Learning Project: Full on transistor/fet curve tracer

[frank freedman]

Well since I learn fastest by doing (not exactly unique I guess) time to start my next trick. Adapting one of the many curve tracer schematics, probably Metzger's circuit from the early '70s with modifications to an ADC capture system and output to a 320*256 ELD panel. (Because I have one to play with) And also add a serial output to enable transfer to a spread sheet of base current, collector current and collector voltage. NPN first, followed by PNP and then FETs. May add the ability to run scroll bars across the screen to enable readout of calc'd hfe. Since I plan to use full wave rectified 60HZ secondary for Vc, I may have to (I hope) use assembler to capture the individual 320 points per 1/4 wave time for each step.

Any comments are as always welcome. I had thought of putting it into oldbitcollectors contest website, but I already have the gadget gangster for my base platform and more than enough T-shirts, that it would be a just for grins entry. If anyone thinks I should post or blog as I go, say so. This is an endeavor that benefits from kibitzing. Of course I reserve the right to ignore any one at anytime even should the release of magic component smoke be imminent. I will check the OBEX and other resources, but if need be I will pester the experts as needed. Promise. And yes, this will take a while to do. Kinda like Mr. Cash's Cadillac, it will get done one object at a time I am sure.

Thanks in advance,
Frank

[prof_braino]

This sound totally cool, mostly because I have no clue about transistor curve tracer.

If its any help there is a logic analyzer implemented in software for the prop as part of the propforth download.

The state of every pin can be sampled
1 cog can collect samples to the size of avaialbe hub memory every 20 cycles,
and 4 cogs can collect samples to the size of available cog memory every cycles.
Sampling can be triggered by combinations of pins states, or other events.

Much of this is in assembler so you might be able to use reuse something.

[Tubular]

Frank, I think this will be a neat little project.
Being able to curve trace other semi's and batteries would be a nice extension too

Good luck.
tubular

[frank freedman]

I've noticed a forum section for projects. would this topic be more appropriate there? This will take a fair amount of time as it is after all my learning project.

Frank

[frank freedman]

Frank, I think this will be a neat little project.
Being able to curve trace other semi's and batteries would be a nice extension too

Good luck.
tubular

Actually when fully completed, it will (should) be able to do diodes and a couple of other things. Right now, I am working on learning to use the MC3201. I will use 3 of these so that the measurements are captured with as little time skew between the points of Ib, Ic, and Vc.

Frank

[Mike Green]

You may want to have the Prop generate one or more voltages rather than using something fixed like the 120Hz signal from a full-wave rectifier fed from the power line. The Prop can generate a nice sine wave using a table and you can use an op-amp to amplify it as needed.

[Leon]

I remember seeing a very simple curve tracer many years ago (I think it was in QST) that just used a low-voltage transformer and a scope. I found a modern version of it:

http://www.changpuak.ch/electronics/Curve_Tracer.php

[frank freedman]

You may want to have the Prop generate one or more voltages rather than using something fixed like the 120Hz signal from a full-wave rectifier fed from the power line. The Prop can generate a nice sine wave using a table and you can use an op-amp to amplify it as needed.

My current plan is to have the Vc generated from a full wave signal w/ a power resistor network to be able to test larger transistors to say Ic~1A and Vc~24V which would be fairly expensive and a bit beyond my skill level with amp design. Op amps become rather expensive in that power range as well. As to the steps, I will either have the prop generate them or as mentioned before, will simply adapt the easiest parts of the Metzger schematic from electronics world 1971 article. I had actually considered simply building that box, and then just instrumenting it with the A/D and conditioning circuits for simple logging to a PC. I may still do this prior to adapting it to the ELD panel and writing the interface and requisite scaling routines.

@ Leon,
I think what you have identified (and I'm thinking most seeing this thread believe) is what we in the service called an octopus. It was esentially a line transformer with series resistors for limiting the current from the low side of the transformer. The curve of which could then be used to perform reasonably good in (or out) circuit testing of suspect components when troubleshooting a problem circuit. A company called Huntron makes one of these as well called the Tracker. The last one I have seen uses a 2"^2 CRT. Surprised they have not released a LCD compact version of this product. Maybe some one could do a prop version.

Here is the output of what I am attempting to do as was done on an Tektronix 576.

http://www.amplifier.cd/Test_Equipme...lications.html

Frank

[Bob Lawrence (VE1RLL)]

Frank,

Interesting project. I built a simple curve tracer a few years ago and it used a 6.3 volt transformer (if I remember correctly) Also, I remember Circuit Ed made one with a Pic and Sword basic:

"The CT100 USB Semiconductor Curve Tracer is 100% USB powered and creates real-time dynamic high-resolution characteristic curves for most popular and common semiconductor devices. It's a great tool for engineers, teachers, and students. Tests NPN, PNP, NJFET, PJFET, NMOSFET, PMOSFET, LED's, DIODES, ZENERS, etc. Can do multiple testing in any category.

No special drivers required, just plug in the USB cable and run. Comes complete with test clips and USB cable. Software and documents are available free online.

MADE IN THE USA!"

http://www.circuit-ed.com/CT100-USB-...cer-P17C5.aspx
================================================== ================================================== ====================================
I'm not sure what happened but I don't think it' s available from Circuit Ed anymore.

[Duane C. Johnson]

Hi Frank;

I have been thinking about the curve tracer connected to a Prop.

Since this is controlled by a micro good improvements can be realized if the current and voltage sources can are pulsed. Pulsed sources with, say, a 1% duty cycle greatly reduce the dissipated power. This allows one to characterize the DUT, Device Under Test, at high power without requiring a heat sink on either the device or source.

Here is a beginning schematic on how to do this.

OK, I didn't enter component values yet as that is yet to be determined.

The basic circuit starts with a ground referenced PWM controlled current source delivering current to the upper +12V referenced PNP current mirror. This reflects current to the -12V referenced NPN current mirror. Both current mirrors develop Voltage and Current outputs of both polarities to the DUT. The output MOSFETs select the desired mode. These floating MOSFETs are turned through simple Blue LED optocouplers.

The Enable Output transistors, in the upper left, enable or disable the current mirrors and consequently the outputs. Note! For a 3 terminal device such as a transistor 2 of these are required.

In addition, a sample and hold AtoD converter is required to read the voltage of the DUT.

The intention here is to make lots of discrete measurement then display them as families of curves on the display device, not as sweept continuous curves. With enough samples this will resemble a continuous display.

I suspect a complete family of samples can be made in 1 second or so.

This circuit can also be used in "Continuous Mode". However, sweeping is done by continuously changing the PWM values. Or, you could simply use a sawtooth circuit, but this would not be as flexible to use.

For me, I intend to use DongleBasic as the logical controller. This is a terminal based program and will output the data values to my computer through the serial port.

FemtoBasic can display the results on a graphics display.

What do you think?

Duane

[Duane C. Johnson]

BTW,How do I post my schematics, linked to my webpage as an image, using the nice thumbnail I see in other posts?
Duane

[frank freedman]

BTW,How do I post my schematics, linked to my webpage as an image, using the nice thumbnail I see in other posts?
Duane

Here is the link to the circuit I plan to "adjust" for starters.

http://www.ka-electronics.com/Images/Curve_Tracer.pdf

There some other variants to try as well. If I was stronger in analog, I would make the driving end capable of +/-30V sawtooth @ max 1A for Vc and replace the UJTs in the stepper ckt with a DAC, but since I have limited time, that is why I plan to start with an existing device and simply instrument it w/ ADCs and logging /display capabilities. The fun will be incrementally altering it into unrecognizability after it initially works. One of my major goals was also hit on by Duane in his comment regarding pulsed operation. The original circuit was designed for a standard scope thus requiring a repetitive waveform to be displayed unless you had either storage or the old polaroid camera unit. On higher power devices, that could be a rather warming experience. If this is eventually triggered for single set of steps, at one half cycle of AC for each of say,8 steps, minimal thermal stress on the device results. Entirely doable as digital capture and logging eliminates the requirement of repetitive display to see and analyze the resultant waveform as would be required for a scope.

Frank Freedman

PS. My thanks to Daniel Metzger for the original article. Good ideas never go away, just morph into another form.

[tdlivings]

Frank
Neat project, this post was good reading over the week end. Enjoyed your link to the original Metzger article and the 576 Tek curve tracer and Leon's link
to two versions of a curve tracer.

Did you build the Metzger project in the past and now want to add digital storage to it or are you starting from scratch?

I have used the 576 curve tracer in the late 70's and there was a newer one that had an op-amp trace plug in, and a lot of poloriod film LOL.

Tom

[frank freedman]

Frank
Neat project, this post was good reading over the week end. Enjoyed your link to the original Metzger article and the 576 Tek curve tracer and Leon's link
to two versions of a curve tracer.

Did you build the Metzger project in the past and now want to add digital storage to it or are you starting from scratch?

I have used the 576 curve tracer in the late 70's and there was a newer one that had an op-amp trace plug in, and a lot of poloriod film LOL.

Tom

Hey Tom,

I am kinda paralleling the whole thing. Learning the prop as I go along, playing with the MCP3201, and building up the core of the Metzger circuit, that being the step generator and base current derivation. After that, I build the full wave circuit to generate the Vc level. Then comes some op amp adaption and the ADC part of the show. Once I can capture the values, they will get stored in the hub ram and be available for display via a 320*256 ELD and serial transfer to a computer for graphing (and maybe auto analysis via spreadsheet.

That is the broad brush plan for the project. I will try to post meaningful parts as I go along. I don't know it this will show up in the projects section of the forums, I never really got back an answer as to where or when that may be a more appropriate forum. Perhaps when I am ready to post chunks, uh I mean parts.

Frank

[frank freedman]

Well its been a bit since I last updated this project, but now I actually have some good progress on the whole mess. On the hardware side, most of the analog guts for generating the IB steps are now about 95% wired. Couple more caps and it will be ready for the smoke check then functional testing. Then wire up the Vc side of the unit. As to the A/D side of the system, I have been working with the MCP3201 from Microchip. Spin is completely out of the question for the project, as even at its most optimized I could not get better than about 1.6mS conversion time under spin. That works out to a bit over 600Hz, and by nyquist sampling theorem, a maximum input of 300hz signal. Great for slow changing inputs, pots, temp / gas / other of these types of senors, but not on this particular project.

On the software side of the house, using pasm, I have been able to set up the framework to the main loop using shared hub memory for each of the 256 samples expected and am feeding test values into the location and have verified the handling of the values is correct. Next thing is get the MPC3201 capturing under pasm in excess of the 60+kcps that I calculated are needed to do the job for the trace capture. So more software and hardware kinks to work on, but progress continues.

If I can figure out how to post pictures here, I have a couple of oscillograms to post showing the conversion timing under spin. The area with the clock pulses and CE low are the conversion time.

Frank

Those of y'all in the way of Irene, stay safe. I slept through Fran back in the day, and that was more than enough for me!!!!!

[Cluso99]

frank: If you go to the advanced section, set your cursor where you want the thumbnail, go down to the upload picture section and click on it you will get a window open. Then, select the file(s), upload them, and click on the bottom button (insert? - beside the done button).

These projects sound interesting

[vettezr1]

Leon I still use that circuit I posted one in the projects about 2 years ago but I still use the circuit you posted to this day it does diodes transitors the only problem is I use it on my old Tek scope becuase I can not for the life of me figure out how to set X,Y on the prop scope

[frank freedman]

Update time:
The project now stands at 5 cogs. Got 'em use 'em. Cog0 init, control, and time base generator. Analog acquisition, cog1 Ib, cog2 ic, cog3 Vce, cog4 ELD panel. More fun later. Time to go do what my CRES says I can do (day job).

Frank

[tdlivings]

Frank
I posted here instead of in your counter question because it makes more sense here.
http://forums.parallax.com/showthrea...unter-Question

Here is a conceptual idea of a zero cross detector to get a falling edge at the
beginning of each collector sweep voltage. The comparators compare oppisite phases
of the low voltage supply transformer. The pull down edge is the one that is the best as the
LM139 style comparators are open collector and pull down hard but are an RC charge going
up. The circuit is right from the data sheet of the LM139 .
Each comparator has a 100k to 10k divider when the input is near zero making that the
switch point.

[frank freedman]

Getting closer to going back to the hardware. Need to finish wiring up the step generator, then the Vc supply. On the software side, the cog count has changed a bit. cog0 for master control, cog1 for clockgen/csel /pulse count/etc. cog2,3,4 for the three A/D chips. cog 5 ELD panel when implemented. 2,3,4 will parallel capture the output of their respective ADC chip data line while monitoring the clockgen pins. Will simply clock into a register the 15 pulse results, and mask with $00000FFF for a 12 bit mask. May still capture to cogram and bulk transfer to hub. not sure which way I will choose yet.

Will need to figure out how to post some code for review once I get it a bit cleaned up. Peer review will be a very good thing to have on this at certain break points in the project. Good for projects an publications both.......

The zero crossing ckt may get put to use as well, need to get the base device going and ready to instrument. Then time to mod for zero det, and other things that will need to be done.

Frank