Well looking at 'the' open loop gain can be misleading. The very high olg of the '797 is caused by the very high gain at DC and very low frequencies, which is not that interesting. We're more interested in the loop gain at the end of the audio band because that's where the amp PSRR is lowest. If you look at 20kHz, the difference between the 825 and the 797 is just 10dB, a factor of three. That's why the difference in measurements is less than you'd expect.
But anyway, of course you can use a MOSFET, but you'll get less performance, which may or may not be an issue for you, about a factor of 10 less performance, roughly, all other things being equal.
Using a high loop gain opamp with a MOSFET and a low loopgain with a BJT is not an apples-apples comparison.
Jan
Hmmm. A power supply is an audio power amplifier whose input signal happens to be a fixed, constant, DC voltage (from a voltage reference zener diode or IC).
Since you're saying that a power supply with a MOSFET source follower output stage, has a factor of 10 less performance than the same supply with a BJT emitter follower output stage .... do you also mean that an audio power amplifier with MOSFET source follower output stage, has a factor of 10 less performance than that same power amplifier with a BJT emitter follower output stage?
This might be a controversial statement; it suggests that power amps with MOSFET source follower output stages are quite a poor idea!
Hmmm. A power supply is an audio power amplifier whose input signal happens to be a fixed, constant, DC voltage (from a voltage reference zener diode or IC).
Since you're saying that a power supply with a MOSFET source follower output stage, has a factor of 10 less performance than the same supply with a BJT emitter follower output stage .... do you also mean that an audio power amplifier with MOSFET source follower output stage, has a factor of 10 less performance than that same power amplifier with a BJT emitter follower output stage?
Mark. If you replace a BJT with a MOSFET in a supply, all other things remaining the same, you must get lower performance because of the lower loop gain.
That's all I'm saying.
Jan
Why don't you apply the exact same reasoning to an audio power amplifier?
The loop gain of a typical 3 stage (LTP, VAS, OPS) power amplifier is gm x Av x Rload x (R3/(R3+R4)). Loop gain is linearly proportional to the gm of the OPS. If you use MOSFET output transistors instead of BJTs, gm falls and so does loop gain. Doesn't performance suffer by the same factor of ten? Why not say so?
And while we're at it, doesn't the same reasoning suggest that Douglas Self's "XD" (crossover displacement) idea is an amazing breakthrough in power amplifier performance? XD installs a 1 ampere current sink that intentionally runs the OPS at 1 ampere of DC bias current instead of the more common 0.1 ampere. When he increases the OPS bias current by 10X, output transistor gm (hence loop gain) increases by that same factor of ten! Don't you conclude that this improves performance by the same factor of ten? (admittedly at low power levels like the FirstWatt folks prefer)
Well you guys are WAY over my head so I am going to really sound like the beginner I am. I am looking for a +18to20vDc for my B1 build and was looking at half of the super regulator board. Does this sound like a practical idea. Can you offer info on choosing right transformer and board hookup considerations? Thanks! I was trying to do with a Peter Daniel universal power supply board that I saw on the B1 thread but it's not available anymore and I don't want to use a walwart.
best regards,
Steve
best regards,
Steve
You seem to miss that other than proposing an increase in current source and hfe, I also suggested moving to an opamp having greater output current capability. If there is one with the other specs required, anyway.
Jan, that is what I meant. Maybe I should replace "feed current" with "output current capability". Sorry for any confusion.
Sorry for any confusion.
I wasn't - but others who are less familiar might. I know, I sometimes split hairs, but communication is difficult enough as it is with only words...
Jan
So that's a piece of cake. The other thing to decide is whether you want to use remote sensing. I believe the article gives all the info.
Jan
Last edited:
You seem to miss that other than proposing an increase in current source and hfe, I also suggested moving to an opamp having greater output current capability. If there is one with the other specs required, anyway.
Jan, that is what I meant. Maybe I should replace "feed current" with "output current capability". Sorry for any confusion.
I have taken note of it.
Thank you.
Well if you want 18-20V DC out, your 12+12 will be too low, unless you treat it as a 24VAC single winding with a bridge rectifier.
The Vout is set by the two resistors that feed the inverting input of the opamp, don't remember the exact reference # right now.
Say you want 18V out. The bottom resistor has the same voltage as the LM329 reference (6.9V) because the inv and non-inv opamp inputs will be at the same voltage.
Then size the top resistor for the remaining 18-6.9=11.1V.
So if the bottom resistor is 1k with 6.9V, the top resistor needs to be 11.1/6.9 times 1k, around 1.6k. Similar for 20V vout, top resistor must be 13.1/6.9*1k= 1.9k.
Makes sense?
Jan
The Vout is set by the two resistors that feed the inverting input of the opamp, don't remember the exact reference # right now.
Say you want 18V out. The bottom resistor has the same voltage as the LM329 reference (6.9V) because the inv and non-inv opamp inputs will be at the same voltage.
Then size the top resistor for the remaining 18-6.9=11.1V.
So if the bottom resistor is 1k with 6.9V, the top resistor needs to be 11.1/6.9 times 1k, around 1.6k. Similar for 20V vout, top resistor must be 13.1/6.9*1k= 1.9k.
Makes sense?
Jan
A very general rule for regulated supplies is:
Use a transformer Vac rating that equals the regulated Vdc you require.
This works well for all output voltages from ~10Vdc to ~30Vdc
At DC voltages below 10Vdc the Vac will need to be a bit higher. eg for 5Vdc use a 7Vac transformer.
At DC voltages above 30Vdc the Vac can be a bit lower. This saves on dissipation in the voltage dropping devices.
eg, for 50Vdc use a 40Vac or 45Vac transformer. You can use 50Vac and it will work. But it will get very hot and need enormous dissipation capability.
The exception is the Salas Shunt Regulators.
They use a CCS as the dropping device and need an extra few volts drop to stay in the correct CCS range compared to series regulators.
Add three to five volts to the transformer rating for a Salas.
eg for 10Vdc (DCB1) use a 13Vac to 15Vac transformer.
Here's a simulation bode plot of the SR with the MJD44H11 pass transistor and an IRFP240 mosfet pass device, AD825AR error amplifier:
I couldn't get the AD797AN to work properly in the simulation with a MOSFET.
An externally hosted image should be here but it was not working when we last tested it.
I couldn't get the AD797AN to work properly in the simulation with a MOSFET.
Did you break the feedback loop and plotted the results?
EDIT: I thought MJD44H11 and D44H11 were different trannies.
Last edited: