bottoming

[skyboltone]

With a 470R screen resistor in a 6L6gc Dumble build, when the amp moves into full feedback and sustain the screens glow like a toaster heating element while the plates remain grey.

Might be a design oversight where you are cutting the Vg1=0 characteristic well below the knee. Could you pass along the B+ value and output transformer primary Z?

This would be at 450Ebb and 4.5K Plate to plate load.

[skyboltone]

Thanks for this Andy. With a 470R screen resistor in a 6L6gc Dumble build, when the amp moves into full feedback and sustain the screens glow like a toaster heating element while the plates remain grey. This is the main reason I'd like to have a bias rite. I'd like to know just how much over on screen dissipation the tubes are running.

Yes Dan, this is one of the great mysteries, I've been fishing a few times in the forum on how to generate screen current curves from general information presented in data sheets. Ocassionaly tube curve data sheets include screen current curves but unfortunately it is rarely included, and never for situations where screens are run outside typical operating potentials. We just get given a few max screen current figures for specific conditions.I guess it didn't occur to engineers of the period that there would ever be such an interest in conditions boardering on what they would term as component abuse. I think I'll end up approaching Kuehnel or someone similar for this information the information/proceedures I seek, if they even exist.

With less resistance at some point the tube changes over to triode mode and with more resistance the amp should have more headroom and screen dissipation should drop; along with compression.

6L6GC has a screen dissipation maximum of 5 watts
6CA7/EL-34 can handle a max of 8 watts.

The plate dissipation ratings are opposite. The 6L6 is a 30 watt plate while the EL-34 is a 25 watt plate.

Said another way, and ignoring overall supply sag, plate and screen supplies that are independant of plate potential are in pentode mode, ranging through UL where screen tracks a % of instantaneous plate potential to triode where the electrodes are strapped.

Well that's what I was trying to say only you said it better. In modulator driver amps, for instance, there was a separate winding on the PT for a separate supply to the screens. When the plates sagged due to transformer burden, (not series resistance) the relative potentials between the two elements remained more or less the same as both shared the same core. But in other cases, with larger Pentodes such as the 813, in CCS the plates ran at 2200 while the screens ran at 1100. Completely separatte supplys. What I was driving at is when the potentials of the plate and screen are very near, but not strapped, the screen starts acting like a plate, drawing more current than it would by the book. I don't know what constitutes too close, (or not close enough) but I know that 470 ohms following a 50ohm or so choke brings out a lot of color in the screens. We hear an aweful lot about tube failure around here. And we do run a lot of screen current.

[Ears]

Actually I should have left the words "plate and" out of the sentence

"Said another way, and ignoring overall supply sag, plate and screen supplies that are independant of ..."

so I didn't really put it so clearly after all.

Guitar amps are probably the biggest end use of current production receiving tubes, I don't know the comparative figures for Hi Fi end use.

Maybe it's about time we pressured manufacturers to provide meaningful data on screen operation this area of operation instead of relying on decades old data intended for hi fi or upon in-house bench tests, which is a avenue not always easily available for hobbyists or small operations.

I bet the manufacturers have this data.

[Andy Le Blanc]

I second the need for more complete family curve for current production tubes.

a screen grid is tied to the cathode dynamically, no change in poteintial
which makes it a shield between the grid and plate.
the screen grid is also kept statically at a positive potential to make the
normal flow of current flow through the tube possible.
The shielding action makes the current flow (cathode) independant of plate voltage.
plate current is NOT independant of plate voltage. And if the plate potential is
not greater than the screen, the current divides between the two.
This is made even worse in the case of pentodes by the virtual cathode
or suppressor turning the tube instantaneously into a triode, with the screen
grid acting as a plate. This is the justifacation of running up the plate voltage
in high power amps.

[Andy Le Blanc]

If one were to measure screen current, DIY, would you measure across the
screen grid resistor? Would you insert a 1 ohm into the circuit?
Before or after the screen grid resistor? Would you need a special meter impedance?

[wscrane]

That operating point is similar to a lot of Fenders so the problem is really common. The screen resistors burn out when pushed. I put some guesses for the 450V Vg=0 curves (dashed lines) and your clipping loadline on the attached GE 6L6 data so you can see the problem when you go below the knee. There is no easy fix. Two possibilities are maybe to drop your screens 50-100V, or get a transformer thats 3800 or lower.

[skyboltone]

If one were to measure screen current, DIY, would you measure across the
screen grid resistor? Would you insert a 1 ohm into the circuit?
Before or after the screen grid resistor? Would you need a special meter impedance?

I think the voltage drop across the resistor is the easiest to (then add ohms law) do but Andy as you know, it's not just DC at this point so you need a very hige impedance measuring instrument here or it will stop the screens from working at all. It's a great place for a VTVM or scope with a 10X probe. No place in there for a DVM.

[Andy Le Blanc]

VTVM probes were tipped with with a 1M resistor that acted with the probes
cable as a R-C to drop the input capacitance to around 1p. This connected to
the top of the multiplier resistances that added up to 10M. From the selector
taps of the multipliers, was connected another 1M, and a cap to ground, to the
grid of a 12au7. One grid for the test, the other to a voltage reference with
the zero and calibration reostats. Both sides are pretty much cathode followers
with the meter between. Its a Wheatsone bridge, the measured DC makes an
imbalance, the current in the circuit is proportional to the value measured.
I have piece a scheme to a Simpson 303. It looks like an easy build.
Im sure theres more to it. The refferance I have says shielded 75uuf input cable. Ill try to post the scheme.

[Andy Le Blanc]

VTVM , Simpson 303

[skyboltone]

VTVM probes were tipped with with a 1M resistor that acted with the probes
cable as a R-C to drop the input capacitance to around 1p. This connected to
the top of the multiplier resistances that added up to 10M. From the selector
taps of the multipliers, was connected another 1M, and a cap to ground, to the
grid of a 12au7. One grid for the test, the other to a voltage reference with
the zero and calibration reostats. Both sides are pretty much cathode followers
with the meter between. Its a Wheatsone bridge, the measured DC makes an
imbalance, the current in the circuit is proportional to the value measured.
I have piece a scheme to a Simpson 303. It looks like an easy build.
Im sure theres more to it. The refferance I have says shielded 75uuf input cable. Ill try to post the scheme.

I got one of these. See that grey fat probe looking deal? Theres a tube in there. Direct connection to the grid. We don't need no cable screwing with our measurements.
http://cgi.ebay.com/Ham-Radio-HP-410B-V ... 7C294%3A50

[wscrane]

I'm all for measuing stuff, but I don't think the existing data is that bad. For one thing its clear that at clipping the screen is going to attract a significant fraction of the tube current. For another, if you have an amp like the toaster above you don't have to read too far between (or past) the lines to see how to get on a less stressful loadline. Exact numbers would be interesting but its already clear what has to change and what direction to go.

Getting data for new production tubes at higher voltages would definitely be cool. Although its hard to imagine Mike Matthews in a lab coat.

[skyboltone]

I'm all for measuing stuff, but I don't think the existing data is that bad. For one thing its clear that at clipping the screen is going to attract a significant fraction of the tube current. For another, if you have an amp like the toaster above you don't have to read too far between (or past) the lines to see how to get on a less stressful loadline. Exact numbers would be interesting but its already clear what has to change and what direction to go.

Getting data for new production tubes at higher voltages would definitely be cool. Although its hard to imagine Mike Matthews in a lab coat.

Granted. And that's why I advocate 1k to 1.2K screen resistors in the subject circuit. It helps. Tone changes, and so does touch sensitivity. Both for the better in my opinion. It depends very much on what you are asking the amp to do. The Jazz fusion guys are hardly pushing this envelope. Going to 3.3K primary OT is a posibility with my amp. It also will allow 6.6K use with KT-66 and EL-34. I have that weird early JTM-45 OT. (Radio Spares clone from MM) I'm currently using the 9K tap with 16 into 8 cabinet.

[Andy Le Blanc]

One of the problems with DIY approach is that you start at the end and then
work backwards to the beginning. The first step in amplifier design is loading
and the choice of an operating point. Fenders and nearly every other amp
made from the beginning were never design to operate in the regions that
our expectations demand. Ultimatly tranny Z means choosing a turns ratio.
Any loading changes also means a change in power out and distortion, which
is a change in tone. Your choices my be very tube friendly but will they go
to far from expectation. With many common trannys for a DIY being based
on our "end" expectations, finding a good value of screen grid resistor seems
like the most accesible option, even useing a three section L-C filter instead
of the more familiar two section L-C to drop the screen voltage some.