PSPICE CIRCLOTRON working simulation
OK, I have a working prototype simulation running of a circlotron amplifier.
This circuit uses the same feedback scheme as the totem pole circuit I posted earlier. I simply changed the output to a circlotron and took the feedback from the side of the load that provided negative (not positive) feedback.
Without the feeback the output impedance was 40 (or 80 - I forgot!) ohms. With the NFB it drops to about 1 ohm (roughly) using parallel triodes of the 6H13C.
The triodes are set up such that each has an independent bias adjustment using a 2 Megohm pot. This should allow some degree of balancing.
Here's the idea: With all tubes in the circuit and the pots all on max (2 Meg), the main bias supply is adjusted so that the tube that draws the most current is at about 75% of the desired Ia. All tubes except one pair are then removed. Then in pairs the 2 Meg pots are used to raise the grid voltages on each triode to the level needed for the correct Ia. Once the first pair is balanced the second pair can be inserted into the circuit and re-balanced, etc., etc.
I'm amazed how quickly this model went. The response looks great, and there is total symetry in the output, unlike the totem pole.
I now need to clean it up and add more tubes.
OK, I have a working prototype simulation running of a circlotron amplifier.
This circuit uses the same feedback scheme as the totem pole circuit I posted earlier. I simply changed the output to a circlotron and took the feedback from the side of the load that provided negative (not positive) feedback.
Without the feeback the output impedance was 40 (or 80 - I forgot!) ohms. With the NFB it drops to about 1 ohm (roughly) using parallel triodes of the 6H13C.
The triodes are set up such that each has an independent bias adjustment using a 2 Megohm pot. This should allow some degree of balancing.
Here's the idea: With all tubes in the circuit and the pots all on max (2 Meg), the main bias supply is adjusted so that the tube that draws the most current is at about 75% of the desired Ia. All tubes except one pair are then removed. Then in pairs the 2 Meg pots are used to raise the grid voltages on each triode to the level needed for the correct Ia. Once the first pair is balanced the second pair can be inserted into the circuit and re-balanced, etc., etc.
I'm amazed how quickly this model went. The response looks great, and there is total symetry in the output, unlike the totem pole.
I now need to clean it up and add more tubes.
An externally hosted image should be here but it was not working when we last tested it.
4 tubes, 60 watts into 8 ohms, 0.25 ohms output impedance
I'm up to 4 tubes (8 triodes, 4 per side).
The output impedance is now less than 0.25 ohms. The power into 8 ohms is about 60 watts. The distortion is low, not sure how low yet..
The LTPs now have CCSs. This further linerarized the performance and eliminated the large negative bias voltage. This is rockin'!!!
Does anyone have an opinion on the method of current balancing? Will it work? The only disadvantage I see is a separate coupling cap is required for each ttiode, but so what? If all the coupling caps are identical I see no problem other than expense.
Here's the latest overall (it's getting big and can no longer be read). It's also starting to appear that 8 tubes per side might be sufficient for more power than I need. A closeup of the power stage to show the biasing scheme follows.
I'm up to 4 tubes (8 triodes, 4 per side).
The output impedance is now less than 0.25 ohms. The power into 8 ohms is about 60 watts. The distortion is low, not sure how low yet..
The LTPs now have CCSs. This further linerarized the performance and eliminated the large negative bias voltage. This is rockin'!!!
Does anyone have an opinion on the method of current balancing? Will it work? The only disadvantage I see is a separate coupling cap is required for each ttiode, but so what? If all the coupling caps are identical I see no problem other than expense.
Here's the latest overall (it's getting big and can no longer be read). It's also starting to appear that 8 tubes per side might be sufficient for more power than I need. A closeup of the power stage to show the biasing scheme follows.
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.
Here's something to think about- as the tubes warm up the bias on them will change. I can think of an easier method to measure the bias. Get a bunch of chassis-mount tip jacks that fit your meter probes. Install 2 by each power tube, and one on the negative side of each power supply (same as the speaker terminals. This way you can measure the voltage drop across the cathode resistors, and thus set the bias.
If you were feeling industrious, you could set up a multi-deck/position switch so you could test and set the tubes on the fly. *That* would be really trick. The easier the amp is to run, the longer you will use it.
You may have a couple of bugs in your sim. Unless you have a higher impedance load and B+, its going to be a trick to get 60 watts out of only 8 tube sections. If you are looking for 60 watts, expect to run a few more tubes.
The chassis mount tip jacks or test points across the cathode resistors was part of the plan - that's how I intended to measure the bias as I adjust it.
I agree the power output of the sim is high. When you say I'll need a few more tubes for 60 watts into 8 ohms, is that based on amps that have used tubes that were not balanced or tubes that were balanced. The better the bias is balanced between the output triodes the more power it will put out.
The multi-position switch was also on the agenda. In fact in another thread I asked if anyone knew where I could source a 4 pole, multi-throw switch so I could display the bias currents of each triode.
The point about the tubes changing characteristics as they warm up is well taken. I plan to let the tubes warm up before I set the "final" bias. As I add each new pair I'll let them warm up. I'm also looking into auto-bias curcuits. If each tube has the DC isolation that a dedicated coupling cap provides I see no reason that each tube couldn't have its own auto-bias circuit. I need to do some more research on this.
Thanks for the input. I'm going to use 12 tubes per channel - that was the original idea and I'm sticking to it.
Any suggestions about fusing and over voltage protection? I have gas discharge tubes, fuses, and MOVs at my disposal and it seems to me if these devices were placed in the right places that the amp could be prevented from hurting the speakers if it failed. I'm examining failure modes but could use some help.
Last edited:
I'm now seeing why I need more tubes. The plate dissipation is being exceeded in the sim at max power. However, it would seem that for short periods of time the amp can deliver the stated power, but not continuously or the tubes will burn up. This assumes the cathode can generate the peak currents needed.
At least that's what it seems like based on the simulation.
At least that's what it seems like based on the simulation.
Shallco makes switches that will do the job...
If you rate the fuse for the B+ transformer appropriately, you will find it very effective at speaker protection. I would avoid fuses in the transformer secondaries- if one blows and the other half of the amplifier has lost bias, the amp can toast a speaker with the resulting DC. OTOH, if you have lost bias on both sides, but the fuse shuts down the B+ transformer, no damage will result to the speaker.
If you are still worried, a fuse in series with the speaker will sort things out. Its worthy of note that if the amp has a problem, like an arcing or shorted tube, the result will be a DC event. So its the woofers that there might be any concern about. With 12 output tubes, the B+ transformer fuse is going to be in the 3.5A to 4A region.
If you rate the fuse for the B+ transformer appropriately, you will find it very effective at speaker protection. I would avoid fuses in the transformer secondaries- if one blows and the other half of the amplifier has lost bias, the amp can toast a speaker with the resulting DC. OTOH, if you have lost bias on both sides, but the fuse shuts down the B+ transformer, no damage will result to the speaker.
If you are still worried, a fuse in series with the speaker will sort things out. Its worthy of note that if the amp has a problem, like an arcing or shorted tube, the result will be a DC event. So its the woofers that there might be any concern about. With 12 output tubes, the B+ transformer fuse is going to be in the 3.5A to 4A region.
Thanks for the tips. I'll look for the Shallco switches. Do you know who would carry those?
BTW, I have 12 tubes now in parallel in the simulation just for kicks. The output impedance is extremely low, but I can't get the output voltage high enough to get as much power as I should be able to get into a fixed 8 ohm load. The total supply voltage is 300 (150 + 150) volts, but I can only get a swing of about 60 volts on the output in spite of the fact the grids are getting a high drive voltage. Adding tubes doesn't raise the max output voltage into a fixed 8 ohm load after a certain point (it seems), it just lowers the output impedance.
This is the latest circlotron although I realize it's too big to be legible.
BTW, I have 12 tubes now in parallel in the simulation just for kicks. The output impedance is extremely low, but I can't get the output voltage high enough to get as much power as I should be able to get into a fixed 8 ohm load. The total supply voltage is 300 (150 + 150) volts, but I can only get a swing of about 60 volts on the output in spite of the fact the grids are getting a high drive voltage. Adding tubes doesn't raise the max output voltage into a fixed 8 ohm load after a certain point (it seems), it just lowers the output impedance.
This is the latest circlotron although I realize it's too big to be legible.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Well, it still appears that the sim has some bugs. So that 60V, is that peak to peak, peak or RMS?
Good luck, keep us posted. I have a pair of Manley mono block 250 watt amp that uses 10 output tube each. As well as some 100 and 120 watt twin mono block that have 8 tubes each with a bridge switch to enable 200 WPC and 240 WPC. I have to say, if you can test and grade a large pool of tubes, and install closley matching sets, the result is impressive. But if you can't do the grading and sorting, you could be in for an ear bleed.
It's 60 volts peak to peak
It's a brick wall. And an even bigger problem is that once you reach clipping it does some strange things - it doesn't just flat top.
I was gaining power up to 4 tubes, but after that subsequent tubes don't increase output voltage (or power) into a fixed 8 ohm load.
I'm not giving up. I know there is a way to get this to work. I plan to model the Kimmel circuit next and see what the results are.
In any event I know I can do something with the totem pole if the circlotron fails (or if I fail to successfully model one).
Last edited:
I could not agree with you more.
"He that hath an ear, let him hear"
or better say, see, the attachment
Currents in two unmatched paralleled tubes
Initial currents are not that far apart.
Design decisions made upon "simulateAbility"...
Attachments
Last edited:
60V P-P is 30 V peak, 21.21 VRMS. That's about 56 watts into 8 ohms.It's 60 volts peak to peak
It's a brick wall. And an even bigger problem is that once you reach clipping it does some strange things - it doesn't just flat top.
I was gaining power up to 4 tubes, but after that subsequent tubes don't increase output voltage (or power) into a fixed 8 ohm load.
I'm not giving up. I know there is a way to get this to work. I plan to model the Kimmel circuit next and see what the results are.
In any event I know I can do something with the totem pole if the circlotron fails (or if I fail to successfully model one).
Take a look at what the grids of the power tubes look like at clipping or whatever it is. Its been my experience that if you want to get power out of these tubes, you want to drive them into grid current- which is easy to do with a direct-coupled driver, but tricky with capacitive coupling. Its my bet that the driver is running out of gas with the huge capacitive load that it has.
FWIW, without changing the B+ voltages, you *do* get more power as you add tubes, because the output section gets more efficient- more of the power is dissipated in the load rather than the output section (another way of putting that is there is less voltage dropped in the output section). I've gotten over 500W RMS (42 power tubes, 84 tube sections) with no need to increase the B+. But if your driver runs out of gas, you're sunk.
You may want to have a look at the the Graaf GM200 circuit. This is an OTL amp that has 150V rails and uses 16 output tubes in parallel on each channel to give 200W/channel. I've just received mine back from the factory in Italy, where they replaced all the tubes. They do not have individual bias controls on each tube but they do match the tubes very closely. They do have a single overall bias control for each set of 16 output tubes that they get the user to set to 750mV - 800mV after about an hour's warm-up.
Some pictures attached - I can nearly do away with my Christmas tree at night...
Attachments
It's 60 volts peak to peak
It's a brick wall. And an even bigger problem is that once you reach clipping it does some strange things - it doesn't just flat top.
I was gaining power up to 4 tubes, but after that subsequent tubes don't increase output voltage (or power) into a fixed 8 ohm load.
I'm not giving up. I know there is a way to get this to work. I plan to model the Kimmel circuit next and see what the results are.
In any event I know I can do something with the totem pole if the circlotron fails (or if I fail to successfully model one).
Sample
Did you get this built yet? I am extremely interested and I hope it goes your way please advise when you get a moment. Happy Holidays
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.