Right. The PCBScope I ordered yesterday arrived lunch time today.

It's a bit rubbish but here's a trace off a 5v 20kHz square wave. I used my scope lead and a few wires so it wouldn't ever be that clean.

Then the Gould's internal cal 1v/1kHz square wave (and the software measured exactly 1kHz)

I'm not quite sure why the horizontal time base shows the wave as nearly identical. Changing the s/div seems to have no effect on the wave but I know it's around the 20kHz as it was measured and calculated on the Gould when it worked. I haven't bothered to learn the software at the moment but it is very buggy and keeps crashing on both PCs I've tried to use it on. Edit: Apparently it does eventually show the wave widening once you get down into the uS region.



Here's the output on the emitter of the TR183. I had to use the x10 on the probe to attenuate the signal for the PCBScope. There is some movement but it can't be more than a few volts of oscillation.

The 'osicllation' was similar when I had the TR183 totally disconnected.

For the moment I have only replaced the electrolytic caps on the oscillator board. So I may just put the originals back on if it eventually comes good.

 

I had to use a bit of wire to bridge the gap.

When you do that, particularly in high V circuits, sleeve the wire.
Don't use bare links like on a logic board.

 

When you do that, particularly in high V circuits, sleeve the wire.
Don't use bare links like on a logic board.

I was undecided what to do about that so left it off. Thinking about it though all the bare wire on the hi-frequency stuff Advance used was sleeved with silicone sleeves. I can get plenty of sleeves from work so will do that next week.



Now that I've got it in a quiet room I've noticed that when I switch on the power the EHT box seems to be making a noise similar to crackling for a few moments, which then gradually dies away.

I can only leave the power on for 10-20 seconds or so before the transistor gets too hot.

 

Its not got a nice square edge that's for sure.

I don't know but I'm guessing it's some sort of exotic material electrolytic.

I have now changed it for a radial 10uF 50v cap I had salvaged from something. The cap itself is tiny compared with that monster I took out. I had to use a bit of wire to bridge the gap.

During the 80's, there was a organic-semiconductor electrolytic fad - haven't heard about them for years, so maybe the novelty wore off for some reason.

If you encounter any, they do have to be replaced with ultra low ESR types - usually I use the best aluminium electrolytics to hand and pad them with a multilayer ceramic chip capacitor, these are often found in computer stuff with values as high as 47uF. One manufacturers brochure was offering values as high as 180uF - a few manufacturers also offer resin dipped leaded parts.

 

before the transistor gets too hot.

Sorry which transistor?

Note, Tealc that lots of us are with you on this one, but only chip in now and again when we have something to say.

 

Thanks Studiot.

Sorry I should have said the transistor that failed short circuit and I replaced initially, the TR815. The top one here with the little skirt on. It gets hot very quickly. I only noticed because the little seat it sits on started to melt.

It stops heating once the EHT box mounted TR813 collector is disconnected.

I'm waiting on a couple of replacements before I swap all four transistors on the board. Hopefully they'll arrive tomorrow. I'm also planning on building a EHT probe using some 20 or so 10M resistors in series inside a caulk tube or similar. I'm also looking at utilising an Arduino to get me some sort of ESR meter.

Thanks Ian.
I think I have a few dozen MLCC 0.1uF axial leaded caps which I could run in parallel with the caps. Maybe that would help them along. I don't have anything higher than that, apart from some rather large 1uF and some even larger but lower value film caps.

 

Here is a diversion - a short video on 'blocking oscillators'. Your Astable is based on these.

does R818 also get hot?

http://www.youtube.com/watch?v=DMu6FQliA_s


Edit a further thought.

Did you disconnect any of the transformer leads from the board. If you did, are you sure you put them back the right way round.
If you get the winding direction wrong on this type of circuitry things can either refuse to work or get hot very quickly.

 

I haven't removed any transformer leads yet apart from one to do a diode check on the multiplier, which I then duly put back on. I would use a white marker pen on the sleeves if I was going to do it.

I haven't checked yet but R818 does look like it's been getting quite warm, you can even see a little splodge on the capacitor next to it. I wonder if it has come from the 390 ohm resistor.

I will check out that video. An astable 555 I can understand but I'm not familiar with discrete transistor oscillator circuits at all.

 

Still trying to get this PCBScope working properly.

I'm picking up this from the collector of TR815.

It's clearly not 20v/div. I'm not sure what the swing is and how accurate that 'waveform' is. I may have neglected to switch into x10 on the probe on that one.

This one is more to scale.

I think this PCBScope is defective as it captures only when it feels like, has frequent USB disconnections and software hangs. A DSO Nano would have been better.


So there is, I think, some ocillation at least. I would expect it to be cleaner though. The PCBscope did pick up my 20kHz signal ok so when it works for the briefest moments maybe it isn't that far off. The manual says the emitter of TR813 was 15v and this swing seems to be around 10v max p-p although mostly 5v p-p.

Any thoughts on what I should be seeing?

 

I managed to get this TR815 collectopr signal by using a wire rather than my probe. I guess the software isn't compensated for the probe. The waveform started fairly spikey but resolved to the waveform shown.

There's a lot of DC content in the signal and not a lot of AC. AC coupling would be really handy as it is tricky to find the AC content with the trigger marker with the huge DC.

I'd prefer to use a proper scope but this is all I have.


Edit:

Edit:

Same signal, just more captures.

At 5uS and a lot of averaging.

At 10uS and not so much averaging

Maybe the waveform is such a mess that averaging just tidies it up.

It looks like a 15kHz wave with 500kHz content modulated on it.

 

Primary resistances.

Pos 13-16 - 1 Ohm
Pos 13-17 - 1 Ohm
Pos 16-17 - 2 Ohm

Pos 14-12 - 1 Ohm
Pos 12-15 - 1 Ohm
Pos 14-15 - 2 Ohm

Test leads together 0.5 Ohm

Confirmed on both meters.

Swapped all diodes for new.

Swapped R819, R820 for new.

Swapped TR812 for new.

Replaced C813, C814, C816.

Tried a D882 for TR813, it melted but I didn't case mount it so it couldn't dissipate the heat.



Checked all resistors for values.

Swapped TR814 and TR815 for new again.

Swapped TR816 - schematic calls for 2N3905, was BC212 fitted, for a 2N3906

Swapped TR811 for a 2N3904, schematic calls for BC107

Here's the results off the secondaries.

Maybe not conclusive but there's no short there and the 100 Ohm and 4k suggest a lot of windings.

Maybe things would be different at high voltage though.

I could go ahead and replace the caps in the mulitplier for around £5 if I'm prepared to wait for a China delivery. I
wonder if it's worth it.

In other news I managed to score myself a working basic 20Mhz Gould OS300 for £27 plus carriage today off eBay. This will help me fix the OS3000, and if I do get the OS3000 scope fixed then I can always flog one or the other off for a bit of profit. £27, surely a steal looking at what even the most basic and old scopes go for on eBay these days.

 

OH NO!!!!!! Has your frustration with this O-scope transformed you into a parts changer now? Please tell me NO! Parts changers = bad ;(. You are going to spend a lot more money replacing components hoping that one you replaced fixes your problem (if it does). And you might end up throwing the balance of components off by continuing to replace multiple components with new, if that makes sense. That in itself might lend to a bigger headache and could even result in the scope ending up in the junk yard.

 

I still have the old active parts and those electrolytic caps safe. I noticed that after changing a few parts the frequency measured with my DMM went up to 50kHz whereas before it was around the 25k mark and things got melty more quickly than before.

Sadly I broke the TR813 transistor by bending the leg too much so unfortunately that's going to need replacing with something either identical or more appropriate than just some medium power transistor I had available (BD435).

 

So I made this thing of beauty today.

20x 10MOhm resistors and a 1MOhm resistor in series. Miliatry grade heatshrink plu 15kV Silicone sleeve plus another 6kV silicone braided sleeve on top, all encased in some tubing. Depending on the range on my DMM I'm getting around a 210-230x reduction in voltage.

So here's the meaurements that I'm getting right now with the few substituted components still in place.

-Vk = -128v
Vg Hi = -128v
Cath (on bright up board) = - 126v
Anode around 700v

So it's around 10% of the normal output.

I suppose looking at the plots I got from PCBscope the other day which were only partially oscillating this makes sense.

 

I would suggest the following procedure, given your resources.

1) Remove Tr814 and Tr815 from circuit.

2) Check that you have about 18 volts at the emitter of Tr813 and that it does not get hot. Supply a looad resistor to draw about 50 mA and check again. This should prove the operation of Tr813.

3) With Tr814 and Tr815 still out of circuit and the power off supply a 1 volt signal (low power) to the transformer, between points 13 and earth. This should generate something around the rest of the transformer that you can safely measure without damaging anything.

 

Thanks, I should have a bit of time this evening so will give this a shot.

I will have to make or modify something to produce an oscillating 1v signal though. Would it be bad to use the 1v/1kHz cal signal of the scope to do this? Otherwise I could add a voltage divider to my 20khz PWM box.

I'm hoping to get the other scope tomorrow.

 

Would it be bad to use the 1v/1kHz cal signal of the scope

No don't use the scope signal this must be done with the scope power disconnected. You are testing the transformer.

Yes a square wave would be better, I expect the inverter is designed for one really.

However a low voltage sine wave from another transformer could be used or the output of a signal generator, anything low power.

The actual voltage does not matter, so long as it is low hence around 1 volt.
0.1 would be better if you could measure it. It brings the step up within reach.

 

I get 17.1v on TR813 with the two 814/5 disconnected. I left it switched on for a bit and it didn't seem to warm up.

So I selected a 330 ohm resistor for the 50mA but where should I put it? Emitter to ground?

...

So I went with emitter straight to ground and the only thing that got warm was the resistor.

 

Well I don't know.

Wired up my 555 pwm box to the wire at point 13 and the ground of my power supply to a ground point and I get nothing, just noise on the meter. I tried with it connected and disconnected, the wire at 13 that is.

I'm using a 555 PWM circuit I put together that uses a 5v regulator but I've reduced the supply to 4.5v. The circuit is operating as the LED I installed, is dimming properly. The output is a pretty decent square wave normally. Of course I can't measure what is actually happening once the square wave hits the transformer but maybe I will be able to tomorrow. I wonder if the transformer fault is sucking all traces of a square wave away. The 555 box isn't particularly high power, especially as I'm coming off pin 7 and not using a transistor output but you'd expect a few mA. I've got a 220 Ohm resistor on the output to protect it anyway.

I measured 1v on the output of my roughly guessed 50% PWM. I'm unsure whether 1v is the output max or a smoothed average so went with my high voltage probe intially just in case 2v became 100v and goodness knows what on the multiplier.

I wonder if there is something inside the transformer primary like a diode that won't allow me to pass a low voltage into it.

I know that an oscillation can be passed through the transformer because of my findings above but either I'm doing something wrong, the transformer doesn't like a square wave or it needs more potential.

I also tried ramping the voltage up on the 555 up to it's max output of 3.5v or so, no effect

 

Well 330Ω is 51 milliamps, which should be plenty to drive the inverter / transformer primary, so the bsic LV rail provided by Tr813 appears to be working. 17 volts seems a tad low - I guessed 18 volts, but not significantly.

I'm sorry to say we seem to be working towards a transformer fault, but keep trying tomorrow as you said.