DIY Curve tracer, exploring options

[Ilya]

I'm in need of the curve tracer. Of course, it's possible to go out and just buy one. But it's much more fun to build one myself.

I'm familiar with most "Octopus" type circuits, but I'd like to make something more precise and controllable. So, while surfing the web I've come across several devices. But only two of them are worth considering at this point:

• Heathkit IT-1121 (or IT-3121, can't figure out the difference between them).
• This curve tracer from Changpuak

Heathkit looks promising, but most transistors are obsolete and I'll have to find a suitable replacements (which should be quite straighforward). The only problem is the power transformer - I'll probably have to order a custom toroid - it seems to be cheaper than buying and combining several transformers from availale stock. There're quite a lot of switches as well. This increases the cost somewhat, because all rotary switches have 2 decks. And I want everything to be mounted on a PCB, so this leaves not that many options.

Changpuak seems to be considerably cheaper and easier. But it can't do high-voltage tests, and it looks like it's less precise (and definitely less documented) than the Heathkit.

I appreciate any thoughs regarding both of these units. Maybe there're other options that I missed, would be grateful for those as well.

 

[PRR]

.

 

[Ilya]

.

Thanks! Looks promising. Doesn't have all the features the Heathkit's got, but you can't beat "cheap"!

EDIT: uA759 is obsolete. What are suggested replacement in this circuit? I've seen OPA547, TCA0372 and L272 suggested as suitable replacement. The latter two are not terribly expensive, would they work here?

 

[benb]

Just an offhand thought, this idea begs for a microcontroller-based design.

 

[PRR]

It only has to supply collector current. Up to 20mA, TL07x is ample. Higher current opamps are easy to find.

I have a strong feeling the lowest Ib sweep is not what it appears.

 

[Ilya]

Just an offhand thought, this idea begs for a microcontroller-based design.

This will be very complicated and will take too much time. Just take a look here (and read all the steps completed so far):
http://www.forembed.com/project-curve-tracer-requirements.html

I'm not ready for such a project.

I have a strong feeling the lowest Ib sweep is not what it appears.

Do you mean the one on the left (2N3053), that goes down below zero? Could you please elaborate on this a bit more?

 

[My3gger]

A friend made modifications to older Tektronix curve tracer. Of course it has display similar to scope, high voltage capability that works for tubes as well, sockets were replaced and added on extension chasis screwed to tester.
Can't remeber model #, it wasn't expensive and modifications were done by young guy without many years of experiences, although with help from father who is repair guy at national tv/radio.
Just a thought, if you need more information pm me and i will ask about model, modifications and capabilty.

 

[Ilya]

A friend made modifications to older Tektronix curve tracer. Of course it has display similar to scope, high voltage capability that works for tubes as well, sockets were replaced and added on extension chasis screwed to tester.
Can't remeber model #, it wasn't expensive and modifications were done by young guy without many years of experiences, although with help from father who is repair guy at national tv/radio.
Just a thought, if you need more information pm me and i will ask about model, modifications and capabilty.

Thanks for the idea. Old high voltage gear is often bulky and thus not very convenient. I'll let you know if I choose this method.

Currently, I'm still considering both PRR and Heathkit options. I like that Heathkit is very "defined" and you know what you're reading on your scope - it's got controls that are properly marked, it's also quite capable of destroing a lot of small (and not-so-small) signal parts.

PRRs suggested circuit is very simple and very cheap. But it takes some effort to figure out what it shows on the screen. And I need to know exactly what figures i'm reading.

Still can't decide..... Quick and crude costs calculation show  <=15USD in parts for PRR circuit (no PSU) , and ~ 150USD in parts (including power tx and enclosure if I get lucky) for Heathkit.

 

[PRR]

I don't even understand why you need a curve tracer.

 

[Ilya]

I don't even understand why you need a curve tracer.

Well.... Mostly to measure and match components. Sometimes there's a special curve to the component that is needed, and it differs between manufacturers. Curve tracer will make it easier and faster to observe than other methods.

 

[PRR]

> 2N3053), that goes down below zero?

That diagram is c0cked because of how illustrations were pasted-in when preparing copy for printing.

The "base drive" circuit puts a variable voltage into a resistor and calls it current. There is no accounting for Vbe.

> match components.

A curve tracer usually plots "all" possible points.

This is dandy for the maker who wants to sell the part into many unknown applications. Trace the whole useful quadrant and let the designer find his happy zone.

For Matching you want a close-up of THE point you intend to work at. You can usually match with 3 resistors, 2 batteries, and a meter or two, for 2% of the labor to build and debug a curve tracer.

Most parts any "curvature" is 99% fixed by basic geometry and physics. You can extrapolate a lot from one point. Two points may be needed in some cases. Often these can be the extremes: in a JFET, V(gsoff) and I(dss) leave little room for doubt over the whole range. (Unless you are comparing two very different fabrications; but that's an odd way to find a "match".)

Yes, I have always lusted for a curve tracer. But never needed one, and never felt the "fun" was a good use of my time.

OTOH I know a guy who his mentor's favorite saying was "Throw it on the curve tracer!" (It helped it was a large operation which bought CTs as a matter of business, and was selling parts which wanted a "all points" graph for the sales-poop.)

 

[mattamatta]

The Digilent/Analog Discovery box you were looking at in another thread has a few curve tracer projects that seem to exist for it:

https://ez.analog.com/community/university-program/blog/2013/12/07/analog-discovery-bjt-curve-tracer-program.
https://blog.digilentinc.com/using-the-analog-discovery-and-waveforms-as-a-semiconductor-curve-tracer/

When I glanced at it it did look pretty cool for doing anything automated, especially using Python (which interests me).

 

[Ilya]

The Digilent/Analog Discovery box you were looking at in another thread has a few curve tracer projects that seem to exist for it:

Yeah, that's correct. And my threads are actually somewat tied together - I've stumbled upon AD while looking for a curve tracer. I liked that you can automate measurements which is very convenient. I don't know Python, but I beleive it's not that difficult to learn. I'm still considering AD, but in the scope of testing dynamic processors, and it will have to wait.

The "base drive" circuit puts a variable voltage into a resistor and calls it current. There is no accounting for Vbe.

///...snip...///

Yes, I have always lusted for a curve tracer. But never needed one, and never felt the "fun" was a good use of my time.

I see now, yes, this makes sence. However, the circuit is very simple and is intended for general measurements.

I'm aiming for something that will give me precise figures, some instrument that has a real value to it. From this perspective, I don't see anything that compares to Heathkit. Maybe I'm crazy, but I'm going to attempt a build of it. I'm sure this will be a great experience and I'll have a powerful tool in my hands in the end. Being not as trained as you, it will be a valuable learning experience and use of my time.

I'm going to open a dedicated build blog thread in the Lab so anybody who's interested can participate and whatch the progress.