801 amp with Power Drive

[Sheldon]

I've built a couple of 801 resistively loaded stereo amps, as describe here: http://www.diyaudio.com/forums/showt...558#post883558

Nice sounding, earth shaking 1/2 watt. So for fun and more power, I'm plannin'g on converting one of these to a regular choke loaded parafeed amp, and to incorporate Georges power drive circuit for some additional headroom. Question: I know the bias supply to the gate of the mosfet needs to be nice and quiet, but how about the supply to the source? I would think that as a follower, it should have pretty good ripple rejection. What kind of ripple limit should I go for here?

Couple of other general questions. I plan to use the same power trans and chassis (cause that's the hardest part of these projects), except for the top plate. So space will be at a premium. I'm shooting for a B+ of about 400V. Was looking at reusing the PS chokes and going with a choke input supply, followed by maybe a mosfet/Maida style regulator, or some type of mosfet follower. Would like to keep voltage drop to a minimum and don't want to burn off more than a couple of mA, so no shunt regs. Don't necessarily require regulation, but would like nice low impedance. I've found some workable designs here, but open to other suggestions too. I notice that Wavebourn had posted a follower design, but the figures are no longer available.

Sheldon

[Sheldon]

I'm thinking along these lines:

[Sheldon]

And the supply:

[mach1]

Sheldon,

I'm not sure about 'power drive' ? - the schematic appears to be a conventional mosfet source follower, which is the solid state equivalent of a cathode follower and does not exhibit very good PSRR. However, I would expect your posted low ripple psu to be fine.

The following link may be of some interest.

http://www.geofex.com/Article_Folder...osfetfolly.htm

[Sheldon]

Thanks Mach1. The power drive name came from the Tubelab site. I had seen the site you mentioned, and it's a good one. It does include the use of a mosfet source follower as an output tube driver. George goes into more detail in their use for tubes driven into A2. So I added the power drive name to the title to maybe attract those experienced with this topology.

As this amp has been "designed" but not yet "engineered" (SY's dictionary of DIY), I was hoping to at least somewhat rectify this current state. And just curious too, so I'll see if I can chase down some approximations of the quantitative values for PSRR of the mosfet source follower. Along the same lines, I was also hoping to entice, in the interest of engineering, some of our bretheren to tear into the power supply design. I may be forced to learn to use Linear's Spice program after all.

Sheldon

[mach1]

Sheldon,

I remember reading the PSRR figure -20db for a 12AU7 follower somewhere, so I assume a source follower

[mach1]

Sheldon,

I remember seeing a PSRR figure of -20db for a 12AU7 follower somewhere, and assume a source follower would be in the same ballpark.

Years ago I used a similar psu to yours for a linestage and it was very quiet. I am sure it would be fine for your intended application if the mosfet was heatsinked. If you want a series regulated psu, then a TL783 with protective zeners would do the job easily (I am currently using these for pentode g2 supplies). If you want to go passive, a second LC section would probably be fine as well.

[tubelab.com]

In theory the mosfet follower has decent PSRR from the positive supply, but not as good from the negative supply. The mosfet and the source resistor form a voltage divider across the two power supplies. The divider is made up of the internal mosfet resistance and the source resistor. The internal mosfet resistance is the dynamic impedance, determined by its characteristics and the current through it (similar to the plate resistance of a tube). The Rdss on resistance (usually small) is added to this value.

If the power supplies to the mosfet were equal, and the output (grid) voltage was adjusted to 0 volts the internal resistance of the mosfet would be equal to the source resistor. If a symmetrical power supply is used in this case there would be perfect hum cancellation in this voltage divider since the output (grid voltage) is taken from the center. In reality the output voltage is adjusted to a non zero value, which is changing with the audio signal, so the hum cancellation is less than perfect, but useful. Tubes with sensitive grids (6L6, EL34, EL84) are usually operated with only a few volts of negative bias, so the hum cancellation helps. Tubes that require a lot of bias, usually are not very sensitive either.


Due to many requests I have been working on a "universal PowerDrive" circuit that can be added to many amplifiers. The power supply is a full wave bridge across a center tapped 230 volt transformer followed by a CRC filter. The positive and negative halves are identical. The supply ripples are equal but out of phase allowing for some cancellation of hum in the mosfet. Output voltage is + and - 150 volts.

I am using a Triad N-68X isolation transformer with the line connected across the secondary and the primaries in series used as a 230 VCT secondary. The diodes are UF4004, the caps are 100uF and the resistors are 220 ohm. This transformer is overkill but I had one already. The other components were also chosen because I had them. Any small isolation transformer that can be wired to produce 230 VCT can be used. The N-68X is cheap from Mouser.

It would seem that the power supply circuit that you posted would be adequate, but I have never tried the PowerDrive with a voltage doubler as a power supply.

[SY]

Sheldon, look at the curves for the MOSFET. Nice and flat, eh? That means that the drain-source voltage can be all over the place without upsetting the gate-source voltage. So the rejection is extremely high. If you want to make it even higher, you could cascode. But the signal level there is pretty high, so it's doubtful that any reasonably low supply ripple would contaminate the output via the followers, so why complicate?

The 12AU7 you mention has worse power supply rejection because its voltage gain is lower. Something less than 26dB would seem right, since its mu is 20.

edit: tubelab's point about negative PSR is a good one and argues strongly for a CCS or using a large source resistor and run the follower AB.

[Sheldon]

Thanks guys, very helpful. I like George's notion of using antiphase ripple to cancel some of the noise. I had thought that this might help. I'll try this first and see how it goes. If I get some residual hum, I'll put a current source in the negative leg, as SY suggests. I like my amps to have no audible hum to within cm of the speaker cone. No doubt overkill, but hey, what is fanaticism if not overkill.

Mach1, I assume you are referring to the cap multiplier approach for B+. Anyone care to comment on the cap multiplier approach for a power amp? I would think that it should have adequately low impedance over a wide enough frequency range, especially with the large plate choke?

Sheldon