Any ideas on Ultra Linear designs using Beam Power Pentodes?

[Redhat]

Hey,

I'm considering making a 20-40 Watt guitar amp. head using an ultra-linear design with something like 6AQ5 beam-power pentodes.

Has anyone any experience using beam-power pentodes in guitar amps? Also, has anyone experienced advantages/disadvantages in using ultra linear power output stages?

Thanks,
Redhat

[pdf64]

They are very hard on the screen grids if the power supply is stiff.
Ensure that you have a saggy supply, ie tube rectifier maybe even a 5AR4/GZ34 would be too stiff for 20 watt amp, or a 100ohm sag resistor with a silicon rectifier.
The screen grids in fender 75s dissipate over 20 watts when cranked (5 watt rating on a 6L6GC)!
Dr Z advises using the KT series tubes for UL operation, with saggy supply.

[Steve Conner]

No reason why you can't run screen resistors in series with your UL taps, like in the old Mullard 5-20. I did that in the old Crown I restored, because I changed it to EL34s.

I recently measured the THD:
0.038% at 1kHz, 1W
0.22% at 1kHz, 10W
1.00% at 1kHz, 30W

[Redhat]

Ah, I think I see what you mean. They have fairly high screen grid current compared to non-beam pentodes. Makes sense I suppose since their current density is higher.

How saggy should the supply be? Would this quantified by a basic voltage regulation measurement?

Thanks, Redhat

[Bob M.]

My most recent build is a 5F6-A with a UL output transformer into 1x12" speaker. It's pretty awesome so far but it's early days. With the design I chose, being overly clean isn't a problem but I certainly have enough sound to play any small to medium club with this amp (the amp is in a very small package). I did alot of work choosing the power supply node voltages and picking the proper voltage dropping resistors to achieve what I was after. The NOS GE 6L6s that I'm running seem very happy in UL mode and of course, sound great. I'm also using screen grid resistors with the UL taps.

A future build I'm planning now is for a very clean non-guitar tube amp with 2x 6550 tubes and 1x 15" speaker. I will most probably use a UL wired output transformer for this project.

I'd say the advantages are a little extra output, no need for a choke and a small savings on the parts count but conversely, one must pay attention to power supply voltages.

Bob M.

[pdf64]

'No reason why you can't run screen resistors in series with your UL taps'

I tried that, they were 470ohms as stock, I had to raise them to 4k7 before the screen grid dissipation got reasonable. Dr Z was kind enough to respond to my query about this, and he confirmed that as being the best way forward, and I paraphrased his advice regarding the best way to design UL amps above.
Unfortunately the amp sounded thin and lifeless, and lost power output, with screen grid resistors that big. I ended up installing a choke and moving to regular pentode operation with mine and my buddy's fender 75.
Hifi amps just aren't going to get cranked in the way that guitar amps do, so it I don't think it's a problem for them - the screen grids only start drawing heavy currents when the tube is overdriven.
The STR387 type 6L6GC seemed to cope with the screen grid dissipation levels in the fender 75, but current production tubes get eaten up.
I'd get as much info as possible about how current UL guitar amps with current production tubes (ie DrZ) operate.

[ThermionicScott]

My dad and I built an ultralinear 2x6V6 guitar amp when I was in high school. It has enough odd things to its design that I don't really know how the UL mode really contributes to the sound. Even with the NFB disconnected, it maintains a pretty clean sound up until it clips. I need to finish up my tone control tweaks and get it scoped for power output!

Some other considerations:
* Because you're running somewhere between triode and pentode mode, you'll get less output than pentode mode. So it might take a quartet of the 6AQ5s to make your 20 watts. (RCA didn't upgrade the 6AQ5 to 14-watt plate/2.2-watt screen under the Design Max system like they did the 6V6.)

* To keep PS hum under control, you'll want to consider a CLC filter before the OT, much like the E-series tweeds.

* It will probably take a wider voltage swing from the phase inverter to drive it to full power.

* There isn't much data out there on running 6V6/6AQ5 in UL mode, so it's hard to say what the ideal load is. Thankfully, UL mode is a lot more tolerant of different loads than pentode mode.

- Scott

[imaradiostar]

I've had good results with EH 7591a's in ultralinear using the Triode Electronics "deluxe reverb" 6k6:8/4 output transformer. My amp matches the operating characteristics on the datasheet almost exactly (around 400 volts on plates) and I haven't had any problems even after extended playing time into a hotplate and/or speaker cabinet. I'm using an antek inc toroid which is pretty stiff- no dropping resistor in series or anything. I seem to recall using 470 ohm screen grid stoppers but I'm not 100% certain.

Sounds pretty great too- might wanna try it!

If screen dissipation is an issue you could try lowering your plate voltage. I recall reading somewhere that this becomes the sticking point for many ultralinear designs- lots of tubes work better with the screens at lower voltage relative to the plates and ultralinear doesn't allow that.

I'd imagine a quad of 6aq5's would be a nice match for the slightly larger Triode Electronics output transformer at 4.2k- each pair of 6aq5's would of course see 8.4k which seems to me like a good match for them. I'd try to stick with pretty low plate voltages- I'd say in the neighborhood of 300 volts if you can swing it.

Cathode bias may be the way to go- built in voltage sag as needed. You can always limit it with a zener to clamp the cathodes at a certain voltage under load.

jamie

[Steve Conner]

Beam power tubes are supposed to draw LESS screen current than pentodes, not more. The whole point of a beam tube is that it "beams" most of the electrons onto the plate and less of them go astray onto the other electrodes.

Pdf64: How did you measure the screen dissipation? It's notoriously hard to measure, and UL only makes it harder. You can't just take the current read by a meter in series with the screen under drive conditions, and multiply it by the DC voltage you measure on the screen.

The real dissipation is considerably less than this figure, because the screen voltage is low at the times when the current is high, and conversely, the current is low when the voltage is high.

This is true even with a pentode amp that drops voltage across a screen resistor, but it's especially the case in UL. In fact some of the power that would be dissipated in the screens and screen resistors (maybe even 43% of it? or 0.43 squared?!) becomes useful output power instead.

The screens on my Crown glow red in patches at idle, but they don't get any worse when I crank it into a dummy load.

For tubes that like a high plate voltage and low screen voltage, I've heard rumours that you can replace those screen resistors with Zener diodes. :-)

[pdf64]

'Pdf64: How did you measure the screen dissipation? It's notoriously hard to measure, and UL only makes it harder. You can't just take the current read by a meter in series with the screen under drive conditions, and multiply it by the DC voltage you measure on the screen. '
Yes, it took something thinking time for me to get my head around it.
I came to look at it because my buddy's amp was going through 6L6GCs in a few hours, I've never seen such dead tubes.
The bias supply was good and they were biased appropriately.
My fluke 187 has a (Vac+Vdc) option (as well as being true rms) which seemed to be appropriate as the screen dissipation under drive will be a combination of ac and dc.
To get the screen current, I measured the Vac+dc across the 470 screen grid resistor, and divided by the resistor. To get the screen grid voltage, I measured Vac+dc from screen grid to ground.
The voltage and current measurements were made at 'no-signal', max sine, and max square conditions, then the screen grid power dissipation calculated for each condition.
The Vac+dc figure across the resistors was the same as the Vdc figure at 'no signal', and about 1.5 Vdc with high signal levels.
I was fairly confident that the power levels I measured were accurate, even though they were far higher than I expected, at least it explained why the amp was eating tubes.
I even borrowed a calibrated 187 from my works to confirm the measurements that I made with mine.
I'd be grateful if you could point out any flaws in the method I used.
Having converted it to a regular fender arrangement, ie installed an additional choke/cap screen supply and taped off the UL OT taps, the power output was about about the same, but the screen grids dissipation were very high, though lower than in UL mode.

[imaradiostar]

Something seems wrong here...why would it be hard on the tubes to run ultralinear unless they were already at unsafe screen voltages to begin with?

Like I said- I've run EH 7591's for a long time now with lots of power amp distortion with no ill effects at over 400 plate volts. Perhaps 6L6 family tubes that weren't designed for ultralinear use just can't take the screen voltage and feedback?

Meanwhile I wouldn't sweat it- build your own ultralinear amp and see how it sounds for you. Just don't use 6L6's if they're problematic. I think KT66's probably sound better in that kind of circuit anyway. For that matter- 6L6's sound pretty good in the traditional Fender pentode arrangement so there's nothing wrong with using what works.

jamie

[Steve Conner]

My fluke 187 has a (Vac+Vdc) option (as well as being true rms) which seemed to be appropriate as the screen dissipation under drive will be a combination of ac and dc.
To get the screen current, I measured the Vac+dc across the 470 screen grid resistor, and divided by the resistor. To get the screen grid voltage, I measured Vac+dc from screen grid to ground.

Yeek... My head hurts... I don't actually know if that would give the right answer or not.

However, the pile of smouldering 6L6s was probably pretty good evidence too.

Wasn't the Fender 75 one of Fender's line of super amps like the 160, 300PS and 400PS? Those all used specially rated tubes, so maybe the 75 needed a special 6L6GC too. (or whack a pair of KT88s into it)

[Redhat]

From the ancient 1953 GE datasheets it would seem pdf64's method is correct for measuring screen current. The screen voltage given is the DC value (from GND). Its RMS value is with respect to the grid. But the AC and DC component of the current through a resistor is always going to be: Vdrop/R (if Vdrop has a DC and AC component).

After all these inciteful comments I am keen to give it try. I'm thinking of going with a Fender Twin based topology with an ultra clean channel and an overdrive channel based on the Marshall DSL. It seems that UL really only provides an advantage for clean mode, but we'll see.

Interestingly, I found through web research that many power pentodes that are being produced are actually beam tetrodes (aka beam-power pentodes). So if nothing else it will be a great educational experience.

[ThermionicScott]

Interestingly, I found through web research that many power pentodes that are being produced are actually beam tetrodes (aka beam-power pentodes). So if nothing else it will be a great educational experience.

Yeah, the datasheets are usually good about specifying whether a tube is a true pentode or not. A lot of us will use "pentode mode" as shorthand for running a pentode or beam power tube conventionally, rather than in triode or UL mode.

- Scott

[mooreamps]

Beam power tubes are supposed to draw LESS screen current than pentodes, not more.

U have some reference for this ??? Do you even know how a screen grid draws current ; in any type of tetrode or pentode ? U ever crack open a power tube and see how it is built ?? Just, u know, asking the questions......


-g