http://www.nostalgiaair.org/PagesByModel/471/M0001471.pdf

The radio has been completely restored. All paper and filter caps were restuffed. All resistors checked and replaced if out of tolerance (with reproduction dog bone resistors). Yards of rotten rubber wiring was replaced. All tubes checked. All voltages are OK. Everything works from the first IF back. But the oscillator is not working on most bands. On band 4 (high short wave), the radio works. On band 3 (medium short wave), it works from 6-10mhz and is dead below 6mhz. Band 2 and Band 1 (broadcast) are totally dead. The oscillator is not running on those bands (listening on my Realistic DX-440 with the BFO on). All coils show continuity on all windings. The fixed (C6) and variable padder (A5) for the broadcast band seems OK (no leakage, measures 750pf).





When the radio was received, all except one of the many flexible wirewound resistors in the oscillator section (R2, R3, R4, R5) had been replaced by either carbon composition or large wirewound resistors. R5 was still there (band 4). Interestingly, THAT IS THE ONLY BAND THAT WORKS! I replaced R2, R3, and R4 with 1/4 watt carbon comp resistor inside rubber sleeving with end terminals in order to mimic the appearance of the original resistors (using R5 as a model), and routed them as shown in the pictorial diagram.

The B+ to the RF section is 240 with 110 volts AC in (spec is 190). Screen voltage is 58.9 (spec is 60).

Interesting thing: if I short out R2 (broadcast band) with a jumper, that band works from 1270 up. It is dead below 1270. And also interesting, once the oscillator fails, I have to go back up to 1500 in order to "shock" it into oscillation. Moving slowly back up from 1270 will not restart the oscillator until I reach 1500. Then I can move back down the dial!

I have tried several 58 tubes in the oscillator position - all test strong. No change.

I checked the oscillator grid capacitor and resistor, and even replaced them with new parts - no change.

I just have a feeling that the 4 wire-wound flexible resistors are somehow CRITICAL to correct operation, since the only band that works has its original resistor (R5, 12 ohms) in place.

I'm not sure how I would go about reproducing one of these wirewound resistors.

And I do not know what to try next! I have restored close to 600 radios, and never had a failure before.

You might have a lead dress problem around the oscillator.

That is a very strange oscillator. It appears the resistors are to adjust the feedback as it seems that they would create a lossy low Q tickler winding sort of artificially. And looks like they are chosen to keep oscillator output constant, hence the decreasing resistance. Other than that I don't have a clue. The very low value of the grid coupling capacitor is odd too. Why would he use 2- 8 pf instead of a 15 or 20 single??? Perhaps, as you suspect, the inductance of the usually inductively wound resistors enters into the operation. I would be tempted to try winding some little chokes on a form about the same size and length as the original resistors, if you know what they looked like, of #36 to #40 wire, maybe vary turns and see if you can get it to work. Use resistors in series to keep resistive component constant. Don't see why this approach wouldn't work. This approach would simulate the wire wound resistor.

Only thing strange here is the fact that with the resistor shorted out on BC, it still won't osc. consistently. Perhaps the resistor and its inductance are really the plate load in the oscillator and the coil is just to couple the osc. inductively to the first Det. grid coils. That doesn't make sense either.

The wire resistors are not critical; they can be replaced with common metal film resistors. You may want to add more ground lugs for the new resistors.

Is the insulation crumbling in the switch/coil area? My 708 had very similar problems as yours and I finally had to throw in the towel and replace every wire. If you want good color documentation on the coils, PM me.

Pete

Did it work before you replaced every part in it ??

Another case where to many parts are replaced at one time .

I have not traced the circuit but the fact that it appears to have four tuning gangs and a seperate ocillator, tends to suggest it only looks weird because the oscillator is more like a colpitts, but is not as Colpitts only uses one coil.

Normally a crook tube will die at high frequency this does not. I would have a serious look at the non working coils, & components fitted to them including switches and what is on them. (Wiring errors etc)

Marc

Do you (or anyone else on this forum) have a working example where the flexible resistors have been replaced with normal resistors?

In my radio, only BLACK, BLUE, and GREEN wires were crumbling (as well as the power transformer leads). The remainder (white, yellow, red etc) were still supple! This meant that I did not have to touch the wiring around the bandswitch and coils - the lead dress was not disturbed in this area. And there was only ONE paper cap replaced in that area also (a B+ bypass cap and the AVC filter cap). I did test the bandswitch - no problems.

I did have to replace some of the leads from the RF subchassis back to the main chassis, including the lead from the oscillator plate to the bandswitch (green).

The radio DID NOT WORK when I received it. Also, whoever replaced the flexible resistors may have ran into the same problem I am having - it may not have worked after the replacements and he gave up!

Hey A A Kent, I sent you a PM awhile back regarding a previous question (not yet picked up).

Looking at the circuit the fault has to be in the coils & switches and the plate circuit of the 58's or the grid switches of the oscillator.

These are easily checked, as voltages will disapear on the plates & grid 1 if its a plate switch.

The wire wound resistors will be inductive and have the potential to throw the oscillator off frequency. Dirty switch contacts cannot be overlooked.

If its an open construction type switch (wafer) and its got greasy type gunk on it? An aqueous solution of Soda Ash, can be most effective.

Do not leave it on for long periods; A small brush is useful, but watch the fine wires and where the water is going to go. Clean water will remove the Soda Ash. Spray it with circuit board cleaner, or something else, to get rid of the water. Do not run the set until you get rid of the water.

A CRO may be better, to see if its the oscillator and not the RF coils, causing the problem?

Marc

The fact that AK had to put in series resistors, and used a different value for each band, suggests that they had trouble keeping this oscillator working consistently.

This isn't a Colpitts, it's a regular Armstrong, with the slight embellishment of putting the padders on the opposite side of the secondary away from the tuning cap. This made it possible to ground them.

jkaetzjr, we don't know the extent to which the ticklers are self-tuned, but if they are, the resistors wouldn't materially affect the Q because they're in series with the already high plate resistance of the oscillator tube. However, their inductance and coupling to the coils may be important. Hmm, judging from the picture they were outside the shield cans; in that case their inductance might be interesting but at least we don't care about coupling.

That's a really small value for a grid coupling cap. Since micas are suspect these days, I recommend you temporarily replace them, and experiment with the value of your substitute.

HTH,
Dave Wise

Yes, this radio is FULL of 8pf caps, many times they use 2 in parallel.

I did try a 22pf silver mica grid cap and 100K resistor (the existing dogbone measured 114K) for the oscillator grid cap and resistor. I will try other values, since I have no clue what to do next.

A A Kent also suggested a better test of C6, the series padder cap (which may be leaky). All sections checked OK on my DVM (>40megs and values close). But my TO-4 would be a better test. And I could also substitute a silver mica for one of the dead bands.

I did replace the wiring for T5, the broadcast band oscillator and mixer coil. NO CHANGE.

I also found a REAL 2K flexible wirewound resistor that was the correct size (but not an AK) in my junk box and tried that. NO CHANGE.

If you get desparate, you might try adding 10% more turns to the tickler coil on BC and see if that helps. The radio worked once and should work again but who knows. Not a big proponent of modifications but sometimes it works! I had an autodyne circuit in a Gloritone that was intermittent starting sometimes if turned on tuned below 700. Once it started, no problem. I messed with it a month. Finally added 4 turns to the tickler in the cathode of the osc. /mixer (it had two, one in cathode and one in series with the IF transformer primary).That fixed it. I later obtained the service data and they recommended trying a different 24A and as a last resort, replace the osc. coil/If transfoirmer combo. I betcha that new transformer had a few more turns on it as they did recognize there was a problem.

Some 24A tubes simply did not work in a few sets, such as Crosley autodyne sets. The sets were designed for the 24 tube, not the 24A and relied on secondary emission to work, and the secondary emission was more controlled in the 24A tube.
Curt

This one was for 24A, written on socket. I tried a half dozen others and none made any difference. I was a real head scratcher. EWQhen you replace every piece in the osc circuit and no help, gets desparate. Funny to me if you build an amp. often it wants to oscillate, and an osc doesn't want to do anything.

I agree with Jack and the others about not enough inductance in the plate circuit of the oscillator.

As a WAG Try using a choke of about 50 to 100 u (EDIT maybe even upto a few hundred uH) in series with the 2k resistor for the BC band to see if it starts to osc. If you have some chokes available it might be interesting to see the minimum inductance needed to make the osc work.

If this gets the BC osc going then you can try lower value chokes for the other bands.

I agree with your original idea that the inductance of the flex resistors is the problem.

I believe the reason for the resistance is to equalize the osc injection level to the mixer for the different bands. The BC band will have the highest osc level, so must have the highest value resistance to lower the current in the plate coil.

If there is missing plate voltage, do not overlook broken socket, seen one this week. Not likely in this episode but I did see one switch make & break on a plate, depending on which way it went.

The small caps may be for dealing with "Miller effect"? I have one oscillator that has a 3.3pF in the grid circuit (not coupling), remove it and the oscillation stops.

Marc

I checked the plate and screen voltages using an old SET TESTER, that plugs into the radio using a plug and cable and accepts the tube in the tester. I then measured the voltages. The plate voltage was about 230 (spec 190) on all bands, and the screen voltage was about 58 (spec is 60).

I did determine the effect of raising the screen voltage, up to about 75 volts - no change. I did notice that the screen voltage did vary considerably depending on whether the set was oscillating and/or actually receiving a station. The screen voltage jumped up to about 80 if actually receiving a station. I assume this is due to AVC action, since the plate and screen voltages to the RF and IF sections go through common dropping resistors.

I determined that R5 (12 ohms, the only original flexible resistor) was nothing more than a straight piece of resistance wire inside a sleeve, about 3" long.

Other tests made:

I experimented with different oscillator grid capacitor (from 22pf to 100pf) and resistor (47K to 220K) values. There was some improvement with higher grid resistor values, but the set never worked. At one combination (100pf and 150K) the set would actually oscillate and receive on the UPPER part of the BC band with R2 in place (previously, R2 had to be shorted out in order to get ANY action on the BC band).

I removed and examined T6 (band 2 oscillator/converter transformer). The oscillator tickler was about 15 turns. The order of the windings from the top was: Osc Tickler, Osc Grid, Converter Grid, RF plate. I added about 5 turns to the tickler (in the correct direction) and reinstalled the transformer after replacing some wiring. Band 2, formerly dead, now worked a little, but only at the TOP of the band. While the coil was out, I heated it with my heat gun to the point that the black wax melted (but did not run).

I now am wondering if the problem may be MOISTURE. Perhaps the coils should be baked? They are all covered by a wax that melts at a low temperature. I will have to do some research as to how HOT and HOW LONG they should be baked.

Thanks for all your advice and suggestions!

As suggested, I did a better test of C6, the padder capacitor(s). I checked the broadcast padder (disconnected it) with my Sprague TO-4. It had ZERO leakage, and the value was reasonable (about 400pf with A5 in parallel).

I also did a test using a type 55 (triode) tube (in a adaptor that had no connections to the diodes). It actually worked better than the original type 58! Except, on band 4 (high short wave) where reception stopped at 10mhz (went to the top with the 58).

Hi Dave...

Did you pull the schematic off of Nastalgiaair? If you did, you'll be missing three pages. Namely page 4-18 (a pictorial of coils T-1 thru T- eight) page 4-19 ( parts list) and page 4-20 (below chassis pictorial)

I'm not one for looking at a pictorial, it's confusing if you're used to reading schematics, but with AK's it can sure help when your back's against the wall.

...Jim

Thanks. I have a complete set of Riders manuals, which I prefer to use (vs the online scans). The schematic for the AK808 (same chassis) contains the additional pages in Nostalgia Air.

Did the three caps marked C6, which are all different, get changed?

I would expect Mica there. American Mica's fail regularly; despite looks some of those could be paper, and may have let go?

Marc