Carvin X-50B with glowing second grids of EL34

[kvimbo]

Hello,
I have a problem with the service of a Carvin X-50B.
The second grids of the power output tubes (EL34) are turning red to orange when the amp is going into overload, never mind from the clean or lead channel. This leads to early failure of the tubes.
The real problem is, that everything seems to be OK.
The output transformer is OK, as seen from the graphic, representing the Raa.

The Anode idle current is correct as the producer defines it (50mA), and a little less, for more distortion (cool biased).
The power supply voltages seem to be in normal condition - Anode (Va) 448v, G2 440v.
The G2 resistors are OK, 350R each G2 and common 350R from the Va.
The Anode currents at full power are 144mA each, average measurement, and the common G2 current is 115mA (2x57.5).
In this situation the G2 heats to red and more, which according to essential knowlege for the right work of the tubes is unacceptable.
The input signals to G1 are normal.
Tried even some weird assumption, if the G3 (suppressor) is floating, due to some connection loose, and the dynatronic effect to take place, but nothing of the kind.
So, my questions are directed to the respected qualified and experienced persons in the forum.
-Is this situation normal for this amplifier (which is very doubtful for Carvin)?
-Does anyone has experience with such problem, not only with this kind of manufacturer?
-What am I not confirming, that can be the reason for this problem?
Thank you in advance!
Tony Manov, M.Sc. AES member

[Jazz P Bass]

350/ 5 watt sounds skimpy.
You could try to bump them up to 1K/ 5 watt.

[Steve Conner]

There is a long tradition of abusing EL34s this way, ever since the original Marshall Plexi. By overdriving the screen grids you can get lots more power out of the tubes than they were designed for, and I'm sure some people claim that you have to do it to get the Marshall tone.

I think it's normal for the screen to glow a little red, but orange, yellow or white is bad.

A more conservative design would have 1k or 1.5k screen resistors. No good having the famous Marshall roar if it explodes halfway through the gig.

[kvimbo]

According to the original schematic R114, R115 and R121 are 350R/5watt.
Of course, I can change a lot of components, but the engineers of Carvin had used just these values.
If there is a mistake due to them, than I want to hear it.
That's why I asked "Is this normal for Carvin" or it is due to any malfunction of the original components.
I also don't agree with the condition of the power stage, as it doesn't correspond to the suggested by the valves manufacturers. But this is what Carvin made, and if it is not technically correct, I must know why.
The guitar amplifiers are crazy creatures, where the sound is leading, not the engineering. And if they had made easily breaking amplifier with great sound, then nothing is to be made! In this case I'll suggest the owner to stocks with a bag of EL34...

[Steve Conner]

I think it is "normal for Carvin" but I have no proof of this. I know that using screen resistors smaller than 1 to 1.5k with EL34s leads to glowing screens, having seen it with my own eyes, but I don't have any data on what this does to reliability of the tubes. In other words, maybe the screens were designed to glow.

Another possibility would be to change to 6L6GC, 6550 or KT88 type tubes, all of which draw less screen current, being aligned grid tubes. Legend has it that Hendrix had to have his Marshalls converted to 6550s to make them last the length of a set.

[kvimbo]

When metal heats, a gas is emitted, and the vacuum inside the glass is impaired. This leads to uncontrollable current flows as a result of the ionization and damages follow. So it is absolutely unfortunate to leave any of the electrodes in the tube to glow...
One set of tubes gave short circuit with sparks inside just in front of me. This is absolutely unacceptable situation - Every moment the amp can blow in fire...
Well, I think to change the supply to G2, with Z diode, appropriate for the purpose, lets say 60-70V/150-200mA in series with the common R121. In such way, the power dissipation at G2 maybe can be reduced to suggested values.
But! This will change the kind of the sound that the owner got used to, which can breed discontent...
I'll be patient to hear more information.

[pdf64]

Schematic http://www.webphix.com/schematic%20h...n_x_series.pdf
Using EL34 from a different manufacturer may help, as some cope better than others with screen grid overload. What have you tried so far? Winged C are well thought of.
However, as advised, the 350ohm R114 & R115 resistor value is lower than normal used (1k) and increasing it may resolve the issue. Pete.

[kvimbo]

[QUOTE... What have you tried so far?

Winged C are well thought of.

[/QUOTE]
Russian Electro-Harmonix/Sovtek and German RFT NOS.

Sorry, what are Winged C? (I am not familiar with all bywords)

[Steve Conner]

Winged C Home Page

They're the tubes that were called Svetlana, before the Svetlana brand was bought by New Sensor, at which point they became rebranded Sovteks and the old Svetlana had to rename themselves to Winged C.

[kvimbo]

Thanks! We continue to call them Svetlana here.

[kvimbo]

Well, the interest to the topic decreased, so I will tell my steps to the decision, after doing some tests.
First, I placed in series with the resistors to G2 Z diodes, to decrease the voltage to approximately 375 volts, as the tube manufacturers recommend for the selected anode voltage and Raa. The grids stopped heating, the maximum peak to peak voltage at the load decreased from 52 to 49 volts, which is acceptable. The arising problem is, that the bias has to be changed, because the decreased Vg2 decreases Ia (a lot of other parameters too) and a lot of crossover distortion appears. This was not acceptable by my point of view.
Second, I changed all the resistors in the supply of G2, as follows - from 2x350R/5W to 2 x 1k/5W for the G2s, and the common 350R/10W to 1k/ 17W. In this case, the problem with the crossover distortion was not available, to the point of clipping the output signal. It is so, because of the dynamic behavior of the voltage at G2s. So far so well. The heating of the grids is not so well removed as in the case with the Z diodes, but is far from the problematic one at the beginning. Now a dark red to little light red color is noticed in case of heavy limiting at the power stage. At lower levels at the output (with the same high drive from the lead channel) , the grids are absolutely dark.
This solution in my opinion is preferable. Not ideal, but more secure to the output valves, and doesn't need any change of the bias control.

Thanks to all, who participated in the topic!

[kg]

kv: something else that can greatly reduce the beating the screen grids take is to reduce the load impedance.

you can do this by "mismatching" the output transformer tap to the speaker load. ie, run a 4 ohm load on the 8 ohm tap.

[kvimbo]

Yes, if the load to the tubes is smaller (as proposed - 4 at 8 or 8 at 16) the anode potential will be higher at the low side in full power, and the G2 will take less current. But obviously this was not enough. I thought about this, that's why I measured the impedance of the output transformer, in a hope, that there is problem in it. The amp has 3 positions - 4, 8, 16 ohm, and I checked also how the change of the load manages the heating of the grids, and the regime of the valves. But the try didn't solve the problem. At least not enough.

[MarkusBass]

Kvimbo,

On the schematic there is a little note: "SELECT R127 FOR IDLE CURRENT OF 100mA, (50mA FOR X-50B AND X-60).". Have you checked this?

Mark

[kvimbo]

Of course, I did.