The topic of Single-Ended OTL's has come up a lot lately, so I decided to give it a go. It still needs work, but all and all what do you think? Should put out about 20Wpk, dissipation per tube is aprox 40W. Wanted to do it open loop, but the Zout would be about 6ohms, which is too high. The output tubes are 6c33c-b's I forgot to add that text to the schematic.
An externally hosted image should be here but it was not working when we last tested it.
I see some problems in the design, frankly.
First of all you shouldn't return feedback directly to the input: in that case the nfb would be influenced by the output impedance of the previous audio device in the signal chain.
Also 10k grid stopper resistors on the 6S33 tubes seem a bit high.
20mA at that voltage in a 6DJ8 also is wrong, go for 10mA. Use two resistors in series for the cathode: an unbypassed one in series with a bypassed one, the sum of the two being the cathode resistor value needed for your biasing. Inject feedback across the unbypassed resistor and not the input.
Watch out for the 30W you'll dissipate in the transistor below the output tubes!
First of all you shouldn't return feedback directly to the input: in that case the nfb would be influenced by the output impedance of the previous audio device in the signal chain.
Also 10k grid stopper resistors on the 6S33 tubes seem a bit high.
20mA at that voltage in a 6DJ8 also is wrong, go for 10mA. Use two resistors in series for the cathode: an unbypassed one in series with a bypassed one, the sum of the two being the cathode resistor value needed for your biasing. Inject feedback across the unbypassed resistor and not the input.
Watch out for the 30W you'll dissipate in the transistor below the output tubes!
I dunno. Looks like a massive cathode follower with some sand across the speaker. If I were trying such a thing I'd be using an iron and copper device there replacing holes with Henries.
You will need a LOT of drive voltage swing. Can that 6DJ8 do it with a sand CCS as a plate load? I'd be using a choke there too, doubles the available theoretical swing.
Best way to know is build it and let us know how it works.
You will need a LOT of drive voltage swing. Can that 6DJ8 do it with a sand CCS as a plate load? I'd be using a choke there too, doubles the available theoretical swing.
Best way to know is build it and let us know how it works.
I First had it going to the cathode resistor with it unbypassed, but couldn't get it to simulate that way. I figured since Rf would be much larger the Zo of the amp. Zout's effect on the feedback would be ignorable.
Thanks for catching that, The curves I was using did not have the Pmax line.
The power transistor I'm using is rated at 120W so I'm not too worried. It will probably be darlington paired with a smaller transistor just to shrink the size of the base resistor.
Theres nothing wrong with using sand for a current source. I couldn't find a piece of iron with an even comparable bandwidth to the SS CCS.
Well, 1.6A peak into an 8 ohm load would be 12.8V peak on the load. The 6DJ8 can supply a 34V peak signal to the grid of the output tubes. With the active load at that B+. As long as the gain of the output is above .37 then it should be capable of driving it
That would be a concern, But I've built a 6c33c SET cathode follower amp. The driver stage was a 6su7gty with active load idling at a mere 2mA. And it drives the 6c33c past 30khz before cutoff. The pull down resitor of the 6c33c was quite a bit larger then the one in this design though. I figure the 6DJ8 idling at 15mA will be adaquate to drive it and keep the Pd at an exceptable level. That's also why I went with larger value grid resistors, since miller effect doesn't plauge cathode followers.
I don't think that a cathode follower is necessary, the input capacitance of a cathode follower stage as is used here is Cgp + (1-G)*Cgk which with a 6C33C will be something like 31 + 10 = 41pF and for 4 tubes in parallell ~120pF is that so high?
Where you have got the information about grid current in 6C33C I don't know, it is however not true, grid current is negligible up to practically 0V and I for instance use 68k grid leaks with very good results.
I would however reduce the series grid resistors to 1k or something similar, anything higher is not necessary.
Regards Hans
Hi Giaime,
Interesting and unfortunately it seems not an unusual experience. It seems that defective 6C33C tubes are still available in Europe and US. I had hoped that the supply of defective tubes or rejects was exhausted by now but it seems that it is not so, I heard recently from someone who bought 4 6C33C from an European vendor and all 4 of them where defective with either low emission or grid to cathode short.
I can really recommend to buy directly from Russia, quality, (speaking from my own experience) is excellent as is customer service and prices are the lowest in the market, the only problem is that there is a minimium order sum but if you have some friends who need tubes at the same time it is not a problem.
I have bought 6C33C from 3 sources, San-Ei in Japan, (they where the first to import 6C33C then relabeled EC33C to the western world), tubes.ru and GS tubes, I had one faulty tube of almost 100 and that one was a grid to cathode short. Viktor Khomenko at BAT say that you should expect about 1% failure not more.
In my OTL amp I use a simple split load inverter to directly drive the output tubes with very good results and I can easily see that there is practically no grid current even close to 0V. You should expect to see a little bit of grid current close to 0V with a but it should be in the uA range and not more. The original russian datasheet says that the grid leak should be 200kohm or less so it is not different from any other power tube,
Regards Hans
Interesting and unfortunately it seems not an unusual experience. It seems that defective 6C33C tubes are still available in Europe and US. I had hoped that the supply of defective tubes or rejects was exhausted by now but it seems that it is not so, I heard recently from someone who bought 4 6C33C from an European vendor and all 4 of them where defective with either low emission or grid to cathode short.
I can really recommend to buy directly from Russia, quality, (speaking from my own experience) is excellent as is customer service and prices are the lowest in the market, the only problem is that there is a minimium order sum but if you have some friends who need tubes at the same time it is not a problem.
I have bought 6C33C from 3 sources, San-Ei in Japan, (they where the first to import 6C33C then relabeled EC33C to the western world), tubes.ru and GS tubes, I had one faulty tube of almost 100 and that one was a grid to cathode short. Viktor Khomenko at BAT say that you should expect about 1% failure not more.
In my OTL amp I use a simple split load inverter to directly drive the output tubes with very good results and I can easily see that there is practically no grid current even close to 0V. You should expect to see a little bit of grid current close to 0V with a but it should be in the uA range and not more. The original russian datasheet says that the grid leak should be 200kohm or less so it is not different from any other power tube,
Regards Hans
Jeb-D,
If you must use a 9 pin miniature as a driver consider experimenting with a 6CL6. I have a P SET amp (eight 6080 triodes in parallel) that has high miller C breadboarded that I need to get back to after quite a few years tying up an entire workbench. I was planning to use the metal octal near equivalent of the 6CL6, the 6AG7 as a driver. Thjese are both video power tubes that can swing a large voltage into a rather low plate load.
I have yet to be impressed with the sonic character of the 6DJ8. Maybe it's OK for DJ work, the frame grid being robust enough to survive life on the road.
If you must use a 9 pin miniature as a driver consider experimenting with a 6CL6. I have a P SET amp (eight 6080 triodes in parallel) that has high miller C breadboarded that I need to get back to after quite a few years tying up an entire workbench. I was planning to use the metal octal near equivalent of the 6CL6, the 6AG7 as a driver. Thjese are both video power tubes that can swing a large voltage into a rather low plate load.
I have yet to be impressed with the sonic character of the 6DJ8. Maybe it's OK for DJ work, the frame grid being robust enough to survive life on the road.
That is because the feedback will be positive in that case, the cathode as input is in phase with the anode, cmp grounded grid so if you feed a cathode follower with a grounded cathode amp as you here, the output will be in phase with the cathode of the input tube.
Connecting the feedback to the grid of the input tube gives correct phase but as Giaime points out the gain and feedback will be dependant on the output impedance of the preceding stage.
Regards Hans
Not a bad idea. I did some simulations with it and in order for a diff amp to work with CCS's for plate loads. I have to attach an AC dummy load to the plate of the second triode(non inverting input), equivalent to the AC load on the input triode(inverting input). Otherwise the it will just be passing a continuous current, which will cause no voltage change on the cathode resistor, which means no feedback. Also, if it's a different value load or biasing, the feedback won't be a symmetrically inverted signal.
For now the input will stay SE. I'm quite confident that the effect of Zout on the feedback will be so small, due to a large value of Rf, that it can be ignored. It's been done on plenty of amps before. If there is noticeable effects once built, I'll go with the differential input.
You have a nice website, by the way.
Yeah, your right. What was happening is, the output was out of phase with the cathode on the 6DJ8 while in open loop. But, once feedback was applied, the non inverted AC signal that was on the cathode, saw the output as ground(since there was only as small resistance path to ground thru Rf). Leaving only the output voltage on the cathode. Which makes the cathode voltage in phase with the plate. It created some pretty crazy looking waveforms.
Oh, my bad. I read it so fast the first time I didn't notice you guys were refering to Zout of the signal source, loading the pull down resistor of the input. Therefore reducing the amount of feedback. I was thinking some other crazyness. I'll definately have to keep that in consideration.
Is there any kind of standardized Zout for signal sources, more specifically CD players? Or can it vary greatly depending upon manufacture.
Thanks for the input guys!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.