Problem with a VFD display driver circuit - no filament voltage

[jjbunn]

This is the VFD display circuit from a Kenwood R5000 receiver.

The relevant part of the schematic (attached) to the problem is top right: Q4/5 oscillate at about 17kHz, the voltage on the secondary 4/5 of transformer T1 is rectified by D1/2/3/4 to produce about -20 Volts at the base of D6, a 3.9 Volt Zener. This part of the circuit works fine.

The secondary 9/10 of T1 is supposed to produce around 3 Volts AC to power the VFD filament across pins 1/2 and 39/40 of the VFD.

With D6 and C7 disconnected, this works ... about 3 Volts appears across the 9/10 secondary winding of T1.

However, with D6 and C7 connected as they should be, the voltage across the 9/10 secondary drops to zero, and then of course the VFD doesn't work. Since the filament has to be biassed a few Volts above the grids, this should be achieved by D6 ... that's my understanding at any rate. The "DC BIAS" measured value, instead of being -16.2 Volts, starts off at about -20 Volts, and then gradually keeps rising towards 0 Volts.

I've checked that D6 and C7 are good. Could anyone offer advice on what might be going wrong here, or other things I can check to help resolve the problem?

Thanks!
Julian

[CountChocula]

Did you check if the filament of the VFD is not shorted and bringing down the -20V rail? What happens if you disconnect the cathode of D6 from the circuit and you load it down with, say, a 1k resistor to ground? Do you get the correct DC bias on D6?


—CC

[jjbunn]

Thanks, CountChocula!

I will check the bias on D6 after disconnecting it and loading it with 1k.

If I apply 3V DC across the filament using an external supply (-ve to D6), the display lights up correctly! I wonder what that implies ?!

[jjbunn]

I disconnected the cathode of D6, and then brought it to ground via 1k. The cathode voltage in that setup is -16.2V, exactly as shown on the schematic, so it's doing its job :-)

[jjbunn]

Update: I think the secondary 9/10 on T1 is either burned out or intermittent. Right now it measures open circuit, so I think that must be the problem.

[CountChocula]

Interesting! I thought you had measured the 3.7VAC you were expecting there. Sorry to hear that this is the problem… maybe check whether it could be some kind of connection problem, like a cold solder joint or something like that. Good luck!

[jjbunn]

Yes, I thought I had measured ~3V AC across it, but I can't repeat that now. Measures open circuit. I'm going to pull it off the board and take a closer look. Hopefully it's a visible detached wire, otherwise I'm in trouble.

Thanks!

[floobydust]

With a magnifier, look very carefully at the transformer pins and lead terminations, to see if a bad solder joint is to blame. It can occur at the wire wrap on the bobbin's pin (top), sloppy removal of the varnish, flux corrosion etc.
If the winding is open-circuit, it's only a few turns of thicker wire and I would pull the transformer- to do surgery.

edit: some digital multimeters are only good to a few kHz and the ACV may not be accurate as this DC-DC runs higher frequency than that. Ohms don't lie.

[elecdonia]

It is rather unusual for this type of transformer to fail. The internal wires are relatively thick and there are only 20 to 50 turns of wire in any of the windings (maybe less). Of course it is possible to have a poor solder connection or even a wire that has broken at one of the pins which go into the PC board.

Check ESR of all electrolytics in this DC-DC converter circuit. Or just replace them all.

Use an oscilloscope to confirm the 2-transistor oscillator which drives the transformer is operating. 

[jjbunn]

No visible disconnect ... and on carefully attempting to pull the transformer, I'm afraid it completely disintegrated :-( The wires are very thin indeed, and extremely delicate. I did measure open circuit on the winding in question, so I'm afraid it's too far gone now :-(

[jjbunn]

Yes, the oscillator ran at ~17kHz, and was producing the correct voltage on the secondary, as mentioned in my initial post. No problems with the few electrolytic capacitors I pulled - all measured good.

I'm a bit puzzled how the 3 VAC secondary winding could have burned out, but this radio was exposed to a very humid environment, so perhaps that might explain it.

[floobydust]

 
you can reverse-engineer it by counting # turns for the windings and getting a new bobbin+core. It's a bit of work but not that difficult. Posting pics of what's left could help too.

[elecdonia]

Can you post a photo of the remains of this transformer?
That 3V winding is probably only 5 or 10 turns of wire.
Back in the day (before my time and I’m old) people often wound their own transformers. At that time inductors, transformers, and other magnetic components were better understood than capacitors or resistors.

[jjbunn]

Here's a photo of the remains. All but two of the pins disintegrated, and the wires detached, but the body of the transformer is intact.

[floobydust]

Can you measure the core size, to see if it's a standard size like EE8.3 or EE10, and if a new bobbin can be procured first. The ferrite core is brittle and easily broken! I would not take the core apart until a new bobbin can be found. The original transformer would be a custom part from a Japanese manufacturer guessing TDK for Noritake or Futaba.

A mains wall wart 12V 2A (larger) transformer teardown to show what is involved. Some youtube videos are kind of funny, people boiling them etc. to unglue the two core halves. I use hot air or my oven.

[elecdonia]

Here's a photo of the remains. All but two of the pins disintegrated, and the wires detached, but the body of the transformer is intact.

Can you post a photo of the other side of the transformer? (The side facing the towel in your first photo).
Also please post photos of the PC board which this transformer was mounted on.

[jjbunn]

Here's a photo of the transformer on the board. It shows the other side (opposite red dot), with pins 10,9,8,7,6. I'll try to get a photo of the disintegrated transformer from the other side.

I think it's too far gone.