Bias Oscillator black hole

[Clay1905]

Hi Folks,
As my first foray into audio, about mid-year I bought a Pyrox wire recording machine. When the machine arrived nothing worked and there was much evidence of meddling. After some time I have brought the amplifier back to life, and freed up the transport mechanism. Now there's only a couple of issues remaining. The big one being that the machine will not record.
What I have found is that when set to record, the B+ voltage doesn't so much sag as plummet. Taking the oscillator valve out of circuit stops the voltage crash.

When in standby or play, B+ is at about 270VDC. Switching to record has the voltage drop to about 160VDC.
At this voltage, the oscillator is barely on the brink of operating. It starts and runs for a moment and then stops. This cycle repeats at about 500Hz. When starting, the oscillator is running at about 48.8KHz.
I have replaced pretty much all of the passive components other than W/W resistors.
In order to find out if the rectifier has a depleted cathode, I knocked up a solid state substitute. Nope. With 2 1N2007's the same voltage drop. (I did include ballast to equal the 50V drop across the 5Y3.)

As this is all rather new to me, I can't help feeling that there's some simple thing I'm missing.
One last note. On then rotary switch there were the tails of some wires that had been snipped off. There is one I couldn't figure the use of. SW1a, 11&12 has one of these tails, but I cannot figure out where else signal on the preamp grid might also be required.

Below is the schematic.

Any help would be greatly appreciated here. Presently I'm feeling a bit

Thanks,
Clay.

[Radio Wrangler]

With many oscillators, the static bias on the valve, without the oscillator running can give a rather high quiescent current, and that when running, self rectification at the grid-cathode generates negative bias, throttling back the anode current and stabilising the oscillation amplitude. That the HT voltage comes down quite a lot with a non-running oscillator isn't necessarily an indication that something is wrong in the HT area. Look first for a reason for it to not oscillate.

The excessive anode current goes with the valve having a lot of surplus gain to start oscillation....

One of my dirty tricks with reluctant oscillators is to use a pair of oscilloscope probes (maybe that should be mis-use!) One is used backwards and mis-matched on a sig gen to inject just a tiny bit of signal to tickle the oscillator. The other one is used conventionally on a scope to look at activity from the oscillator. Sweep the sig gen tuning across the likely frequency range and look for a peak in the response on the scope. Is it where it should be?

David

[Clay1905]

I think the penny dropped just after the nick of time. I'll make checks, but I may well have made a rod for my own back.
I had replaced the anode dropping resistors, and had no 250K's. Plumped for 220K, which might be draining current to the extent that adding the oscillator pushes the voltage below a satisfactory level. I'll try adding 27 or 33K to each and see how thing go then.

[Clay1905]

Oscilloscope probes being misused Nah! I've seen the Red Green show.

"Any tool is the right tool".

[Clay1905]

Something in what Radio Wrangler said made me remember that at the beginning of this project I didn't have 250K resistors for current limiting on the anodes. I plumped for 220K as I had plenty of these.
Adding 33K to each of the anode resistors has stopped up the massive voltage drain, down from 100VDC to less than 6VDC. Quite a satisfactory result.
The oscillator is running very strongly now. I don't know that it's running correctly though. I read that the oscillator should, ideally, produce a constant amplitude and frequency sine wave. This oscillator makes a pulsating signal. I don't know enough of the principles used in early magnetic recorders to say whether this is OK or not.
The trace on the 'scope is a rapid rise in amplitude of a 50KHz sine wave, that as quickly dies away to a flat line. This rise and fall is repeated at about 500Hz. If I didn't know better, I might think those numbers were related in some way. Is this even significant?

To add insult to injury, the oscillator output disappears as if by magic at the end of a wire.
Between the DC blocking capacitor on the anode of the oscillator valve, and the oscillator output transformer, there is a short yellow wire. The end of the wire soldered to the capacitor lead has the signal on it. The other end has no trace whatsoever of that signal. The wire is continuous. But the signal is not. I would have thought that a short to ground would put the entire wire at ground potential. Both of the primary coils are continuous, with <2 ohms DC impedance each.
I'm a litte beyond the limits of my knowledge with what's going on here. It's all quite mysterious. Suggestions on how ro proceed from here would be welcome.

Clay.

[kalee20]

I'm finding it hard to follow the circuit diagram through all the switching! Then I saw that the second 6J7 valve has a capacitor directly in series with its anode, so I lost confidence in it anyway...

However, what you are describing is a classic squegging oscillator, and yes you're right, it should be a steady sinewave.

What is often the case when an oscillator does this, is excessive resistance in the grid circuit (or sometimes excessive capacitance feeding it). As earlier, I don't have confidence in the circuit and its values - but this is where I'd start looking. The oscillator, although awkwardly drawn, looks like a tuned anode tank circuit with phase inversion tue to the tapped winding. The grid circuit has a mystery inductor there, but the 0.01uF capacitor feeding it looks rather over-large from what I'd expect. If there's anything to check it against, do so - otherwise I'd try changing it for a 0.001uF (or even 470pF) component.

[Al (astral highway)]

If I've read right, now that the anode resistors have been brought back up their intended value, the worst of the B+ collapse has gone away, but what’s left is the oscillator being dragged out of continuous operation by something.

That something could be excessive load or disturbed grid bias, and the transformer / grid-bias network look to me to be the main suspects.

So I'd measure the two 500 k resistors and make sure they really come out to about 250 k in parallel. As well, check the 5 nF capacitor as it critically affects grid bias.

The other two capacitors that tie from the transformer side of the circuit to ground / back into the oscillator loop are worth checking for leakage as they affect feedback and potential loading.

And what about in the bias/erase transformer? Poss shorted turn? (This is only speculative as I don't know whether this fault condition ever arises!)

May be worth putting some magnification on to check for carbon tracking or contamination around the B+ and transformer contacts too?

[Clay1905]

Below is a fresh schematic with component names added, and amendments made.
Sorry about the switches. These were an absolute nightmare to map out. Please tell me about any anomalies you spot. They may be either my error, like the misplaced anode capacitor, but may also reflect the true state of the circuit and be something I have not picked up.
Component values are all correct. The only deviation was the too small resistors on some anodes, which are corrected.
The blocking capacitors, C4&5, are the right size. I've kept all the originals for reference. This pair were not of the same make, which I thought a little odd seeing they both looked undisturbed and original. One is a Ducon wax and paper job, the other a Technico mystery black cylinder.

Next is the matter of C3, 5nF. Another apology required. You know those mica capacitors that are always "good as gold"? Yeah well....shouldn't have trusted it.
A little trap here. I recall reading some while ago about Australian made mica's being prone to a slightly higher failure rate than overseas types. This has something to do with there being too much iron within the mica. The failure rate was calculated I believe, to be exactly just enough to catch you out.

OK, with the 5nF capacitor replaced, not a sausage. Just a weeny little bit of AC hum. However, before I go on with that, I noticed that R14 is getting quite hot and I think that should be attended to first. I can't really say if this resistor was getting hot previously, but I doubt it. Today it would be more evident owing to the cool day for this time of year. I could feel the warm air rising. On the other hand, I haven't changed anything around the resistor to change the current drawn through it.
Apparently there's another current leak somewhere around V4&5, or so I think. Once I discover what's the go here, I'll return to the matter of no oscillation.

Clay.

[Al (astral highway)]

hi Clay, I mentioned C5 in my post but I don't understand what you're saying here. You replaced it and it was duff or you replaced it and it didn't make any difference?

Originally Posted by Clay1905

Next is the matter of C3, 5nF. Another apology required. You know those mica capacitors that are always "good as gold"? Yeah well....shouldn't have trusted it.

Originally Posted by Clay1905

OK, with the 5nF capacitor replaced, not a sausage. Just a weeny little bit of AC hum.

Originally Posted by Clay1905

However, before I go on with that, I noticed that R14 is getting quite hot and I think that should be attended to first. I can't really say if this resistor was getting hot previously, but I doubt it. T

There's no component numbers on the schematic, only values!

[kalee20]

R14 is feeding the 6V6 which drives the loudspeaker. At 3k it seems a bit high, as the 6V6 will suck a few tens of milliamps - it'll get hot but this will be the case whether or not the oscillator is running.

I'd check L2 (for continuity) and R1, R2 together for resistance (should be 250k). You've already checked C3. I do confess to some bafflement as to what L2 and C3 are for - unless some way to improve waveform purity? They form a series-tuned circuit, so actual values are significant.

[Clay1905]

Good (insert time of day here) folks,
Al, I'm, confused too. Did the new schematic not come with my last post? It see that Kalee20 got it.

OK. I replaced C3 with a known good one. Any oscillation that was present before is not there now. Just a tiny bit of AC hum.
R14. I was concerned about too much current passing through something downstream (I know not what) and cooking it. However, the current passing through R14 is not enough to worry about. Only 37mA. As this resistor is providing power to 2 valves and the speaker when in play, this is not too great a current. I worry especially about burning out a transformer. I just wanted a pause to be sure.

R's 1&2 are at 252K.
(Typing while tinkering now)
L2 is continuous. Can't get a meaningful reading for value on the bridge though.
Now this could be of import: In checking L2, I found that the lead on the tab strip was a cold solder joint. Very old too. The tinning on the lead was typically grey and ancient looking. I must have disturbed the joint when checking C3. That's cleaned up.

Right. The oscillator is running again but still doing the stop-start thing.
Now I've led you on a merry dance from square one back to square one, where to now?

[Clay1905]

I think I should say how I've been interpreting the oscillator circuit. Remembering that audio is a new area for me, and oscillators are not a familiar concept.

I have been taking C's 4&5 to be DC blocking capacitors only, keeping B+ off the grid. C3 & L2 form the tank. R's 1&2 are the grid leak, for bleeding accumulated charge off.
That's about it. As I'm unable to establish a value for L2 yet, I can't work out what it's resonant frequency should be.
I should add that the stop start rate is now about 250Hz (4mS/ peak to peak.) I'll pop this in as my arithmetic can be woeful before coffee.

[kalee20]

Originally Posted by Clay1905

I have been taking C's 4&5 to be DC blocking capacitors only, keeping B+ off the grid. C3 & L2 form the tank. R's 1&2 are the grid leak, for bleeding accumulated charge off.

I'd partly agree - R1 and R2 are grid leak, for bleeding accumulated charge off, yes.

C4 alone is a DC blocking capacitor, however. C5 (you've drawn it rather awkwardly!) is across the anode transformer TR2, and on the basis that this is adjustable (you've drawn an arrow right through it) I'd have said that this pair is the frequency-determining network.

I'd still go for C4 being too big - try reducing it to 1,000pF. If it makes little difference to the 250Hz stop/start squegging frequency, then the cause is elsewhere and you can go back to your 0.01uF. If it does make a difference, then examine where the 0.01uF value came from - original documentation, ir just what was fitted, and might have been changed erroneously in the past, and what's so far not been known to work?

[Clay1905]

Here's a redrawing of the oscillator in isolation. I invite constructive criticism. Might as well get more clues on how to best represent common networks. I see the drawings are coming out sideways. Tried rotating them, but something is persistent.

[Cobaltblue]

A quick snip there is plenty of definition to not require a pdf.

Cheers

Mike T

[kalee20]

First, as a circuit diagram, it's loads better! Really clear.

It's so clear that it's possible within a second to see it's wrong! It can’t oscillate.

But if you take the top connection of C4, remove it from HT+ and connect it to the top connection of C5/TR2, you'll have a circuit that does.

If you have a signal generator, it's worth connecting it between chassis and anode of the valve (the whole thing unpowered), and then 'scope on each extreme of TR2 so we can compare the feedback voltage with the anode voltage. You'll want the generator set to about 30kHz and a few volts output.

[Clay1905]

Hip hip hooray gentlefolk! Throw your caps in the air. By making C4 ten times smaller, and Kalee's other suggestion of connecting C4 in the right place, and Presto! An oscillator that is running hot and strong. Close enough to 50KHz for jazz, (48.8KHz), and 2V peak to peak. Very clean trace, and not a bad shape. A bit pointy at the crests, but otherwise very nice.
I have now found that the erase/bias coils are open. Rhubarb! Early on I found that the head leads were pretty rotten, the wire strands were all black and crumbly. These were replaced. Of course during the operation the voice coil wire broke off its lead, and the end went into hiding. Four attempts later, a new coil was wound and fitted. It works, I can play wires.
I'm a bit loathe to open the head case again. It's just a tin-plate box, and it's getting to the much more of this and the tabs will break off stage. Should I be sending it to a psychiatrist? They say amateurs shouldn't interfere with fragile head-cases.

Thank you all for your input thus far. It's been very helpful and enlightening too. In case anyone is curious, once I've done the brain surgery, I'll post on how things went.

Thank you again,
Clay.

[kalee20]

Wow! Super news.

Yes, it would be interesting! As would a 'scope photo of the waveform.

Good work!

[Clay1905]

Originally Posted by Cobaltblue

A quick snip there is plenty of definition to not require a pdf.

Cheers

Mike T

Umm,? I don't know what you are alluding to with "a quick snip". I can export drawings to pdf, or jpg, but the jpegs are pretty awful. I get that you are trying to steer me towards a smaller file to upload, but what is a quick snip?

[Al (astral highway)]

Originally Posted by Clay1905

Hip hip hooray gentlefolk! Throw your caps in the air.

Nicely done! A vote for what Kalee says, be good to see a scope trace!