• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Screen drive and other P-P experiments

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Several months ago I started a thread about the creation of a new "universal driver board". My immediate need is for a board to rebuild my "300Beast" amplifier which has died. I also want the board to be capable of driving the typical audio tubes in A2 or AB2 mode. Screen drive and cathode follower output circuits are a given too, hence the term "universal". That thread went cold when lifes little interruptions got in the way and the board went on the shelf.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=137063&highlight=

I recently found the time to finish the board, and start some testing. I realized that I have been testing several different things and posting into several existing threads that happen to be relevant. Unfortunately I can't even find my own posts, so I started this new thread to put them all in one place. There will be a few more experiments this weekend, then two weeks of silence (I won't be here). The experiments will then resume.

The experiments will serve to refine the driver board, test it with a wide variety of output tubes and topologies, and pick three or four diverse designs to be built into completed amplifiers. One will obviously be a 300B or 307A push pull amplifier to replace my 300Beast. Another will be a push pull amplifier using the typical audio tubes. Power output to be determined. The other designs have not been determined yet, but I still have those 400 watt OPT that need a home!

Some of the relevant threads are:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=143703

http://www.diyaudio.com/forums/showthread.php?s=&threadid=133034

http://www.diyaudio.com/forums/showthread.php?s=&threadid=143620

http://www.diyaudio.com/forums/showthread.php?s=&threadid=143135

I included a photo of the driver board driving a pair of 6L6GC's.
 

Attachments

  • 6l6ampcranked_a.jpg
    6l6ampcranked_a.jpg
    96.9 KB · Views: 1,009
***DON'T DO THIS UNLESS YOU KNOW WHAT YOU ARE DOING!***

There was some debate about whether the power measurements that I showed in the 6L6GC in AB2 thread were real or even possible. I believe that they are and provided examples from the 807 data sheet, but I am in the process of verifying the data.

There was some speculation as to whether my 6600 ohm load was really 6600 ohms. The transformer that I initially used was from an unreliable source, and used non standard wire colors, so I substituted a known transformer that I had previously tested. This resulted in even higher output power numbers.

How could I verify the load impedance, and the power measurement capabilities? While I was outside mowing the lawn, it came to me. All I had to do was to simply lift the center tap from B+ to totally isolate the primary, and plug the plate leads into the wall outlet! The 8 ohm load and the power meter would remain connected to the 8 ohm secondary. If the primary was indeed 6600 ohms, it should draw 2.29 watts from my 123 volt wall outlet. This power should be dissipated in the 8 ohm load resistor and read on the power meter (minus the transformer losses). The watmeter reads 2.22 watts indicating that the load is very close to 6600 ohms at 60Hz.

It has been pointed out by someone much smarter than me that my big 500 watt load resistor is "wound" and somewhat inductive. This may make it effectively a higher impedance at the higher audio frequencies. This led me to purchase a bunch of non inductive Caddock power resistors and build a non inductive load bank that I use for frequency response measurements. I however am not connecting my expensive load bank up to these amplifiers that are more than capable of blowing it up.

I made some comparisons between the big resistor and the load bank and found that they provided similar power readings. I took the big resistor to a HP component analyzer and measured it. It has 14 uH of inductance which is roughly .088 ohms at 1KHz. This could lead to a 1% power measurement error at 1KHz, but it could be about 10% at 20KHz. This means that it is perfectly useful for power measurements at 1KHz, but should not be trusted for frequency response measurements.
 

Attachments

  • walloutlet_a.jpg
    walloutlet_a.jpg
    64.4 KB · Views: 874
After warming up a pair of 6L6GC's in conventional grid drive I decided to "test" some sweep tubes in screen drive. First up two different tiny 9 pin sweep tubes that I obtained cheaply. I am not revealing the type numbers yet (soon). The first pair looked like they might be useful with 10 watt plates, and real published screen curves. This proved not to be the case. No matter how biased these tubes I either got lots of distortion, or low power. Best case was a whopping 4.5 watts (tried 2 pair). I punted these tubes and hooked up the second set.

These were much better. I could get 30 watts out of a pair of these guys but I had to run 450 volts to get there. Idle current was set at 7 mA. Load was 6600 ohms. There was no hint og glow from plate or screen. No blue either.

Yeah these little guys are really sweep tubes. They mus have been for a tiny TV set. OK, I will move up to some bigger tubes.
 

Attachments

  • little_tubes_a.jpg
    little_tubes_a.jpg
    81.9 KB · Views: 919
I got out the same pair of 6BQ6GTB's that I tortured last year in the "tube sale at AES" thread. We already know that these guys can make some big power, but now I have a bigger power supply!

They were connected up in screen drive with a seperate power supply for the plate capable of going to 660 volts at 1.7 amps. I tried 6600 ohm and 3300 ohm loads. I tested at several supply voltages until the inevitable happened, I fried one of the tubes.

I tried several plate supply voltages starting at 500 volts with each load. At 600 volts and a 6600 ohm load the tubes were well behaved with 90 watts of output. The idle current was 15 mA per tube, and cathode current at 90 watts was about 125 mA per tube. There was no plate or screen grid glow.
 

Attachments

  • 90w no glow._a.jpg
    90w no glow._a.jpg
    98.4 KB · Views: 893
Operation with a 3300 ohm load was problematic and led to eventual damage to one of the tubes. The amp was relatively stable at 100 watts, but any attempt at increasing the drive met with a runaway condition.

If I increased the drive to produce 110 watts of power I noticed a faint redness on the plate, and the cathode current would start to climb slowly. Operation at 110 watts is possible for 10 to 20 seconds is needed for the current to climb.

120 watts of power was possible, but a distinct red stripe formed on the side of the crooked tube. It seems like the screen grid was glowing red too, but it is hard to see, and if the amp was left at this power level for a few seconds the cathode current would take off.

Could I just leave it alone? No, I had to turn things up just a little more. I hooked up a meter to each grid, and a scope probe on the screen. Then I proceeded to explore the runaway zone to find out what was going on. I noticed that an unusual thing was happening. As I pushed the tube into the "red zone" and let it stabilize the screen grid voltage was where I had set it, but as grid started to glow the DC screen voltage would increase. Where was this"extra voltage" coming from?

The driver board is currently set up with three bench power supplies. There is a 10 volt negative supply for the LTP tails. It is also tied to the output tubes control grid. There is a second negative supply for the PowerDrive only. This is currently set at -150 volts (good for 200 mA) and is tied to the output tubes screen grid through a 10K 5 watt resistor. There is a third supply which feeds the B+ to the board and the PowerDrive mosfet. It is set at 500 volts. A fourth supply feeds the plate all it can eat. It is connected to the output transformer only.

I decided that I would run the drive up to the runaway point then kill the third supply which would remove the drive AND the positive voltage supply for the screen grid. This would leave the screen grid connected to a -150 volt source through a 10 K resistor. The control grid is still at -10 volts. This should shut down the tubes. I did this 4 times. The first 3 times the tubes shut down and cooled off immediately. The scope trace associated with the screen grid dropped to -150 volts. On the fourth try I let the tubes run until they were cranking out 126 watts, as the screen grids glowed the current took off. I shut down the third supply. One tube shut off, the other took off. The screen grid glowed brighter and its voltage climbed in the positive direction! I shut the plate supply off just as the fireworks started inside the tube.

Why did the screen voltage go positive when it was connected only to a negative supply? There is only one explanation that I can think of. The screen grid had become hot enough to emit electrons and begin to act like a cathode. These electrons traveled to the plate causing some serious "secondary emission" raising the screen grid voltage, which in turn caused the cathode to plate current to increase leading to runaway.

I believe that this may be a limiting factor in screen driven amplifiers. Operation at high power extremes causes the screen to be driven highly positive at the same instant the plate voltage is approaching zero. This causes serious screen current on the peaks of the signal. Is it a real problem. No, I don't think so. First off most users of sound mind (that excludes me) will not try to extract 120+ watts from 6BQ6's. Second, even if they did, normal music would not demand full power operation for extended periods of time. It takes time to make the screen grid glow and start the avalanche.

Bob Carver made a career out of exploiting this property of music. I still have an old Carver M-400 amplifier that bears the warning not to test the amplifier with a sine wave, or use it as a musical instrument amp. Why? It puts out 400+ watts and weighs about 10 pounds. It uses an undersized power transformer and minimal heat sinks. It will shut down if you feed it some Pink Floyd and turn it up all the way, but who does that? :D

The tubes glow at 126 watts before shut down. The tube that blew was formerly known as the leaning tower of power since is came bent! For 98 cents I am not complaining.
 

Attachments

  • 3.3kload126watts_a.jpg
    3.3kload126watts_a.jpg
    99.5 KB · Views: 874
Is it possible we might see an universal push-pull PowerDrive board for screen-drive in the near future?

Yes, I plan to sell these after I have the time to test them with just about everything that anyone could possible connect them to, and refine the design. My current work (and other stuff) schedule along with thin financial resources make "the near future" unlikely.

I'd love to see what could be had with screen driven 813's......iirc, its been said before that the 813s are not suitable for screen drive.

Yeah, the 813 is one of the last tubes that I would think about for screen drive. I have found that most tubes need a bit more P-P drive voltage than their maximum screen grid voltage rating. The max G2 rating for an 813 is 750 volts, that would be a whole bunch of drive voltage. On the other hand the 813 could rock in conventional grid drive, and I got a few 813's and sockets to test with this board.
 
I have explored the upper limits of the 6BQ6's and found them. I killed a 98 cent tube in the process. Since I have limited time before my trip I decided that the simplest next step would be to simply plug a pin compatible tube into the sockets and test them. The 6DQ6 is pin compatible with the 6BQ6, yet is a bigger tube. These are not 98 cent tubes and the ones that I have here are NOS, so I must be nice to them.

I cranked them in my usual 50 volt increments starting at 500 volts with the 6600 and 3300 ohm loads. I started to notice that the tubes were becoming unhappy with the 3300 ohm load. There was no runaway, but the distortion went up faster than I would like, and there was some evidence of current increase when left running at 100 watts on 550 volts. I have plenty of "testable" 17DQ6's in the warehouse, so I decided to abandon 3300 ohm testing for now.

These tubes were unconditionally stable with a 6600 ohm load. I turned the power supply to max at 660 volts. I cranked the drive up to max. The output was a near perfect square wave and the power meter read 148 watts. This reading is likely inaccurate because the analyzer expects something more sinusoidal.

I set the power to a more realistic 110 watts and left it there for about 10 minutes while I took pictures. This picture is at 110 watts.
 

Attachments

  • dq6_100watts_a.jpg
    dq6_100watts_a.jpg
    75.8 KB · Views: 487
I took this interesting picture. The amp is operating as before at 110 watts into a 6600 ohm load with a 660 volt supply. I attached two scope probes.

The bottom trace is the screen grid with the scope set for 100 volts per division. The zero line is the second graticule line from the bottom. This shows that the screen grid swings from -140 volts to +250 volts.

The top trace is the plate with the scope set for 200 volts per division. The zero line is the second graticule line from the bottom. This shows that the plate swings from about 60 volts to about 1250 volts.

For about 1/5 of the audio cycle the plate voltage is less than the screen voltage. Many tubes really don't like this, but this one didn't seem to mind.

I only have a little experiment time left, so I must leave the sweep tube experiments for later. I really want to see some DHT's (or triode wired DHP's) hooked up to this thing.
 

Attachments

  • dq6_110w_sg_p_a.jpg
    dq6_110w_sg_p_a.jpg
    61.5 KB · Views: 490
What kind of instrument are you reading distortion with? Its definitely from HP...

It is an old HP8903A audio analyzer that I got from a hamfest about 10 years ago for $75. At that time there were several local firms closing their factories and the test equipment flooded the surplus market driving down the prices. Within a few years these things were fetching $300 to $500. I don't know what they go for now.

I use it because it is push button easy to read distortion (THD only) and power. It is a given for this kind of testing. I use a sound card and some FFT software to do real amplifier development and tweaking since it can show me the distortion spectra. Often I have both hooked up.
 
I have exceeded my allotment of fun for this weekend, and I didn't even blow anything up, since the fried 6BQ6 happened during the week. I hooked up a pair of triode wired DHP's. The only identification on these guys is SC899A. These tubes came wrapped in newspaper marked "special". They both rattle and have loose glass pieces inside, so that must be what makes them special. Otherwise they look and work like 307A's or RK-75's or VT225's.

I wired them into the amp in my usual triode mode which connects G3 to ground and G2 to the plate through a 100 ohm resistor. Pete Millett connects G2, G3, and the plate all together. In my usual total disregard for the specs, I dialed the power supply to 380 volts and set the tube current to 60 mA. Where do I get these numbers? It is where these tubes sound best in a Tubelab SE.

The tubes came to life, and respond very well to AB2 operation. I see the same thing that I saw in SE operation. These tubes don't rudely clip when overdriven. They clip gracefully. These guys may make a better 300B than a 300B does.
 

Attachments

  • sc899_a.jpg
    sc899_a.jpg
    87.6 KB · Views: 508
The amp remains in class A (both tubes conducting all of the time) up until about 15 watts, it does however enter A2 (the control grid goes positive) at 12 watts. After 15 watts the amp enters AB2 mode (one tube cuts off for part of the cycle). The distortion remains below 3% up to 25watts.

The photo is the output at 25 watts.
 

Attachments

  • 25watts_3pct_a.jpg
    25watts_3pct_a.jpg
    59.9 KB · Views: 469
I cranked the drive up to the 40 watt output point. Some real clipping is now evident, but it is still not abrupt. The distortion is now 15%.

Y'know, you could easily run the 813s triode connected in AB2, 1000V.

I will try that at a later time. I also have some 211's and 845's to try in P-P but I don't have a good enough OPT to see what they can really do.

I think that the fun is over for now. I must start gathering all of the stuff I need for a 2500 mile road trip. If I have some time I want to hook some speakers to this thing and see if it rocks like my old 300Beast did. It might have to wait until I return though.

It still looks like I will get at least one day at the Dayton hamfest (next weekend). Who knows what I will find.
 

Attachments

  • 40watts_15pct_a.jpg
    40watts_15pct_a.jpg
    59.4 KB · Views: 263
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.