What's eating my 5U4s for lunch?

[Tom Wells]

Standard fullwave tube rectifier with 5U4GB, 16mfd 600v input filter cap (old), BIG old choke, then 220 mfd @450v. second filter.
Loaded output is 105 volts .
This powers a pt 15AM with a 6V6 and VR105 on the oscillator, 3-6SN7s, a 78, and a 6V6 on the modulator.
I set this 'way back with a 1-amp fuse on primary that rarely blows, but sometimes has.
This has been in service 18 years with normal rectifier tube life. (3-4 years).
Recently it has begun to open filaments on the 5U4 in a week or so....but never popped the close-set 1-amp fuse on the primary.
Filament voltage measured 5.4, a little high, but should be no problem.
I measured both caps after each "lunching", no meaningful leakage..no hum, .06v AC at supply output.
New tubes plugged in don't show red plates, purple glow or any angry signs.
I'm right back on the air, powersupply transformer is normal temp...output and sound normal, 2-second "fall-off" on power loss as normal.
Been through 4 of these tubes in the past 6 weeks. Ate my last spare yesterday, so tonight I grabbed two new/old stock
12.4 kv diode sticks and soldered them into an octal base, put the thing back on the air, everything is normal,
etc, transformer is cooler due to not running the fil on the 5U4, can't figure what's going on.

None of the failed rectifiers show the "reduced to ash" effect of serving as the fuse in a high-DC current condition.
I thought I was seeing solder melted out of pin 2 on some failed tubes, but corrected the loose/oxidized contact, and the
tubes ARE failed just the same, even after flowing new solder into the "funny" looking pins #2.
Maybe I'll stay with solid state rectifiers, but I really have to expect fuse popping soon as increased current is the only way
I know to lose filaments on a rectifier tube so soon.

I'm still on the same 1-amp fuse as before all this started. 4 hours and counting....
Is 5.4 volts on a 5.0 volt tube really enough to kill them so fast?
Incoming voltage is 116, low if anything. I don't know if we've had surges, but this is a new problem.

Would you dig further, or wait for the next 'effect" to hopefully point to the failing component?

 

[speakerman]

Your choke may be leaking to ground. Shotgun a 10-25 ohm resistor in series to the filter bank and measure the voltage drop across the resistor with a Fluke to determine the amount of current flowing through the tube rectifier.

I built a power MOSFET regulated power supply a while back for my TMC SBE-3 exciter out of old parts I scrounged from the warehouse. The first choke blew up on me and spewed black vaporized tar all over the place. What a mess.

 

[Tom Wells]

The power supply (ahem) is "thrown together" in the wooden base from something like a 1940's slide projector.
Everything is mounted on wood, so no leakage path is possible in this case.

The tube that first expired after a "long" time was an ancient Westinghouse, then a 5T4 out of a Johnson Viking Ranger (rather old, as a metal Cunningham ) , then 2 NOS Amperex's. It's the short life on the beautiful amperexes that made me worry.
Maybe I just had bad luck of the draw on tubes?
Im inclined to order a dozen used from fair radio in Lima Ohio, but still wonder what's going on.
All voltages still normal at 5 hrs. Audio good, no hum or weakness in modulation....

Lifting either filter cap out of circuit shows each is doing its job fully. 102.1 v and steady. Transformer barely warm.
Maybe I'll throw in a heat loss resistor in the filament circuit before putting a tube back in, and really get it to 5.0 or a smidgen under.

 

[Tom Wells]

Ate both stick diodes at 01:00. Fuse still good. I have something to work out in "night engineering".
zzzz
Ideas?

 

[radiosaur]

It's rare, but I've seen fuses that won't open even well over thier rating. That's an outside chance, but try another fuse, maybe?

 

[Tom Wells]

It's rare, but I've seen fuses that won't open even well over thier rating. That's an outside chance, but try another fuse, maybe?

That's a good point and I'll do that, but I still can't believe the transformer ran so very cool.
I need to paw around for some diodes to sacrifice while checking, before I can test again. I'm not eager to eat up any more tubes.
I guess I'll run the oscillator alone and compare to previous conditions, then the modulator only, getting rf from an outboard source.

Now I'm angry that my Fluke 189 has bad jacks and I wont be able to current clamp on different sections.

Perhaps I overdid forward current on the stick diodes, ECG 502, extras from a past oscilloscope repair.
At least that eliminates any question about heater voltage being too high. There weren't no heater this time.
I better go find my old GE or ECG reference.

Today is an official snow mess here in Chicago, and therefore a perfect day to pull full maintenance on this.
I was going to go to Baton Rouge to pull some new fiberoptic sensors into a press and swap phase legs on a blower running backward.
Now I get to see a weak point or part in my own design, and I have the pleasure of finding it.
Guess I'll start with the usual suspects. I recall very few 'lytics on either section at truly high potential, but put lots of RF bypassing on the
modulator. I think grid bypass caps are more likely to do this than low-voltage cathode bias AF bypass cap service.

 

[ncradioeng]

5.4 volts is a little high for a 5.0 volt filament. Maybe there are some shorted turns on the transformer primary that are causing higher secondary voltages which may have caused your solid-state rectifiers to blow also.

 

[Doctor_Technical]

I was going to go to Baton Rouge...

I'll catch some beads for you at the Mardi Gras parades.

 

[Tom Wells]

Maximum foward current on ECG 502 is .6 ma. As in 6/10 ths of one milliamp. Whoops. I think I overtaxed them.
Time to plug in a 5Y3 that fell readily into hand while looking for silicon diodes....
I will be adding some heat loss R to the filament ckt.

I won't make any bet on how long the 5Y3 will live.

 

[Chuck]

Maximum foward current on ECG 502 is .6 ma. As in 6/10 ths of one milliamp. Whoops. I think I overtaxed them.
Time to plug in a 5Y3 that fell readily into hand while looking for silicon diodes....
I will be adding some heat loss R to the filament ckt.

I won't make any bet on how long the 5Y3 will live.

Tom, I don't think the 5Y3 will last very long if 5U4's are popping. It's only capable of handling about 2/3 the amperage of a 5U4. I like the idea or replacing the 5U4 with some big diodes. You ought to be able to find some rated at 5-10 amps very easily. I'd also look for some with a PIV of 600 volts or more. I realize you are running at much lower voltages, but in my old TV repair days, anything less was usually a "temporary" fix. In a power supply like this, "brute force" is a good thing.

I know a lot of guitar amplifier manufacturers went to silicon diodes because they were having a lot of trouble getting 5U4's that actually worked. I think you may be experiencing the curse of the 5U4 like a lot of other people did when good reliable RCA's GE's Sylvanias etc. went away. Using two diodes seems to be a good solution, but they will make your B+ voltage go up. You might want to check that. I'm also surprised that you are only using 16 mfd filter caps. 40-60 would be more normal, but I guess the choke does a great job of eliminating the ripple. Since this thing has worked for a very long time, the solution must be simple.

It is possible that the 5 volt winding has some shorted turns which might be shorted to something else in the transformer. I'd recommend using the diodes, but move them off of the 5 volt winding so B+ can't leak through. The good news with the diodes is the transformer doesn't need to supply the 2 amps or so that the filaments of the 5U4 draws, so it should be much happier (and cooler) in the long run.

 

[Tom Wells]

Added .15 ohms to bring filament voltage to 4.9. The 5Y3 did flash around a little until the emission started, but didn't fail.
Transformer feels exactly 2 amps warmer. I like the idea of lifting off the filamant winding entirely for solid state rectifiers.

I'm putting two 400v piv in series, two branches for full wave. I like lots of reserve, too. The 5Y3 lived just fine for an hour.
Here goes...

 

[mr_elbows]

Tom

Tried to e-mail you, the bin is full

I've got some rusky-made 5U4's here, e-mail me if you need one or more

G

 

[Tom Wells]

All seems well with the new solid-state replacement. Transformer cool, etc.

Maybe it was just 2 bad new tubes, and 2 really old tubes.
VR105 on oscillator stage is glowing in a larger area of conduction than previously with tube rectifiers, so I must be seeing the
higher voltage always expected with such a conversion.

 

[Tom Wells]

I was going to go to Baton Rouge...

I'll catch some beads for you at the Mardi Gras parades.

I'll be arriving in B.R. latelate Sunday, and hopefully have only about 3 hours of real work, which must happen earlyearly Monday.

Naturally, other things will need adjusting or repairs, just because it's new. This morning, the trainer calledfrom B. R. saying the press main
screen was reporting a main motor overload immediately upon press power-up.
I explained that modern motor drives wake up "funny" 1% of the time when perfect. It used to be that a technician or someone needed to know
to open the electrical panel and press the "reset" button on the motor drive keypad.
In acceptance of this flaw the press manufacturer has moved this drive keypad to an outer wall of the cabinet, with a press/click door affording
easy access from a convenient and normal operator area. " Oh yeah, that... well, they do that."
Don't you love technology?

I hope to see something as in 20 years of travel, this has been my first assignment (3 weeks in Jan, now a return) in Louisiana.
They have me returning Wed morning 9-ish out of N.O. That is going to be an interesting flight, betcha.


I'm inclined to watch the parade on Perkins road, if only because that's how I've been going to the customer.

 

[littlejohn]

That's 8% high on the filament, >if< you used a true RMS meter. Might be even higher if tyou didn't.
Does anyone but the Heathen Chinee make 5U4s anymore?
1N <2amp at 1KV> diodes are cheeeep, why not add a couple and let the tubes rest?

 

[Tom Wells]

That's 8% high on the filament, >if< you used a true RMS meter. Might be even higher if tyou didn't.
Does anyone but the Heathen Chinee make 5U4s anymore?
1N <2amp at 1KV> diodes are cheeeep, why not add a couple and let the tubes rest?

Yes it's an RMS reading meter.
The cheap diodes are holding just fine and I see no reason why they won't last.
Maybe it was just a string of bad luck, but 2 (two?) NOS American-made Amprexes each have life of 1 week?
I always did like the feature of using a filament-type rectifier as an "additional fuse", but this is the first time that feature has become
annoying.

 

[Chuck]

That's 8% high on the filament, >if< you used a true RMS meter. Might be even higher if tyou didn't.
Does anyone but the Heathen Chinee make 5U4s anymore?
1N <2amp at 1KV> diodes are cheeeep, why not add a couple and let the tubes rest?

Yes it's an RMS reading meter.
The cheap diodes are holding just fine and I see no reason why they won't last.
Maybe it was just a string of bad luck, but 2 (two?) NOS American-made Amprexes each have life of 1 week?
I always did like the feature of using a filament-type rectifier as an "additional fuse", but this is the first time that feature has become
annoying.

I'm thinking you just got some bad tubes. A while back, I had a NOS Tung-Sol 12AU7 last about a week before it crapped out. It's a scary thought, but maybe tubes don't have the infinite shelf life we though they had.

Did you remove the diodes from the 5 volt filament taps? Just curious, but it could be a leakage problem in the power transformer.

 

[littlejohn]

Now if you bridge each diode with about a .01 and a 200K or so resistor to equalize the drop across them, it ought to go forever, or until the iron shorts. At which point you can gin up a Foolkiller (line violtage doubler or tripler) and continue on untikl someone touches the B+.
I think yer right, given my luck with topday's receiving tubes, you got a pocket full of poor ones.

 

[Eng.Mike]

If you replace your 5U4 with a pair of silicon diodes, your rectified voltage will become significantly higher.
The 5U4 has a voltage drop of about 44 volts and the silicon diodes drop what? 0.6 volts? So with silicon, your plate voltage for all the other tubes goes up and that changes the opperating point for all of the tubes which is a bad thing. You need to compensate for the voltage increase or you may start loosing your other tubes from over-dissipation. The circuit will likely not work as designed tubes running at a different opperating point.

About Chineese rectifiers:
I've only used the Chineese 5AU4 since old stock can't be found at a reasonable price. The ones I've used worked great, lasted longer than the Russian ones and looked like they had copper plates. The wierd thing is that the printing on the glass reads "Election tube" instead of electron tube.

 

[Chuck]

The wierd thing is that the printing on the glass reads "Election tube" instead of electron tube.

The Chinese aren't the only ones with a spelling problem. A few years ago at a NAMM Trade show, TEAC/Tascam was proudly showing off prototypes of their new offerings. One device had an especially attractive front panel that clearly said it was an "Amprifrier." ;D