To start with, many thanks to Norm, Curtis and plenty others for the valuable help and advice given on this forum.

I have a few radios, most of them multi-band, that have front-end trouble - the local oscillator isn't working.
This seems to be the circuit I have the hardest time troubleshooting, and the circuit with the most variations from one radio to the next. I mean an audio amp or a power supply or IF amp is a lot easier for me to diagnose. But the books generally go over one or two simpler examples of the converter stage.

So I'll ask here for somebody to look over my shoulder again, and sooner or later I hope to absorb some of this by osmosis. A critique of my work flow, logic and troubleshooting methods would be most appreciated.

I really want to learn how to fish... so getting this radio fixed is one thing, but knowing how to find and fix the trouble is more important to me at this point.

Case #1 - an Airline model 04BR-720
Schematic is located at http://www.nostalgiaair.org/PagesByModel/920/M0011920.pdf



Got it through the classifieds here, one-post seller had inherited a lot of stuff from a repair shop. I knew there might be problems with the set when I bought it, but that didn't stop me. I should never buy radios that have been stuck at a repair shop. It might mean another tech couldn't fix it.

Checked the transformers for continuity, changed the electrolytics out. Two voltage dividing resistors in the B+ supply had drifted high so those got replaced. Had crumbly wires to both IF cans, replaced that. Powered it up and got just a faint hum from the speaker. Proceeded then to replace all the other caps. Still not working.

Touch finger to volume control, got hum. Inject modulated 465KC at first IF, good response. Inject modulated RF at grid of 6A8G, nothing.

Swapped out the 6A8G with a spare, no difference.

Took resistance readings around the oscillator circuit. R3 was okay, R4 and R5 had drifted high, so I replaced them.
On the oscillator coil...
The winding from O to K should be 1 ohm - I got 0.2 not including the meter lead resistance.
From K to P should be 7, I had 6.6
From Q to G should be 1 ohm, I got 0.8
And from G to H should be 2 ohms, I had 1.4

Continuity on the bandswitch was normal all around. Confirmed that none of the trimmers/padders were shorted.

So, I pulled the coil and had a look, visually no indications like green stuff. But since there was such a low reading on the O to K segment, I decided to rewind. This winding was underneath the Q to G coil so that had to go too. MAde some good sketches and notes, and did the deed. Put everything back together and got the same result. Swapped the leads on the O to K coil but no difference there either.



After rewinding, not much difference in the resistance of that coil. Should be 1 ohm, I had 0.2 but then after rewinding it only went to 0.3 ohm - so it appears this was a wasted effort.

At this point I've run out of steam - not sure what to do so I'm asking for help.

Here are voltage readings at the tube pins to chassis, note everything related to B+ is high. These were taken with exactly 120V power input, not 115V as stated.



Do the higher plate voltages just mean the tubes aren't drawing current because there's no signal going through? Or because I'm using a digital meter with high impedance?

Thanks in advance
-Sam T.

Yes, the higher plate voltages are caused by using a digital meter with high impedance. When using shortwave band you need an external antenna, but that's not the problem. Lead dress is critical for everything associated with higher band. If you have before and after pictures, look for anything you replaced in a different position. You can probe with a chopstick to see if moving a component even a little bit changes the pitch or intensity of the oscillations. Could also have a drifted high carbon resistor, I'd just change 'em all.

I should have stated - the radio doesn't work on either band. Hopefully if I could get it working on BC then I could manage getting SW to work. There is no oscillation at all.

Hi

Important to wind oscillator coils in the same direction as original. If opposite may need to reverse wires going to one winding.

A coil is usually good or open. Resistance readings are given for a quick ohm meter check and aren't accurate.

Pin #5 on a 6A8 must be several volts negative when the osc operates. You will also hear the osc in a second radio tuned IF frequency above dial setting.

You are doing the right things, checking bandswitch and voltages. Also be sure tuner plates don't short.

Higher voltages may also be caused by 125 volt line? Tubes also lose emission over time and draw less current. This causes voltage to be higher.

I was sure I had wound the coils in the same direction and number of turns as original, but when it didn't work I swapped the ends of the O to K coil and that didn't make a difference.
Good point about these coils being either good or open, I'll remember next time that a low resistance doesn't equate to a bad coil.
I'll check the tuning capacitor for shorts, then double-check the resistors in the circuit.
You say pin #5 on the 6A8G should be several volts negative, but I have +2 volts there.
This negative voltage is derived from where?
I see R3, one end grounded. Wouldn't that make pin #5 positive with respect to ground?

-Sam T.

You did all the right things, and you're right, a failed oscillator is hard to fix.

It's possible someone replaced the oscillator coil and hooked it up wrong, these coils are NEVER labelled very well and it's easy to get things hooked up wrong. The phasing has to be just right, if one winding is reversed you have the opposite of an oscillator-- you have a notch filter. Try reversing ONE of the windings.

If that doesn't help, you may have a shorted turn. All it takes is one shorted turn and the coil will not have enough Q to oscillate. You won't be able to see this on an ohmmeter. You need to use a "Q" meter or grid-dip meter or a signal generator and RF voltmeter to check out an RF coil.

If you have patience you could rewind the coil. Not all that many turns, even on a BC oscillator coil.

You'll have to try both polarities again, with luck one of them will work.

The negative voltage would actually be caused by the oscillator "oscillating" (if it was). The oscillating voltage is coupled via C7 and in conjunction with that grid resistor causes the grid to float negative a few volts. However if the oscillator is not working (which yours isn't) it will just go to the +2 volts you are seeing. So it is just confirming what you already know (the oscillator isn't oscillating).

As to why it isn't oscillating, there as so many ways to go wrong and everyone seems to be hitting them all. Shorted turn, miswired coil, etc. Your basic coil form "looks" OK, but at least one person around here found that their coil form was was so humidity saturated that this was killing its Q and preventing oscillation (they ended up baking it in an oven for a while to dry it out). Another person found their coil form apparently was satuated with something besides water that would not bake out, and ended up having to replace the whole thing to get rid of the problem (theirs was only failing on SW, and worked on AM Broadcast).

One last thing. Have you tried another 6A8? How does the current 6A8 test? Testing good doesn't necessarily make it a good oscillator, but testing bad almost always makes it a bad oscillator. Also, any of the caps (even micas) that are part of the tuned circuit, if they are leaky, will kill the oscillator.

Curtis Eickerman

Okay that kinda makes sense, in my mind there would be an AC waveform there at the oscillator frequency but perhaps it has a negative DC offset.

I only have an emissions-type tester. It shows the original and the replacement tube to be okay, not great but useable. I have yet another 6A8G I can try in the circuit tonight as well.

C7 is 50 PF, a micamold block but that was changed out with a ceramic disk type.

I had a chance to buy a working Q meter at the Nashville swap meet a couple weeks ago for $100 but didn't for fear it would sit idle.

Another question... to concentrate on the BC band first, I'm looking at coils O to K and Q to G, right? The bandswitch shorts the other two on SW?

From what I understand, AC being coupled to the grid causes a grid leak negative bias to occur if the oscillator is running.

It's not the easiest schematic to read, but I think in the AM Broadcast postion all the coils are in use and then when you switch to Short Wave it shorts the lower coils leaving O-K and Q-G. I base this mostly on the wafer switch "blob" that is supposed to be in the Broadcast position and appears to show G unconnected just like the schematic.

This would make sense since more coil would be more inductance and lower frequencies (i.e., Broadcast Band).

Curtis Eickerman

So then I would start with the SW band as it may rule out a few candidates, sort of. Absolutely sure tuning caps and trimmers are not shorted? You have to disconnect to test through range of movement.

Tried a third 6A8G tube, no luck. This one tested good too.
Confirmed no shorted variable caps, confirmed the wiring from the 2-section tuning cap was going to the right places.
Fixed value caps have all been changed, resistors changed where not within 15% or so.

Later tonight I'll try swapping ends on coil Q-G, this is delicate though because it means moving the actual coil wires.

Unless it looks like someone has been working on them before, I would tend to leave them as is.

If you have a signal generator handy, it might be possible to check out the coil without swapping a lot of wires.

Curtis Eickerman

Actually Curtis I have two signal generators on the bench - dunno why, just never got rid of one when I got the second. A Precision E200C and an Eico 315. Also a B-K #1477 scope, although at higher frequencies the display is really dim. A Senior Voltohmyst, a cheap digital multimenter, AC P-P VTVM, Knight signal tracer, a B-K #1801 freq counter and the Eico #625 tube tester.

I thought at one time about feeding in say 1000 khz modulated at the antenna and then 1465 to the oscillator, just to see if a mixed signal could be forced through the front end. But not being sure exactly where to inject the 1465, and also not sure if this would really prove anything, I abandoned the idea.

Is there a method/procedure you could walk me through to prove out the coil?

-Sam T.

The IF frequency of this radio isn't at 455 Kilocycles?

Rider says 465kc.

-Sam T.

OK, this may just work.

Start with the radio turned off, unplugged and tuned to the bottom end of the AM band (probably around 540 KHz). Connect the signal generator ground to the chassis of the radio and the signal lead to Q on the coil. Also connect one scope probe to the same location.

Tune the signal generator to about 1005 KHz (540+465). Put another scope probe on point O. Set the scope sweep rate to something around 1 uS/Div. Hopefully you will be able to see your signal generator on the trace from point Q. Sync on this trace so it is nice and stable.

Rock the signal generator back and forth in frequency and you should see a nice peak in the signal amplitude coming from point O. This signal should be exactly in step with the generator but inverted. That is, as the waveform from the generator at Q crosses zero going positive, the signal from O should cross zero and go negative.

The peaking of amplitude should be pretty sharp (a high-Q response). If the peak is real broad that would be an indication that something is not right with the tuned circuit (losses causing low-Q).

If the signal from O doesn't look inverted from the signal at Q this would indicate an unintended signal inversion (like swapped wires on the coil).

You can do this same thing in the Shortwave frequencies by switching the bandswitch, and setting the generator to the shortwave frequency + 465 KHz. The scope will have to be set to a higher sweep rate to again see the waveforms.

We are trying to establish whether or not the tuned circuit is low-loss (High-Q) and essentially rings like a bell and whether the signal sent back to the control grid via C7 will be the correct phase (polarity).

In all honesty, I am not sure this is going to tell us the definitive answer, but am hoping it leads us in the right direction.

Curtis Eickerman

Curtis has the right idea-- check out the coil to see if it resonates, just a few minor details:


You need to LIGHTLY couple to the coil, so hook your signal generator to the coil through a like 22pF capacitor or so, and hook up the scope through a 10X probe. Other wise the 50 to 250 ohms of the signal genersator is likely to swamp the coil's response.

Another sideways thing to try is to hook your signal generator to point "O", the grid, and inject a ton of signal at the freq of a strong local station + 455Kc-- that's using your signal generator as a substitule local oscillator-- if everything else is perking well you should be able to hear the station! Of course you'll have to rock the signal generator frequency a bit and rock the tuning too to peak up the rf coil to the station frequency. If you manage to hear things, that tells you that EVERYTHING but the local oscillator is working, that will help morale.

Thanks Ancient_Hacker, I'll go back and try coupling the generator through a capacitor if needed.
Meanwhile I connected it directly and the result doesn't look too good to me.
Upper trace is the input, 0.2 volts/div - lower is output, 0.01 volts/div
Varying the input frequency does not change the waveform height...

Hmm, yeah, if you have a lot less output than input, the coil or LC circuit is scrozzled, somehow. Maybe a shorted turn.

That is what I'm beginning to suspect too, but it's probably well worth taking each trimmer and cap loose at one end and doing a resistance test before throwing in the towel on the coil. Especially make sure the LO section of the main tuning cap isn't shorting. I'm just not used to running into shorts on oscillator coils, but I guess it's possible.

In the mean time injecting the signal generator right at the 6A8 grid is certainly worth trying too just to make sure the converter can actually "convert". Of course that injection also needs to be through a cap with a value like C7 so the bias works correctly.

Curtis Eickerman