Golborne Vintage Radio

Ages ago someone did an article about using an external PSU on the uTracer3 for high current low voltages, but the idea only work if the valve had an isolated heater. This because the Cathode voltage and default H+ is essentially at the laptop PSU positive rail (typically 18V to 20V).

Previously I used an isolated PSU with + connected to H+ and Cathode and negative at about 0.7 V, 1.4 V and 2.1V via 1N4007 shunt regulators to the H-.

I sometimes used the bench PSU for 4V and 6.3V valves as the internal heater supply seemed low. I thought this was because the internal is PWM and I had too many ferrites on all the cables.

So recently I removed a lot of ferrites (leaving one small bead per wire), but the 6.3 V still seemed poor. My suspicion then was the IRF1404Z FET driven by push-pull transistor buffer from the CPU. It's about 8 nF gate capacitance and likely needs more than 5V drive on a 33% duty cycle at 19.1V to 6.3V. This became evident when trying to drive an opto-isolater from the drivers. Due to tight THP I couldn't unsolder it so snapped it off. Then the opto-isolater was much better though still not good enough.

The idea of the isolator is so an external PSU up to 120 V can be connected to the +H / Cathode (a fixed +19.1V) and the FET gate drive driven by the isolater. A buffer using a PNP and NPN transistor allows more capacitance to be driven, but has no gain, thus doesn't sharpen the isolated PWM signal. I need 8V to 12V there, so a 74HC74 is no use. I tried a 4049 (not the 4049U) with one gate as inverter and the other 5 gates in parallel and lovely sharp pulse on the scope. Before connecting an IRF740 FET I loaded it with a capacitor. Rise & fall time is sad. I'll recheck capacitor value vs gate capacitance of the IRF740 FET, and change the 8V supply to 12V or more. I found another small transformer. Not sure if my 4049A (not the later more buffered version) is 15V or 18V max. I'll maybe use the 8V one for a floating supply meter with its "OV" referenced to the FET drain to measure heater voltage.

More testing needed.

I'll fit a switch now to select internal or external PSU.

I've decided even 5 CMOS gates in parallel isn't enough to drive the IRF740. The issue is charging and discharging that gate capacitance of nearly 1700 pF.
The previous IRF1404Z is about 8 nF in the datasheet though statically measures at about 6700 pF on the component tester.

Typically I use 2N3904 and 2N3906 for most things, but they are Ic max =200mA. The BC3xx family parts I have loads of are only 100mA.

I seem to have loads of TO92 MPSA05 (NPN and 500 mA, 60V) and 2N5401 (PNP 600 mA, 150V), so while not at all a matched complementary pair the gain looks OK at typically over 100 (measured nearly 200) and fT both about 100 MHz. The PWM is 19 kHz.

I've soldered them up and mounted an ext/int huge toggle switch (10A or more?). The IRF740 drain is mounted to the H- panel socket and source goes to toggle switch, either main PCB 0V or external supply negative. The toggle switch connects the external supply positive to H+ panel socket. A 1uF 250V met poly goes from FET source (H- supply) to H+ (the laptop PSU +19.1 for the μTracer) on the rear off the panel.

Perhaps it will work. Maybe test tomorrow. I'll need to fit a connector for external supply.
Is this the highest voltage heater? 117Z3 and similar tubes with heater running direct off USA mains. https://www.radiomuseum.org/tubes/tube_117z3.html

Why don't you message Ronald directly about this. He's extremely helpful and knowledgeable.

I've discussed it previously.

Well, I looked at the circuit and it's grand without driving the IRF740 FET. Then it oscillates during FET on time. The 4049AC is oscillating. It is decoupled at pins with 100 nF. Adding another 100 nF across the MPSA05 & 2N5401 driver pair helped. The little 11V AC transformer drops to about 8V AC. It's too small!

Also the ILD74 dual opto-coupler is really slow (even in data sheet). I've found some single opto-couplers that measure 1/2 the LED capacitance (Not found data yet), so I will try one of them and bigger 12V to 16V transformer for PSU and a dual PNP transistor Schmitt trigger (so output goes to zero) to drive my output driver. About 10V to 12V pulse is recommended and I'm only getting 8V. Though that's better than the original barely 4.5V drive .

I recall the Siliconix D469 quad gate driver ic of yore: it was capable of driving and sinking over an amp into the gate.

That sort of thing is a good deal snappier than any number of CMOS gates.

Dunno if paralleling the 4 drivers would give 4 amps of gate source/sink.

It worked quite well as a driver for small dc motors too.

Other gate drivers seem to be available these days.

Yes. I've used gate drivers, but first I'm trying to do it with what I have.
The 2N5401 & MPSA05 gate driver isn't bad, but is only a buffer. A regular logic Schmitt trigger hasn't the o/p swing and any of my comparators are too slow. So meanwhile I'll put an IRF510 instead of the IRF1404Z (lower drive voltage and only 160 pF gate). The previous IRF1404Z is about 8 nF in the datasheet though statically measures at about 6700 pF on the component tester and really needs more drive than a buffer running off 5V (which doesn't give 5V drive).

I may need to order a gate driver. Likely someone does one with the opto-coupler built in!

See this also about heater control: http://tronola.com/html/tube_analyzer_yo...page1.html
I'm not doing that.

The IRF510 does work very well instead of the IRF1404Z, but that is only for the internal supply. However the pulse width and thus the voltage seems wrong. With 19V PSU and setting of Vh = 9.5V it should look like 50% and it's much less. I'll try a different tracer.exe as there are many available and the Cal file / settings look correct.

There is also an alternate 3rd party GUI I've never tried.

I do now have a massive toggle switch for a fixed level external PSU rather than separate 2mm patch cables direct from valve base patch to PSU.

My bad. 50% duty cycle UNFILTERED of 19V is not equivalent to 9.5V DC (or sinewave AC), because power is V squared and we want the same power as 9.5V DC.
See http://tronola.com/html/tube_analyzer_yo...dices.html

I would pull an old TV or something else that has still got a good SMPSU in it from the street.
Look up the controller chip and see if the input to the final drive comes out to a pin so that you can use it as a buffer.
If you are placing an order for something else look couple of chips up and add a few to your order.