I have one of those EA counters & they all seemed to have issues with the highest frequency. Like the Fluke it does not like modulated signal.

The generator here takes its output off of a cathode. It puts out volts which can be varied and is also why I tend to use pF series caps. You can wind the output down enough to stop the counter counting.

This is the second attempt at posting this as MS have once again made an update which cause loading issues and games to stop working. Obviously with the retrenchments they followed Australian tradition & sacked the ones that knew what they were doing.

Mine lacks the prescaler to make the x100 range work. The rest of the circuitry is there and considering finding one.

I didn't associate difficulties in posting here with the recent update as everything else OK.

I put a bypass cap on the oscillator anode supply right up next to the filter electrolytic, but this had no effect on the oscillation.

Have been getting used to my CRO again. Managed to sort out a distortion problem in its display which I tried to fix a few years ago. Found the problem was the four spring steel strips inside the handle. Demagnetising didn't work but the simple expedient of turning one of the four strips end to did, which seems to have cancelled out its magnetic field. Now an almost perfectly square display when showing the square wave calibration signal.

With probe set on X1 it loads down grid circuits and the AGC; better on X10.

PS: Blue sparks mentioned in Post #59 were from me, not the radio.

I would be using the CRO to see the wave form, its shape can be a hint as to what is generating it.

I have had a Fax get into a 175kHz IF. Switch mode power supplies Electronic Balast's: CFL's & LED mains lighting have been known to cause massive RFI . I often short out the antenna at the set (0.01cap) to see if the crap is not coming in that way?

I did have an issue with a hacked STC where the parasitic oscillation was at the mixer and was caused by factory lead dress. That was spotted with an oscilloscope. I actually ended up replacing the screen wire with shielded cable to shut it up.(No AGC).

It might be an idea to run calibration frequency into it as per alignment & see what the signal looks like at the output. Any attachment to a tuned circuit is liable to pull if of frequency.

Marc

I need some best practice advice.
Attaching the CRO probe ground lead to the radio chassis lowers the gain of RF circuits even further than the probe itself. I guess that this does not matter where just wave shape is being observed, but what about amplitude? Would using a mains isolation transformer help in this respect?

I have always thought that the problem was centred around the mixer, especially because taking the RF stage out shifts the MW frequency at which the oscillation occurs from about 650 to 800 kHz; also following the advice to put the radio on the brink of oscillation and prod for sensitive wires finds among the most sensitive those in that area.

I have been getting used to what waveforms are to be found at various points around the radio. Oscillator seems to be working OK right through the MW range which seems to contradict my idea above about the mixer. We will see.

When oscillating, a lot of signal on the AGC circuit. This getting fed back to the RF and IF and no wonder it oscillates. AGC volts gets to -150V at its worst. Makes me wonder whether external voltage applied to AGC will defeat it.

I have decided to purchase a bench power supply to test this; should be here in a few days.

I will use the CRO without the aerial connected just to see exactly how the IF varies without modulation, and with a calibration signal from a frequency generator. Input impedance of the CRO is 1 meg, ~30pf which is worse than my digital multimeter. I expect it is 10 meg with the probe on the X10 position.

Photos uploaded to Post 51.

After many interruptions due to health and other life happenings, I have finally got to the bottom of this problem.

The main clue came from measuring the oscillator frequency at various stations using the CRO. Useful connections to the CRO were made at x10 probe settings connected through a gimmick made of a low value mica cap and a crocodile clip - not too much disturbance of the circuit requiring only slight retuning of the station tuned to. The oscillator frequency measure at various stations indicated the IF to be about 750-800kHz, just at the point where audio oscillations were occurring. So this indicated that the IF was wildy out.

It looks like as I did rough alignments as the IF transformers were repaired I succeeded in wildy misaligning the radio! This was done at the time to answer the question "How is it working". Got to be a lesson there somewhere.

The next issue was how to get the IF down to the right value. Shorting an oscillator transformer winding and feeding in 450kHz from the mixer end of the IF train did not work as discussed above presumably because nothing was going to get through such wildy mis-set transformers. So I tried an idea I read somewhere months ago which was to first start at the other end of the IF train. So take out the mixer valve to completely disable the oscillator then connect to the detector to make sure that was working - the expected 1kHz whistle there. Then to the 2nd IF valve anode and set the 3rd IF transformer to the IF, just by ear. Then connect to the 1st IF valve anode and adjust the 2nd IF transformer to the IF. Finally, reinstall the mixer valve and connect to its grid and adjust the 1st IF transformer to the IF.

Connecting an aerial, I now had stations at much better sensitivity right across the dial, and no oscillation around 750-800kHz!

Tracking was then tackled using the padder and oscillator trimmers. Perfect tracking could only be achieved at the low frequency end by reducing the cap parallel the padder trimmer by about 60pF. I don't know why this is. Tracking was precise before the IF transformer physical repairs were made, and when the padder trimmer was disassembled, cleaned, and reassembled its value did not change.

A precise alignment is still needed using a swept signal and CRO, and then careful measurement of the sensitivity of the various RF stages. And short wave needs alignment. But I now know that the IF physical repairs to stop them wobbling about in their cans was OK.

And now I don't have to listen to various gas-baggings (alright if it is the Science Show) but can listen to music on 2UE!

For alignment there are many traps and AGC is one on SW it will be disconnected as it destabilises all but one of the octal Pentagrids.

You should be able to, with a very low strength radio station, peak the IF trimmers for maximum signal first. If it does not cooperate you need to fix that stage. Running any 6U7 without its shield will end badly. One of them oscillating can generate a huge amount of AGC. As can bad lead dress in the detector 1st AF area.

If you feed in signal from a generator It is usually around 200pF here, in series with the coupling lead. No grounding wire. That signal input should be around 50uV, or lower, as having the AGC cut in will ruin any chance of getting it right. That is why its volume high, to be able to hear the low signal. That injection goes into the top cap of the 6A8.

Another trap is calibrating the generator with a counter and having the tone/modulation ON.

Oscilloscope if its an old one is best connected to the output tube plate. Otherwise the speaker if it cannot handle the volts many modern ones can't.

Where there is an RF stage preceding the mixer tube (6A8). If the two are not on the same frequency that will attenuate the signal.

With the bands spreading note on BC the oscillator is set at 600kHz and the aerial trimmed at 1400kHz on a weak station.

Thanks for the tips Marc.

This radio has no AGC on the mixer - plenty of that already with the RF and two IF valves. AGC level is presently about 10V on a strong local station. Should be higher but only aligned by ear - the scope and sweep should improve that.

My sweep generator is the Silicon Chip micromite gadget, but will monitor that with a CRO. This agrees very well with my counter, so won't be using the counter. I would connect the CRO to the first audio plate (6J7) as that is only about 80V or so.

No need to have the shield off any valve. Valve shield continuity has been well and truly fixed by replacing one of their base rivets with a brass screw and nut; though in retrospect the oscillation was not due to feedback from that cause in any case - just ham handed rough alignment.

Aerial and RF trimmers are a bit vague in adjustment but should be able to sharpen those up using instrumented alignment.

Time to devise a better antenna as the best I have now is a downpipe connected to the corrugated roof. Probably not much good for shortwave.

There are things that stop IF calibration. One it was pulled to bits & mis wired, or something was broken; It was actually broken. The stage is oscillating; There is absolutely too much signal and the AGC has cut in.

If the AGC cuts in you have not got a hope in hell of calibrating it. feed a low signal into the grid of the tube feeding it &see if it calibrates: Series cap 100 -200pF and check the wiring & shielding.

Marcc

Thanks for those reminders Marc. I am now rereading various contribitions here on the micromite before I get busy.

It is ideal to not have any external antenna connected during calibration. Often hooking up the CRO earth causes more issues than enough at times.

Parasitic oscillation in a stage, can generate a huge amount of AGC

Got my head around the Silicon Chip sweep generator gadget. I might write up a guide for that as a seperate post as I had trouble getting it to produce the correct sweep output graph - often it would invert for no apparent reason. In that case the shape of the plot is OK even though inverted, but the mean and 3dB frequncies are not calculated correctly.

The sweep signal was applied to the mixer valve grid through the gimmick mentioned in post 67, with the oscillator primary winding shorted out to shut down the oscillator. The audio response for the vertical axis was taken from the output of the RF choke which supplies AGC for the RF valve and indicator valve, and also the audio for the 1st audio valve.

Once it was working I could see that my by ear alignment was surprisingly good. I decided to align at 455kHz instead of 450kHz, as this is not a multiple of Australia's 9kHz station separation, and reduces the likelyhood of whistles on some stations.

The mean IF was 456.7kHz, and the frequencies at the 3dB level either side were 454.7 and 458.9kHz. So that is a 3dB spread of 2kHz one side and 2.2kHz on the other for a total spread of 4.2kHz. I think that is pretty good for the result of the by ear alignment described in Post 67 - the mean only 1.7kHz high, and 3dB either side nearly equal. I will submit a photo of the sweep generator screen to Brad for consideration to post here.

I will try to get this closer to 455kHz using the micromite in sweep mode and maybe get a flat topped plot. My strategy for this will be to apply the RF sweep signal to the grid of the 2nd IF valve, and peak shape the 3rd IF transformer response. Then apply sweep to grid of 1st IF valve and try and get the same transformer response with the 2nd IF transformer. Then sweep signal to the mixer grid again and peak and shape the 1st IF transformer response.

Hope this makes sense. If there is a better strategy hopefully someone here will know it.

Photo uploaded to Post 73.