MOSFET Dissipation (for Power Scaling / VVR)

[R.G.]

Completely off topic, here's a great read:
http://www.lhup.edu/~dsimanek/museum/patents.htm

[R.G.]

OK, Mr. Simulator says that two CMOS 555's generate the required timing and drive polarity. Also runs from full wave rectified 6.3Vac. Just need to figure out an isolated MOSFET drive for either a high-side MOSFET or a low-side MOSFET. There are quirks with this positioning on both approaches.

I'd use an LED to PV driver and have done, but PVs take as much as 5mS to turn on a MOSFET. Too slow for this app. I'll have to come up with something else.

[R.G.]

... and I found a sub-100uS turn-on LED-PV isolated MOSFET driver, and a 1kV MOSFET that ought to work in this app, 4A continuous, 16A peak, 2 ohms, 120W.

So if anyone wants to try it out, I'll sketch up a schematic and a bill of materials. Right now it looks like:
1A full wave bridge - $0.50
1x 1N400x
470uF cap - $0.35
2x LMC555N - $2.00
Vishay Opto isolated driver, $2.50
STM 1kV/4A MOSFET $2.00
The usual suspects for resistors and caps.
Oh. A pot. It needs a 50K pot to set output B+ between full and way, way down there if not zero. Probably $1.00.
So a raw guess at parts is maybe $9.00-ish, in units quantities. The MOSFET will need some heatsinking, but far less than it would doing a linear regulator. Maybe chassis-mount low. The mechanical job of putting it on a bit of board and wiring it right will be interesting.

[Gingertube]

mhus,
I did a quad of 6V6 powerscaled. Picture from during the builds is attached.

THe MOSFET is bolted to the side of the piece of aluminium C section. The powerscale board is the early London Power one.

As far as temperature rise goes I can only give guidance. In the day job I did a 3 x parallel mosfets to deliver 100Amp 100us pulses into a pair of laser diode arrays to optically pump the main Nd:YAG laser slab.
On those I put thermal switches to shutdown the circuit if they went overtemperature. I used 90 degrees C thermal switches.

Cheers,
Ian

[Tom Phillips]

The attached photo isn't working for me. Please try again.
Thanks,
Tom

[Gingertube]

Take 2 using a previously uploaded image from that 4 x 6V6 build. Did'nt have any luck trying to upload a new image.
See Right Hand Side - PS board in the ally channel with the MOSFET bolted to the side - not yet connected to the PS Board.
Sorry,
Ian

[MarkusBass]

... and I found a sub-100uS turn-on LED-PV isolated MOSFET driver, and a 1kV MOSFET that ought to work in this app, 4A continuous, 16A peak, 2 ohms, 120W.

So if anyone wants to try it out, I'll sketch up a schematic and a bill of materials.

I'm interested. Especially, that there is a long discussion about a circuit but actual schematic was never shown. I need to put power scaling circuit into my latest amp (50W, 2x6L6).

Mark

[mhuss]

So if anyone wants to try it out, I'll sketch up a schematic and a bill of materials.

I'm interested.
I never liked linear power solutions because they're intrinsically wasteful. This seems much more elegant, the tricky bit will be managing the layout and noise...

[mhuss]

Hiwatt was doing this back in the 70s (1/2 power switch that just put a large resistance inline with the screens). I never liked the way it sounded, although theoretically it should work fine.

[R.G.]

OK, here's a schematic:
Phase Shift Voltage Reducer

Notes and warnings:
- This is a design only. It has been simulated to the extent possible in a circuit simulator, and the timings seem to work. I have NOT wired up a prototype. There may be errors, and there may be oddities about the parts datasheet that might make it work less than perfectly. Can't tell until one is actually built.
- This design involves working with hazardous high voltages. DO NOT TRY THIS UNLESS YOU UNDERSTAND THE RISKS AND ALREADY HAVE THE SKILLS AND EXPERIENCE TO DO THE WORK WITHOUT ENDANGERING YOURSELF AND OTHERS.

Oh, yeah:
I remembered the original purpose of the thread. Finally. The RMS current in the MOSFET will be the same as the RMS current in the high voltage rectifiers. This will be at most about 1.8 times the DC current leaving the first rectifier. Call that half an amp max, so the MOSFET is heated by about 1A. The MOSFET datasheet says it will saturate to about 3.5 ohms; call that a "datasheet fact" and estimate it at 5 ohms. The MOSFET is then heated by 1A through 5 ohms, and dissipates on the order of 5W. It still needs a heat sink, as a TO-220 can only get rid of about 2W without a heat sink, but a fairly modest heat sink will keep it well cooled.

... if the circuit works.

[Gregg]

Thanks fro sharing the idea however it will work only on cathode biased amps. Since most guitar amps are fixed bias the device must include one more similar section for the bias voltage linearly tracking the high voltage.

[R.G.]

Thanks fro sharing the idea however it will work only on cathode biased amps. Since most guitar amps are fixed bias the device must include one more similar section for the bias voltage linearly tracking the high voltage.

There are lots of things it won't work for. It was an answer to the OP's question about cutting down the power in a MOSFET when reducting B+. The OP mentioned nothing about scaling bias simultaneously. I would assume he has his own answer to that issue, if he's interested.

It was not intended to be a complete voltage scaling design, or I'd have put in bias (and possibly other) scaling. Even then, I'd probably opt for other means to scale down bias voltages, not phase control. Phase control is a bit of a blunt instrument for bias voltages.

[Dave H]

Since most guitar amps are fixed bias the device must include one more similar section for the bias voltage linearly tracking the high voltage.

The bias voltage is very low power. Why not use a ganged pot and use the second section for the bias voltage?

[Gregg]

The pot sections usually don't track each other well and down in the "bedroom level" region even tenths of volts would cause the operating point to shift. You can fix that by selecting from several pots or just live with it.

[Tooboob]

1. The NFB loop - An NFB loop will try to "correct" for the power scaling. Until the power tubes clip, the volume control will work much as it does unscaled. It sounds like loss of headroom with an abrupt transition to distortion, not the fully powered amp at lower volume. There are other ways to reduce crossover distortion and improve linearity. One is to use cathode biasing. Using a resistor, bias current won't change as you scale, but it's possible to make an active circuit to sub for the resistor that tracks the scaling. This costs max output power and adds complexity. An ultralinear output transform is simpler and effective. Removing the NFB also improves the performance of a Post-PI master volume, which is still a nice thing in addition to power scaling.

2. Following R.G.'s plan, most stock amp designs will also scale the preamp supply voltages, and the triode biasing won't scale (much), so the preamp behavior changes dramatically. His circuit goes at the power supply rectifier. The only way to fix this is to have two sets of rectifiers/filter caps. The preamp's current needs are low, so these items can be smaller. Scaling of the PI is a judgement call, and with two supplies available, you get to try both. Also, power scaling won't affect the effects loop level with dual supplies if you run it off the unscaled supply.