AIMS 6550 red-plating question

[ilyaa]

is any red-plating bad? always?

this amp sounds great but when its cranked that evil red glow starts to come out a bit from deep within one of the tubes.....

tubes are brand new!

the story: when i got it the amp had EL34s - the stock is 6550s, so the owner asked me to return it to stock. i did - not much had been changed by whoever put the EL34's in, except they had changed the 47K resistor in the bias network to a 23K and a 10K trimpot, and they had 470 ohm CONTROL grid resistors (??) on both tubes -

(also the tremolo wasnt working but i figured that out and its not relevant to my question here)

i returned things to stock (i took off those 470 grid resistors.....whats their deal, anyway?) and put in 6550s - with a 47K resistor in the bias network they were running like 80+ mA's each though, and the B+ is like 550V (!!) so i tamed them down to about 55mA (by putting a 68K resistor in series with a 20K trimpot instead of the 47K resistor in the bias network) each so like 30W each - that should be fine for 6550s, riight?

BUT i saw some red-plating - noticed there was a large-ish 20Vp-p ripple on the B+ so i replaced the first filter caps but the ripple is still there....hmm?? could that cause the red-plating?

someone added 220k bleeder resistors to the big caps, too, but i dont think those are related......

any thoughts?

oh yeah: https://www.flickr.com/photos/73628360@N00/14628046201/

[Enzo]

30w a tube at idle sounds like a lot to me.  If it were mine, I'd probably back off from that.

Have you checked current on each tube individually to see if the one red plating is drawing more current than the others?  REmember, just because a tube is new doesn't mean it can't be bad.

There is a good chance your red plating is causing the B+ ripple, not the other way around.

[LateDev]

 The screen grid resistors need lowering from 1K to 470R.
6550 has a higher grid leakage current than EL34

6550 needs more negative bias than the EL34 therefore the typical higher values for the resistors should be lowered to around 47K.

Only reading from old notes on Marshal mods so bias resistor may not be quite correct value.

[g1]

  The control grid stoppers are probably not a bad idea.  Marshall's with 6550's used 1K5's, same as Fender uses in most tube amps.

LateDev:  why change the 1K screen R's?  That is the stock value for this amp with the stock 6550's, also used by Marshall's with 6550's.
The EL34's in this amp weren't stock, the schematic is for 6550's.
And which resistors did you mean to lower to 47K ?

[ilyaa]

the 47k im referring to is the one in parallel with the 10k biasing the grid -

i increased that resistance to about 89k or so to decrease idle current -

i havent checked the coupling caps off the PI for leakage - that might cause red plating, right? if that DC voltage sneaks onto the grids?

[LateDev]

  The control grid stoppers are probably not a bad idea.  Marshall's with 6550's used 1K5's, same as Fender uses in most tube amps.

LateDev:  why change the 1K screen R's?  That is the stock value for this amp with the stock 6550's, also used by Marshall's with 6550's.
The EL34's in this amp weren't stock, the schematic is for 6550's.
And which resistors did you mean to lower to 47K ?

Sorry I was just going by the Marshall cct diagrams they use either 75R or 470R depending on the model. The 47K was for the bias as is shown in the diagram.
The screen grid current is a lot higher on the 6550 which is why the resistor value was lowered from 1K5 USA which was for the EL34. If however you have more up to date data, (mine was 1990 ish) I will have to go by what you say. As we know manufacturers do modify their ideas over time.

If the circuit is stock for this model with 6550's then it should work as designed, so why change the bias ?

All things being equal and if the circuit diagram is correct, then you are indeed looking at a possible fault.
Leaky caps were common on many older valve amps, however I would check out the bias chain first. Think that is a Zener which really should be checked first.

[g1]

  From your description, the red-plating is not occurring at idle, but only when pushed hard into load, (assuming full power or clipping).
  In these cases, it can often be cured by decreasing idle current.  I agree with Enzo, is is still idling a bit hot and could be reduced.
  Idle current can often be reduced below 60% with no negative impact on tone.  Many amps are biased around 50%, so I usually suggest biasing as cold as possible without having a negative impact on tone.  This will increase tube life.
  Another reason amps sometimes red-plate at high volumes is load mismatch, where the load impedance is lower than what the power tubes would like to see.
  What load impedance is the amp supposed to have, and what are you using for speakers?

[ilyaa]

yeah, its only red-plating when driven really hard -

i reduced idle current even further (about 50mA per tube - at 550V) and the red-plating seems to have abated.

loads are matched - the PI coupling caps were okay but i replaced them anyway -

i think it was just running the tubes too hot - we'll see what happens once its tested for a while under real-world conditions -

[LateDev]

What are the Voltages around the tube.
Screen Grid
Grid
Anode

Quiescent current of 50mA @ 550V
Can you double check this ?

Did a search for an earlier Marshall for comparison, which may help.

http://www.drtube.com/schematics/marshall/1987mk2u.gif

Old chart for 6550 quiescent/idle current, for a given Voltage of 550V.
6550 SED Min 32mA Max 45mA
6550 EH Min 38mA Max 53mA

Hope info helps

 

[g1]

  If biasing at that level has taken care of the red-plating, then you are probably ok.
But I would caution that you may have a higher dissipation (%) than you think. 
Old production 6550's were rated for 42W, some new production ones are only rated for 35W, depending on brand.  Check the specs for the brand you are using.

[DrGonz78]

Are you also checking your plate voltage after adjusting the bias? I just noticed that at 55ma and 50ma of your measurements that plate voltage was still just at 550v. Or at least that is what was posted.