single resistor, I can compensate for the drop across the resistor. Or just wait and do it when I modify my circuit to use this 65 V... Though something tells me that using a resistor to isolate the MOSFET from the power rail at the other end of the amplifier may actually provide better isolation against noise induced on the rail as it crosses the amplifier.

 

[Goobley]

Following what nikongod said (thanks for the information about the mu metal). In my testing I'm fairly sure that noise is only coming through the power lines as I kept the power supply at least 1.5m away. I'm not going to be putting a regulator after the switching ps, I'm going to make a whole new linear regulated power supply instead (see note 1). The main reason for this is the fact that even measuring from ground to ground with this power supply I can see noise, and I really don't like the way these things switch (square waveforms 50v amplitude compared to mains ground with 180° phase difference between V+ and V-, it's hardly surprising it sounds like *****). I'm doing a bit of simming now, trying to decide how much regulation is too much (from a cost to performance point of view). At least I'm getting there, you guys' help has been invaluable even if I didn't always listen to you
 
Note 1: Reading what Kim said about increasing the power supply, I may well try this too, as far as I can see more plate voltage is a good thing (I don't think the mosfets wold need increased heatsinking, would they? as I believe it depends only on the current actually flowing through them).
 
I also have a few decent sized capacitors around and may try decoupling the mosfets... hmm... lots of things to try (including quite a radical change with a different ps) but not enough time. Ah well I'll try and get my stuff ready for a large diy session over the christmas break.
 
Cheers

 

[KimLaroux]

Quote:

Following what nikongod said (thanks for the information about the mu metal). In my testing I'm fairly sure that noise is only coming through the power lines as I kept the power supply at least 1.5m away. I'm not going to be putting a regulator after the switching ps, I'm going to make a whole new linear regulated power supply instead (see note 1). The main reason for this is the fact that even measuring from ground to ground with this power supply I can see noise, and I really don't like the way these things switch (square waveforms 50v amplitude compared to mains ground with 180° phase difference between V+ and V-, it's hardly surprising it sounds like *****). I'm doing a bit of simming now, trying to decide how much regulation is too much (from a cost to performance point of view). At least I'm getting there, you guys' help has been invaluable even if I didn't always listen to you
 
Note 1: Reading what Kim said about increasing the power supply, I may well try this too, as far as I can see more plate voltage is a good thing (I don't think the mosfets wold need increased heatsinking, would they? as I believe it depends only on the current actually flowing through them).
 
I also have a few decent sized capacitors around and may try decoupling the mosfets... hmm... lots of things to try (including quite a radical change with a different ps) but not enough time. Ah well I'll try and get my stuff ready for a large diy session over the christmas break.
 
Cheers

 
Yes they will. Heat is power, and power is Current times Voltage drop.
 
P=UI
 
P is power in watts
U is voltage drop
I is current in ampere (not mA)
 
A 12AU7 MSSH has 35 V across the MOSFET, and 0.15 A. 0.15*35= 5.25 W. If you raise the drain voltage to 65 V, you'll have 52 volts across the MOSFET. 52*0.15=7.8 W. And that's only one MOSFET, together they'd dissipate 16 W. If you count the tube heaters, the whole amp would dissipate 20 W. That's just ridiculous for an headphone amplifier. I would not recommend it.
 
The only reason I'm considering it is because the transformer I use is 30-15-0-15-30. I'm currently only using 30-30. I could use the lower taps, 15-15 for a second rail at a lower voltage. That way I won't be wasting more heat. At this point I could create a voltage doubler to get over 100 V on the tubes, and still have a 30 V rail for the output MOSFET. My build would actually be cooler that way.
 
But yeah at this point I won't call it a Starving Student anymore. 
 
Another thing to keep in mind is capacitors voltage rating. High capacity capacitors are very expensive in high voltages. They also get bigger. Most of the caps in my build are 100 V, except the big cap at the output of the PSU which is rated for 63 V. So I would not be able to use this one if the rail was 65 V. I have buckets of capacitors I salvaged over the years, but none of them had the correct voltage rating and capacity needed for the 48 V SSMH. I had to buy all of them new. (Which is probably preferable anyways, as caps degrade over the years...)
 
And you can't just raise the voltage on the stock amplifier. I still haven't done it mainly because I haven't learned everything about the design yet. College have kept me busy, so I didn't get the chance to learn how tube circuits works. I have a feeling you'd have to replace a couple of transistors for it to work properly at a higher voltage. As long as I don't know why each of the resistors are at their current values, I'm keeping it at 48 V.