It's not just the voltages from the HV power supply to the PC via whatever interface -- it's also leakage of line voltage from the PC back to the tester -- as strange as this sounds, it does happen. I read about it in an application note from one of the folks who makes IC's for medical instrumentation. Every time I see an article in the popular electronics magazines about a DIY EKG I have to point out what should be obvious safety issues.
Mmmmh, no doubt, that's probably true . . . Do not listen music with an headphone plugged anywhere 🙄
BTW, a tube tester/tracer is a thingy that uses HI voltage that can be applied to any pin of any tube.
It should be enclosed in an hermetical box automatically switched off when opened . . . as are ALL our loved tube amplifiers.
Be cautious. 😎
Yves.
BTW, a tube tester/tracer is a thingy that uses HI voltage that can be applied to any pin of any tube.
It should be enclosed in an hermetical box automatically switched off when opened . . . as are ALL our loved tube amplifiers.
Be cautious. 😎
Yves.
Excellent points about the safety of interfacing to the PC. PC switching supplies are usually extremely noisy/cheap too. Hardly medical grade isolation. (one could put a medical grade isolation xfmr in the PC power line, but the external devices like printers, monitors .... LAN? DSL? Tele?... would all have to be isolated too.
The external USB approach probably solves the noise problem, but not the safety problem. I looked at USB earlier, and it is designed in such a way that opto's cannot be used with it. (the idiots used DC levels and line impedance for revision compatitbility issues) Although it might be workable with optos in a limited, known environment. But would not meet USB standards then.
The power supplies manufactured for automatic test systems (ATE), like Xantrex (seem to be AMETEK/Sorensen now), HP (or Avago, Agilent), Sorensen and Kepco generally have taken care of the isolation issues from day one. Usually using opto GPIB to the PC. Sometimes opto RS-232 or isolated-analog. Be careful with some older analog interfaces, many were not opto isolated. Digital parallel port ones too. Safest to check the manu. specs first. The Xantrex remote analog prog., RS-232, and GPIB interfaces are all opto isolated or use the Burr-Brown isolator, good for 1500V I think.
Using ATE supplies also solve the HV issues since you just use a HV model. Except for the programmable power resistor issue, ATE equipment is just a software development effort. I'm planning on using a spare isolated Xantrex RS-232 option card to drive my programmable power resistor unit. Will provide floating V and I readbacks also.
OT:
I just got a Xantrex supply delivered today from Calif. (EBAY) and the inside was clogged with mud! Landslides lately in Calif? Or dumpster pickings. Amazingly it still worked!
Also, this new forum software is making it very difficult to post anymore lately. It always logs me off while typing a response, then dumps the response after logging in again.
The external USB approach probably solves the noise problem, but not the safety problem. I looked at USB earlier, and it is designed in such a way that opto's cannot be used with it. (the idiots used DC levels and line impedance for revision compatitbility issues) Although it might be workable with optos in a limited, known environment. But would not meet USB standards then.
The power supplies manufactured for automatic test systems (ATE), like Xantrex (seem to be AMETEK/Sorensen now), HP (or Avago, Agilent), Sorensen and Kepco generally have taken care of the isolation issues from day one. Usually using opto GPIB to the PC. Sometimes opto RS-232 or isolated-analog. Be careful with some older analog interfaces, many were not opto isolated. Digital parallel port ones too. Safest to check the manu. specs first. The Xantrex remote analog prog., RS-232, and GPIB interfaces are all opto isolated or use the Burr-Brown isolator, good for 1500V I think.
Using ATE supplies also solve the HV issues since you just use a HV model. Except for the programmable power resistor issue, ATE equipment is just a software development effort. I'm planning on using a spare isolated Xantrex RS-232 option card to drive my programmable power resistor unit. Will provide floating V and I readbacks also.
OT:
I just got a Xantrex supply delivered today from Calif. (EBAY) and the inside was clogged with mud! Landslides lately in Calif? Or dumpster pickings. Amazingly it still worked!
Also, this new forum software is making it very difficult to post anymore lately. It always logs me off while typing a response, then dumps the response after logging in again.
you could use opto's between the logic and the high voltage. and use optics between the PIC and a RS232 interface. There are plenty RS232-USB driversboards in the marktet for a couple of bucks.
Here's a curve ball that might rile some, though from Yvesm's screenshot he might already be ahead of the game, consider outputting Spice variables instead of curves. Iterative algorithms to tune polynomials should be a relatively easily achievable give some prior art out there. Another option is to create a full data matrix of plate current/voltage against grid for import into spread sheets.
/flame on!
/flame on!
I succed to export captured (measured) data to "Curve Captor" wich itself generates Spice models in various formats.
Seems perfect for triodes but I've problems with pentodes.
I admit that, not being a Spice fervent afficionado, I didn't go further . . .
OTH, data are stored in simple text files that can easily be imported in a spread sheet.
Here are some (with .tdf extension for Tube Data File):
Index de ../Traceur/TDFiles/
They were taken at different design levels and I can't guarantee they contain exact data.
Each "space tabbed" line contains:
(Voltages in Volts (!) and currents in Ampères)
and so on as long as there are other grids and hardware able to drive/measure them.
Yves.
Seems perfect for triodes but I've problems with pentodes.
I admit that, not being a Spice fervent afficionado, I didn't go further . . .
OTH, data are stored in simple text files that can easily be imported in a spread sheet.
Here are some (with .tdf extension for Tube Data File):
Index de ../Traceur/TDFiles/
They were taken at different design levels and I can't guarantee they contain exact data.
Each "space tabbed" line contains:
Code:
AnodeVoltage AnodeCurrent Grid1Voltage Grid1Current(or "?") Grid2Voltage Grid2Currentand so on as long as there are other grids and hardware able to drive/measure them.
Yves.
"consider outputting Spice variables instead of curves"
Or just make the Spice model call the tester to take a real measurement during simulation. A quite simple approach.
Or just make the Spice model call the tester to take a real measurement during simulation. A quite simple approach.
No, not joking. Although only making sense if you have a tester available, and the tube in question.
When spice calls a tube model, it passes some voltage parameters (cath., g1,g2,p) to the model. The model then does its polynomial calcs. and returns some currents to Spice as return parameters.
So just have Spice call the tester with the parameters, the tester outputs the voltages to the tube and measures the currents, then returns the currents in the return parameters of the subroutine call. I think Spice generally has hooks provided for calling external routines. Like say calling a Mathcad model. Or Vis. Basic model. The model would be a link to the tester in this case. So for a simulation, Spice would be calling the tester repeatedly at each time increment of the simulation to take a real measurement. This might slow the simulation down though if the tester runs through slow RS-232 lines.
When spice calls a tube model, it passes some voltage parameters (cath., g1,g2,p) to the model. The model then does its polynomial calcs. and returns some currents to Spice as return parameters.
So just have Spice call the tester with the parameters, the tester outputs the voltages to the tube and measures the currents, then returns the currents in the return parameters of the subroutine call. I think Spice generally has hooks provided for calling external routines. Like say calling a Mathcad model. Or Vis. Basic model. The model would be a link to the tester in this case. So for a simulation, Spice would be calling the tester repeatedly at each time increment of the simulation to take a real measurement. This might slow the simulation down though if the tester runs through slow RS-232 lines.
Thanks Vyes. Accept this with the sincerest respect when I say you're wonderfully mad. Excel doesn't see the tab delimiters in these files, is it a special encoding? I've always wanted to generate a 3D map of slope across a set of curves. The flattest region should yield the highest linearity.
Gaaaah ! !
I see ...
"but what if more than one type of tube used 😱 😕 "
... just use more testers 😀
Time to sleep by here 😴😴
I see ...
"but what if more than one type of tube used 😱 😕 "
... just use more testers 😀
Time to sleep by here 😴😴
"but what if more than one type of tube used "
... just use more testers "
Well, that's one way.
Or use some Mosfets to multiplex multiple tubes using one tester. A resistor to each grid1 with a Mosfet switch to pull it down to -60 V or so. A Mosfet follower to each grid2, with a resistor in series with the gate, and another Mosfet switch to pull it down to 0 V. The tube plates and cathodes would be wired in parallel to the load resistor and ground. The Spice models would then select which tube to enable in the tester by logic signals to the Mosfet switches.
... just use more testers "
Well, that's one way.
Or use some Mosfets to multiplex multiple tubes using one tester. A resistor to each grid1 with a Mosfet switch to pull it down to -60 V or so. A Mosfet follower to each grid2, with a resistor in series with the gate, and another Mosfet switch to pull it down to 0 V. The tube plates and cathodes would be wired in parallel to the load resistor and ground. The Spice models would then select which tube to enable in the tester by logic signals to the Mosfet switches.
Found a few moments to work on the project. Yvesm what did you use for the transformer on the output of the audio amp in your circuit? I have been looking over circuit and studying it. It's funny how many different ways one can come up with to measure the specs of a tube. Interesting approach, I would be interested how it was conceived or wanted it to work. Such as " I was thinking about not only providing audio but a grid bias and read the result". This is a nice idea you came up, and simple. I have to make a trip to the radio store for some ic sockets so I can start on the tube tester project.
- Status
- Not open for further replies.