UL taps hookup?

[Madison]

I have an OT for my Plexi type build.
It has the 40% screen taps.
I want to put them to use on a switch.
-I have 1.5K screen grid resistors in place.When in UL mode,should they be bypassed?
-Pretty obvious, if the output wires are reversed and become PFB I assume I have to reverse all of them?
-Anything else to worry about/tips?
Thanks

[triode]

I would leave the 1.5ks in. I think most would disagree with me.

If you get the squal, probably easier to swap the inputs to the grids instead
the plate and UL taps. Two wires is easier than four.

[GroundhogKen]

I agree.  If it's EL34s leave the 1.5K resisters on or near pin 4 of the tube sockets.

Wire the other end of the resisters to the common legs of a DPDT switch.
Run the UL taps to one half of the switch.  Bus the other half of the switch to a power supply node.

Now you have switchable Pentode/Ultra Linear.  I prefer UL (at least with beam power tetrodes like 6L6 and 6V6).  The switch is nice because you can quickly decide which mode you like better.  My unscientific impression is that 87.9112% of guitarists prefer pentode.

Make sure the plate supply is well filtered.   In UL mode a small amount of ripple on the screens will cause hum.


Ken

[HotBluePlates]

My unscientific impression is that 87.9112% of guitarists prefer pentode.

My unscientific hypothesis is that 85.573% of guitarists have never tried UL.  

Actually, I can't recall if I have ever tried it. I'll have to put that on the to-do list...

[FYL]

I would leave the 1.5ks in. I think most would disagree with me.

I'd even say that screen R's are a *must* with current EL34's.

[DummyLoad]

start with 100ohm resistors - too much resistance and you may as well not use the taps, but then again UL is hard on the screens when the output stage is over-driven.

buttery and i were experimenting with SE UL a while back with some KT88s and IIRC, we were seeing up to 40mA on the screens when driven hard. we bumped up to 500ohm then to 1Kohm on the screens and overdrive distortion did not improve any.

the consensus was that we preferred pentode mode if we were to overdrive the output stage. EL34s may or may not exhibit similar characteristics.

if you are shooting for clean with low distortion, then UL mode might be the tone that you seek.

[Madison]

>>Wire the other end of the resisters to the common legs of a DPDT switch.
Run the UL taps to one half of the switch.  Bus the other half of the switch to a power supply node.

Oh man!
I wired it up wrong.
No wonder.
Will correct and report.
Thanks

[HotBluePlates]

start with 100ohm resistors - too much resistance and you may as well not use the taps, but then again UL is hard on the screens when the output stage is over-driven.

buttery and i were experimenting with SE UL a while back with some KT88s and IIRC, we were seeing up to 40mA on the screens when driven hard. we bumped up to 500ohm then to 1Kohm on the screens and overdrive distortion did not improve any.

I wonder about this some...

In pentode mode, it is normal for screen current to skyrocket when the grid is run positive. So in those instants where the peak input signal exceeds bias voltage, plate current plummets (left of the knee on the plate curves) and the screen winds up passing much more current.

But, if you think about a class A output tube, it dissipates less when there is a signal than at idle. Part of the power input is being transferred to the load by the OT rather than heating the plate. So I wonder if the same effect happens for the screen in UL. It's probably time for me to check out some data sheets and old books...

[PRR]

> In pentode mode, it is normal for screen current to skyrocket when the grid is run positive.

No, actually when plate "bottoms", falls as low as it can for a given current.

Or more accurately: the cathode-G1-G2 interaction defines total cathode current. This splits between plate and G2. Because electrons have large velocity and G2 is open (and in 6L6, G2 sits in G1 shadow), normally most current flows to plate. But if plate voltage gets VERY low, it doesn't suck any more electrons. Cathode current is already happening; the excess finds its way to G2.

Put a 6L6 in a standard Champ. Put 1K 10W resistors in series with plate and screen. Rig a switch to break the plate connection. In normal operation (say B+ at 350V, G2 near there, G1 at zero, cathode at +20V, about 30mA plate 5mA G2), you find about 310V (350V-40V) on plate, maybe 345V on G2, due to drops in the 1K resistors. Now break the plate. Obviously it falls to zero. But all that "plate" current now has to exit through G2. G2 will fall from 345V to 315V, showing that it is getting 35mA, its normal 5mA plus the 30mA which was in the plate. (And will melt quickly! 35mA*315V= 11 Watts!! Measure fast!!)

If your loadline is high impedance, it will tend to "bottom below the knee", and G2 current will spike. This is of course when G1 is nearest zero; but for a very-very-high Z load, G1 does not have to be close to zero for this to happen.

This may be one reason old RCA sheets for 6L6 show quite low load impedances. With a very lo-Z load the plate does not reach low voltage, G2 current won't spike.

> I wonder if the same effect happens for the screen in UL.

UL is tricky. For one thing, when plate is low, G2 is also low, so it has less tendency to suck the excess when plate bottoms. And if it does, its load connection makes G2 sag more, limiting its suckage. Some argue that this G2 current contributes to output; the leverage is so bad (40% tap is about 6 times off the optimum load) that I doubt it adds much.

[jojokeo]

...so it has less tendency to suck the excess when plate bottoms. And if it does, its load connection makes G2 sag more, limiting its suckage. Some argue that this contributes to output; the leverage is so bad that I doubt it adds much.

  sounds like me & the wife   (sorry, it's late & she's wanting sack time)

[HotBluePlates]

If your loadline is high impedance, it will tend to "bottom below the knee", and G2 current will spike.

This is the idea I didn't think/say clear enough. That when load impedance is low enough that the loadline cuts below the knee of the plate curve, then screen current can be excessive when G1 is near 0v. If the load impedance is higher, the loadline cuts the G1=0v line above the knee, and there is less chance for excessive screen current.

[Madison]

Reporting "IMVHO" findings.
I suppose I am in the 87.9112 percentile.
I do like UL.
It's a bit more "focused" and cleans up some of the mud.
Clean tones are very nice.
On the flip side it seems, to my ears, to be slightly brighter.

I find it a bit more effective with 6V6s in this current Plexi type build I am making that can run EL34, 6L6, or 6V6.
Not sure I'd personally own an amp with UL only (though a SF Pro Reverb lived and gigged with me for a spell), nice to have the switch and compare both Pentode/UL settings.

Peace

[JayB]

Reporting "IMVHO" findings.
I suppose I am in the 87.9112 percentile.
I do like UL.
It's a bit more "focused" and cleans up some of the mud.
Cleans tones are very nice.
On the flip side it seems, to my ears, to be slightly brighter.

I find it a bit more effective with 6V6s in this current Plexi type build I am making that can run EL34, 6L6, or 6V6.
Not sure I'd personally own an amp with UL only (though a SF Pro Reverb lived and gig with me for a spell), nice to have the switch and compare both Pentode/UL settings.

Peace

I need to try it one day. I keep contemplating it for a bass amp or a high gain metal amp where a clean power amp is needed. What is the head room compared to non UL?

[sluckey]

All of the old Sunn amps were UL. They had a good sound to me. Like Madison said, clean and focused. Loud too.

[HotBluePlates]

What is the head room compared to non UL?

Bias voltage for UL is similar to the same tube run as a pentode, judgiung based on the Tung Sol 5881 data sheet. That means that input headroom, os how big a signal you need to/can feed it, is about the same.

Total distortion is a bit less, and the characteristic curves appear the similar/same as triodes, so it will distort differently. I'd expect more even harmonic distortion, which then gets cancelled by the push-pull out. So the apparent headroom would increase somewhat. How much? You'd probably have to build and test to find a meaningful answer.

[JayB]

I thought about the same, bias and voltage is basically the same. Other than screen voltage would track closer to the plate? I'll just have to try it one day.

[HotBluePlates]

Other than screen voltage would track closer to the plate?

Yes. The csreen voltage has to track closely with plate voltage, because the only resistance across which to drop voltage is the relatively low dc resistance of the OT.

But you could figure this out another way. Look at the normal plates curves for a pentode/beam power tube. They rise quickly at low voltage, then turn nearly horizontal, which is typical of pentodes. Look also at a good data sheet (like the one I cited earlier), and you'll see that the triode operation curves look very much like small-signal triodes (like a 12AU7). Triode-operated pentodes have the screen connected to the plate, so screen voltage has no choice but to equal plate voltage. Now, look at the UL curves on that Tung-Sol sheet. They are essentially the same as the triode curves, and without thinking along any other line, you could assume from the shape of the curves that screen voltage equals plate voltage at all times.

"But what about amps where screen voltage and plate voltage are the same, with the output tubes run as pentodes?"

The screen and plate voltages are essentially the same at idle, but while the plate voltage swings up and down with the signal, the screen voltage is kept essentially constant by the seperate screen supply node. If the screen voltage dropped at the same time the plate voltage dropped, the screen voltage would essentially be tracking the plate voltage, and triode characteristics would result.