GU81M

LT spice, Curve Captor, PSUDII and whatever other sims you can think of.

Post Reply
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

GU81M

Post by jarek »

I've figured out 833A would be to weak to electro-cute well enough, even in A2,
but GU81M, pentode mode, should do the job.
Anyone (Stephie :):):)?) would have a model: A1 + the onset of A2, G1 and G2, G2 at 500V
I've tried contacting this Italiano:

http://www.diegobarone.it/ModelloGU81.htm

but to no effect.

Thank you!
jarek


sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Been Busy at work. I'll see what I can do. I did download the datasheet. 12.6V 11A filament, it should be nice and bright. Low gi-g2, so triode connection, although not particularly linear, would be low plate resistance.

Stephie <3
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

Stephie, I was thinking of it in a pentode mode rather, G2 can handle only 600V max. The goal is a powerfull, 150W+ single ended electric chair...eeee....bass amplifier. The point I was looking at is 1500V/300mA
3.2k load. Would be interesting to see how non-linear it is and how it matters given the application (ULF 10-100Hz).
Thank you!
jarek
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

I thought you would use it pentode mode. Should have a model available in a few days.

Stephie <3
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Drop down menu for tetrode.txt with the GU81M has been placed in the "sticky" folder. It's reasonable for both A1 and A2 operation. Optimum with 500 volts on G2, but relatively OK. Fair to middlin in triode mode.

Should be useful for electrocution purposes. :)

Enjoy,

Stephie <3
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

Thank you Stephie!!!!
Let's see how linear this electrocution is ;-)
I don't want to be electrocuted by a distorting , non-linear
chair :evil:
I'm also trying to see if a triode connected 6E5P DC cpoupled will drive it (<100Hz)...

Cheers,
jarek
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

There is a discussion on Romy's site about a powerfull >100W SET amp to drive a dedicated bass channel 50-100Hz max. Romy has imposed a series of constraints, e.g.: 2 stage design, no IT coupling, no parallel drivers, no cathode followers, pref. DC coupled, etc etc. Since I have some heavy HV iron I'm for fun trying to see if I can meet those criteria. Triodes are gone out of the picture for the moment, so I tried to see how GU81M behaves.
I've simulated 6E5P (triode) choke loaded DC to GU81M. At 2Vrms input it gives respectable 185W staying in A1.

FYI, I attach the schematics and some simulation data. To me it all looks very non-linear but I have no idea how to properly relate those FFT's to sound. I don't get why the higher the freq. the worse the spectrum looks?
What do the dips in the FFT signalize?
If anyone's able to analyze those data, I'd love to hear some comments!
Cheers,
jarek
Attachments
20Hz waveform
20Hz waveform
GU81M_20Hz_2Vrms_wave.JPG (44.49 KiB) Viewed 30243 times
20Hz FFT
20Hz FFT
GU81M_20Hz_2Vrms.JPG (25.19 KiB) Viewed 30243 times
50Hz FFT
50Hz FFT
GU81M_50Hz_2Vrms.JPG (25.93 KiB) Viewed 30243 times
100Hz pentode hell FFT
100Hz pentode hell FFT
GU81M_100Hz_2Vrms.JPG (28.58 KiB) Viewed 30243 times
The schematics
The schematics
GU81M_Spice.JPG (100.06 KiB) Viewed 30243 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

For FFT to be reasonably accurate, you have to have a LOT of cycles. If you're simulating 1kHz for instance, run the simulation for 500ms or so so it accumulates many cycles. (in this example, it would be 500 cycles of 1kHz)

Enjoy,
Stephie
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

me idiot :oops: thank you Stephie!

Below are the plots for more cycles. Stephie, form your experience, how to interpret them? To me looks like guitar amp, but i have no experience.

what are the dips right before some harmonics?

thank you!
jarek
Attachments
20Hz
20Hz
GU81M_20Hz_2Vrms.JPG (28.89 KiB) Viewed 30233 times
50Hz
50Hz
GU81M_50Hz_2Vrms.JPG (32.38 KiB) Viewed 30233 times
100Hz
100Hz
GU81M_100Hz_2Vrms.JPG (33.13 KiB) Viewed 30233 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Notice the odd order distortion (3rd) is showing the clipping that your waveform is also showing. Back off a little on the drive to see what the structure out of clip is. If you're driving at 2 volts, back off to about 1.5 volts (slightly less than 3 dB) and look at the harmonic structure. It will give you a better indication of the linearity of the system.

On the dips, partly "artifacts" of the number of samples etc, and partly due to "start up transients". That can be minimized by altering the transient analysis command.... in the spot where to "time to start saving data" start that at perhaps 200ms, and make your stop time however long AFTER that point, say 1200 ms for your 100Hz signal. (That would give you 1 second of data, or 100 cycles of 100Hz for the FFT to chew on.

Sometimes it also helps to add a "minimum stepsize"... for 100Hz signal 1e-5 would be reasonable.

Stephie <3
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

It seems that it just wants to clip...the bias was always
negative but the cut-off is ugly. I've followed your
instructions, below is the FFT for 20Hz @ 1.4Vrms
with 2x less load set at 2k. It gives only 90+W, which
makes it unattractive as big triodes like 833 can do that too.
And without so much distortion.
Seems this tube is good only for tesla coils :(

Cheers,
jarek
Attachments
GU81M_20Hz_1.4Vrms_2kload.JPG
GU81M_20Hz_1.4Vrms_2kload.JPG (30.61 KiB) Viewed 30193 times
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Ultralinear?

Post by jarek »

I'm wondering if there would be a way to put into the ultralinear mode?
The problem is 1kV difference between the G2 and the plate...

Cheers,
jarek
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

That might make it more linear, I'm seeing a couple hundred watts even in pentode mode before things get too out of hand. I adjusted biasing slightly on your circuit. Do a DC analysis first, before the transient analysis.

The way to run UL on this tube is implemented most straightforward using an additional winding on the output transformer. That would be fed from the lower voltage point, and would have fewer turns than the "primary".

Incidentally, at 20 Hz, if you start data collection at 500ms (10 cycles of 20Hz) and stop at 10500ms (total of 10 seconds of data- 200 cycles of 20Hz), your FFT will look prettier. :)

Enjoy,
Stephie <3
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

sbench wrote:That might make it more linear, I'm seeing a couple hundred watts even in pentode mode before things get too out of hand. I adjusted biasing slightly on your circuit. Do a DC analysis first, before the transient analysis.
Hmm...I thought I've done it. What operating point do you get? And what load?
sbench wrote:The way to run UL on this tube is implemented most straightforward using an additional winding on the output transformer. That would be fed from the lower voltage point, and would have fewer turns than the "primary".


Bingo! That's the way I've been looking for. A separate 500V screen supply and then
an additional winding with user-adjustable ratio!
sbench wrote:Incidentally, at 20 Hz, if you start data collection at 500ms (10 cycles of 20Hz) and stop at 10500ms (total of 10 seconds of data- 200 cycles of 20Hz), your FFT will look prettier. :) .


thanks for educating a moron!!

cheers,
jarek
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

In pentode mode 6E5P 3.5 volt bias. I changed the resistance of L1 to 3.32k. This puts the grid of the GU81 at -89 volts.
6E5P total current =27mA, 5 watts (23+4mA)
Gu81 at 1600v, 267mA=450 watts. Changed primary ind to 130H, 1 ohm load from 32.5mH.
2v peak in produces equal 2nd and 3rd harmonics at a total of 10% THD and 187 watts. Above 2v peak obv iously produces more power and distortion, but the 3rd is now above the 2nd. At less, than 2V peak in, second predominates.

If you label the output "out" and add a spice directiv .four 20Hz v(out) and look at the .log file, you will see the Fourier (not FFT) output showing you the magnitude of the fundamental and harmonics. Of course, you can still call for a FFT if you like.

Stephie <3
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

sbench wrote:In pentode mode 6E5P 3.5 volt bias. I changed the resistance of L1 to 3.32k. This puts the grid of the GU81 at -89 volts.
6E5P total current =27mA, 5 watts (23+4mA)
Hmmm...my motivation was that in the triode mode, run hard to eventually supply a bit of a grid current, 6E5P would give a more linear and robust driver. It seems so beautiful in the triode mode!
sbench wrote: Gu81 at 1600v, 267mA=450 watts. Changed primary ind to 130H, 1 ohm load from 32.5mH.
Stephie, your load seems the same as mine in the first run 130H to 32.5mH => 4k or have you changed it? I'm simulating now 5k load. Your OP seems slightly below 450W at 430W. A bit unexpected as lowering the current should give uglier cutt-off? The clipping you see
on my waves is exactly from the cutt-off I've been trying to fight.

My thinking was since I burn 120W of heater, to get the damn max out of it!
Google-searching points to amateur radio forums where some people claim running GU81 it past 450W max for years with no trouble. I'm wondering how much one can squeeze out of it with FAC? On diyaudio there is (was?) a guy claimimg 1300V/400mA OP and 5k load. Unfortunately he does not reply. Maybe the tube did not like that and took a revenge on him :(?

I'm very curious how will the UL idea perform. Briliant solution with an additional winding, Stephie!

Cheers,
jarek
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

sbench wrote:In pentode mode 6E5P 3.5 volt bias. I changed the resistance of L1 to 3.32k. This puts the grid of the GU81 at -89 volts.
6E5P total current =27mA, 5 watts (23+4mA)
Gu81 at 1600v, 267mA=450 watts. Changed primary ind to 130H, 1 ohm load from 32.5mH.
2v peak in produces equal 2nd and 3rd harmonics at a total of 10% THD and 187 watts. Above 2v peak obv iously produces more power and distortion, but the 3rd is now above the 2nd. At less, than 2V peak in, second predominates.

If you label the output "out" and add a spice directiv .four 20Hz v(out) and look at the .log file, you will see the Fourier (not FFT) output showing you the magnitude of the fundamental and harmonics. Of course, you can still call for a FFT if you like.

Stephie <3

Stephie, I can't reproduce your results. I'm getting 154W with ugly clipping and the 3rd harm. approx. 0.9dB above the 2nd. Moreover the FFT is very noisy and ugly at 500m and 10500m. Would you please post your schematics? Thank you,
jarek
jarek
Posts: 62
Joined: Wed Feb 02, 2005 10:02 am

Post by jarek »

I must be a complete Spice idiot as the mentioned by Stephie configuration gives me only this:
Attachments
2V peak, 155W
2V peak, 155W
GU81M_20Hz_1.4Vrms_4k_wave.JPG (78.65 KiB) Viewed 30104 times
Ugly FFT, 500ms start, 10500ms stop, 10^6 samples
Ugly FFT, 500ms start, 10500ms stop, 10^6 samples
GU81M_20Hz_1.4Vrms_4kload.JPG (88.77 KiB) Viewed 30104 times
Post Reply