X hits on this document





3 / 5

IRF820's are rated for a max Vds of 500V. You have to have a B+ supply that can supply all the necessary current to the 820's and still make the drain-source voltage under 500V even under moderate overvoltage, say about 10-15% to allow for line voltage variation. This is the wrong place to start the design, though, as the 820's need to be designed to make the output tube grids happy, then make the drain voltage right.

2. What is the best way to derive the high negative voltage for the cathodes? Note: I get 425V B+ from the rectifier plate. Also, the PT's 2A/5V filament winding is unused, if this can be used at all. 3. I adjust the -ve supply so I get -36V on the cathodes, right?

*This* is the place to start the design.

You have to have the source of the MOSFET be able to sit at -36V, and swing to several volts positive to get into positive grid voltage on the output tubes. That means that they have to be able to swing the same amount of voltage negative on the other half cycle of signal to be able to turn the off- side tube off.

So the minimum minus supply for the MOSFET sources has to be at least twice the existing bias supply, plus a bit more to cover the odd cases. The minus supply needs to be at least -80v (about) and maybe a few more, so -100V would be good. -90 might work, depends on exactly how far you have to drive the output tubes to get them to cut off.

Note that this analysis is for a marginal-Class-A amp. The deeper you run the amp into AB, and the closer to B you get, the less negative bias swing you need. A pure class B amp would only need about -40V. But for average class AB, lets say -75 to -80V.

Having done that, we have eaten 80V from the available Vds on the MOSFET. So the drain supply for the MOSFET can't be more than 500V-80V = +420V without killing the MOSFET from overvoltage. You could use a higher voltage MOSFET (they're available up to about 1kV) or just a lower B+. Anything safely under 420 will work. The minimum is probably about 40V-50V to allow the MOSFET to pull positive. So, any drain voltage from 40V up to 300V is OK. The higher this voltage, the more heat wasted in the MOSFET.

This voltage supply has to provide enough current to pull the source-follower resistors and positive grid current up to the most-positive drive voltage. For the 10K resistors listed, that's going to be about (-Vs -10V)/10K. Typical values might be (-80V-10V)/10K = 9ma plus grid current of a few ma. Call it 12ma. The grid resistors dissipate (-80V - (-36V))* 0.012A = 0.528W, so you need 1W resistors there. This happens once for each MOSFET, so you could need 24ma of supply current. Notice that the source resistors are the majority of the current, so you might be able to up the source resistors to 22K or even 47K. The higher the resistor, the lower the off-state drive; however, those would probably be good enough. I selected 10K as a good high performance driver. 47K source resistors would require only about 5ma per side, 10ma total. Actually, that's conservative, as only one side will be at max current at a time. You might get by with 50-75% of that if you're on a tight current-budget.

That current has to come from the + supply on the drain of the MOSFETs. So you need something that can provide 10 to 12ma of current at no less than 40V at max current, and no more than 300V at minimum current. The simplest way to do that is with a resistor-zener supply from the main B+. You make the resistor drop your 420V to 40V under max current: (420V-40V)/24ma = 15.833K. You have the zener eat all the current when the voltage hits 300V: 420-300 = 120V; Izener = 120V/15.833K = 7.6ma; so you need a 300V, 2.3W zener minimum. Any zener between +40V and +300V that can eat the no-load current at that voltage will do. I would use an amplified zener with another MOSFET.

Which brings up another way to do this. Use a second MOSFET with its drain tied to +420, its gate tied to one of the lower preamp voltages, like maybe the +260V preamp section, and its source tied to the drain of the actual source follower MOSFET. Now the upper MOSFET does the heavy lifting on the power supply. Hmmm. I hadn't thought of that one. I'll have to draw that up. It's a neat trick. The reference supply is just that, a reference. The gate of the power supply MOSFET will not load your preamp power supply. I'll have to draw that one up.

The -Vs supply has to be able to eat all the current that the B+ supply lets through, too.

4. How do I bias the grid (gate) of the MOSFET?

Use the old bias pots where the pots are shown in the article. In this case, you do the same job you used to do - measure the power tube current and twiddle the knob until the power tube currents are right. However, because the MOSFETs are enhnacement devices, the actual voltage on the MOSFET gates will be about 1-3V higher than the voltage used to be when it drove the power tube grid directly. You won't notice this; just bias for power tube currents.

5. I seem to recall some previous posts here about blowing MOSFETS in this application if not done properly. Is there anything else I need to add that's not mentioned in the R.G.'s article?

Document info
Document views15
Page views15
Page last viewedFri Jan 20 09:46:00 UTC 2017