New Production 6550 - any thoughts on reliability?

[talbany]

It would be wise to run the load lines on those 6550's and make sure your not over dissipating the screens..They have a very wimpy screen supply and are really designed to be run lower than the plates..UL (regulated)..300 might be pushing your luck!

Good Luck
Tony

[Milkmansound]

Easy solution - switch to KT88, I guess

[matt h]

(deleted)

[Milkmansound]

It would be wise to run the load lines on those 6550's and make sure your not over dissipating the screens..They have a very wimpy screen supply and are really designed to be run lower than the plates..UL (regulated)..300 might be pushing your luck!

Good Luck
Tony

The EH data sheet:

http://www.thetubestore.com/lib/thetube ... 6550eh.pdf

I'm definitely within those specs.

[matt h]

(deleted)

[martin manning]

KT88s also have wimpy screens compared to EL34. They (the screens) do crazy things as drive voltage approaches the bias voltage.

edit: the issue isn't that the screens can't handle more than 300V, it's that to ensure the average dissipation doesn't exceed rated spec involves biasing the tubes colder than other guitar amps traditionally do. If you're not running UL, this involves either monstrous Rg2's, or a lower voltage ps node for the screens. At least for KT88's. I imagine 6550s are quite similar, as the two are often (not always) repackage versions of the other.

The EH 6550 data sheet shows similar ratings to the EH KT88 data sheet, but the curves are clearly different- less kinky at low Va and Vg1. The funny thing is that the EH 6550 sheet doesn't say what screen voltage goes with the anode curves. Notice too that the max grid resistance in fixed bias is only 50k (grid stop plus bias feed).
In fixed bias, the idle bias has little to do with the screen dissipation since it is far below the limit at idle conditions. In cathode bias, Va-k and Vg2-k sag together with current, so again I don't think there is much effect. What does matter is where the load line is with respect to the knee in the Vg1=0 curve when the plate voltage is pulled down (at high power). I agree completely re series resistance in the screen supply. It can be used to reposition the grid curves and limit screen current peaks.

[statorvane]

I agree completely re series resistance in the screen supply

Yes. Take a look at the AX84 SEL The screen voltage supply is further filtered and dropped quite a bit after the B+2 node. This is a cathode biased SE amplifier; max grid leak ~470K. I am using this as a power amp to a 4-hole Marshall preamp and tonestack / cathode follower back end. Power tube screens are around 285 volts, plates around 335. For ~20 watts this thing roars. The Tung-Sol data sheet suggests max screen voltage of 400 volts.

[talbany]

"knee"
200 mA | ___________________ Ip @ 0 bias
|* /
| * /
| *
| / *
| / * <-AC load line
100 mA | / + *
|/ + *
| + *
| + *
| + _ +__________* ____ Ig2 @ 0 bias
5 mA | + _____________* _______ Ig2 @ quiescent operating point
+----------------------*-----------------
0V 500V
Plate Voltage


Minimum instantaneous anode voltage is usually not allowed to
swing as low as the dc screen potential, which where screen dissipation
tends to become abnormally high...

This (crude)drawing shows graphs for plate and screen current (KT-88) when the grid is at zero bias and for screen current at your quiescent (no-signal) operating point. It also shows an AC load line that intersects the plate grid curve below the knee. Look at what happens to the zero-bias screen current at plate voltages below the knee: it starts to spike. So when your plate voltage is at a low value on that load line (because the grid signal has swung all the way up near zero volts), the screen current is very high. When the plate voltage rises and passes the "knee" of the curve (such as happens when the control grid goes more negative), the screen current falls off to very close to your quiescent screen current.

It is that spiking of screen current that you want to manage and why the average screen current is more important than the quiescent screen current.In cathode bias (like Martin mentions) Va-k and Vg2-k sag together so the curves will look much different..
Some of the older GE 6550's (like what was loaded in the early 80's Marshalls) were a rugged tube and had some sag so they could handle those spikes..with new tube's all bets are OFF!!
Good Luck!
Tony

[Rat6v6]

I built a 100 watt (2x6550) bass amp. I am running 680 volts on the plates and 430 volts on the screens. I used 1k5 5w screen resistors.

I have been using the amp weekly (left at practice space) for 3-4 hour session practices with no problem. 6 months and going strong.

I am using the Tungsol RI 6550 (Sovtek).

I did bias the tubes for 55-60 % max dis.

[martin manning]

I built a 100 watt (2x6550) bass amp. I am running 680 volts on the plates and 430 volts on the screens. I used 1k5 5w screen resistors.

I have been using the amp weekly (left at practice space) for 3-4 hour session practices with no problem. 6 months and going strong.

I am using the Tungsol RI 6550 (Sovtek).

I did bias the tubes for 55-60 % max dis.

Rat, sounds very interesting. What primary impedance are you running? Can you share the schematic for your bass amp?