Plate current and red glow

[dscottguitars]

Hi,  I'm having red glow in my 6L6 EH tubes, new just put in.  I had the same issue with some used GE's and thought the new tubes would solve that.  I am a novice and have a few questions about this I'm hoping someone can enlighten my knowledge a bit.

First the specs:  in fixed bias mode-

It's a home build similar to the Fender Twin, four 6L6's.

Plate voltage is around 398v
Screen voltage is around 390v
bias voltage: -26v on one side and -28 on the other
plate current: 45mA on one side and 40mA on the other; per tube.  Meter reads 90 and 80 respectively.

1.  The amp's hum is much louder if I have both sides with equal current.  At this difference, whether more or less current the hum is at it's minimum.  Is this normal?  The tubes were advertised to be a matched quartet from Yen audio, bought on ebay.

2.  The plate dissipation calculates to: 17.9% on the hotter side.  That seems well within the limits I read about even close to the 'medium' side from Weber's Bias calculator.  Why are the plates a bit red?  Everything I read states shutdown immediately.  The glow is very slight, in the corner of the fold on one side.  One tube is a bit redder than the other.  I've tried switching tubes around but get the same thing.

3.  Is that slight redness a problem considering the plate dissipation not being so high?

Thanks....

Daniel

Below is the schematic:  It is just the power amp.  No preamp-I'm building that separately because this chassis was too small for everything.

Also, the high/low power switch has a heat issue so I'm changing that, but I only use the high setting.

[jojokeo]

Can't open the file you attached.

[dscottguitars]

Sorry, I saved it in the wrong format...

[jojokeo]

That schematic looks good but doesn't as far as accuracy is concerned and it's a bit confusing too. You list Bias and B+ on same circuit area along w/ (360vac)? w/ an improperly facing diode as it appears it's supposed to be your Neg bias? Then the PPIMV, I'm scratching my head here too. Is that what it is supposed to be? Your signal is supposed to run into a 250k dual pot and out through the wipers to your grid stoppers/power tubes, not grounding the signal like that through resistors and a switch. The other side of the pots go to neg bias which I think is what you mean by the small "b". Do you have these things drawn wrong from the actual layout? What's the point of the switch and resistors? I'd get rid of all the switching stuff. This area is likely your biasing issue?

[HotBluePlates]

So much wrong here.

Hi,  I'm having red glow in my 6L6 EH tubes ...  I had the same issue with some used GE's ...

If the glow is on the plate and not from the filament, you have too much current caused by too-little bias voltage. The tube swap seems to confirm it's not a tube-thing, it's a bias-thing.

bias voltage: -26v on one side and -28 on the other

For a typical Twin Reverb, that's not enough. Your B+ is lower than a typical Twin, but the range of adjustment is too small, and centered at not-enough negative voltage. I'd like to see an adjustment range around -25 to -40v, although you probably don't need it to be that wide. I'm guessing you need to be around -30 to -35v on pin 5.

The plate dissipation calculates to: 17.9% on the hotter side.  ...

3.  Is that slight redness a problem considering the plate dissipation not being so high?

That's 17.9w, not 17.9%. If you have old-school 19w 6L6, that's 94%; if you have a true 6L6GC, that's 59.7%. Modern tubes might vary between these limits.

There is an error in your measurement/calculation somewhere. The tube doesn't lie; if it's red at all, it's too hot. You can usually go over 100% dissipation and still not have red-plating.

plate current: 45mA on one side and 40mA on the other; per tube.  Meter reads 90 and 80 respectively.
...  One tube is a bit redder than the other.

That seems to imply a common resistor from which you're measuring the current of two tubes. If you consider it for a moment, 80mA total might be 40/40 or 70/10 or 80/0 or any such combination. Don't assume they divide the current equally. That one is more-red than the other is proof, as it is more-hot.

...  The tubes were advertised to be a matched quartet from Yen audio, bought on ebay. ...

I don't trust anything a seller says on ebay, unless I can verify it for myself and have a return policy. Even then, you never know if they'll live up to it.

Even assuming "ultimate matching" from a dealer of the utmost integrity, it's nearly impossible to get a pair or quad of tubes that are exactly the same in every characteristic at all times and under all conditions. It just doesn't happen. So assume there's slop.

So either there are problems in your bias circuit as-built, or not enough bias range, or both.

Don't bother doing hum-chasing until you resolve the red-plating issue. Red-plating tubes typically overburden the power supply and cause much hum at the output.

[stratele52]

+1 HotBluesPlates


redscottguitar you do some mistake somewhere for sure

[HotBluePlates]

redscottguitar you do some mistake somewhere for sure

Note that mistakes aren't bad, but an opportunity for learning.

My last amp build was a full-power oscillator at first power-up. There's no negative feedback around the output transformer, so it wasn't a matter of swapping output tube plate wires. I just made a dumb mistake and had 1 wire in the wrong spot. Took about an hour to shake off the frustration, pour over the layout and the built amp, and find the wire in the wrong spot.

I usually do at least one silly thing in every build. The more complicated the circuit, the more places I might do something silly...

[dscottguitars]

"You list Bias and B+ on same circuit area along w/ (360vac)? w/ an improperly facing diode as it appears it's supposed to be your Neg bias?"

I have the bias voltage coming off of the secondary winding from the power transformer.  The diode is backwards because that's the way I copied the schematic.  Is that not how the voltage gets to be negative?  There is nothing wrong with the drawing or the way it is wired, I'm pretty sure of that, because I've use it before.

What is PPIMV??  The two pots I use to adjust the bias voltage are 25K and are correct.  The bias circuit is switchable from cathode to fixed.  And the two different tail/load resistors are for the high/low power switch.  Otherwise the plate current would be really low.

"so much wrong here"

I meant 17.9W dissipation, not sure why I used percent, other than not going over 70% or 21W.  But, at the current measured, high side, divided by two that's 45mA per tube times say 400v and that calculates to 18W dissipation, correct?

If the amount of current is dependent on the bias voltage, then why should it be at -30v or more?  That would be less current and I am trying to get to 19W dissipation.  I agree it is less than other schematics show, but the current is what I am adjusting for. 

I measure the current by the shunt method, not the cathode resistor method.  I have no problem doing this and it is listed in Hoffman's how to bias link as the method used my many professionals.  I switch the meter to 200mA and touch the black probe to the plate pin #3 and the red probe to the B+ supply.  I get the reading of 90mA on the one side and according to the instructions: divide by two.  I don't see anything wrong with my procedure.  But I do agree the plate voltage is a bit low, but maybe that's the transformer.  I get 460 volts on the main filter caps before the standby switch is engaged and then about 400 with it on.

I am quite confident the meter is not lying to me either.  I can raise the bias voltage to about -35v but that lowers the current to around 70mA and the plates are still red.  The plate voltage goes up too.

What do you mean the range is too small??  I did not list that in OP, only that one side is -26v and the other-to balance the hum-is -28v.  The pots adjust from -22v to -35v for the bias voltage.

How can one know if the two tubes are measuring equally or not?  You say it could be 40/40 or up to 70/10.  If there's another way please let me know.  Would the imbalance be due to unmatched tubes?  They are brand new Electro Harmonix 6L6 GC.  I have a tube tester that has a meter 0-50 but did not test them. 

Could there be anything with the screen voltage and screen current?  I have not measured it, but will after I'm done here.  About the only thing I don't have that the Twin has is a choke.  Is that an issue?  I have the 470 ohm resistor to get the screen voltage a little less that the plate voltage. 

Other things:  I have about 195v on one plate of the PI and 215v on the other, with about 375v on point C-the supply to the plates of the PI.  Usually I don't see that much difference and think that could be the imbalance issue with the current and hum.

I wasn't really hum chasing, only noticed it was less when the current on both sides were not equal.

I have been all over the layout of this amp.  I've rewired the rectifier switch and the high/low switch-had stupid problems that I fixed and still have the same redness issue.

Thanks again for all the input...

Daniel

[dscottguitars]

Quick update:

The screen current measures between 5.9mA and 6.7mA for the four tubes.

The 5.6K resistor in the power supply was reading very low so I replaced it with a 4.5K.  The voltage a point C is 385v and the plates on the PI are 230v and 233v.

The plate voltages for the 6L6's are about 404v, and screens are 400v now.

I pulled two of the 6L6's, the inner ones and the plate voltage went up to  420v for one and 421v for the other, the current is 47.1mA and 46.3mA respectively with -35.2v bias for each.  Screen current for each is roughly 4mA

That calculates to 19.78W and 19.49W dissipation and the plates are still slightly red.

Also, the wall voltage is 119.6v AC and the heater voltages is 3.37v AC each side-adjusted with the 100 ohm pot.

[stratele52]

dscott ,

Your schematic are not easy to read , électronical and mechanical symbols are mixed. But to me there is an error on bias wiring pot ; the pot wiper go to the ground ; this shouldn't . Look at your Bias circuit .

 Also sorry I  can't see how bias is connected to output tubes grids.

IMO your problem is there.

But your drawing look pretty nice I can't do that myself.

[stratele52]

A fender amp schematic could help you to see how to build bias circuit

http://www.ampwares.com/schematics/vibroverb_ab763.pdf

[dscottguitars]

symbols are mixed??  How so?

The pot wiper does not go to ground, it goes to the junction at the 1500 ohm resistors going to the grid through the 200k resistors.  This is standard as I understand.  

Fender amp schematic shows exactly what I have:  But, I have two pots, one for each side for balance.  I have a 25K pot vs a 10K, and a 12K load resistor vs a 27K.  25K +12K=37K and 10K+27K=37K.  Pretty close I'd say.  The only difference I can see is that bias supply comes off a direct bias supply voltage tap which I don't have.  

My schematic comes directly from Gerald Weber's book on how to convert a non adjustable bias supply to an adjustable one.  The resistors I chose are to get the voltage to what I need.

IMO, there is no problem with it...

[dscottguitars]

see legend for symbols...

[sluckey]

There's nothing wrong with your drawing. The non standard symbols are confusing some people. It might be clearer if the pot wipers (little arrows) actually touched the body of the pot (like in the legend you posted).

[stratele52]

I talk about the pot with the 68 MFD 100 volts . Wiper go the green arrow . This is ground no ?

I can't understand why one side of this pot go to output tube's cathodes. You mix ground's cathode with bias negative volts with this pot ,

I never see a bias circuit like that, I will be glad to see the Fender schematic build like that.

[sluckey]

I talk about the pot with the 68 MFD 100 volts . Wiper go the green arrow . This is ground no ?

That ain't a pot. It's a switch.

[dscottguitars]

Sluckey: I thought that would be implied.

Stratele: It is a switchable bias circuit.  From fixed to cathode bias.  That is what I have there.  I didn't label it as a SPDT switch though.  I thought it would be obvious with the DPDT ones that are labeled.

Now, back to the OP, redness on the plates??  Any other thoughts?

[stratele52]

Stratele: It is a switchable bias circuit.  From fixed to cathode bias.  That is what I have there.  I didn't label it as a SPDT switch though.  I thought it would be obvious with the DPDT ones that are labeled.

Like I write before ; " Your schematic are not easy to read , électronical and mechanical symbols are mixed. "

[dscottguitars]

I hope this is this better...

[dscottguitars]

Wow!!!??? What a great use of help this forum has been....

With that ugly criticism from jojokeo I cannot believe this schematic is that difficult. 

This amp has:

Switchable recitifer
Switchable bias
Switchable power

Why??  Why not??

The schematic is just a representation of what I have, not an exact draftsman copy intended to be 100%.  I tried to eliminate wording and such to be quick, but, no that's not good enough.  If a switch is labeled as A and B they are the same switch, just drawn in different places for ease and compactness.

If a switch has no A, number, etc. it is independent and alone. 

And, the red plate problem still persists...

[dscottguitars]

new drawing...

[jojokeo]

Hey Scott, I was joking w/ you and you took wrong. Get over it.

[dscottguitars]

Some joke,  if you can't help, then shutup.

[HotBluePlates]

... the current measured, high side, divided by two that's 45mA per tube times say 400v and that calculates to 18W dissipation, correct?

If the amount of current is dependent on the bias voltage, then why should it be at -30v or more?  That would be less current and I am trying to get to 19W dissipation.  I agree it is less than other schematics show, but the current is what I am adjusting for.  ...

You're using OT shunt method; if there's more than 1 tube per push/pull side, you can't know how the current is split among the tubes. You can only know this by a method that allows the measuring of individual tube currents, such as using 1Ω or 10Ω cathode resistors.

Bottom-line: you're saying tubes are red-plating. If that's happening, the only way it can occur is too much current. If it's not happening, there is nothing to discuss further in this thread.

The reason for -30v bias (or more negative) is this:
Theoretically, if V_b = -(V_g2/mu_g1-g2), then the tube flows zero current. "V_b" is bias voltage, "V_g2" is screen voltage and "mu_g1-g2" is the mu of the tube when operated as a triode (that is, from the control grid to the screen grid).

Cheat: 6L6 mu is around 10, and you have about 400v on g2. In theory, -40v would bias the tube to cutoff. Really, it will be some small current (few-mA's) but not cutoff. You're using less than -30v now, and you describe red-plating. Therefore something between -30v and -40v should be ideal bias.

...  I can raise the bias voltage to about -35v but that lowers the current to around 70mA and the plates are still red.  The plate voltage goes up too. ...

The rising plate voltage is perfectly normal, and what always happens as you reduce plate current. That the plates are still red with the reduced current again points to either a big problem, or something which is not actually redplating.

You can't have it both ways; either tubes are redplating and must have reduce plate current, or there's nothing wrong.

What do you mean the range is too small??  I did not list that in OP, only that one side is -26v and the other-to balance the hum-is -28v.  The pots adjust from -22v to -35v for the bias voltage.

Sorry, I misread that as voltages when the bias pot is at either extreme.

-22v to -35v is quite low for a Twin-style amp.

I have been all over the layout of this amp.  I've rewired ... and still have the same redness issue.

I'm a broken record: only too much plate current causes red-plating. That's either not enough bias, or some fault (such as a leaking coupling cap or bias supply fault) which has a net result of too little bias voltage.

[jojokeo]

Some joke,  if you can't help, then shutup.

With an attitude like that...you're on your own. You already have shown your character which is why I joked with you. But you can't take it and it shows how small minded and insecure you are.

[punkykatt]

Now that the switches on the schematic are drawn in it makes the schematic much easier to read.

[sluckey]

And, the red plate problem still persists...

Increase the negative bias voltage or change tubes.

[dscottguitars]

I am confused still...

I put the bias voltage at -35.2v, that's the most at this point unless I change the bias load resistor which I can do.  I am getting a reading of 76mA per side and 408v on the plates, 404v on the screens.  Red plates still, on all four tubes.  It is a small spot, about one inch long in one fold of each tube. and I really need the light off to see it.  It's not so bright with the light on, but still there.

I'm not sure how to check if the capacitor is leaking.  There is 223v on one side and the -35v on the other.  I've checked for AC and there is none.

I disconnected the rectifier switch for now as it's too close to the PI and there is virtually no hum and no buzz at all.  It's really quiet.  And that's with the negative feedback pot at it's extreme, putting very little back into the PI. 

If I just increase the bias voltage the amp will be running at low plate dissipation.  It calculates to less than 16W with the above parameters.  I don't see that as a solution.  I have also changed tubes.  These are brand new Electro Harmonix, and have pretty good reviews.  The others were used GE's and the plates of those were the same.

I did pull two tubes mentioned earlier and measured 44mA with 420v on the plates.  That's still only 18.5W dissipation and should not red plate.

Could it be the red I am seeing is not really a problem?  It is definitely on the plate.  I'll try to get a picture of it and post that...

[sluckey]

I put the bias voltage at -35.2v...   Red plates still, on all four tubes.

Change the resistor so you can set the bias higher, something like -40 or even -45. Do the red plates go away?

[dscottguitars]

here are a couple of pics with the settings at 389v: plates, 95.2mA current, and -28v bias.

The flash doesn't show the redness at all...

It's redder at this setting.

[sluckey]

It's redder at this setting.

Yes, that's red plating. Increase the bias voltage.

[dscottguitars]

I increased the bias voltage to -40.1 and no red plating.  But, the current is only 56mA at 408v on the plates.  That's 11.4W dissipation.

Any thoughts??

[dscottguitars]

I do have two 1 ohm resistors that measure 1.5-1.6 ohms on my meter.  But I'm not sure exactly what to do.  If I were to put one on each side going from the cathode to ground I could verify my ammeter. If I do that I would measure the voltage and multiply by 1.5 and then divide by 2 for the two tubes, correct?  Or I could just put them both on one side and measure each tube's cathode voltage and multiply by 1.5 ohms.  Would one way be better than the other?

Thanks...

[HotBluePlates]

Use one resistor per tube. That's the only way for sure to know how much current each is passing.

If your 1Ω resistors are 10%, assume that they read no more than 1.1Ω even if your meter says 1.5-1.6Ω.

Why? Most meters include the resistance of the meter leads when measuring a resistance that low. Touch the leads together, and you'll probably get 0.2-0.4Ω. If your meter has a "relative" setting, you can null out that lead resistance before measuring the 1Ω resistors.

And I think you meant to say "divide" the voltage by the resistance to obtain the current measurement.

In my opinion, it is worthwhile to assume the 1Ω resistors do indeed measure 1Ω and mentally convert mV's directly to mA's; any error means you have slightly less current actually flowing through the tube than indicated. There's almost no point in being more precise than about 5% with tubes anyway, as they will very certainly drift during use and over their lifetime by much more than that.

[dscottguitars]

I'll do that, and you're right, I meant divide.  Thanks...

[dscottguitars]

HUGE DIFFERENCE!!

Using the 1 ohm resistors, assuming 1.1 ohms shows 54.8mV and 53.7mV on the tubes.  That's 49.8mA!  @ 419V on the plates to calculate to 20.9W dissipation.  I would have never guessed the OT shunt method could be so far off.  I have a Velleman multi meter which I thought was a pretty good quality one.  I will use the cathode resistor method from now on...

thanks for all your help everyone...

[stratele52]

HUGE DIFFERENCE!!

Using the 1 ohm resistors, assuming 1.1 ohms shows 54.8mV and 53.7mV on the tubes.  That's 49.8mA!  @ 419V on the plates to calculate to 20.9W dissipation.  I would have never guessed the OT shunt method could be so far off.  I have a Velleman multi meter which I thought was a pretty good quality one.  I will use the cathode resistor method from now on...

thanks for all your help everyone...

Congratulation.
A good bias is needed  ; that's everybody write since the beginning.

OT shunt method is not as good as cathode 1 ohms resitor. Meter impedance and quality could do a false reading on OT shunt method. And it is a very dangerous and lethal test .

[HotBluePlates]

Maybe just the current ranges of your meter need calibrating; it sounds like the voltage ranges are tracking more accurately with what we see the tubes doing.

My point with my earlier posts was that for whatever reason, your meter readings didn't agree with the reality of redplating tubes. That, and an unfortunately reality of using the OT Shunt method with a 4 output tube amp is that you don't know what each individual tube is doing.

So now, if you use a 1Ω resistor per tube, you will measure cathode current. Yes, this is higher than plate current by the amount of screen current. You can calculate true plate current if needed by measuring the resistance of the screen resistors, measuring the voltage drop across the screen resistors, and dividing the voltage by the resistance. Screen current at idle will be pretty small.

I was getting very frustrated that I couldn't get you to trust what you saw the tubes doing over the readings you were getting. Sometimes we have to question our equipment or ourselves if readings don't jive with physical evidence.

[stratele52]

  I was getting very frustrated that I couldn't get you to trust what you saw the tubes doing over the readings you were getting. Sometimes we have to question our equipment or ourselves if readings don't jive with physical evidence.

+1000

[dscottguitars]

I do understand, and I was frustrated too.  But you never mentioned that my meter could be off, only the bias voltage needed increasing.  So, with my limited knowledge, I didn't see that as a solution because my meter would say that the current was too low.  I also was not too sure that the redness was really current generated because it really only showed in the dark.  However, I won't be mistaken by that again.  Most of my knowledge comes from reading, and I have little experience.  Now I have a bit more experience, thanks to everyone's help...

Thanks again,

[stratele52]

Most important ; you learn . Now you are a bias graduate ; congrat  ; )

IMO shunt method is very dangerous and do not show each tube .  Bias probe or a bunch of 1 ohms 1% resitors are the best tool.

[Willabe]

Most important ; you learn . Now you are a bias graduate ; congrat  ; )

IMO shunt method is very dangerous and do not show each tube .  Bias probe or a bunch of 1 ohms 1% resitors are the best tool.

Yep to both things stratele said.


            Brad     

[Jack1962]

Shunt Method works for me 

[HotBluePlates]

The OT Shunt method works, but it doesn't tell the whole story on a 4x output tube amp.

It's also very easy to have a slip or forget to change a meter setting, which will damage your amp or meter near-instantly.

I've used it, but prefer safer methods now.

[stratele52]

The OT Shunt method works, but it doesn't tell the whole story on a 4x output tube amp.

It's also very easy to have a slip or forget to change a meter setting, which will damage your amp or meter near-instantly.

I've used it, but prefer safer methods now.

+1000