I would like to make a computer tube tester to test all functions, trans-conductance,Ip, and even graph load lines etc. The software can even be written as license free and allow for more development.
I started a thread about another subject that turned into talk about a computer tube tester so I am starting a thread just on this subject. The end result hopefully will be someone has already done this and save us a few years on our life.
I can visualize how I would put one together. I saw the RAT circuit and seems to be ok. Maybe use that circuit and add to it. Need a data base of tubes in your program with input to add more. That way it knows what switching to make automatically when you select a tube. An approach to change the voltages could be realized by simple switching using CD4066's or similar scheme controlled by the computer of course. Switching to each resistor a part of a regulated adjustable power supply. Selection of the proper resistor sets the adjustment on the regulator. That is to say each resistor you select or combination of is the same as turning a dial to dial the voltage in. Then program the computer to take voltage readings, change voltage and go for it again. Even use the computer audio to inject the signal in the tube at varying levels and input the response. Now for switching the tube configurations. I think, I am going to look at high side load switching a mosfet, for the B+. More of course can be done and it never works the way one wants its too and always have to tweak on it to get it right. At least for me anyway.Maybe these ideas can work. I know there is such a thing on the market I saw it. Its always fun and challenging to make one to make one. What's your thought on a computer based tube tester and your ideas?
I started a thread about another subject that turned into talk about a computer tube tester so I am starting a thread just on this subject. The end result hopefully will be someone has already done this and save us a few years on our life.
I can visualize how I would put one together. I saw the RAT circuit and seems to be ok. Maybe use that circuit and add to it. Need a data base of tubes in your program with input to add more. That way it knows what switching to make automatically when you select a tube. An approach to change the voltages could be realized by simple switching using CD4066's or similar scheme controlled by the computer of course. Switching to each resistor a part of a regulated adjustable power supply. Selection of the proper resistor sets the adjustment on the regulator. That is to say each resistor you select or combination of is the same as turning a dial to dial the voltage in. Then program the computer to take voltage readings, change voltage and go for it again. Even use the computer audio to inject the signal in the tube at varying levels and input the response. Now for switching the tube configurations. I think, I am going to look at high side load switching a mosfet, for the B+. More of course can be done and it never works the way one wants its too and always have to tweak on it to get it right. At least for me anyway.Maybe these ideas can work. I know there is such a thing on the market I saw it. Its always fun and challenging to make one to make one. What's your thought on a computer based tube tester and your ideas?
Here is another similar project that is worth looking over for ideas. A Microprocessor Controlled Vacuum Tube (and other) Curve Tracer
This should be a nice project to play around with - might get me back into tubes after 40 some odd years.
This should be a nice project to play around with - might get me back into tubes after 40 some odd years.
Last edited:
Hello ,
RoeTest - Röhrenprüfgerät Röhrenmessgerät
and here the description in the german forum :
Computer-Roe-Pruef-und-Messgeraet, von Helmut Weigl
Kind regards , Alexander .
RoeTest - Röhrenprüfgerät Röhrenmessgerät
and here the description in the german forum :
Computer-Roe-Pruef-und-Messgeraet, von Helmut Weigl
Kind regards , Alexander .
I like the people on this forum, most interesting. I have been to other forums where there is a one man band so to speak, a lot of people participating but only one person who has nothing else to do, answers the questions and makes you feel like you don't know stink. Well I don't know stink, but at the same time I finally got him one day and proved him wrong. What a waste of time and effort and I ended up being just as bad as him by getting back at him. I cant live like that! I like this forum because I am not seeing that here, and I don't have to feel intimidated. I see people who are willing to share their knowledge to others and I don't have to feel the need to try and protect my feelings and thoughts or try to lash back at someone to save myself from ridicule. Don't get me wrong. I like to hear all sides of the story and a better way to do things or different ways to do things. To all of you I thank you. I like the enjoyment of all the articles and the learning all the post provides. Keeps my mind busy. Just seems like a great group of people here, definitely in a higher class, and of good nature.
Dang! did you see all the relays in the picture? about 70 relays. The guy really did alot of work on this project. I don't know if I need one that bad. I like the program though. I am going to look at that a little more closer and see if it is easy to hook up to with other circuits. I did have to drop the url in Google translate to see, since I don't understand the language. Just alot of work went into that machine. you have to eat and sleep it in order to get done by Christmas. I still think relays are too costly and bulky. I think if a guy came up with a good circuit, just send it to a pc board making place and save some time making a board. I have read some on using the midi interface, then I also read it was not accurate. I don't have enough data on that yet to know for a surety. I suppose for tube testing it doesn't have to be that precise. Does one have to have precision or is just some simple tests good enough for testing tubes. You know the basics. I guess if you wanted to know if the tube starts oscillating you could just simply have a frequency counter sniff that out. What are the basic tests you would consider as being enough? of course one could add to it later. Is just testing the emission, Gm, Ip, Rp, with choice of voltages at your disposal for plate, screen and grid enough? What about a load of some kind? Ac and Dc selection of filament and variable?
Dang! did you see all the relays in the picture? about 70 relays. The guy really did alot of work on this project. I don't know if I need one that bad. I like the program though. I am going to look at that a little more closer and see if it is easy to hook up to with other circuits. I did have to drop the url in Google translate to see, since I don't understand the language. Just alot of work went into that machine. you have to eat and sleep it in order to get done by Christmas. I still think relays are too costly and bulky. I think if a guy came up with a good circuit, just send it to a pc board making place and save some time making a board. I have read some on using the midi interface, then I also read it was not accurate. I don't have enough data on that yet to know for a surety. I suppose for tube testing it doesn't have to be that precise. Does one have to have precision or is just some simple tests good enough for testing tubes. You know the basics. I guess if you wanted to know if the tube starts oscillating you could just simply have a frequency counter sniff that out. What are the basic tests you would consider as being enough? of course one could add to it later. Is just testing the emission, Gm, Ip, Rp, with choice of voltages at your disposal for plate, screen and grid enough? What about a load of some kind? Ac and Dc selection of filament and variable?
I think that one should simplify first to make progress on such a complicated project. The pin switching network is almost frivolous, but very complex with many HV issues. Just put bananna or pin jacks on the unit case labeled cathode, g1, g2, g3, plate (grid stoppers, ferrite beads etc inside the unit). Then make up some general 7,9 pin, octal, .... sockets with bananna or pin plugs on each pin. (use short leads, the pin plugs are easier to insert/ remove than bananna ones) You have to consult the tube base diagram each time, but is quite adequate for testing a box of similar tubes.
Then for the next step up in convenience, one can make sockets per the common tube base types, with the pin plugs mounted on an insulating strip that plugs in to the jacks correctly.
For plate resistors, one can avoid a collection of switched-in power resistors by using a programmable power resistor:
EDN PDF (bottom article)
Power supplies are available off the shelf with programming ports that can connect to the PC. I use some Xantrex units with serial ports. (GPIB ports are more common than the RS-232 ones, but are more complicated to get interfacing set up to the PC and to use with cheap software) (I don't have software designed yet to use them, but I plan on using Visual Basic).
In the mean time (probably my main excuse for not getting the above mentioned programming done yet), I use a Tektronix 576 (or 577) curve tracer that has been modified to do tube curve tracing. I have the Xantrex power units (for the above project) rigged to supply g2 voltages (separate for each test port - 1 and 2), and separate filament supplies too. Just have to toggle the selector switch to display curves for either tube, which can be totally different tube types if desired.
The curve tracer is a giant leap over any tube tester, and for a similar price too, and can match tubes over their whole characteristic. But a PC automated unit, will clearly be the ultimate, allowing further linearity analysis and storage and printouts.
Don
Then for the next step up in convenience, one can make sockets per the common tube base types, with the pin plugs mounted on an insulating strip that plugs in to the jacks correctly.
For plate resistors, one can avoid a collection of switched-in power resistors by using a programmable power resistor:
EDN PDF (bottom article)
Power supplies are available off the shelf with programming ports that can connect to the PC. I use some Xantrex units with serial ports. (GPIB ports are more common than the RS-232 ones, but are more complicated to get interfacing set up to the PC and to use with cheap software) (I don't have software designed yet to use them, but I plan on using Visual Basic).
In the mean time (probably my main excuse for not getting the above mentioned programming done yet), I use a Tektronix 576 (or 577) curve tracer that has been modified to do tube curve tracing. I have the Xantrex power units (for the above project) rigged to supply g2 voltages (separate for each test port - 1 and 2), and separate filament supplies too. Just have to toggle the selector switch to display curves for either tube, which can be totally different tube types if desired.
The curve tracer is a giant leap over any tube tester, and for a similar price too, and can match tubes over their whole characteristic. But a PC automated unit, will clearly be the ultimate, allowing further linearity analysis and storage and printouts.
Don
Consider using USB sound board as DAC and ADC, they work fine, precise and fast !
Look at:
Tube Curves Tracer
(Click on the blank frame to read more !)
I've improved the idea more than a lot since I've wrote that, added G2 control and more ...
And I can share Visual Basic more recent (dirty) source code ... at your own risks !
At present time, I'm able to do that:
Yves.
Attachments
Hi Yves,
Looks rather interesting. This might be a good solution for some of us as it stands. Any caveats you care to mention? In the grid supply can you provide details on the transformer?
Kevin
Last edited:
Check out Broskie's tube tester design. He's thought alot about how to implement one with all the tube variations
Heatsinks & Tube Testers
Heatsinks & Tube Testers
"Check out Broskie's tube tester design."
Seems to be a work in progress yet. I only see the tube biasing design so far.
But I see some SERIOUS deficiencies in the plan outlined.
Depends on what one is going to use the tester for, but measuring (and graphing) effective "Mu" or gain versus input signal voltage swing, at a selected operating point and plate load is the MOST important feature in my opinion, if one is designing with a tube.
(and a plot of gm and rp versus plate voltage and current too for pentodes, but even there, a g1/g2 "Mu" plot is paramont for construction/design quality checking. Varying "Mu" comes from mis-aligned grids, bent grid wires, bad grid shape versus plate shape design like overly aggressive hi gm ones. A high gm reading may sound good on a simple tester, but it may be just telling you the grid is too close to the cathode someplace.)
And while an FFT distortion plot will give you a good indication of tube linearity problems at some specific op point, it does not give you any indication whether another operating point would work better without performing a massive testing effort. Effective Mu plots will also give the designer a good means to determine if another tube is complementary in its gain profile (for distortion cancelling in subsequent stages).
And Mu is not a derived parameter, it is geometrically determined by the ratio of grid and plate capacitances. Variation of the gain with signal swing will show how well the tube is designed as well as actually constructed.
Ordinary gm testers (and worse, emission testers) are just go/ no-go testers for dime stores. Not even suitable for matching really. Get a curve tracer, at least, if you want to match tubes. With my Tek setup I can easily determine if a tube can be matched with another by a bias change or screen voltage change on one of them too. No need to go thru hundreds of tubes to find a match at some silly fixed grid bias. (many tube "mis-matches" come about due to simple contact potential differences in the cathode construction materials or processing or spot welds/materials/oxide surface finish. These are easily remedied by simple DC bias changes if you can "see" them.)
Don
Seems to be a work in progress yet. I only see the tube biasing design so far.
But I see some SERIOUS deficiencies in the plan outlined.
Depends on what one is going to use the tester for, but measuring (and graphing) effective "Mu" or gain versus input signal voltage swing, at a selected operating point and plate load is the MOST important feature in my opinion, if one is designing with a tube.
(and a plot of gm and rp versus plate voltage and current too for pentodes, but even there, a g1/g2 "Mu" plot is paramont for construction/design quality checking. Varying "Mu" comes from mis-aligned grids, bent grid wires, bad grid shape versus plate shape design like overly aggressive hi gm ones. A high gm reading may sound good on a simple tester, but it may be just telling you the grid is too close to the cathode someplace.)
And while an FFT distortion plot will give you a good indication of tube linearity problems at some specific op point, it does not give you any indication whether another operating point would work better without performing a massive testing effort. Effective Mu plots will also give the designer a good means to determine if another tube is complementary in its gain profile (for distortion cancelling in subsequent stages).
And Mu is not a derived parameter, it is geometrically determined by the ratio of grid and plate capacitances. Variation of the gain with signal swing will show how well the tube is designed as well as actually constructed.
Ordinary gm testers (and worse, emission testers) are just go/ no-go testers for dime stores. Not even suitable for matching really. Get a curve tracer, at least, if you want to match tubes. With my Tek setup I can easily determine if a tube can be matched with another by a bias change or screen voltage change on one of them too. No need to go thru hundreds of tubes to find a match at some silly fixed grid bias. (many tube "mis-matches" come about due to simple contact potential differences in the cathode construction materials or processing or spot welds/materials/oxide surface finish. These are easily remedied by simple DC bias changes if you can "see" them.)
Don
Last edited:
Yves:
"Consider using USB sound board as DAC and ADC, they work fine, precise and fast !"
I agree, as long as you can get DC coupling from the sound card hardware. Many (most, if not all, actually) sound cards have AC coupling. But if the Op. Amp circuitry is accessible, one can modify it to handle DC. (have to be carefull when using that for audio purposes later though). Easy to find analog programmable power supplies around (or just build a DC coupled amp.) that could be controlled this way then.
I went with the Xantrex approach because serial port control was available off the shelf (well, Ebay actually), and Visual Basic has easy interfacing with serial ports. The Xantrex also have 16 bit D/A control and A/D monitor features and auto calibration.
-------------------------------------------------------------------
For design purposes, (maybe not needed for tube matchers or tube checkers) it is also desirable to be able to configure the test setup to mimick the actual intended use. For example, for ultra-linear use, one could program the screen voltage to follow the plate voltage at some %, maybe with some DC offset too.
Or a cathode degenerating resistor could be included in the test model (which makes special demands on the grid 1 voltage generator for range and float ability.)
One might like to "Schade" a pentode with a plate to grid feedback path and observe the resulting "triode" curves. (this would require a current generator for grid 1 drive) Or, one might even want to measure multi-tube stages, like differential pairs or beam deflectors. (so extra control signals needed)
Then there are computed features to consider. In theory, one could just take an extensive curve set measurement for a tube, and then have the software compute the characteristics for some more complex (like those mentioned above) configuration. This would avoid the need for special or extra signal generator hardware.
Various voltage, current and wattage limits can be also be built into the software to protect the tube elements during test.
Then there are things like "forced" responses to consider. These are cases where some specific response is desired, and the software (either off-line from tube data, or using online live tube operation) is required to find some control signal (say a screen grid signal or tail current) to make some specific response occur (like linear output versus g1 voltage). So the software itself is effectively in a feedback loop minimising the error with the specific response wanted. For example, one could find the required common mode feedback for linearity using a P-P WE Harmonic Equalizer setup.
Don
"Consider using USB sound board as DAC and ADC, they work fine, precise and fast !"
I agree, as long as you can get DC coupling from the sound card hardware. Many (most, if not all, actually) sound cards have AC coupling. But if the Op. Amp circuitry is accessible, one can modify it to handle DC. (have to be carefull when using that for audio purposes later though). Easy to find analog programmable power supplies around (or just build a DC coupled amp.) that could be controlled this way then.
I went with the Xantrex approach because serial port control was available off the shelf (well, Ebay actually), and Visual Basic has easy interfacing with serial ports. The Xantrex also have 16 bit D/A control and A/D monitor features and auto calibration.
-------------------------------------------------------------------
For design purposes, (maybe not needed for tube matchers or tube checkers) it is also desirable to be able to configure the test setup to mimick the actual intended use. For example, for ultra-linear use, one could program the screen voltage to follow the plate voltage at some %, maybe with some DC offset too.
Or a cathode degenerating resistor could be included in the test model (which makes special demands on the grid 1 voltage generator for range and float ability.)
One might like to "Schade" a pentode with a plate to grid feedback path and observe the resulting "triode" curves. (this would require a current generator for grid 1 drive) Or, one might even want to measure multi-tube stages, like differential pairs or beam deflectors. (so extra control signals needed)
Then there are computed features to consider. In theory, one could just take an extensive curve set measurement for a tube, and then have the software compute the characteristics for some more complex (like those mentioned above) configuration. This would avoid the need for special or extra signal generator hardware.
Various voltage, current and wattage limits can be also be built into the software to protect the tube elements during test.
Then there are things like "forced" responses to consider. These are cases where some specific response is desired, and the software (either off-line from tube data, or using online live tube operation) is required to find some control signal (say a screen grid signal or tail current) to make some specific response occur (like linear output versus g1 voltage). So the software itself is effectively in a feedback loop minimising the error with the specific response wanted. For example, one could find the required common mode feedback for linearity using a P-P WE Harmonic Equalizer setup.
Don
Last edited:
Guys,
More possible source material - JB had some good ideas on tube testers.
TCJ 10 Year Anniversary
Heatsinks & Tube Testers
Cheers,
Ian
More possible source material - JB had some good ideas on tube testers.
TCJ 10 Year Anniversary
Heatsinks & Tube Testers
Cheers,
Ian
just one word of caution with respect to USB, RS-232, GPIB, parallel port etc and high voltages -- you can get some really nasty surprises between your computer interface and the B+ -- that's why they sell isolated IO cards -- your motherboard can go up in smoke in a few milliseconds. Isolated RS-232 is probably least expensive, but nowadays probably least convenient. I guess the analog opto-isolators I used in the article are now manufactured by Avago, formerly Agilent, formerly Hewlett Packard.
I started with an integated Sound Board because it costs nothing, all PC having one inside.
Later I went to an external USB one because of so many different impementations of Windows Mixer along with unpredictable Output level and/or sensitivity among various manufacturer caused programming nightmare.
Your right about AC coupling, but it's an easy to solve this problem using "good old clamping methods" (as routinely used in pre-sand era for DC restoring video amplifier) here implemented by software ... and this kills -may be- DC drift issues.
This is made possible because the plate voltage is "pulsed" (in fact simply full wave rectified but not smoothed) and we know that the most negative value IS zero.
Of course, the program does not control the plate voltage, it just measures it simultaneously -using Left and Right stereo Inputs- with plate current.
It'is still a "work in progress", another hardware implementation -with embarqued TI's PCM2904 USB chip- being planned for next weeks ... stay tuned !
Software "add ons" are just limited by the imagination (and the lazyness) of the programmer (me)
Yves.
I cant wait to read more about JB's tester. Interesting, I read that over and over again to make sense of it all. For the idle current I would like to make that variable. even have an elimination of the cathode bias or a combination of both. I heard that was a secret to have both grid and cathode bias going to get specific sound quality some look for. It is great to see all the info. Yes about the isolation, I some opto isolators I bought for another project one time never got built. actually for port control. I was starting and stopping a vcr at times I programmed to play commercials on a low cost tv station. I'd like to experiment with the circuit JB has. The curves Tracer looks good too. Looks a little scary where it hooks to the computer though. I might have not seen the whole circuit in detail as I quickly gazed at it. I suppose I need to put together a computer just in case I mess it up I wont have to worry. Sounds to me like one might need an interface board to protect the computer from nasty voltages as stated by Jack. I think the project should have plug in cards so its easy to repair and add more ideas. If you blow something remove the card,fix, and modify if needed. I certainly learned my lesson with ics. Use a socket, if you don't you will wish you did.
The cost is minimal compared to the work it takes to undue solder connections etc. Not sure which op amp JB used yet, but needs to handle the 20 volts on the one rail. I wish there were more hours in a day. too bad one couldn't create a week within a week, so I can play all week long and work too. Now that I see some diagrams to work with I cant wait to get started. Al I was going to do is measure some lousy 6AQ5's so I could try and get a few dollars more out of them off ebay. cant say this is helping, wont cover the time or cost, oh well what the heck. I do have a Sencore T-162. How in the world did I get twisted around and now wanting to build a tester? good grief no wonder I never finish anything! I have made it an oath to myself to try and complete something I started. Then move on to something else. I started something so I must continue on. Its more fun to build stuff with someone else and their ideas and then make of it what you can.
The cost is minimal compared to the work it takes to undue solder connections etc. Not sure which op amp JB used yet, but needs to handle the 20 volts on the one rail. I wish there were more hours in a day. too bad one couldn't create a week within a week, so I can play all week long and work too. Now that I see some diagrams to work with I cant wait to get started. Al I was going to do is measure some lousy 6AQ5's so I could try and get a few dollars more out of them off ebay. cant say this is helping, wont cover the time or cost, oh well what the heck. I do have a Sencore T-162. How in the world did I get twisted around and now wanting to build a tester? good grief no wonder I never finish anything! I have made it an oath to myself to try and complete something I started. Then move on to something else. I started something so I must continue on. Its more fun to build stuff with someone else and their ideas and then make of it what you can.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.