We are taunted with pictures of warp drives, there are even two flux capacitors in the picture. These are not capable of jumping to hyperspace.
I built the proper equipment for the job, and in my usual style, I set the controls for light speed, and made the jump. The voltage pot on the warp drive (power supply) has always been loose. I was exploring the upper limits (above 500 volts) of a very well used Chinese 6L6GC when the voltage pot broke. The voltmeter on my power supply slammed all the way to the right (it goes to 600 volts) and a serious fireworks display erupted inside the 6L6. I hit the kill switch, fast!
I disconnected the power supply form the test amp and fired it back up. The voltage knob has no effect, but the voltage range switch (selects transformer taps) still gives me 50 volt steps. The 250 volt range now gives me 470 volts, and the 300 volt range gives 520 volts.
I set my camera to record video and decided to plug the amp into the 480 volts and make video of the meltdown. There was none. That tube just won't die! I set the power supply to the next range. Still no fireworks.
What do I do now? I decided to find out how much power I could get out of a 6L6GC in UL mode. Given a limit of 5% distortion, but ignoring all tube specs and its cries for mercy, I sterted cranking things up. I was getting 23 watts at 5% distortion when I took this picture. For a 6L6GC this IS light speed! The power supply was no longer in regulation and I was getting about 490 volts under load and the current meter read 110 mA. The poor tube just kept going!
I built the proper equipment for the job, and in my usual style, I set the controls for light speed, and made the jump. The voltage pot on the warp drive (power supply) has always been loose. I was exploring the upper limits (above 500 volts) of a very well used Chinese 6L6GC when the voltage pot broke. The voltmeter on my power supply slammed all the way to the right (it goes to 600 volts) and a serious fireworks display erupted inside the 6L6. I hit the kill switch, fast!
I disconnected the power supply form the test amp and fired it back up. The voltage knob has no effect, but the voltage range switch (selects transformer taps) still gives me 50 volt steps. The 250 volt range now gives me 470 volts, and the 300 volt range gives 520 volts.
I set my camera to record video and decided to plug the amp into the 480 volts and make video of the meltdown. There was none. That tube just won't die! I set the power supply to the next range. Still no fireworks.
What do I do now? I decided to find out how much power I could get out of a 6L6GC in UL mode. Given a limit of 5% distortion, but ignoring all tube specs and its cries for mercy, I sterted cranking things up. I was getting 23 watts at 5% distortion when I took this picture. For a 6L6GC this IS light speed! The power supply was no longer in regulation and I was getting about 490 volts under load and the current meter read 110 mA. The poor tube just kept going!
Attachments
Test Data at last!
OK, first block of test data in (25 spectra ).
These are all using 6L6GC SE with 385V plate B+ and 50 mA plate current and 8 OHM resistor load for a spec'd 3200 OHM primary Zload. And all are using B++ fixed supply for screen current return (thru MOSFET drain).
In the upper left corner of the charts are specifiic parameters as follows:
Pent/UL%/Triode, Screen Volts, G1 volts, Output AC Volts, Screen return point (B++ for this set) , Input db level
I can't get the images accepted, they are only 68KB JPGs, says image must be less than 1000 X 1000 !
Any help on this?
Don
OK, first block of test data in (25 spectra ).
These are all using 6L6GC SE with 385V plate B+ and 50 mA plate current and 8 OHM resistor load for a spec'd 3200 OHM primary Zload. And all are using B++ fixed supply for screen current return (thru MOSFET drain).
In the upper left corner of the charts are specifiic parameters as follows:
Pent/UL%/Triode, Screen Volts, G1 volts, Output AC Volts, Screen return point (B++ for this set) , Input db level
I can't get the images accepted, they are only 68KB JPGs, says image must be less than 1000 X 1000 !
Any help on this?
Don
Now, back on earth......
I decided this morning to back up, clean all of the circuits off of the breadboard, and build a new test amp. The Tubelab (my breadboarding system) had two complete amplifiers still wired up on it and I had started on a third. I have about 10 different experiments that I want to do here, so I needed to start fresh. I documented the two amplifiers on the breadboard, and removed them.
The first step was to build a basic SE amplifier. I built an amp very similar to my SimpleSE, but used fixed bias in the output stage. I left plenty of room around both stages for experiments. I will experiment with different driver tubes with CCS and resistive loads. I plan to explore A2 operation using PowerDrive, and of course experiment with adjustable UL and G2 bias. I have a few other ideas that work in LTspice, but have to be proven in the real world. I also have a big box full of different mosfets to try out.
The basic amplifier was built and tested. I used an Edcor OPT for initial testing because it has UL taps, and I have plenty of test data on the SimpleSE amps using this transformer. This amp does indeed work like a SimpleSE. The included photo shows the basic amp. The screen grid is simply connected to the UL tap on the transformer for this testing.
I decided this morning to back up, clean all of the circuits off of the breadboard, and build a new test amp. The Tubelab (my breadboarding system) had two complete amplifiers still wired up on it and I had started on a third. I have about 10 different experiments that I want to do here, so I needed to start fresh. I documented the two amplifiers on the breadboard, and removed them.
The first step was to build a basic SE amplifier. I built an amp very similar to my SimpleSE, but used fixed bias in the output stage. I left plenty of room around both stages for experiments. I will experiment with different driver tubes with CCS and resistive loads. I plan to explore A2 operation using PowerDrive, and of course experiment with adjustable UL and G2 bias. I have a few other ideas that work in LTspice, but have to be proven in the real world. I also have a big box full of different mosfets to try out.
The basic amplifier was built and tested. I used an Edcor OPT for initial testing because it has UL taps, and I have plenty of test data on the SimpleSE amps using this transformer. This amp does indeed work like a SimpleSE. The included photo shows the basic amp. The screen grid is simply connected to the UL tap on the transformer for this testing.
Attachments
Here is a picture taken at almost full crank. The audio analyzer is set to display "watts into 8 ohms". I backed the current and drive level off until the tube was barely glowing. I was still getting 21 watts at 5%. The bottom scope trace is the 6L6 grid and the top is the amp output. I don't have the FFT hooked up yet.
Tomorrow I will perform an autopsy on the power supply since it doesn't work very well like it is. With no signal into this amp, I get 140 mV of 120 Hz at the output. This is messing up the distortion readings at low power levels.
Tomorrow I will perform an autopsy on the power supply since it doesn't work very well like it is. With no signal into this amp, I get 140 mV of 120 Hz at the output. This is messing up the distortion readings at low power levels.
Attachments
Thanks SY
The upper trace is plate current, the lower trace is screen current
The upper trace is plate current, the lower trace is screen current
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Tubelab, you murderous $%@#&^***
And it is not even a 6L6GC! Poor old OLD 6L6G! (Also looking at the PS on your posts - doesn't work, does it!)
Horseplay aside, I enjoy these contributions; I am still not in a position timewise to do some own measurements - perhaps after all this they might not be necessary. There was something in the L-F articles that I wondered about, but that for later.
Regards
And it is not even a 6L6GC! Poor old OLD 6L6G! (Also looking at the PS on your posts - doesn't work, does it!)
Horseplay aside, I enjoy these contributions; I am still not in a position timewise to do some own measurements - perhaps after all this they might not be necessary. There was something in the L-F articles that I wondered about, but that for later.
Regards
more data
Note: The screen current measurements (lower traces) used a 5x more sensitive current probe setting than the plate current measurements ( in case someone is trying to compare the two absolutes, this goes for the previous batch too )
I generally tried to measure each %UL or Screen voltage case with the input signal level set where it just begins to provoke 3rd and higher harmonics. Then I upped the input level in increments to raise the higher harmonics as well as the power level. Its pretty clear from testing that one can get comparable distortion levels with any of the schemes by just adjusting the input level, but power output then varies too. For each case I readjusted the Vg1 bias to get 50mA idle plate current.
Next cases to do will use the 80/80 (half of primary) tap on the primary, for the return of the screen current (MOSFET drain actually). I will use a +250V floating regulated (60 pF common mode to the power line) power supply from the tap to the Mosfet drain.
Don
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Note: The screen current measurements (lower traces) used a 5x more sensitive current probe setting than the plate current measurements ( in case someone is trying to compare the two absolutes, this goes for the previous batch too )
I generally tried to measure each %UL or Screen voltage case with the input signal level set where it just begins to provoke 3rd and higher harmonics. Then I upped the input level in increments to raise the higher harmonics as well as the power level. Its pretty clear from testing that one can get comparable distortion levels with any of the schemes by just adjusting the input level, but power output then varies too. For each case I readjusted the Vg1 bias to get 50mA idle plate current.
Next cases to do will use the 80/80 (half of primary) tap on the primary, for the return of the screen current (MOSFET drain actually). I will use a +250V floating regulated (60 pF common mode to the power line) power supply from the tap to the Mosfet drain.
Don
If you look carefully at the glowing tube, you can make out "6L6GC" on the front of the tube. Just below that it says "Made in China". I pulled a pair of 6L6GC's out of a Fender Bandmaster when the customer wanted to try the new Sovteks. I have used that same pair of tubes for many life threatining experiments since then. This is one of those tubes. There are truly hard to kill! They show up on Ebay from time to time in the $3 to $5 range, and I have bought a few more just for experiments like these.
I am not sure which PS you are referring to, since there are 3 in the picture. All were working at the beginning of the day. The Fluke 407D developed the issue described above during yesterdays experiments. I am in the process of taking it apart now.
Tubelab,
No, PS in its grammatical sense - postscript! "Turn it up until it explodes, then back off a little". The 6L6 won't oblige.
I judged the 6L6 purely by the bottle - only 6L6G's used to come like that. So the Chinese use the Coke-bottle enclosure also for the -GC! Not to again digress, only how does one know that is a 6L6GC? Firstly the direct mounting on thick pins in the base lets out a little heat compared to mounting on a tongue. I think I notice the latter in the photo (will check again). Then comparatively large cooling fins on g1 frame prongs needed, larger anode "fins", etc. I did not notice that either. Anyway, back to ...
Smoking-amp,
Most interesting. I will have to print and study at my leisure. Fine work.
Regards
No, PS in its grammatical sense - postscript! "Turn it up until it explodes, then back off a little". The 6L6 won't oblige.
I judged the 6L6 purely by the bottle - only 6L6G's used to come like that. So the Chinese use the Coke-bottle enclosure also for the -GC! Not to again digress, only how does one know that is a 6L6GC? Firstly the direct mounting on thick pins in the base lets out a little heat compared to mounting on a tongue. I think I notice the latter in the photo (will check again). Then comparatively large cooling fins on g1 frame prongs needed, larger anode "fins", etc. I did not notice that either. Anyway, back to ...
Smoking-amp,
Most interesting. I will have to print and study at my leisure. Fine work.
Regards
Many of the new production tube suppliers have been known to use some loose interpretation of the tube numbering system.
In this case I had a friend with a Fender Bandmaster. He had gone through several sets of 6L6 type tubes, with some of them only lasting a few months. Several years ago Antique Electronics Supply was advertizing some Chinese "Coke Bottle" 6L6GC's. Their ad claimed "we could not blow up this tube no matter how hard we tried". I took this as a challenge, and ordered 4 of them. Two went into that Bandmaster where they lived for over a year. They were still alive when my friend wanted to try the Sovtek 6L6WXT which is also hard to kill.
I now know that these tubes were made by Shuguang and are not commonly sold in the US any more. They have been replaced by the 6L6GCR which is the same tube in round glass. The round glass must hold more heat because the new tubes are easier to glow red.
Both of these tube types have no problem operating at power levels that will melt an ordinary 6L6GB. I have run this same pair of tubes in a simpleSE amp for extended periods of time at 35 watts of plate dissipation. Are these true 6L6GC's, I don't know, but they act like it.
I have been working on the Fluke 407D power supply. The voltage knob problem was easy, a broken wire. I tested it and then put it back in the case, which doesn't fit right. I then put it back on the shelf and fired it up. As I was turning up the voltage again, the needle slammed to the right, and the poor old 6L6 got the treatment again.
The power supply is now back apart, and I found that I cracked the voltage range switch when I put the cover on. I am not sure where I can get a switch for this thing.
Revising test procedures
I have been thinking over the first 25 test results and have decided that some improvements need to be made.
The input signal level listed in dB on the charts was taken from the M-Audio sound card monitor. It doesn't allow uniform 1dB increments, and I don't know how accurate it is. Also, there is an additional gain factor unaccounted for in the SS audio amp used for grid drive ( around 21X). So from now on I am going to have a DVM on the 6L6 grid drive signal (before bias cap) so that accurate and meaningful AC drive levels in volts are recorded.
Next problem is the plate current setting at 50 mA used for all the tests so far. It was taken from the B+ supply's current readout. This supply is really too big for single tube stuff and its bottom digit is in 10 mA increments. This could lead to 20% errors in plate current setup, and who knows how accurate the bottom digit is anyway. (1.7 Amp 620V supply)
Also, for power out comparisons, I think the total DC cathode current being held constant from test to test is more useful than plate current constant. So I will install a DVM current meter in the cathode and another in the plate supply return lead too (for P-P xfmr DC current balance purposes).
I will also add a parameter for effective load resistance to the charts, and then use 4, 8, and 16 Ohm load resistors which will give effective loads of 1600, 3200, and 6400 Ohms.
And I will do a more complete scan from pentode 0%, to triode 100% in 10% increments since my 600K UL% adjust pot is a ten turn one.
I will also use a more systematic set of input drive levels, starting from a level that is just threshold for 3rd harmonic, and increasing thru several fixed % increases. This should make for more uniformly comparable charts.
I also want to set up a switch for the MOSFET drain connection (screen current) to change between the B++ supply, the 50% primary tap, and the plate tap (+250V floating boost on these last two taps).
This will obviously be quite a large set of test spectra, maybe 100s. But its important to get meaningful results that can be compared apples to apples. So consider the first 25 tests as a warmup exercise.
Much work to do! This will be spread out over the next week.
Don
I have been thinking over the first 25 test results and have decided that some improvements need to be made.
The input signal level listed in dB on the charts was taken from the M-Audio sound card monitor. It doesn't allow uniform 1dB increments, and I don't know how accurate it is. Also, there is an additional gain factor unaccounted for in the SS audio amp used for grid drive ( around 21X). So from now on I am going to have a DVM on the 6L6 grid drive signal (before bias cap) so that accurate and meaningful AC drive levels in volts are recorded.
Next problem is the plate current setting at 50 mA used for all the tests so far. It was taken from the B+ supply's current readout. This supply is really too big for single tube stuff and its bottom digit is in 10 mA increments. This could lead to 20% errors in plate current setup, and who knows how accurate the bottom digit is anyway. (1.7 Amp 620V supply)
Also, for power out comparisons, I think the total DC cathode current being held constant from test to test is more useful than plate current constant. So I will install a DVM current meter in the cathode and another in the plate supply return lead too (for P-P xfmr DC current balance purposes).
I will also add a parameter for effective load resistance to the charts, and then use 4, 8, and 16 Ohm load resistors which will give effective loads of 1600, 3200, and 6400 Ohms.
And I will do a more complete scan from pentode 0%, to triode 100% in 10% increments since my 600K UL% adjust pot is a ten turn one.
I will also use a more systematic set of input drive levels, starting from a level that is just threshold for 3rd harmonic, and increasing thru several fixed % increases. This should make for more uniformly comparable charts.
I also want to set up a switch for the MOSFET drain connection (screen current) to change between the B++ supply, the 50% primary tap, and the plate tap (+250V floating boost on these last two taps).
This will obviously be quite a large set of test spectra, maybe 100s. But its important to get meaningful results that can be compared apples to apples. So consider the first 25 tests as a warmup exercise.
Much work to do! This will be spread out over the next week.
Don
Another weakness in the tests, so far, is that one cannot tell whether the harmonics in the screen current and plate current are in phase or anti-phase. I am thinking of adding another FFT trace for the sum of these currents (cathode current), at least for selected tests. Three traces all the time will probably clutter up the graphs too much.
The reason for my interest in the phase, is to see if screen distortion is effective in siphoning off 3/2 power law distortion harmonics (due to the load Z) by routing screen current to the B++ (or to a lesser extent to the UL taps). Screen current is after all, siphoning off plate ( or really cathode ) current.
One might expect that for some setups, some null could occur, where screen current harmonics and cathode current harmonics effectively leave zero resulting plate harmonics. Looking at the tube's plate characteristics, this would show up as rounding of the knee on the left side, and with just the right amount, could linearize the load line current/voltage versus grid drive.
Don
The reason for my interest in the phase, is to see if screen distortion is effective in siphoning off 3/2 power law distortion harmonics (due to the load Z) by routing screen current to the B++ (or to a lesser extent to the UL taps). Screen current is after all, siphoning off plate ( or really cathode ) current.
One might expect that for some setups, some null could occur, where screen current harmonics and cathode current harmonics effectively leave zero resulting plate harmonics. Looking at the tube's plate characteristics, this would show up as rounding of the knee on the left side, and with just the right amount, could linearize the load line current/voltage versus grid drive.
Don
Tubelab,
Thanks for a lot of information. Fair enough, if they work like 6L6GCs, they probably are. I am fond of using 6L6s but with the manufacturing spreads found nowadays one really does not know whether any make stands out. Gone are the days when RCA etc. could simply be trusted. Our main local distributor stocks largely 6L6WXT or ...EH. 90% of his clientele are disco and guitar owners.
Smoking-amp,
Envying your test equipment and good work in the offing. Since my retirement I unfortunately don't have what I would like at my disposal any more. Looking forward to your results. (Where do you find the time?)
Regarding minima in distortion I seem to remember that because of basic tube physics 2nd and 3rd harmonic generation cannot be minimised at the same operating point; but don't ask me to prove that now. I would also like to know more about generation of high order harmonics at h.f. in the transformer steel, but will have to do that myself - nothing found elsewhere.
Good luck!
Thanks for a lot of information. Fair enough, if they work like 6L6GCs, they probably are. I am fond of using 6L6s but with the manufacturing spreads found nowadays one really does not know whether any make stands out. Gone are the days when RCA etc. could simply be trusted. Our main local distributor stocks largely 6L6WXT or ...EH. 90% of his clientele are disco and guitar owners.
Smoking-amp,
Envying your test equipment and good work in the offing. Since my retirement I unfortunately don't have what I would like at my disposal any more. Looking forward to your results. (Where do you find the time?)
Regarding minima in distortion I seem to remember that because of basic tube physics 2nd and 3rd harmonic generation cannot be minimised at the same operating point; but don't ask me to prove that now. I would also like to know more about generation of high order harmonics at h.f. in the transformer steel, but will have to do that myself - nothing found elsewhere.
Good luck!
Lets just say they are closer to being a 6L6GC than some other tubes that are sold as 6L6GC's. The Electro Harmonix 6L6GC appears to be the same tube as the Sovtek 6L6WXT. The large rectangular plate structure doesn't look like any traditional 6L6GC that I have ever seen. I have used these in guitar amps because they will live in situations that fry some tubes. I have several of them, but have yet to spend some time listening to them in a HiFi amp.
After messing with it for most of the day, my Fluke 407D power supply is back on the bench, working. It turns out that the voltage range switch was broken. I glued it back together with 5 minute epoxy. There were some wires pinched between the voltage pot and the front panel. These were probably there since the unit was new. The loose pot finally cut through the insulation causing erratic operation and random 600+ volts of output. There were some burnt wires that were apparently damaged by someone tweaking the current meter shunt. Other than these issues the unit seems to work fine with all of the original capacitors. All of the tubes are IEC brand. These are either the originals, or the unit was re-tubed with the same brand of tubes.
The 6L6 got shocked with 600+ volts again, no sparks this time but it was glowing really bright when I hit the kill switch. It is still alive! Now I can get back to the experiments.
Just before ducking:
In my 100W amp I use 600V h.t. for plates, regulated down to 500V for screens and the rest. But there is 42V odd on the cathodes. So actually the 6L6s see only 560V and 460V screen. Maybe I should go to 640V and 540V! But I get the 100W easily so let us not tempt fate. (That is 4 x 6L6 per amp)...... although 150W would be nice .....
See what you have started now!
In my 100W amp I use 600V h.t. for plates, regulated down to 500V for screens and the rest. But there is 42V odd on the cathodes. So actually the 6L6s see only 560V and 460V screen. Maybe I should go to 640V and 540V! But I get the 100W easily so let us not tempt fate. (That is 4 x 6L6 per amp)...... although 150W would be nice .....
See what you have started now!
Crystal tetrodes!
At last! I can finally contribute something practical to this thread (my lab's still not up and running from moving house). My FQP1N50 MOSFETs have finally arrived, and clutching them cold from the hands of the postman I popped a couple in the curve tracer. Now, we'd want to keep at least 20V across these devices in order to minimise capacitances, but it turns out there's another reason to keep Vds > 10V. These are crystal tetrodes! Take a look at the curves and note the behaviour around the 6V region...
At last! I can finally contribute something practical to this thread (my lab's still not up and running from moving house). My FQP1N50 MOSFETs have finally arrived, and clutching them cold from the hands of the postman I popped a couple in the curve tracer. Now, we'd want to keep at least 20V across these devices in order to minimise capacitances, but it turns out there's another reason to keep Vds > 10V. These are crystal tetrodes! Take a look at the curves and note the behaviour around the 6V region...
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.