Western Electric 121A, power supply time

[NewYorkDave]

OK, I have a few free minutes. I'll try to answer some questions.

1. Neither of those meters (0-5mA and 0-50mA) will do the job. As it says in the manual, you need a 200uA (FS) meter. This was a fairly common type, but you'll probably have to make up your own scale.

2. You're not specific about what you mean when you say "as the wiper approaches zero." Which end of the pot are you talking about--zero attenuation or zero signal? Change the filter caps before you get into any kludgey fixes with series resistors, etc. If the motorboating is happening toward the "zero attenuation" end of the wiper's travel, the filter caps are the prime suspects. Also, check the pot, make sure there's continuity between the wiper and either terminal at all points in its travel. Weird things can happen when a grid goes "open", although motorboating isn't usually one of them.

Assuming you meant "grid" when you typed "screen", the volume pot arrangement should be perfectly usable and is, in fact, what I'd consider an improvement in this design.

 

[Butterylicious]

Thanks Dave. I think it's great that you copied my scans. I'll try to add some more stuff soon. My scanner is old, slow, and connected to a different computer so I have to poke myself to work on my archives.

Yes I did mean grid. I'm really awful with names, people, parts, & otherwise even though I understand who are what I am talking about. I've mastered the art of, "hey mang how's it going" since I'm so awful with names.

I'll try to explain the gain mod a little better even though I think you've got it right on. R5 was replaced with a pot of the same value with the fixed ends connected as shown in the schematic. The lead going to the grid was poped loose from the point where R5 and C3 meet and is connected to the wiper. When the pot is turned down (decreasing gain) but befoe it is turned all the way down is where all hell breaks loose. Motor boating is not really an apt description, it really more like a 1Hz oscillation. Not pretty sounding and I didn't hang out there long enough to take any measurements.

My evilBay crap-shoot of can caps showed up last night and I did get 4 of the 6 multi-section cans I was looking for. Now I have a huge cache of canisters. I ordered the 50uf x 4 cans from Antique Electronic Supply last night, as well as some Solen coupling caps. It was that or orange drops. Finding something that resembles the original caps just ain't gonna happen. The .4 uf caps were replaced with .47 caps. I was told that I should increase capacitance where specified values can't be had. I hope this is correct. My other choice was .33 uf.

Too bad about the meters. Hopefully I can find something along the way that will work. What is your opinion on padding the input? Also I'm considering adding phantom mic power to the front end. A penny for your thoughts.

Thanks!
-Richard

 

[Butterylicious]

For those of you keeping score at home, I just finished the cap job.

1 channel is flat from 30Hz to 18KHz. It's -1dB at 20Hz and -3dB by 18Hz. On the high end there is audible ring modulation but it doesn't show up on the scope. Long & short of it, it's operating better than spec.

The other channel is still acting goofy. At about 400Hz it starts gently rolling off and is down 3 dB by about 30Hz. After that rolls off hard as would be expected. High end is flat up to 18KHz and again there is audible ring modulation but it does not show on the scope.

 

[PRR]

> as the wiper approaches zero, all hell breaks loose

Exactly where does the grounded side of the pot return to?

If it is randomly connected to any old "ground", specifically ground around the output stage, sure it will motorboat.

Seems to me the best place is probably "where it was": the commoned side of R5.

Lacking a 0.2mA meter (which needs enough series resistance to make it 1K ohms total), use a 0.2VDC meter with a 1K resistor in parallel. 0.2V needle-meter is rare; 0.2V digi-meter is a very common item. You still have to do the math as indicated in the manual to find the exact current. The first stage should show 0.085V, second 0.09V, third 0.15V, all +/-10% more or less.

 

[Butterylicious]

Actually the cap job fixed the "all hell breaking loose" issue. I'm betting one of the caps on the screen or supressor of the 2nd stage was letting some bad juju through and the stage was actually oscillating at about 10Hz. It was ugly but it's gone. The pot is grounded at the shell of C2. I agree with your theory of "where it was" and when I move the pots to the front panel, I will ground it accordingly.

It's a pitty about the meter. I really would like to have vintage diagnostics built in. I even found a cool old selector switch I could use.

I'm hoping my remaining frequency resopnse issue is tube related. The side in question does have a 5693 installed in place of a 6SJ7. I wouldn't think that should be a problem since they are pretty much the same cigar, but swapping their tube complements will take all of a minute. Seems like a good place to start. If that isn't it, I'm going to have to start checking resistors which I don't really want to do. Those things are old & delicate. I've already cracked (and replace) 1 just from looking at it wrong.

 

[Butterylicious]

OK 1 side is working perfectly. The other side has a gradual (almost linear) roll off starting at about 1.5k and the signal is down 3dB at about 40Hz. I swapped tube complements and it didn't change anything. So I took a whole bunch of A vs B measurements and wrote them down on the schematic here . The measurements are all listed good side over fussy side. The only noticably differences are the screen & plate voltages for V2. Current wise the 3rd stage is 4 mA less on the fussy side. Scoping the signal path & sweeping the input frequency, I did not notice any differences until after the 3rd stage.

A penny for your thoughts.

Thanks!
-Richard

 

[CJ]

Go to the bad side, put the freq where it starts dropping compared to the good side, put a cap across the catrhodes one at a time to see if anything changes. Us a 50 uf at 50 volts. Ground one end with a black alligator cable to the chassis.

You might have a (gulp) output transformer that has a few shorted turns.
You can swap them, or just un hook them and do a sweep test with a generator.

cj

 

[Butterylicious]

I would like to think it's not an output transformer since I am measuring the roll off just ahead of it. Is it OK if I use red alligator clips? :?

 

[CJ]

Red is oK I guess.

Impedance can be reflected back.

 

[Butterylicious]

OK I did the cap thang.
Stage 1 = no change after initial charge.
Stage 2 = pegs the VU meter
Stage 3 = nochage after initial charge.

Both sides reacted identically under this test.

As far as the possibility of a dieing tranny, can I just pop the leads loose and compare DC resistance? Will that tell me, or do I need to set up a more elaborate experiment?

 

[CJ]

Sorry I should have said put the cap across the other caps on the cathodes. When you bridged the stage 2 cathode, you shorted out all your neg feedback, which is what caused the osc.

An inductance bridge would be nice to have to measure the primary inductance, as a few shorted turns won't show up as much of a difference.And transformers are wound by the turns, not by the ohms, so there will alwys be a little difference.

Did you also rerplace that C6 feedback cap?
Too lazy to read all the posts.

 

[Butterylicious]

OK, again no change other than the momentary increase in output the instant the cold cap hits the circuit.

Yes, I changed all the caps.


I found a tutorial on inductance bridges here. I don't have a null meter or a known inductor.

:sad:

 

[Butterylicious]

in my best Stimpy voice
JOY!


I did the flip flop on output trannies. The problem is in the electronics and not the transformers.
JOY!


Of course CJ is correct about the DC resistance. From doing the flip flop I now know that both transformers are good. It's a parallel primary. Here's what I measured:

Fussy Side          Happy Side
1488                  1620                 Blue Primary
1489                  1605                 Green Primary
371                   365                  Red Secondary

I guess it's on to checking resistors. Them thangs are so old, they'll break if you look at them wrong. I tried my hardest to keep my iron off the resistor turret board.

 

[Butterylicious]

Man I'm banging my head on my desk.Let me sum up where I am at. I have replaced all the caps. 1 side is perfect, the other is not. Here is a frequency response graph (a zillion times better than before the cap job)

Here is the schematic with some voltage and current measurements.

I've tried flip flopping the tubes and nothing changed. At CJ suggestion, I swapped the output trannies and I determined that they are both OK. I took it 1 step further and swapped the input transformers. They're both hunky dory too. I looked at the individual stages with a scope and they seem to be equal up until the output of the final stage. I'm just completely stuck now.

 

[PRR]

10V/10V on V1 cathode can't be right. Maybe R2 drifted, but both the same? I'm thinking measurement error.

It seems unlikely that V2 has 70V/64V at the plate, 50V/43V at screen, yet both units have the "same" 2V at cathode. Resistor drift? It would have to be implausibly balanced to end up the same at both cathodes. I know I decry excess precision in tube design, but when diagnosing mismatched "matched" units it is better to have excess precision. 2.34V, not 2V.

Your graph shows equal gain at 1.5KHz. Is that correct? Or does "fussy" have lower gain all across the band, approaching "happy"s gain at the highest frequencies?

I'm thinking C5 is ~500pFd, or open, instead of 0.5uFd.

Are both load resistors identical, not one wire-wound and one carbon?

 

[Butterylicious]

My graph is fairly accurate. Fussy is crossing 0 dB at about 1.5KHz. It's almost a linear increase starting at -3dB at 30Hz and ending at +2 dB at about 18KHz.

It's quite possible I have errors in my measurements. I was unorganized and in a hurry during that bench sit. I took these measurements without any input signal. The load resistors are identical and a close match.

I'll change out C5 again tomorrow and see what happens. I'll also take a better set of measurements.

Thanks for the advice!
-Richard

 

[Butterylicious]

I change out C5 and there was no change. I took a whole new set of measurements paying much closer attention to detail.

Here's what I am wondering now:
Terminal 13 goes to the center tap of the fillament power supply. Both preamps are running from the same power supply. Could this be the source of my problem?

 

[Butterylicious]

GOT IT!

C7.2 was bad. I was retracing my steps and I'm guessing CJ assumed I was smart enough to pick up on strapping across that cap too when making my tests, but maybe not... didn't he tell me to use a 50uf/50v? I should have checked this right away after seeing congruent operation up until the output of V3. It's not like I don't have a cool old Hickok tester that'll check caps too.

Thanks to everybody who chimed in. I really appreciate you sharing your knowledge. I still feel like a stranger in somebody elses house here, but you guys are awfully nice. I've got quite a bit of chasis work to do. I'll post pictures when it's done.

 

[CJ]

Did you try bridging C7?

Got past the radar. Good perseverance. I would have had it bolted into the rack at this stage, writing it off as "vintage sound".

 

[PRR]

> C7.2 was bad.

That's novel.

I see why it gives a near-flat response, but can't explain the exact curve.

Generally, the choke is lower impedance than the transformer, so adding it in series has only "small" effect. I think the residual slope is iron-loss: the transformer laminations a little better than the choke core.

Good work.