In respect of stages oscillating, lead dress is of utmost importance. Modern caps do not have outside foil albeit that there are easy ways of finding the outside layer. Parallel wires induct and so do some caps. I have found it necessary to shield wires in the det 1st audio area.

Wires away from the metal chassis can act as antenna's & radiators. An oscillating RF stage will often generate considerable AGC voltage.

As usual; Beware of calibrating a signal generator with a counter: Tone must be off.

Fred: going on about alignment with the micromite was more to do with me getting used to using that and making sure that I can set the IF in the most precise way. I couldn't get it to work your way (low AGC) but this may be due to to the fault in the radio and/or my incorrect set-up. But feeding MW into the front end will do until the problem is fixed.

I've been going over 3rd IF to detector circuit connections, components and shielding with a fine tooth comb with the to make sure everything is as it should be, which it seems to be. So I have then turned to the IF3 transformer, which peaks OK on the secondary but not quite on the primary. I think that the primary is oscillating, just not peaking (maybe deteriorated with age). So took the can off and removed the transformer and trimmer, and then disassembled the trimmer and gave it a good clean - micas with isopropyl alcohol followed by oven dry at 100C. Have just reassembled the trimmer and the minimum capacitance is slightly reduced which might be enough to peak the primary - time will tell. Will now reassemle the trimmer/transformer combination and reinstall keeping leads away from the can.

I have thought of resonating the transformer out of the radio with the micromite. Is this worth doing?

I had a go at bypassing stages in Post #21. It didn't stop the oscillation, but I didn't check peaking of IF2 with IF3 bypassed. Will look at that when IF3 back in.

Marcc: I have also tried pulling the RF valve and feeding the aerial into the mixer. This did not stop the oscillation but did shift it from ocurring at about 650kHz to about 800kHz.

My frequency counter is only good for up to about 4MHz when checking the micromite but up to there they agree well; this counter loses the plot when the micromite is modulated with 1kHz - reads about an order of magnitude low. Checking the micromite with a PLL digital radio (a Sangean ATS800, MW and SW) indicates the micromite is spot on.

I have continuously warned about calibrating with counters. There are two here and both will go berserk, should you feed them modulated RF.

I did look at the Micromite and would not expect it like most signal generators, to suffer that issue, as its likely to be reading the oscillator.

The issue is wholly with the counter.

With a superheterodyne, be it Pentagrid with separate exciter, or an Autodyne (Screen grid) with a Pentode, the IF signal of around 50uV is applied to the signal grid. That is common usually to the antenna section of the tuning gang that goes to the converter.

IF signal will be severely attenuated if fed to the antenna. If the IF cans have been fiddled with, you start by injecting the IF signal (series cap) into the grid of the tube driving it. If it does not calibrate repair.

STC, pulling the IF and bench testing it is a good exercise.
I have swept IF's and other tuning coils with the micromite source signal and used the CRO as the viewing tool.
You feed the coil under test the signal via a 100k resistor, the CRO hooks across the coil being tested.

The CRO then sees the response as an RF signal as no detector diode is involved to give a DC output.
The actual resonating frequency will be different to that when hooked into a real circuit as the total capacitances are different.
The coil under test usually peaks at a higher frequency without the circuit loading.

Try this test with a couple of spare IF coils to get the hang of it.
You can do a back to back test with two coils to pick if one is bad.

Cheers, Fred.

I agree this is entirely a counter issue. It is a counter not a frequency analyser. The counter is a Dick Smith/EA kit and I managed to find the original articles for it (August/September 1978). The required signal input for it increases as the frequency goes up. I think that the micromite just doesn't have enough output to satisfy the counters needs at higher than about 4MHz. It should work up to 40MHz without a prescaler and 200MHz with a prescaler which hasn't been installed (95H90 ECL). But the fact that it agrees with the micromite up to 4MHz was enough for present purposes.

I'll get the 3rd IF reinstalled and see if it will calibrate with the slightly reduced trimmer cap minimum. A fond hope I think.

Fred in the end I didn't bench test the IF because I didn't think that I would be able to get realistic loading conditions, and it is now back in the radio. Will definitely try your set-up if it has to come out again.

The story is having decided that the very small change in trimmer minimum value was unlikely to help much, I decided to take the plunge and put the transformer in the oven (after cleaning the dirt off with isopropyl alcohol and a cotton buds) and slowly bring it up to just under the encapsulating wax melting point to drive moisture out of the wax and coils. In this way hopefully reverse 85 years of deterioration. The wax starts to glisten at this point and if it looks like wax drops are starting to form, open the oven door immediately and drop the oven set temperature a bit. The critical temperature was about 90-95C for the STC green wax. Constant vigilance is needed for a half an hour. My experience with getting the wax off old capacitors and an AWA 429MA choke is that the wax goes from glistening to running off in a very narrow temperature range.

Not having a ready means of measuring inductance the only way to test this was reinstallation. And it peaked nicely.

This leads you to wonder why this transformer need this treatment and the others don't, and how long the fix will last. But taking it as a win anyway.

Unfortunately the oscillation wasn't fixed. However the increased sensitivity brought in stations (faintly) below the oscillation range about 650kHz whereas previously there was nothing. So now can align by ear at low MW frequency where it should be done.

An interesting observation was that with the trimmer fully closed, ie low 3rd IF gain, there was no oscillation, and it came back as the transformer approached peak.

Will send in photos of transformer before and after heat treatment. 1st and second show state of the transformer before cleaning and remelting, 3rd cleaned with isopropyl alcohol and in the oven, and 4th remelted, reassembled and reinstalled in the radio.

"An interesting observation was that with the trimmer fully closed, ie low 3rd IF gain, there was no oscillation, and it came back as the transformer approached peak."

STC that makes perfect sense. Something is "spilling over" at a critical gain point.
It would be interesting to use the peaking as a tool.
Set the peaking so that the instability just sets in.
Then do the Marcc thing, try moving wires around, shielding valves from each other, try to find something sensitive that reduces the instability.

By the way, what is the AGC level now?
Presumably you should have a low value with no signal and rising as a signal level increases?
You must have a working AGC system to get stable running.
Fred.

To answer your 2nd question 1st, AGC at either end of the RF choke on the detector output is about -8V on the strong local TAB racing station.

The RF choke feeds to the RF valves through 1.5Mohm; about -1.5v gets to the RF valve grid measured with my 10MOhm multimeter.

The voltage divider output of 1.16V feeds to the IF valves through 1MOhm; 0.36V gets to the 1st IF valve and 0.45V to the 2nd, again measured with my 10Mohm multimeter which loads the grids a bit going on drop in volume.

So AGC not as it should be. I have checked all three AGC bypass caps; OK for resistance and capacitance. Yet to try an insulation meter.

Have tried reducing gain with the 3rd IF trimmers to threshold of oscilation. A number of sensitive wires behind the 6A8G, the worst of which is the HT feed for the oscilator anode grid. This wire in Mike Melzer's radio, of which he has photos on facebook, is routed via a more indirect route. I will try this and see what happens.

Another question, that HT feed to the 6A8.
Is it bypassed for RF?
That would be a 0.1 μF normally on the downstream side of the dropping resistor.
Fred.

It is bypassed with it own 16μF electrolytic, doing a dual job of HT filtering and bypass, located as you describe.

Electrolytics are not used to bypass RF.
So, should have a 0.1μF or something similar across the electro as well.
JJ

Jonny is right the electro may look open circuit to RF.
Try a 0.1 poly cap and note what happens to the RF level with the CRO as you do it.
In a set like that I would trawl through various HT feeds noting any signals that should not be there.
Long HT wires carrying signals that should not be there are great sources of instability!
Fred.

Hi STC, I looked at the circuit diagram and did a bit of thought rambling ignoring any thing that has been covered already:


I have been looking at the circuit for the STC 830 and can see why a set could be unstable. Having scratch built a few superhets and in particular one with two IF stages I can see the 830 could be a “box full of kittens” to get stable. Essentially it has four pentode valves in a straight line with a potential colossal overall gain. I know the problems I made for myself with uncontrolled feedback loops that were a result of either coupling by radiated signals, or, poor bypassing at each stage socket.
Also equally important was getting the AGC loop working properly. The fact that each stage is working at a different frequency means nothing. The coupling devices are not perfect and a front end like that can sit there barking mad unstable at some frequency that goes through the coil traps by capacitive coupling only. That can be physically in the coil sets and outside between grid and plate wirings.
I note this set uses the build practice of placing a lot of components on tag boards, quite common for production line building but not very good as some crucial bypass caps are placed well away from where they are really needed. That makes bypassing of signals on valve elements like the screen grid and cathode in efficient where remote mounting is done.
You would be amazed at the radiated signals under a chassis. Using a coil pick up probe with the CRO or even just moving a probe near things picks up all sorts of signals. Plate wires are just radiating transmitter antennas and a grid wire a couple of stages up is just a receiving antenna! Now you have an Oscillator!
The set also used unshielded valves which then relay entirely on the external tin shields for effect. If there is any short coming with the grounding of the tin cans, the set will be unstable. The circuit also has a feature that I could not get away with higher gain valves; the screen grids are all looped together and fed from a common source. That is just asking for instability. As you push the stage gain up decoupling resistors and ‘on the socket bypass caps’ are mandatory. The original design may have been marginally stable even when new but I fear the accountants have got in here and saved a few shillings. I don’t think the engineers had some esoteric “neutralising” scheme here, it’s just penny pinching.
When I was floundering trying to make a front end stable at high gain I could find all sorts of signals on wires that should not have had any signal at all. My poor bypassing and isolating was the root cause of those. Corroded tin shield cans failing to make good chassis contact was another shortcoming.

Another thought, does your “magic eye” run from fully open on no signal to fully close on a strong signal?
that gives you a good clue as to how the AGC voltage is varying.

Cheers, Fred.

Thanks for the comments gents

Johnny: I wondered about that and will certainly make that easy addition.

A point about the oscillator anode HT feed is that on the circuit it comes form the first HT filter cap; on my radio it comes from second cap. Radiotron Designers Handbook says that that the first cap feed is so that the oscillator anode feed is as stable as possible and more immune from output valve draw. But it also says somewhere that a lower voltage gives better oscillator performance. So I have left it as it is, even though Mike Melzer's radio on Facebook has it as per circuit, with a long wire around the periphery of the chassis from the first cap, an obvious modification because it has a join in it. Tried that without success wrt oscillation, but then no extra poly cap bypass.

Rather than having a long wire across the diagonal of the chassis as it is now, with second cap feed, it could also be very short off the 1st IF transformer HT. Will try that too.

Fred: I have said elsewhere that this radio has wires all over the place and one of the reasons is the two tagboards, evidently used to facilitate production. For example the aerial wire runs above the back of the output valve. They also make following the circuit very difficult at times. I thought yesterday that I had found a wiring error wrt the AGC bypass caps; but it turned out that STC had snuck in a link going diagonally across one of the boards that is not shown on the circuit which meant that it was right after all. Twenty minutes of head scratching and blue sparks.

Re the susceptibility to feed back I was astounded to see David Tipton at one stage of his restoration of an STC830 on Utube run the radio without any valve shields from RF to 2nd IF. "Might howl a bit" he said - but it didn't. Mind you this was before he installed new old stock valves and did the final alignment. My experience is much the same with my low gain radio, except that it does not like having no shield on the RF valve particularly, in which case putting a screw driver or hand near it produces whistles. To my way of thinking this points to a particular fault rather than susceptibility. But will meticulously follow shielding as designed nevertheless.

The magic eye does not make a full deflection except when oscillating or radio fed high signal generator output. The local TAB radio which is about 1k away should I think top it out. Lack of gain for some reason.

Will see how I go and try the CRO.

Good move.
Try to assess whether each stage is actually amplifying by coupling in a signal to a grid and seeing the result on the plate.
A good pentode biased around -10 volt should give a gain of about x15 at RF and IF frequencys.
The gain should vary as the AGC level varies.
Clamp an external DC supply to the right hand side of R7 and you can manually vary the AGC from zero to -30v.
that way you can prove each stage is working.
With signal flowing scope each element and look for signal where there should be none, say at the cathode, screen grid, suppressor grid.
If so why? Feedback or bypassing?
Its back to basic design 101.
Good luck.
Fred.