Bias range not cold enough

[Sansteeth]

Hello everyone,
I'm currently working on a Sovtek MIG-100 for a general check up.
I noticed the plate voltage is pretty high, 513VDC. Screens are at 512VDC with a 750 Ohm/3W resistor. I've read on here that it wouldn't hurt to go for 1K/5W there which is going to be the next thing together with a full cap job, however is a 1K screen resistor going to be enough to lower the screen voltage to healthy levels?
My actual problem for now is that the bias won't go low enough, at -51.3VDC I get a plate dissipation of 26.7W (for around 50mA per tube).
My question are, is that common to have a bias supply that doesn't go low enough? I see people all over the Internet having a bias pot range that doesn't go hot enough, but I have the opposite issue. Is that a bad design or a failing component in the bias supply?
Can I modify the bias supply to bias the output tubes colder considering the plate voltages?

Thanks a lot!

Sansteeth

[sluckey]

I don't consider your screen voltage a problem.

Look at the simple bias circuit. DECREASE the size of that 10K resistor (don't be afraid to go to 1K if necessary) to increase the bias voltage which will make the tubes run cooler.

[Sansteeth]

Hey Sluckey, thanks a lot for your response, I thought that more than 500VDC on the plate of a 6L6GC was quite a lot but it was fine with a lower voltage on the screen, (I'm not yet too familiar with the plate and screen voltage relationship).
Also why have a 750 Ohm screen resistor if the voltage drop is non-existent?

Look at the simple bias circuit. DECREASE the size of that 10K resistor (don't be afraid to go to 1K if necessary) to increase the bias voltage which will make the tubes run cooler.

Alright, I will experiment with the value of that resistor! Thanks for the tip! I have two questions left though, why the 10K resistor and not the 47K or 15K, do they have a different function in the circuit?
Interestingly enough, on the MIG-100H schematics, they dropped the 10K resistor entirely from the bias supply and use a 63k pot instead of the 100K pot on the MIG-100.
I do have a MIG-100 but its power section looks more like the schematic of the MIG100H, I'm going to have to figure out which bias supply I have.
My last question is: why could I not get it cold enough in the first place? bias supply that is not beefed up enough to counteract more than 500V on the plates?

Thanks a lot

[sluckey]

Maybe you need to replace the bias caps.

[2deaf]

Hey Sluckey, thanks a lot for your response, I thought that more than 500VDC on the plate of a 6L6GC was quite a lot but it was fine with a lower voltage on the screen, (I'm not yet too familiar with the plate and screen voltage relationship).
Also why have a 750 Ohm screen resistor if the voltage drop is non-existent?

Alright, I will experiment with the value of that resistor! Thanks for the tip! I have two questions left though, why the 10K resistor and not the 47K or 15K, do they have a different function in the circuit?
Interestingly enough, on the MIG-100H schematics, they dropped the 10K resistor entirely from the bias supply and use a 63k pot instead of the 100K pot on the MIG-100.
I do have a MIG-100 but its power section looks more like the schematic of the MIG100H, I'm going to have to figure out which bias supply I have.
My last question is: why could I not get it cold enough in the first place? bias supply that is not beefed up enough to counteract more than 500V on the plates?

The 10K is mounted in eyelets over by pin 39 while the other resistors are wave-soldered on the printed circuit side of the board.  Those things didn't always come with a 10K resistor there.  It looks like they had to fire the thing up to see what voltage they were getting and then select an appropriate resistor.  Fender was like that also, so they didn't always have a 470 ohm resistor there.

That amp originally had Russian 5881 tubes in it.  These weren't your grandfather's 5881's, but instead they were some kind of a Russian military tube.  They could handle very high plate and screen voltages and they could take a lickin' and keep on tickin'.

   

[tubeswell]

Russian 5881 (6P3Se) are rated at 20.5W and were designed for use in ICBMs and soviet supersonic bombers to withstand specific vibration/shock/G ratings, so they can take a lickin alright.


750R eats up a bit more g2 current under heavy signal conditions

[2deaf]

I opened up a MIG 100U and the bias range was from -38.7 to -61.8.  This one has a 10K resistor mounted on turrets instead of eyelets.

I don't know how your's is, but mine had the capacitors solidly glued to the boards so that I had to skin the capacitors to remove them.  The bias capacitors are little metal cans and the original ones are still in the amp.

[HotBluePlates]

... the bias won't go low enough, at -51.3VDC I get a plate dissipation of 26.7W (for around 50mA per tube).
My question are, is that common to have a bias supply that doesn't go low enough? ...

The thing to understand is the bias supply outputs a range of voltage.  That range is based on the designer's guess for what an "average output tube" will need to keep the tube from overheating when driven.

However, you & I buy tubes that may not closely match the "average tube" the designer expected.  So any given tube might idle a lot hotter or a lot cooler than what we think it should when fed the voltage from the bias supply.  So your problem is not that the MIG-100's bias voltage is "wrong" but the set of tubes you're using now happens to idle too hot and needs more bias voltage to cool them off.

     - I tested a batch of ~30-40 6L6s not long ago, and my tube tester shows me the idle plate current (among other things) for a precisely-set test condition.  Out of that large batch, a single pair of output tubes were "defective" in my opinion, as they ran very much hotter than every other 6L6 tested.  Meaning they had twice the plate current of all the other 6L6s.  Obviously, these weren't "average tubes" or anything remotely close.

[shooter]

MIG

Russian ICBM, now MIG fighters, give me a Tomcat any day 

[Sansteeth]

Hey, thanks everyone for chiming in! That's really helpful

The 10K is mounted in eyelets over by pin 39 while the other resistors are wave-soldered on the printed circuit side of the board.  Those things didn't always come with a 10K resistor there.  It looks like they had to fire the thing up to see what voltage they were getting and then select an appropriate resistor.  Fender was like that also, so they didn't always have a 470 ohm resistor there

Found it! Thank you, so I do have the 10K resistor like on the MIG100 schematics, it's soldered the same way the rest of the resistors are on the board. Will mess around with the value and report back.
When you measured the range of the pot, you did it without tubes, am I right? 

I don't know how your's is, but mine had the capacitors solidly glued to the boards so that I had to skin the capacitors to remove them.  The bias capacitors are little metal cans and the original ones are still in the amp.

I have all original caps, they've put a lot of hot glue in there (not as much as in my Sovtek MIG60....) but it's easily removable with mine, all caps are the original aluminum cans with Cyrillic markings, I was planning on changing them, it's long overdue I think.

That amp originally had Russian 5881 tubes in it.  These weren't your grandfather's 5881's, but instead they were some kind of a Russian military tube.  They could handle very high plate and screen voltages and they could take a lickin' and keep on tickin'.

When I bought it 10 years ago, it had the Sovtek 5881 in there. Don't know how early '90s 5881 fared back then, but I dropped a quartet of Svetlana Winged C in there and they didn't survive. Now I have modern production Tung Sol 6l6GCstr in there, pretty hot if anything.

So your problem is not that the MIG-100's bias voltage is "wrong" but the set of tubes you're using now happens to idle too hot and needs more bias voltage to cool them off.

     - I tested a batch of ~30-40 6L6s not long ago, and my tube tester shows me the idle plate current (among other things) for a precisely-set test condition.  Out of that large batch, a single pair of output tubes were "defective" in my opinion, as they ran very much hotter than every other 6L6 tested.  Meaning they had twice the plate current of all the other 6L6s.  Obviously, these weren't "average tubes" or anything remotely close.

Okay I understand what you mean, what is it about the tube that makes them idle too high for a given grid voltage? My guess would have been that the transconductance is too high? As in, the tube is "too" efficient in translating a change in grid voltage into a change in plate current? That doesn't sound right because idle is a point on the curve, not a change over a range. What am I missing here?

Russian ICBM, now MIG fighters, give me a Tomcat any day

I'll take a hot amp any day over a cold war 

[shooter]

 


Nice comeback!!
and they are Hot amps, I worked on one, that was enough fun for a week 

[2deaf]

Found it! Thank you, so I do have the 10K resistor like on the MIG100 schematics, it's soldered the same way the rest of the resistors are on the board. Will mess around with the value and report back.
When you measured the range of the pot, you did it without tubes, am I right? 

Tubes were all in and powered.  I set the plate voltage to 513V with the bias at -55V.  I could smell the 5881's at the hottest setting, but they weren't red-plating.

The plates averaged 30mA at 525V with the bias at -55.6V.

[Sansteeth]

Tubes were all in and powered.  I set the plate voltage to 513V with the bias at -55V.  I could smell the 5881's at the hottest setting, but they weren't red-plating.

The plates averaged 30mA at 525V with the bias at -55.6V.

So I went straight to 1K on the first bias supply resistor (because I thought that's all I had in the 1k-10k range) and while the 10K was giving me a -34.8VDC to -52.7VDC, the 1K was giving me -54.3VDC to -68.2VDC.
Somehow I started getting wildly different plate dissipation on both sides (19W and 11W per tube), while I was getting pretty similar dissipation stock.
In the meantime I found a 4.7K that could give me a better range for my needs and dropped it in.
My measurements looked a lot better, and then I did something REALLY STUPID. I removed the probe from PIN 3, it arced with PIN 2 and the amp put on a little show for me. Tube in question looks fine, heater was still glowing normally after the sparks and the smoke. However, learnt my lesson.
I'm leaving for a month and will pop a new standby fuse in and look more into it then!

Thanks a lot for your help folks!

[HotBluePlates]

... what is it about the tube that makes them idle too high for a given grid voltage? ...

Current flows from the cathode to the plate, unless there is a counter-acting negative voltage at the grid repelling those electrons back towards the cathode.  So when two tubes pass differing currents when the same electrode voltages are applied, we could guess that either their cathodes are emitting differing numbers of electrons, or the distance from grid-to-cathode is different. Or the distance between adjacent turns of the grid-wire is different.

Ultimately, we're seeing the effects of manufacturing tolerance causing individual tube samples to perform somewhat differently.

I doubt the pair of 6L6WGBs have super-awesome cathodes resulting in double-plate-current, so my guess is the position of the screen and/or control grids don't conform to "normal 6L6WGB tubes."

... tube is "too" efficient in translating a change in grid voltage into a change in plate current? That doesn't sound right because idle is a point on the curve, not a change over a range. What am I missing here? ...

You're thinking "point on a (data sheet) curve" while I'm talking about tube samples whose curves are entirely different (due to manufacturing slop).