Just kidding, I've used QBasic. But how are you going to integrate in QBasic? You'd have to write all math stuff from scratch.
I'll try to solve them symbolically in Maple one of these days.
BTW, MarcelvdG suggested this one quadrant analog multiplier circuit; I can use this to multiply voltage and current samples for negative feedback and create a voltage-to-power amplifier:
What do you think about a MOSFET current source above the anode? I don't really want to have one end of the discharge directly connected to the power supply.
I'll try to solve them symbolically in Maple one of these days.
BTW, MarcelvdG suggested this one quadrant analog multiplier circuit; I can use this to multiply voltage and current samples for negative feedback and create a voltage-to-power amplifier:
An externally hosted image should be here but it was not working when we last tested it.
What do you think about a MOSFET current source above the anode? I don't really want to have one end of the discharge directly connected to the power supply.
Um, something's going to have voltage across it, one way or another. Hanging it from +V is easiest. A CCS will not work (unless bypassed) because...current has to change...
I can probably write an integration subroutine. I've made a differentiator... Having it rewrite the equations would be a bit harder, 😉 but that's why I have paper.
Tim
I can probably write an integration subroutine. I've made a differentiator... Having it rewrite the equations would be a bit harder, 😉 but that's why I have paper.
Tim
Oops, I should have said constant voltage. Otherwise the power supply voltage varies as the load changes, since it's just a couple of pi filters after the rectifier. But how do you make such high power regulator??
I'm wondering if there should be something above the driving tube's plate (besides the discharge cathodes, so the cathode is between two devices).
BTW, I posted about this on Slashdot, and I got modded to +5!
http://slashdot.org/comments.pl?sid=121863&cid=10253926
Can it be? Not all Slashdot moderators are f#cknuts? No... 😀
I'm wondering if there should be something above the driving tube's plate (besides the discharge cathodes, so the cathode is between two devices).
BTW, I posted about this on Slashdot, and I got modded to +5!
http://slashdot.org/comments.pl?sid=121863&cid=10253926
Can it be? Not all Slashdot moderators are f#cknuts? No... 😀
BTW, when looking for a tube, what sort of characteristics am I looking for, other than power handling? For example, I was looking at this 300 W Svetlana designed for audio applications:
http://www.svetlana.com/graphics/products/pdf/3CX300A1.pdf
Now, I don't know anything about tubes, but since we are looking at current output, I assume a high transconductance is needed. So is that tube's too low for this application, given it's a triode (it's about twice that of a 6DJ8)? Also, on some website I read this:
"Transconductance in triodes and pentodes increases quite dramatically with plate current, which is a source of distortion because it increases as the signal swings negative and decreases as it swings positive (at the plate). This is the basic distortion mechanism in all active devices, vacuum and solid state."
What types of tubes would then have smaller distortion? Tetrodes?
http://www.svetlana.com/graphics/products/pdf/3CX300A1.pdf
Now, I don't know anything about tubes, but since we are looking at current output, I assume a high transconductance is needed. So is that tube's too low for this application, given it's a triode (it's about twice that of a 6DJ8)? Also, on some website I read this:
"Transconductance in triodes and pentodes increases quite dramatically with plate current, which is a source of distortion because it increases as the signal swings negative and decreases as it swings positive (at the plate). This is the basic distortion mechanism in all active devices, vacuum and solid state."
What types of tubes would then have smaller distortion? Tetrodes?
Like I said, the basic 3/2 power law that exists at all times in all tubes. The idea being, if you take a certain operating point, a differential increase in grid voltage will cause a respective increase in plate current. This increase is a percentage of the previous current (how so depends on the characteristics; tubes are 3/2 power while SS is exponential), thus Gm increases without bound, in step with current... of course, dissipation limits this from being infinite.
If NFB is applied, tetrodes have lower distortion than triodes. (Lots of transistors have even less, but that's sheer gain at work.)
I wouldn't worry about absolute parameters, stick to class A1 current capacity and dissipation. Gain (mu or Gm) can be amplified by other stages.
Tim
If NFB is applied, tetrodes have lower distortion than triodes. (Lots of transistors have even less, but that's sheer gain at work.)
I wouldn't worry about absolute parameters, stick to class A1 current capacity and dissipation. Gain (mu or Gm) can be amplified by other stages.
Tim
linear current
For linear current output from a tube stage see this thread:
http://www.diyaudio.com/forums/showthread.php?threadid=40595
For some low cost HV tubes with high Gm, 30 to 35 Watt P diss., consider the beam triode voltage regulator tubes: 6HZ5/6JD5,6JH5,6JK5,6HS5/6HV5A
Don
For linear current output from a tube stage see this thread:
http://www.diyaudio.com/forums/showthread.php?threadid=40595
For some low cost HV tubes with high Gm, 30 to 35 Watt P diss., consider the beam triode voltage regulator tubes: 6HZ5/6JD5,6JH5,6JK5,6HS5/6HV5A
Don
Unless I want to be paralleling tubes, something with more dissipation is needed, say 200-300 W. That's why I was asking about the Svetlana 3CX300A1, as it has that dissipation and is also designed for audio (or so they claim).
BTW, I want to avoid capacitor coupling. I like the voltage shifting use of a common-base amplifier as in this design. What, you guys didn't think I was going for all tubes now did ya?
BTW, I want to avoid capacitor coupling. I like the voltage shifting use of a common-base amplifier as in this design. What, you guys didn't think I was going for all tubes now did ya?
I've been simulating the power supply some more. Because it is a couple of pi-filters and I have limited capacitance to put in, despite low ripple, it's output impedance is quite high, around 600 Ohms. That's why I don't think it's good to connect the HV directly to the anode.
Well geez, 300 ohms is an awful lot at 3kV. 10 amperes "lot".
Plus you're only drawing current at higher frequencies (what, 500Hz at the lowest?), impedance will be negligible. And finally, you've got some megohms below the discharge (well, pentode or triode?), I highly doubt it'll care about .0003 megs above it!
Tim
Plus you're only drawing current at higher frequencies (what, 500Hz at the lowest?), impedance will be negligible. And finally, you've got some megohms below the discharge (well, pentode or triode?), I highly doubt it'll care about .0003 megs above it!
Tim
I don't know, but in the simulator I see harmonics is as high as -30dB.
Also, don't forget that the MHCD configuration may not be constant voltage, but resistive. How does that change things?
Since voltage would only vary a bit at 500 Hz, a floating MOSFET circuit can be used as a regulator.
BTW, is there any advantage to a configuration where the anode is grounded, and the HV is now negative?
Also, don't forget that the MHCD configuration may not be constant voltage, but resistive. How does that change things?
Since voltage would only vary a bit at 500 Hz, a floating MOSFET circuit can be used as a regulator.
BTW, is there any advantage to a configuration where the anode is grounded, and the HV is now negative?
Well, then you have the required voltage across the discharge, it has the same absolute voltage with respect to the environment (-2kV vs. +2kV), and the tube is at -3k-some odd volts, which needs some rather chunky coupling capacitors. Not to mention well insulated heater windings.
Tim
Tim
What about a regulator, then? A shunt regulator would only have to sink current as large as whatever the audio amplitude current is, and maybe a solid state device could be used.
Indeed, in a configuration where a minimal ballast resistor (ground to cathodes) was used instead of a tube, with the discharge dropping most of the voltage, then the active device could be set to shunt from the top of the resistor -- much less power to dissipate, and again solid state could be used.
Indeed, in a configuration where a minimal ballast resistor (ground to cathodes) was used instead of a tube, with the discharge dropping most of the voltage, then the active device could be set to shunt from the top of the resistor -- much less power to dissipate, and again solid state could be used.
Sch3mat1c, now that I've finished the HV supply, I have some questions:
As a goal is to have most of the voltage drop being in the discharge rather than the voltage controlled current sink that drives it (more power dissipated in the plasma means higher output), what are the practical limits of how low the current sink voltage drop can be (once the DC operating point is reached), for either valves or solid state designs (assuming no audio linearity is sacrificed)?
Also, what about my last post question? I was thinking that for testing modulating the discharge, I could just use a ballast resistor and then shunt away some more current with some HV MOSFETs I have and feed them an audio signal. They are good to 1500 V so I think if I switch them in once a discharge has started they shouldn't get overvoltaged.
As a goal is to have most of the voltage drop being in the discharge rather than the voltage controlled current sink that drives it (more power dissipated in the plasma means higher output), what are the practical limits of how low the current sink voltage drop can be (once the DC operating point is reached), for either valves or solid state designs (assuming no audio linearity is sacrificed)?
Also, what about my last post question? I was thinking that for testing modulating the discharge, I could just use a ballast resistor and then shunt away some more current with some HV MOSFETs I have and feed them an audio signal. They are good to 1500 V so I think if I switch them in once a discharge has started they shouldn't get overvoltaged.
I still don't like the idea of far more supply volts than Vds.
Triodes are the worst due to internal NFB; holding current constant, minimum voltage depends on perveance and current drawn (and if you are allowing grid current). A 12AX7 (low perveance) bottoms out at 2.1mA at 100V, 6SN7 at 10mA and 6KD6 at 720mA. Generally speaking it is proportional to heater power and inversely proportional to mu.
Screen grid tubes have Vsat as low as 30V (6V6; or less for signal types) on up to 500V or more for transmitter pentodes. Tetrodes (particularly beam tets) are better because the curve has a sharper knee; compare EL34 to 6L6, for instance. Because of the very limited internal NFB, they have a much higher impedance (read: better CCS characteristic, requiring little grid compensation). That's the flat part of the plate curve.
All transistors have the same characteristic (except "plate resistance" is called "Early effect"), but at lower voltages (down to 0.1V or less) and higher currents (a nice MOSFET can easily do 40V saturation at 40A, i.e., 1 ohm Rdson). (There is no BJT, no FET, anything which has the low impedance characteristic of a triode.) As such, a CCS will perform better, but cannot handle as high voltages - I haven't seen *any* transistor capable of more than 1.5kV.
Tim
Triodes are the worst due to internal NFB; holding current constant, minimum voltage depends on perveance and current drawn (and if you are allowing grid current). A 12AX7 (low perveance) bottoms out at 2.1mA at 100V, 6SN7 at 10mA and 6KD6 at 720mA. Generally speaking it is proportional to heater power and inversely proportional to mu.
Screen grid tubes have Vsat as low as 30V (6V6; or less for signal types) on up to 500V or more for transmitter pentodes. Tetrodes (particularly beam tets) are better because the curve has a sharper knee; compare EL34 to 6L6, for instance. Because of the very limited internal NFB, they have a much higher impedance (read: better CCS characteristic, requiring little grid compensation). That's the flat part of the plate curve.
All transistors have the same characteristic (except "plate resistance" is called "Early effect"), but at lower voltages (down to 0.1V or less) and higher currents (a nice MOSFET can easily do 40V saturation at 40A, i.e., 1 ohm Rdson). (There is no BJT, no FET, anything which has the low impedance characteristic of a triode.) As such, a CCS will perform better, but cannot handle as high voltages - I haven't seen *any* transistor capable of more than 1.5kV.
Tim
So what do you recommend for best performance? Hopefully something that won't drop more than something on the order of 500 V at the average current (between 150 and 200 mA for a channel).
Triode, tetrode, pentode, BJT, MOSFET, or some kind of hybrid?
Keeping in mind that the audio frequency modulation is going to be less than 10 percent, what about having most of the current dropped by some device as a constant current sink, and just enough to do audio frequency modulation by a controlled current sink in parallel with the other device. This would allow the audio device to be chosen for linearity rather than power handling.
BTW, since I don't know anything about tubes, how can I calculate what voltage it will drop to get a given current, using data in the standard datasheet?
One more thing. My output capacitors are four 40 uF tin can film in oil types, but they are only rated at 3 kV, which is barely above the no load output of the supply. To prevent frying them, is it OK if I add some electrolytics in series? The simulations shows that using large enough electrolytics will barely decrease capacitance, and the voltage across the electrolytics and film is respectively 300 V and 2700 V, which is well within the ratings of both.
Triode, tetrode, pentode, BJT, MOSFET, or some kind of hybrid?
Keeping in mind that the audio frequency modulation is going to be less than 10 percent, what about having most of the current dropped by some device as a constant current sink, and just enough to do audio frequency modulation by a controlled current sink in parallel with the other device. This would allow the audio device to be chosen for linearity rather than power handling.
BTW, since I don't know anything about tubes, how can I calculate what voltage it will drop to get a given current, using data in the standard datasheet?
One more thing. My output capacitors are four 40 uF tin can film in oil types, but they are only rated at 3 kV, which is barely above the no load output of the supply. To prevent frying them, is it OK if I add some electrolytics in series? The simulations shows that using large enough electrolytics will barely decrease capacitance, and the voltage across the electrolytics and film is respectively 300 V and 2700 V, which is well within the ratings of both.
Bah.... any film cap I've seen is rated for 150%.
I'd go with a good transmitter tetrode or pentode... although if dissipation isn't too high a sweep tube would handle it (rated upwards of 5kV, at least in cutoff). Look at the plate curves, set voltages as shown and the plate responds according to whichever grid line you're on. You'll want to cathode bias it to get additional degeneration which increases Zo. If you have a screen voltage (instead of grid voltage being traced multiple times) graph, you can find a reasonable screen voltage to reduce required grid bias. Remember screen is just a control grid with so little transconductance that it needs to be at a positive voltage to keep current flowing; you can throttle the tube just as well with both, minus Gm that is.
Tim
I'd go with a good transmitter tetrode or pentode... although if dissipation isn't too high a sweep tube would handle it (rated upwards of 5kV, at least in cutoff). Look at the plate curves, set voltages as shown and the plate responds according to whichever grid line you're on. You'll want to cathode bias it to get additional degeneration which increases Zo. If you have a screen voltage (instead of grid voltage being traced multiple times) graph, you can find a reasonable screen voltage to reduce required grid bias. Remember screen is just a control grid with so little transconductance that it needs to be at a positive voltage to keep current flowing; you can throttle the tube just as well with both, minus Gm that is.
Tim
OK, but if it fries I'm coming after you!
There are a few small dents in the old caps, and I'm wondering if I could have trouble if I don't keep them upright while running.
Cornell Dubilier replied to my inquiry that the specific model was custom made for some customer and they no longer had the data. About the only thing I learned is that temperature should not be too high, about 45*C. Since they are inside the enclosure I'm putting a fan.
Prune said:
OK, but if it fries I'm coming after you!
Alright, just wash your pants first.. 😀
(E = 1/2 C V^2 = .5 * .00004 * 3000^2 = 180J 😀 )
Tim
Total capacitance is about 320 uF, so 1440 J.
I'm still planning to add a regulator, but I'll leave that for later.
I wanted to experiment with a ballast resistor, but can't find anything rated at several hundred W, so I guess I can make a salt water resistor.
I'm still planning to add a regulator, but I'll leave that for later.
I wanted to experiment with a ballast resistor, but can't find anything rated at several hundred W, so I guess I can make a salt water resistor.
Did you mean that I should have the tube in grounded cathode? What about other configurations, like grounded grid?
As I said before, I'm not familiar with tubes, so I'm not sure how to figure out the plate current vs grid voltage from the types of graphs I see in datasheets. For example, how would do this and figure out the setup for the tube in the attachment?
As I said before, I'm not familiar with tubes, so I'm not sure how to figure out the plate current vs grid voltage from the types of graphs I see in datasheets. For example, how would do this and figure out the setup for the tube in the attachment?
Attachments
- Status
- Not open for further replies.