The All-New WA1GFZ MOSFET Audio Driver - Pics and Working!

[K1JJ]

Hello All,

After a couple of months of design work by WA1GFZ and building/testing by me, we have a working MOSFET Audio Driver.  I have it running now on my 4-1000A plate modulated by a pair of 4-1000A's.

One volt of audio in, hundreds of clean watts out, with negative feedback if desired.

The results appear to be quite good with this first prototype.  It will pass a 50hz triangle waveform well with sharp points and near perfect slope. This is not easy to do unless direct coupled. My version uses 2.2 ufd coupling caps which allows response down to about 10hz. The high end is easily 12 khz and very clean.  It uses a single + - 300V supply for everything. It will drive any set of modulator tubes. My 4-1000A  tetrode-connnected modulators work well.  The 10hz lows limitation is good to prevent mod transformer burn out. Even 30hz is a better limit. This is set by the coupling caps... bigger = lower response.

Because there is only one transformer in the system now (mod xfmr) I am able to tap the negative feedback (NFB) off the modulation transformer secondary winding. Normally this is impossible to do with two transformers in line. (Like with a driver transformer, etc)

The driver uses five 11N90 900V MOSFETS at about $3.00 each.  I have them mounted on a heatsink and the Fets show no signs of getting warm.  Compared to my older 100w 8 ohm amplifier driving a Linear Standard transformer audio driver, this unit is an improvement. The low end is cleaner and more intense. I now have difficulty choosing which phase to run and can almost justify the "shark fins" phase. This is tough to do with transformer rigs. Before, the opposite phase was very obviously the correct choice. This requires a great low end response to pull off.

The two pots shown in the pictures set the bias for each modulator tube.

The layout does not seem critical, as I have it temporarily mounted in the field of Fabio's RF deck (with no case) and see no instability problems.

Anyway, Frank will make some comments and we will post the schematic tomorrow. The present .wmf format does not upload here.  I know a bunch of you were patiently waiting for this design and it's now working FB. There are no plans for a circuit board at this time. I personally plan to run my handwired prototype indefinately.

** Be sure to use fuses in the grid circuit as shown on the schematic. MOSFETS will sink a lot of current and you might end up with a pile of grid wires in the bottom of your modulator tubes, otherwise...  


Hope to see some built soon.

73,

Tom, K1JJ


WORKING  PROTOTYPE:

[WA1GFZ]

We are trying to convert a WMF file to something that will load. I just sent Tom a ms word and PDF file. Our scanner here at work doesn't do jpg.
A PC board might be as large as 6 by 6 inches so might be cheaper to do point to point. When I get some time I'll post component values for lower voltage operation. This configuration will do close to 500 volts grid to grid and can be biased negative. I'm pretty sure it will motivate tubes like the 833. I think Tom's biggest issue was finding 600 volt film caps. The input cap is two electrolytics minus to minus to get the high value required to pass lows. In simulation with a 50 K local feedback resistor the circuit has less then 30 degrees phase shift 10 hz to 10 KHz. The trick was to make the bandwidth very wide to hold the phase shift stable. Lower voltage operation requires some gate bias resistor changes.
the cap values are not critical. The simulation does not have a pot component so I made 2 resistors where pots are required to set bias for each tube and NFB level. Tom and I have a lot of hours in this but in the end the test data matches the simulation. I didn't add current limiting so use fuses. Current limiting tends to oscillate at the threshold so decided not to use it. 1 transformer can be used to generate positive and negative 300 volts. I have simulations that go down to +/- 100V. This allows the use of lower voltage caps.

[K1JJ]

Here's the WA1GFZ MOSFET Audio Driver SCHEMATIC in .pdf format.



Frank did a great job in designing it for stability and proper bypassing.  As you can see in the pictures above, the handwired version is not wired tightly like an RF unit would be -  but still works perfectly.


BTW, Mouser has 2.2 ufd 600V film caps to use for the four coupling caps.




T

[steve_qix]

I guess a few comments...

I'm absolutely *AMAZED* you don't get a parasitic with the gate zeners connected directly from gate to source without a small resistor (usually a few hundred ohms is plenty) in series with the gate, between the gate and everything else (including the zener).  I have seen parasitics without the resistors (which is why I use them in all analog designs like the class H modulators).

A tube arc-back to the grid can cascade through the circuit.  I would humbly suggest 3 things:  a) strategically placed transzorbs,   b) back biased diodes going from the output to each rail and   c) fuses (as suggested on the circuit).

What are the peak voltage excursions that can be developed looking into a load that equals the peak grid current of an 833A ?  I have to dig out my RCA transmitting tube manual and check the curves for the 833.  I seem to recall the peak grid current is fairly high when the tube is driven hard.

For non-grid current applications (like AB1), things should work ok.

Regards,

Steve

[K1JJ]

Steve,

Tnx for the suggestions.  Frank and I are always looking for ways to improve things, especially for stability and the safety of devices.

It would be easy for me to make some additions to my prototype board here.  What value resistor and wattage do you recommend for each gate input?

Where would you put the safety transorbs and what values?

As far as drawing grid current for tubes like 833A's, etc... what current limit do you see in the design?  I am running my 4X1's AB1, so don't need current there, but others may. Frank may chime in his reasons in this area.


T

[WA1GFZ]

Steve,
A positive transient will conduct through the FET body diode. a reverse diode across the FET wouldn't hurt but not sure if it helps. The body diode can conduct a lot of current.
A negative transient be current limited by the drain resistor. This value can be adjusted higher depending on the required current. Still it wouldn't hurt to put transorbs rated for the peak output voltage. We tested the driver with a 1K load and has no problems making current. The 833 needs about 400 volts grid to grid biased at -60 to -80 volts. The driver can do negative bias. Simulation shows I can do close to 500 volts grid to grid. At zero bias it will do about -125 to plus 125V. I have always put the zener directly gate to source with many designs going through mil qualification. The Body diode is rated 11.4A SS and 45.6 amps pulsed. So a reverse diode across the 100 ohm drain resistors (anode side to Drain) and transorbs on the power supply rails is a good protection idea. A negative pulse would turn the tube off.

[WA1GFZ]

OK tube experts you get a flash back through the tube to the grid.
What is worse. Provide low impedance protection and melt the grid or let the voltage swing and let the rocky point resistor limit plate current.

[W2VW]

Tube arcs from gas are short in time so heating shouldn't be a problem. The arc could be sustained due to a high source impedance on the grid bias without trip protection.

Hi Mu tubes should have the grid voltage clamped. Grid to cathode voltage is more of a problem.

An 833 falls somewhere else and I'm not too sure if you could ruin it either way.

Just thinking out loud.

[W1RKW]

Frank,
What do you use for a simulation program.
Bob

[WA1GFZ]

LTspice IV free download from Linear Technologies.

Dave,
Do you think Tom should have a resistor between each grid and the heater transformer CT?  Maybe a high voltage transorb?
He is running close to zero bias so it would not be much extra loading on the driver.
The present configuration will pull the grids to -300 through 50K if a fuse blows.  I think I would be more concerned with the screen in a tetrode.
I wonder what broadcash rigs did when using the 4X 807 driver.

[Opcom]

That is a very useful circuit! Is it sensitive to the power supply or can do with an unregulated one such as a FWB and C input?

[K1JJ]

It produces a relatively steady load on the power supply so regulation requirements are minimal.   The fets don't even get warm when in use.  I use an unregulated +- 300V supply using a variac.  Capacitor input, no choke.

T

[WBear2GCR]

Wondering what the real world bandwidth of the driver is? Oh yeah you said 12kHz... gotta read! Guess that flys fine for this application!
The cap coupling to the gates adds something to the existing capacitance of the
gates, I didn't see what the values are in the schematic? But I think you said 2.2ufd...

Wondering where the output devices are biased to??
What class are they running in...?

              _-_-bear

EDIT - forgot to ask, with the 4.7ufd cap to ground in the feedback leg, what frequency is that turnover point, and at what frequency is it open loop?

[WA1GFZ]

Tom put a .47uf in the feedback leg to limit feedback above 7 KHz. The phase shift through the modulator was causing it to take off. He is running about 10 dB NFB.
the output devices are biased to zero in Tom's rig he added pots to the three resistor node neat each gate to adjust it. I have simulated bias at -60 for an 833 and seems to work fine.

The 2.2 uf coupling caps were to reduce low frequency phase shift so there was less than 30 degrees delta 10 hz to 10 KHz. This way feedback source could be the output of the modulation transformer.

[WBear2GCR]

Yeah, I always think that it is important to not have the feedback corner frequency higher than the -3dB HF rolloff of the secondary of the transformer... good deal!

I am not asking about the voltage on the gate or WRT ground at the output end, but the quiescent current through the output Mosfets...

                 _-_-bear

[WA1GFZ]

I simulated it at 1K load. The FETs are rated 11 amps so as long as they stay cool there is no reason why it would not make a lot of current as long as the 100 ohm drain resistors are sized to limit voltage drop.
The only current through the FETs is the  current through the 50 K resistors to -300 volts. At zero volts bias there is less then 2 watts dissipated in the 50K
300/50000= 6 ma at zero bias
The biggest dissipation issue was the power resistors not the 300 watt FETs.
At 1 K load to -300V and Zero bias the FET would dissipate under 100 watts. I don't think there is ever a time when you will need 300 ma of grid current at zero volts. Maybe if Tom triode connected the tubes  you might hit 1/2 that value.
I'm not sure how much voltage swing is required to triode connect a 4-1000A and get it to saturate. The driver will do almost 500 volts grid to grid. He thought the design should be able to handle 833 to make it universal. As we found once you step into 600 volt caps the price starts taking off. 600 volts across a 900 volt FET is an acceptable derating. Not like it is a class e final connected to an antenna where you want more safety factor.
This whole thing could be scaled to operate at higher voltages if it were converted to high voltage IGBTs 
Tom could not run much NFB until the coupling caps were increased to control phase shift through the driver. At less than 30 degrees all he needed to worry about was the phase shift through the mod iron.

[WA1GFZ]

REV A
I pulled up this old post so all the information is in one place.

The fuses on each output seemed a good idea at the time but if a modulator with say 833s needing 100 volts of negative bias there could be a problem if one blew open. This would float the 833 grid and it may act like a saturated switch hauling lots of current. I think fuses should be placed in series with the drains of each output source follower in series with the 100 ohm drain resistors. This way if a fuse blows the positive drive would be removed and 50K source resistors would pull the grids to the negative supply voltage cutting off the modulators.
Tom, this is not an issue with your rig since you are doing cathode bias to protect the tubes.

[W2XR]

I recall someone was seriously considering making the bare p.c. board available for this driver circuit. He had solicited interest here on the board, and there seemed to be a number of guys who wanted to build this thing, myself included. I offered to purchase two of these boards in the event they became available.

Has there been any further movement in terms of offering the bare pcb?

73,

Bruce

[WA1GFZ]

Bruce, I looked into having boards made.
I figured it would cost me about $40 to $50 each to build after the time spent for doing art work and docs. This is a simple circuit. A quad 807 driver would be more work to wire point to point.  The problem is the power resistors and big coupling caps eat up PC board area, driving cost up.

[W2XR]

Bruce, I looked into having boards made.
I figured it would cost me about $40 to $50 each to build after the time spent for doing art work and docs. This is a simple circuit. A quad 807 driver would be more work to wire point to point.  The problem is the power resistors and big coupling caps eat up PC board area, driving cost up.

Hi Frank,

Thanks for the reply. I hope all is well at your end, and best wishes to you and your family for a happy and healthy New Year!

Yes, I recall you mentioning earlier that you would probably not be interested in fabricating these boards.

There was a fellow site member (on the West Coast or in the Midwest IIRC) that was soliciting interest in having these boards made. A number of prospective builders of this driver circuit, myself included, responded that they too would be interested. Perhaps it was W7SOE??

I hope whomever was considering making these bare boards, that they follow through with this. A number of us really want to give this circuit a go in our rigs, and it would be very easily constructed if the PCB was made available. And since you are now at the latest (and hopefully final) revision level of this circuit, now is the time to release the board for fabrication.

In the typical plate modulated transmitter, the driver board size should not be a major issue. There is usually plenty of real estate available in such a rig for this circuit, even if the board is made larger to accomodate the larger-sized coupling caps, power resistors, etc.

Although I have that quad 807 audio driver unit from a Gates BC-1G, I have no plans to use it in my homebrew transmitter or my GPT-750. I agree; once you consider the requirement for multiple operating voltages and the attendent power supply design to make the 807 circuit sing, it is just not worth the time and trouble. Performance-wise and from a simplicity standpoint, the FET design blows it away. And most significantly, I want to be able to wrap the NFB loop around the secondary of my mod transformer. I can't do that with my present push-pull 845 xfmr-coupled driver, or the quad 807 configuration either.

73,

Bruce

[K1JJ]

Yes, it wud be nice to see some working boards get out in the field. Maybe that's what ya have to do, Frank. Make up six, charge what you must and get them out there. After they are on the air and the buzz starts, it will probably gain momentum as new guys duplicate the same board on their own.

But as we have seen, very few have gotten built based on my point to point layout. I don't blame them...    I just HAD to get one going regardless of looks.

Bruce, as I told Frank, he deserves his own "GFZ legacy design" claim to fame out there.   

T

[W2XR]

Yes, it wud be nice to see some working boards get out in the field. Maybe that's what ya have to do, Frank. Make up six, charge what you must and get them out there. After they are on the air and the buzz starts, it will probably gain momentum as new guys duplicate the same board on their own.

But as we have seen, very few have gotten built based on my point to point layout. I don't blame them...    I just HAD to get one going regardless of looks.

Bruce, as I told Frank, he deserves his own "GFZ legacy design" claim to fame out there.  

T

Hi T!

Yes, I looked back at some previous correspondence on this topic, and indeed it was W7SOE who was considering building a run of these bare boards for himself and others. He had solicited interest from this forum back in late October of 2010, in terms of who desired to have this FET driver board fabricated. A number of OPs replied in the affirmative.

Rich, W7SOE: if you are watching this thread, there are guys (myself included) that are hopefully still ready to place their orders. Just tell us how much and where to send the bucks to.

73,

Bruce

[WA1GFZ]

I getting some requests to do a board by a few guys who know what it takes to do this type of a job. I figure it will take a few days of work to do artwork.
and than a couple days to look at it for errors. Then look at options. The issue here is the large components take up a bit of board area driving costs up. I think I would need to sell boards for around $100. I have zero interest in selling kits. Tom JJ and I Mouser the best source for 630 volt coupling caps.
There is a big price jump going to values above the ones he used.
So if there is any real interest let me know.
I would provide support. The circuit is easy to troubleshoot and I would also provide value changes (2 if I remember) for different operating voltages.
Simulation shows it would drive a pair of 833s biased at -80 or 90 volts runing plus and minus 300 volts. 600 volts on a PC board requires careful design.
This circuit can be wired point to point and as tom shows layout isn't critical.
Send me a PM if you are really interested.

[K1JJ]

Good show, Frank.

The GFZ MOSFET audio driver will make a BIG difference and BIG bang for the buck in any BIG plate modulated AM transmitter. That's BIG news! Get rid of the tube drivers, backwards audio transformers, interstage transformers, etc.   Not only is it very heat-efficient and has little phase shift, but is small with just one supply required. Excellent performance with negative feedback built in - and easy to build, esp with a PC board.

Considering how much time and effort is needed to put a big plate modulated rig on the air, $100 for a custom board like this is a bargain in comparison.  No way to duplicate it for less once time and actual cash outlay is figured in for a small run of boards.

One volt of audio in = audio drive and fixed bias to the modulator grids, with a negative feedback loop back to the input..

If this board gets put to widespread use, it could revolutionize audio driver technology in the AM community. We might even be able to take over the whirl!...

Disclaimer: Mr Vu has no financial involvement in Mr Carcia's proposed PC board project - your mileage may vary - the individuals portrayed here are actors - offer expires Dec 31, 2011

T

[WA1GFZ]

I would have to be on brown acid if I thought this was a money maker.
I would only do it if I can at least break even. This will easily end up a week of solid work for me.