Tube scope supply board

[Sch3mat1c]

Thought I'd share my new power supply.

http://webpages.charter.net/dawill/I...TubeScope3.jpg

http://webpages.charter.net/dawill/I...TubeScope4.jpg

Haven't had much time to test it, but it seems to start up and make voltage. Importantly, it didn't explode on startup.

Specs:
Startup supply: blocking oscillator running 114kHz. Output: 6.3VAC CRT filament (I hope it doesn't couple too much noise into the electron beam), 15VDC for TL494, -2kV (zener regulated) 1mA HT supply.
Main output: half bridge MOSFETs, transformer coupled drive, capacitor coupled output. TL494 running at 100kHz (output, so internal is 200kHz). Outputs: 6.3VDC 10A, 125 and 250VDC 0.2A.
Total input: about 100W. Efficiency should be fairly high, zener stack notwithstanding.

If you look closely, I got one of the zeners backwards, so it's actually -1800V, and running a bit too much current. The blocking oscillator is also running a bit hot at 160VDC input, which puts a lot of power into the zener stack and dropping resistor (a shame I don't have any 470k 2W resistors!). I should probably put a zener in the base circuit, so it regulates itself somewhat.

Feel free to comment on whatever, layout, circuit, etc.

Tim

[Sch3mat1c]

Update:

The circuit works, except it's consistently blowing a diode.

http://myweb.msoe.edu/williamstm/Ima...pe_Supply2.png

After about a minute of on-time, one of the negative HV rectifiers (UF4007) dies. Output drops to 0V and the supply goes into current limiting.
The transformer has plenty of leakage, but this is damped by snubbers as shown:
http://myweb.msoe.edu/williamstm/Ima...pe_Snubber.png
and only causes 120V peak overshoot, for a peak reverse voltage of maybe 840V, which is still within limits.

The diodes do not warm up appreciably over a short time. I have not observed the waveform for long enough to actually see the failure on the 'scope.

Any ideas?

Tim

[danzup]

Originally Posted by Sch3mat1c

Update:

The circuit works, except it's consistently blowing a diode.

http://myweb.msoe.edu/williamstm/Ima...pe_Supply2.png

After about a minute of on-time, one of the negative HV rectifiers (UF4007) dies. Output drops to 0V and the supply goes into current limiting.
The transformer has plenty of leakage, but this is damped by snubbers as shown:
http://myweb.msoe.edu/williamstm/Ima...pe_Snubber.png
and only causes 120V peak overshoot, for a peak reverse voltage of maybe 840V, which is still within limits.

The diodes do not warm up appreciably over a short time. I have not observed the waveform for long enough to actually see the failure on the 'scope.

Any ideas?

Tim

Those UF4007 are not quick enough !
Use faster diode instead !

[Sch3mat1c]

How so? 50ns should be plenty at 120kHz.

Tim

[Sch3mat1c]

Update

Heavily modified the power supply. High voltage/auxiliary supply remains the same, all the half bridge stuff completely cut off. Replaced with a point-to-point / dead-bug wired flyback controller (UC3843, DCM). Besides using only two UF4007s to supply the high voltage, reducing rectifier losses in general, recovery losses are also limited by recovery occuring at low current (as the transformer returns to 0V), only once per cycle. To reduce noise, a dV/dt snubber was used on the switching transistor. This burns some power at all times, degrading efficiency, but noise is more important in this application. Overall, losses are acceptable for the 160W rated power output.

Picture time!




CRT is running at -2000V, which takes a little more deflection than normal; I'm still adjusting the deflection amps (7KY6 x 4) to improve performance. The deflection at clipping is currently not much larger than the waveform shown, so I have some work to go.

Dang, I should take some pics of the boards inside the box.

Tim