Hello Patrick,
I want to DIY a curve tracer. I've read Transistor Curve Tracer - ELEKTOR.com | Electronics: Microcontrollers Embedded Audio Digital Analogue Test Measurement but I want to have one with higher current. I have already on hand some 8bit MCU with AD, PL2303 USB-serial chips, TLV5618 DAC, etc. Can I make reference to your design?
Cheers,
WK
Thanks for your quick reply. Actually, I got an very early version (serial number 18) from locky_z in 2007 Dec. I want to make one because I no longer have computer with real parallel port.
Cheers,
WK
Attachments
I believe that the TEK 576 will source 2A (200ma/ vertical division) and 1500V (200V/horizontal division).
More current -- just build what looks to be the output stage of an amplifier. See the last page of the datasheet for the Linear Tech LT1990.
I want more current not more voltage.LT1990 datasheet Boosted Bidirectional Controlled Current Source is very interesting. Thanks.
Cheers,
WK
This LT1990 circuit needs bi-polar supply to be able to source & sink current.
On top of that if we want to measure a BJT at say 3A with a hfe of say 20, we need to be able to source or sink 150mA. Then a booster stage is required anyhow, and we are back to a circuit not dissimilar to the original by Locky_z.
But interesting anyhow.
Patrick
On top of that if we want to measure a BJT at say 3A with a hfe of say 20, we need to be able to source or sink 150mA. Then a booster stage is required anyhow, and we are back to a circuit not dissimilar to the original by Locky_z.
But interesting anyhow.
Patrick
If you're talking about a curve tracer's gate/base drive, I once used a Howland-type voltage-controlled current source type of architecture, with a push-pull (BD139/BD140) current-booster amp inside the feedback loops (driven by the Howland's opamp's output). Input was a calibrated voltage staircase waveform of switchable polarity, with selectable step size. Output was up to 150 mA or 15V, and could be switched from current steps to voltage steps by switching in a Caddock precision 100 Ohm power film resistor to ground, right at the output. It eventually worked very well, and was able to get to 150 mA, but just barely. If I did it over I would probably use something like an LM1875 (or whatever) for the current-booster amp stage.
by the way, National has a "how to" of Howland current sources on their website. AN-1515 -- must reading by Bob Pease. Also -- Apex Ap Note AN13 on VCCS, Tex Instr datasheet on the REF200, and the ADI ap notes on the AD8276.
I've used the LT1990 with 2SD669A/2SB649A up to 100mA, and in the other direction, it's accurate down to the tens of picoAmps.
I've used the LT1990 with 2SD669A/2SB649A up to 100mA, and in the other direction, it's accurate down to the tens of picoAmps.
Tom,
May I pick your brain ?
1. What device should I include as DUT for a CT, if I design from scratch ?
(FETs, BJTs, diodes, ZDs, Triac, passives, what else ..?)
2. What other features would you like if we were to make a USB version, other than outputting data to text or excel files ?
Thx in advance,
Patrick
May I pick your brain ?
1. What device should I include as DUT for a CT, if I design from scratch ?
(FETs, BJTs, diodes, ZDs, Triac, passives, what else ..?)
2. What other features would you like if we were to make a USB version, other than outputting data to text or excel files ?
Thx in advance,
Patrick
Hi Patrick,
OK. But make sure you also ask some people who know a lot more than I do. And remember that it's been about three years since I even thought about this stuff.
This is bound to be way more than you wanted to know, but here goes.
SCRs? IGBTs? All passives. ICs? Vacuum tubes? Electromechanical? Speakers?
Hey, since you're using a computer, what about things like pulse or step response?!
One thing I wanted to add to mine was a DUT-comparison fixture. I wanted to have two physical DUT ports, and electronic relays that would switch all three of the connections back and forth at a relatively slow user-settable rate, probably centered around about twice-a-second. But now, with a computer doing the displaying, you could do it all with display-mode options (software only!) and not have to add hardware. e.g. User could select a mode to compare to saved "master" DUT's data and then have sub-modes available, like: 1) regular, i.e. "switching between the two", or, 2) always keep the master's display on and plot any newly-inserted DUT overlayed on the same plot, in a selectable color, optionally flashing on and off. You could even have a mode where it would automatically increment to a new color each time aother DUT was inserted and then keep the last "n" plots on the display.
Heck, you could even save all of the plots when in a "batch statistics" mode, where it would plot the mean, deviation, variance, extrema "envelope" and "master" together, or whatever. Back to comparing just two DUTs: you could also plot only the delta between them.
That all brings up another idea that's "just software": Given a saved plot for a "master" DUT of some type, enable the user to specify an acceptable "error band distance" to be applied as a pass/fail criterion for a "select matching DUTS" mode. And you'd probably want the acceptable-error bands to be plotted automatically, immediately, whenever they entered an acceptable error-size. Then when each new DUT was inserted, the software would automatically check to see if the if the specified curves were in-bounds or not and signal the operator.
You could also apply that idea to calculatable parameters, like gain of a BJT, resistance of a resistor, and merely display how far out of tolerance the DUT is compared to the operator's specified value. That way you could use the thing to sort parts into tolerance bins.
The software possibilities are looking endless so I'd better get back to the more-basic stuff. What about being able to provide DC and AC independently, so you could do things like test with DC bias or offset? That just occurred to me and I don't know if that would be useful or not (But it sure sounds good.).
And gee, since it's software-controlled, I guess your sweep waveform could be anything you want! With that in mind, I would try to not preclude the ability to later enable it to sweep the frequency, so you could add some sort of network analyzer functionality.
Gosh, couldn't you also rig it up to look at amplifier output versus your sweep of the input and do all KINDS of cool stuff??
Ok, those weren't "more-basic stuff". Sorry, the "software thing" really has me going.
The sweep waveform could be almost anything, though! Since it's all sent and received with software, it wouldn't even need to sweep sequentially! Wow, you could even inject RANDOM voltages or currents, gather the data for a while, and then just build the display, or let the display "accumulate" as the data points come in.
I remember that one thing I wanted to try to keep in mind, for my curve tracer, was that maybe, just maybe, people would want to strap a big honkin' high-powered amplifier to the main sweep output, so they could test stuff I never dreamed of testing, or even just vacuum tubes. You might want to try to make that as easy as possible for them (and for you, later). Sort of along those lines, maybe you could even make it so that there would be a mode where you didn't supply the sweep, and they could also plug in their own sense resistances (or even sensing device outputs?). i.e. Find a way to give external connections for all or most of the voltage and current sensing, so they could use your software and measurement engine however they wanted. Ah, well, maybe that's going too far...
You might also want to think about haveing an external connection for an extra selectable (user-supplied) current-limiting resistor (assuming you will have a range of selectable ones built in).
And I always wished I could have just used something derived more-directly (or maybe directly!) from the AC Mains, to sweep the DUT, at least as an option, maybe with some step-up transformer voltages, too. It would have been a relatively easy way to get the high currents and voltages that some people will invariably want or need. At any rate, I would want to err on the side of more power than you think is needed.
Well, sorry to have blathered-on for so long about all of that. I guess I still didn't get to the basic stuff. And I gotta get some sleep. Sorry. I'll try to think about it a little more-sanely and get back to you again in a day or two.
Cheers,
Tom
OK. But make sure you also ask some people who know a lot more than I do. And remember that it's been about three years since I even thought about this stuff.
This is bound to be way more than you wanted to know, but here goes.
SCRs? IGBTs? All passives. ICs? Vacuum tubes? Electromechanical? Speakers?
Hey, since you're using a computer, what about things like pulse or step response?!
One thing I wanted to add to mine was a DUT-comparison fixture. I wanted to have two physical DUT ports, and electronic relays that would switch all three of the connections back and forth at a relatively slow user-settable rate, probably centered around about twice-a-second. But now, with a computer doing the displaying, you could do it all with display-mode options (software only!) and not have to add hardware. e.g. User could select a mode to compare to saved "master" DUT's data and then have sub-modes available, like: 1) regular, i.e. "switching between the two", or, 2) always keep the master's display on and plot any newly-inserted DUT overlayed on the same plot, in a selectable color, optionally flashing on and off. You could even have a mode where it would automatically increment to a new color each time aother DUT was inserted and then keep the last "n" plots on the display.
Heck, you could even save all of the plots when in a "batch statistics" mode, where it would plot the mean, deviation, variance, extrema "envelope" and "master" together, or whatever. Back to comparing just two DUTs: you could also plot only the delta between them.
That all brings up another idea that's "just software": Given a saved plot for a "master" DUT of some type, enable the user to specify an acceptable "error band distance" to be applied as a pass/fail criterion for a "select matching DUTS" mode. And you'd probably want the acceptable-error bands to be plotted automatically, immediately, whenever they entered an acceptable error-size. Then when each new DUT was inserted, the software would automatically check to see if the if the specified curves were in-bounds or not and signal the operator.
You could also apply that idea to calculatable parameters, like gain of a BJT, resistance of a resistor, and merely display how far out of tolerance the DUT is compared to the operator's specified value. That way you could use the thing to sort parts into tolerance bins.
The software possibilities are looking endless so I'd better get back to the more-basic stuff. What about being able to provide DC and AC independently, so you could do things like test with DC bias or offset? That just occurred to me and I don't know if that would be useful or not (But it sure sounds good.).
And gee, since it's software-controlled, I guess your sweep waveform could be anything you want! With that in mind, I would try to not preclude the ability to later enable it to sweep the frequency, so you could add some sort of network analyzer functionality.
Gosh, couldn't you also rig it up to look at amplifier output versus your sweep of the input and do all KINDS of cool stuff??
Ok, those weren't "more-basic stuff". Sorry, the "software thing" really has me going.
The sweep waveform could be almost anything, though! Since it's all sent and received with software, it wouldn't even need to sweep sequentially! Wow, you could even inject RANDOM voltages or currents, gather the data for a while, and then just build the display, or let the display "accumulate" as the data points come in.
I remember that one thing I wanted to try to keep in mind, for my curve tracer, was that maybe, just maybe, people would want to strap a big honkin' high-powered amplifier to the main sweep output, so they could test stuff I never dreamed of testing, or even just vacuum tubes. You might want to try to make that as easy as possible for them (and for you, later). Sort of along those lines, maybe you could even make it so that there would be a mode where you didn't supply the sweep, and they could also plug in their own sense resistances (or even sensing device outputs?). i.e. Find a way to give external connections for all or most of the voltage and current sensing, so they could use your software and measurement engine however they wanted. Ah, well, maybe that's going too far...
You might also want to think about haveing an external connection for an extra selectable (user-supplied) current-limiting resistor (assuming you will have a range of selectable ones built in).
And I always wished I could have just used something derived more-directly (or maybe directly!) from the AC Mains, to sweep the DUT, at least as an option, maybe with some step-up transformer voltages, too. It would have been a relatively easy way to get the high currents and voltages that some people will invariably want or need. At any rate, I would want to err on the side of more power than you think is needed.
Well, sorry to have blathered-on for so long about all of that. I guess I still didn't get to the basic stuff. And I gotta get some sleep. Sorry. I'll try to think about it a little more-sanely and get back to you again in a day or two.
Cheers,
Tom
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.