Hello all-
I wish to make a stereo tube amplifier for my copper speakers, used in my camp at Burning Man. I would like to share my first steps here as this is the first PP amp I have built. I would very much appreciate your comments. Please go to this thread on the copper speakers. They were outside at Burning Amp 2009.
This amplifier will need to be simple and robust. It will need to generate ~70-100W of audio power per channel. To easily obtain this, I have decided to go Push-Pull.
Output tubes: I can go two basic routes- parallel push-pull using quartets of more common tubes (807, 6550, etc), or transmitting tube pairs (211, 813, GM70, etc).
My camp has power. I can get up to ~6 amps provided to run the amplifier.
I am favoring the transmitting tube route. Thoriated filaments would fit the 'Burning Man' theme. They would look very cool!
Of the thoriated tubes, the GM70 and 813 are easy to obtain and relatively inexpensive. I had some GM70's on hand and built up mono breadboard. B+ set at 900v. I chose cathode bias. I biased the GM70's for -70vdc. Each tube is drawing ~70 mA at this bias point (140 mA total). Total power consumption for this breadboard is ~255 watts.
I am using Jack Illiano's idea of using the TDA2030 amp chip and purchased his driver/splitter transformers he designed for this chip. This driver might not be used in the end, but it was easy to build for experimentation. Attached are the amplifier and power supply schematics. I am using a borrowed Edcore 10K 100W output transformer. No neg feedback yet.
Here are the THD measurements at 1kHz, into an 8 ohm load.
5Vpp (~0.38w) 0.48%
10vPP (~1.6w) 0.54 %
20vPP (~6.0w) .72%
40vPP (~24.8w) 1.9%
55vPP (~47.3w) 2.5% -just before clipping.
It takes a ~2.5v peak to peak signal to take the amp to max output (55V peak to peak) before clipping.
I hooked it up to one of the copper speakers and it sounds OK! I did a quick scope frequency response at mid-power (25w). Flat up to ~16kHz, then slowly tapers off.
The power supply- I got a tip from a local builder (Thanks, John!) to use a bi-polar power supply. It works very well. It also makes for easier parts sourcing. I get ~2vPP ripple under load. I am also using AC on the filaments. Yet hum is negligible in the speaker, thanks to push-pull operation.
Other driver ideas would be to do the normal phase splitter route. I would like more efficiency and power, I might try 813's or 828's in UL mode. Experimentation continues.
Thank you- Kent
I wish to make a stereo tube amplifier for my copper speakers, used in my camp at Burning Man. I would like to share my first steps here as this is the first PP amp I have built. I would very much appreciate your comments. Please go to this thread on the copper speakers. They were outside at Burning Amp 2009.
This amplifier will need to be simple and robust. It will need to generate ~70-100W of audio power per channel. To easily obtain this, I have decided to go Push-Pull.
Output tubes: I can go two basic routes- parallel push-pull using quartets of more common tubes (807, 6550, etc), or transmitting tube pairs (211, 813, GM70, etc).
My camp has power. I can get up to ~6 amps provided to run the amplifier.
I am favoring the transmitting tube route. Thoriated filaments would fit the 'Burning Man' theme. They would look very cool!
Of the thoriated tubes, the GM70 and 813 are easy to obtain and relatively inexpensive. I had some GM70's on hand and built up mono breadboard. B+ set at 900v. I chose cathode bias. I biased the GM70's for -70vdc. Each tube is drawing ~70 mA at this bias point (140 mA total). Total power consumption for this breadboard is ~255 watts.
I am using Jack Illiano's idea of using the TDA2030 amp chip and purchased his driver/splitter transformers he designed for this chip. This driver might not be used in the end, but it was easy to build for experimentation. Attached are the amplifier and power supply schematics. I am using a borrowed Edcore 10K 100W output transformer. No neg feedback yet.
Here are the THD measurements at 1kHz, into an 8 ohm load.
5Vpp (~0.38w) 0.48%
10vPP (~1.6w) 0.54 %
20vPP (~6.0w) .72%
40vPP (~24.8w) 1.9%
55vPP (~47.3w) 2.5% -just before clipping.
It takes a ~2.5v peak to peak signal to take the amp to max output (55V peak to peak) before clipping.
I hooked it up to one of the copper speakers and it sounds OK! I did a quick scope frequency response at mid-power (25w). Flat up to ~16kHz, then slowly tapers off.
The power supply- I got a tip from a local builder (Thanks, John!) to use a bi-polar power supply. It works very well. It also makes for easier parts sourcing. I get ~2vPP ripple under load. I am also using AC on the filaments. Yet hum is negligible in the speaker, thanks to push-pull operation.
Other driver ideas would be to do the normal phase splitter route. I would like more efficiency and power, I might try 813's or 828's in UL mode. Experimentation continues.
Thank you- Kent
Attachments
Last edited:
And I bet they'll look even better after 30 days of burn-in
Funny, not really. Childish comment, possibly. Idiocracy, definately.
What's your problem???
Last edited:
Hey folks- thanks for the comments!
Yup, I intend to increase the temp on the playa...that will be another project using GM100 tubes!
The speakers worked great- all burnt-in and I am confident they will work well again this year. Getting an amp to drive them properly is another matter.
I am still trying to plot out the AC loadline and taking measurements to see if I can squeeze a bit more power. My goal is at least 70 watts, and not to do class B (I wanna use them for hi-fi down the road). More info as I figure it out.
If it proves too hard to get 70+ watts w/o running the tubes at 1200v+, I might try 828 tubes and run ultralinear at 750V B+. Anyone ever played with that tube?
Yup, I intend to increase the temp on the playa...that will be another project using GM100 tubes!
The speakers worked great- all burnt-in and I am confident they will work well again this year. Getting an amp to drive them properly is another matter.
I am still trying to plot out the AC loadline and taking measurements to see if I can squeeze a bit more power. My goal is at least 70 watts, and not to do class B (I wanna use them for hi-fi down the road). More info as I figure it out.
If it proves too hard to get 70+ watts w/o running the tubes at 1200v+, I might try 828 tubes and run ultralinear at 750V B+. Anyone ever played with that tube?
You might want to look at screen driven P-P sweep tubes. The voltages will be easier to deal with. You can get ~100W/channel with 5k iron and 550-600V B+. SY and George will probably have something to say about it. I will be working on something like that some time this year, but I will be building a custom SMPS to power mine, something not easy to duplicate.
That could be a viable solution down the road. Yet I want to explore thoriated tungsten filaments for now. More 'Burning Amp' feel for Burning Man!
I have been following George's powerdrive with some interest. I have some 2SK700 about. I dunno if his can drive GM70's directly (I need min 250VPP), but if I go UL then his looks very attractive. I would like some tubes in the front end....
I've sourced some 828's so I hope to play with them soon.
I have been following George's powerdrive with some interest. I have some 2SK700 about. I dunno if his can drive GM70's directly (I need min 250VPP), but if I go UL then his looks very attractive. I would like some tubes in the front end....
I've sourced some 828's so I hope to play with them soon.
It's no doubt that the thoriated filaments look cool, but they are a real power hog - a GM70 P-P rig will be blowing 240W in filament power alone. I'd rather use that power to displace some air. I plan to do a GM-70 amp myself, but a 30W or so SE beast is pretty much the limit of my ambition. I was thinking of going the monoblock route , again with a custom SMPS for power and HV sequencing.
I was able to crank up the bipolar supply to +550v, -550v and achieved 70w just before clipping. But the amp is sucking serious power to achieve this, approaching 300w, and this is for just one channel!
I have some 828's on their way, I'll see how they work and post.
Ya know, wrenchone, the screen drive idea might be implemented in the end, it would weigh less and consume alot less power. Ive been following the PP tubelab threads. I could use the TDA 2030/interstage trans combo to drive just about anything (within reason).
-Kent
I have some 828's on their way, I'll see how they work and post.
Ya know, wrenchone, the screen drive idea might be implemented in the end, it would weigh less and consume alot less power. Ive been following the PP tubelab threads. I could use the TDA 2030/interstage trans combo to drive just about anything (within reason).
-Kent
828 tube experiments.....
Been quite a while since I posted, haven't had much time lately.
I experimented with 828 tubes. First, ultra-linear; then in pentode mode. Circuit is the same as the GM70 PP, just replaced with 828's and hooked up the screens and suppressors as necessary.
The UL setup sounded grainy, and had limited power (~35w), though bass response subjectively sounded better than the GM70's. Seems the 750v screen limit restricts power too much. I did not take detailed measurements.
Pentode mode sounded better. Screens fed with 375V (the max my other supply will go), plate ~980v. Set up the bias to -34v. 1khz signal 2vPP input. Plate current for both tubes was ~180mA. Screen current for both tubes ~22mA. I got ~60vPP (~56w) into an 8 ohm load before clipping. At this point, my screen supply developed a problem and I could not take distortion measurements.
Subjective listening impressions-
Ultralinear sounded 'sharp' and 'tizzy'. More bass that the GM70's. In pentode mode, the bass control is even more pronounced, and the tizzy-ness has decreased somewhat. Still, the GM70's had a smoother sound to them.
Attached is a pic of the circuit and some scope shots. Though you can barely see it, there is some slight 'lumpiness' to the sinewave. The peaks become slightly 'triangulated'. This seemed to occur when the control grids started drawing current. Looks like the 828's don't like that; at least in this circuit. Also attached is a scope pic of the setup w/ a 10kHz signal; one can see this effect better.
At this point, I am considering wrenchone's advice and go with more modern oxide cathode tubes. I found out that my amp also will do sound duty for a nightly mini-cinema. As the amp will be under the movie screen in front, bright tubes don't work too well with projection!
Pete Milletts' sweep tube amp <http://www.diyaudio.com/forums/tubes-valves/151206-posted-new-p-p-power-amp-design.html> circuit looks like a possible route if I abandon TDA2030 drive. I still like the TDA drive for it's simplicity. If it dies, I can easily swap in a fresh driver board.
I picked up some 829B's recently. I will experiment with these next, tying the two sections together. Maybe play with a few sweep tubes as well.
-Kent
Been quite a while since I posted, haven't had much time lately.
I experimented with 828 tubes. First, ultra-linear; then in pentode mode. Circuit is the same as the GM70 PP, just replaced with 828's and hooked up the screens and suppressors as necessary.
The UL setup sounded grainy, and had limited power (~35w), though bass response subjectively sounded better than the GM70's. Seems the 750v screen limit restricts power too much. I did not take detailed measurements.
Pentode mode sounded better. Screens fed with 375V (the max my other supply will go), plate ~980v. Set up the bias to -34v. 1khz signal 2vPP input. Plate current for both tubes was ~180mA. Screen current for both tubes ~22mA. I got ~60vPP (~56w) into an 8 ohm load before clipping. At this point, my screen supply developed a problem and I could not take distortion measurements.
Subjective listening impressions-
Ultralinear sounded 'sharp' and 'tizzy'. More bass that the GM70's. In pentode mode, the bass control is even more pronounced, and the tizzy-ness has decreased somewhat. Still, the GM70's had a smoother sound to them.
Attached is a pic of the circuit and some scope shots. Though you can barely see it, there is some slight 'lumpiness' to the sinewave. The peaks become slightly 'triangulated'. This seemed to occur when the control grids started drawing current. Looks like the 828's don't like that; at least in this circuit. Also attached is a scope pic of the setup w/ a 10kHz signal; one can see this effect better.
At this point, I am considering wrenchone's advice and go with more modern oxide cathode tubes. I found out that my amp also will do sound duty for a nightly mini-cinema. As the amp will be under the movie screen in front, bright tubes don't work too well with projection!
Pete Milletts' sweep tube amp <http://www.diyaudio.com/forums/tubes-valves/151206-posted-new-p-p-power-amp-design.html> circuit looks like a possible route if I abandon TDA2030 drive. I still like the TDA drive for it's simplicity. If it dies, I can easily swap in a fresh driver board.
I picked up some 829B's recently. I will experiment with these next, tying the two sections together. Maybe play with a few sweep tubes as well.
-Kent
Attachments
Amplifier update- driver problem
Amplifier Update
I had some time to play with the amp in early May. I was having some fun tube rolling; Eimac 100TL's, Svetlana SV811-10's, 829's; even tried some old 852's (couldn't, one turned out to be gassy).
I realize now that the limitation of power is the TDA2030; it doesn't put out enough voltage swing. After the interstage/splitter transformer, I get 240V PP per tube. For driving big triodes into A2, I need more. So a bigger chip amp? Well...
I think I discovered a problem using a chip amp as a driver: it doesn't like the abrupt impedence change the tubes present it going into A2.
I put a current sense resistor from the interstage tran primary low side to ground. Connected my scope ch2 probe on the high side of the resistor to see waveform when the output tubes went into A2. Ch1 connected to my output load resistor.
Please refer to the 'kinks' in the scope pics in the post above.
Looking at these two waveforms together on my scope, I noticed that the kinks in output waveform happen at the same time when the tube goes into A2.
To describe image in words: As the sine wave trace goes up, it changes angle abruptly, peaks, then changes abruptly again as the trace goes down. These kinks in the sine wave occur on the same y-axis as the current starts to flow to the grids. This repeats in the negative part of the output waveform; the effect is symmetrical.
Upon closer examination, the downward slope even has a little oscillation.
When turning down the input signal to where grid current stops, these 'kinks' disappear.
I am sorry my description is so amateurish, but I'm not an EE and I didn't take pics. The amp is apart now. I don't know why I didn't notice this effect before. I could have damaged the TDA 2030, but it worked fine as long as output tubes stay out of A2.
So it appears that there is a flaw in using chip amps to drive big tubes. I will explore other driver ideas. I might try tubelab's source-follower power drive.
This project is on hold as I have no time to get it done for Burning Man in two months. Next year...! Pic below running 100TL's.Nice and bright!
Amplifier Update
I had some time to play with the amp in early May. I was having some fun tube rolling; Eimac 100TL's, Svetlana SV811-10's, 829's; even tried some old 852's (couldn't, one turned out to be gassy).
I realize now that the limitation of power is the TDA2030; it doesn't put out enough voltage swing. After the interstage/splitter transformer, I get 240V PP per tube. For driving big triodes into A2, I need more. So a bigger chip amp? Well...
I think I discovered a problem using a chip amp as a driver: it doesn't like the abrupt impedence change the tubes present it going into A2.
I put a current sense resistor from the interstage tran primary low side to ground. Connected my scope ch2 probe on the high side of the resistor to see waveform when the output tubes went into A2. Ch1 connected to my output load resistor.
Please refer to the 'kinks' in the scope pics in the post above.
Looking at these two waveforms together on my scope, I noticed that the kinks in output waveform happen at the same time when the tube goes into A2.
To describe image in words: As the sine wave trace goes up, it changes angle abruptly, peaks, then changes abruptly again as the trace goes down. These kinks in the sine wave occur on the same y-axis as the current starts to flow to the grids. This repeats in the negative part of the output waveform; the effect is symmetrical.
Upon closer examination, the downward slope even has a little oscillation.
When turning down the input signal to where grid current stops, these 'kinks' disappear.
I am sorry my description is so amateurish, but I'm not an EE and I didn't take pics. The amp is apart now. I don't know why I didn't notice this effect before. I could have damaged the TDA 2030, but it worked fine as long as output tubes stay out of A2.
So it appears that there is a flaw in using chip amps to drive big tubes. I will explore other driver ideas. I might try tubelab's source-follower power drive.
This project is on hold as I have no time to get it done for Burning Man in two months.
Next year...! Pic below running 100TL's.Nice and bright!Attachments
Late to this thread
I was referred here from a thread I started on AudioKarma about pentodes. Coincidentally, I have an article in Issue #23 of Make Magazine which is a build project for a stereo chip amp. I used the TDA2030A in some of the prototypes.
I have not experimented with Jack's driver transformer, but I wonder if the sudden impedance change is what's to blame. Maybe the transformer needs resistive loading on the secondary so the chip provides current all the time instead of sudden conduction on part of the waveform. A bigger chip like LM3886 might be required. I would also look at the chip's NFB circuit. They usually run 60-80 dbs of NFB and this should overwhelm any impedance change in the load.
I've worked with the GM70 before on the bench but never finished a usable 2-channel amp. Too heavy! I did run it in Class A2. IIRC, 1500V B+ -150V bias, 110ma, 9K plate load. The power output was 32W from 1 tube at 6% THD and the THD at 10W was very low. I was impressed. Of course when the grid swings + it wants a huge slug of current so I had to drive it with a direct coupled CF (1626 IIRC or 12BX7).
Right now I'm working with GU-50s, which I'm coming to like a lot. If this weekend's experiments work out I may have an interesting SE topology to talk about.
I was referred here from a thread I started on AudioKarma about pentodes. Coincidentally, I have an article in Issue #23 of Make Magazine which is a build project for a stereo chip amp. I used the TDA2030A in some of the prototypes.
I have not experimented with Jack's driver transformer, but I wonder if the sudden impedance change is what's to blame. Maybe the transformer needs resistive loading on the secondary so the chip provides current all the time instead of sudden conduction on part of the waveform. A bigger chip like LM3886 might be required. I would also look at the chip's NFB circuit. They usually run 60-80 dbs of NFB and this should overwhelm any impedance change in the load.
I've worked with the GM70 before on the bench but never finished a usable 2-channel amp. Too heavy! I did run it in Class A2. IIRC, 1500V B+ -150V bias, 110ma, 9K plate load. The power output was 32W from 1 tube at 6% THD and the THD at 10W was very low. I was impressed. Of course when the grid swings + it wants a huge slug of current so I had to drive it with a direct coupled CF (1626 IIRC or 12BX7).
Right now I'm working with GU-50s, which I'm coming to like a lot. If this weekend's experiments work out I may have an interesting SE topology to talk about.
Regarding drivers, I'll share today's experiment, excerpted from the AK thread:
It's been all pentodes all day here. Here's some promising results on the Russian 6J9P (similar to 6688/E180F):
400V B+
200V plate
160V screen
1.6V bias (105 Ohm unbypassed Rk)
15ma plate
13200 Ohm plate load.
Pdiss = 3W
1VRMS input
71VRMS output
Gain = 71
THD = 1.1%
This is about where I need to be for a driver. It's pretty amazing to be swinging 100V peak output with 1.1% THD off of only 1VRMS input. I cold probably get the THD under 1% with a little more tweaking but this will do for now. You could drive just about ANYTHING with this.
Pentodes rock. Of course I had to carefully wire up a special socket with all of the resistors slammed right up to the pins to stop oscillations, but that's the price of high Gm.
** EDIT **
Comparing to Pete Millet's results wit hteh similar 6688 pentode, he uses a 300V B+ and 8K load and gets 5% THD at 70VRMS out compared to my 1.1% THD. I'd be worried if my results were higher THD, but I'm going in the right direction.
I'm running the tube at max Pdiss, so I'll probably increase the plate resistor and Rk and bring the Pdiss down about 15 - 20% for longevity.
BTW, this tube is NOS USSR and is $2 in small quantities.
** EDIT**
This setup:
2.2% THD at 100VRMS out
5% THD at 130VRMS out (185V peak out of a 200V plate voltage)
I love triodes but you just can't run them to the rails like you can a pentode.
It's been all pentodes all day here. Here's some promising results on the Russian 6J9P (similar to 6688/E180F):
400V B+
200V plate
160V screen
1.6V bias (105 Ohm unbypassed Rk)
15ma plate
13200 Ohm plate load.
Pdiss = 3W
1VRMS input
71VRMS output
Gain = 71
THD = 1.1%
This is about where I need to be for a driver. It's pretty amazing to be swinging 100V peak output with 1.1% THD off of only 1VRMS input. I cold probably get the THD under 1% with a little more tweaking but this will do for now. You could drive just about ANYTHING with this.
Pentodes rock. Of course I had to carefully wire up a special socket with all of the resistors slammed right up to the pins to stop oscillations, but that's the price of high Gm.
** EDIT **
Comparing to Pete Millet's results wit hteh similar 6688 pentode, he uses a 300V B+ and 8K load and gets 5% THD at 70VRMS out compared to my 1.1% THD. I'd be worried if my results were higher THD, but I'm going in the right direction.
I'm running the tube at max Pdiss, so I'll probably increase the plate resistor and Rk and bring the Pdiss down about 15 - 20% for longevity.
BTW, this tube is NOS USSR and is $2 in small quantities.
** EDIT**
This setup:
2.2% THD at 100VRMS out
5% THD at 130VRMS out (185V peak out of a 200V plate voltage)
I love triodes but you just can't run them to the rails like you can a pentode.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.