can I build this curve tracer circuit with an AWG?

[k8943]

Found this lab exercise which corresponded to exactly what I wanted to practice: use XY on scope to visualise characteristic of MOSFET using a function generator.

https://inst.eecs.berkeley.edu/~ee100/sp06/labs/guide08.pdf

On reading through it seemed to make sense but when assembled the first circuit (just to test passives) at the bottom of page 2, it occurred that this wasn't  going to work: my scope (DS1054z) and AWG (SDG1032X) are both grounded.

Attaching probe 1 to the "-" output of the AWG is not going to (and doesn't!) give a great result.

Have I misunderstood something or would the students have had different kind of equipment? Any suggestions for an elegant way to do this experiment?

[rstofer]

It certainly looks like the FG is floating and that the scope ground clips would go to the ground point.  There are some cheap battery powered function generators around.

If you are thinking in terms of something much more elegant, try SLab:
http://r6500.blogspot.com/2018/02/slab-first-release.html

I built up a PCB for this project and it came out well.  I have some great curve traces but I'll point out that it was just for giggles.  I haven't used it for anything important because I can do similar things with the Digilent Analog Discovery.  But the fact that the software is all Python based is pretty compelling.

The thread that got me started:

https://www.eevblog.com/forum/microcontrollers/small-lab-(slab)-release/

[soldar]

Yes , it requires the signal generator to float.

Better get a wooden one

[John B]

Why can't you swap probe A1 and ground so that the scope and AWG share the same ground, then subtract A1 from A2 on the scope? Can the rigol source one of the XY axes from the math function?

[rstofer]

There's a pretty popular curve tracer called an Octopus.  I haven't built one but here is a video:



You can also Google for 'octopus curve tracer'.

[rhb]

You just need a low frequency triangle wave generator with an adjustable voltage range. Op amp datasheets are full of circuits for this.  So look for an op amp that runs off a split supply of +- 30 V. 

For convenience, build it dead bug style with something like the MSP430G2553 to create  a staircase generator that controls the drain-source voltage.  Stick it in a box and you now have a useful piece of test gear that you built yourself. 

But *don't* make the mistake that I made and fail to label the device and  connections and put a schematic on the back.  I recently found a bunch of similar devices I built 30 years ago.  No idea what they are now.  After puzzling over some, I decided to just recycle the box, connectors and switches.  Not worth spending several hours reverse engineering my own work to salvage $1 of electronics which I might not have a need for now.

[Mortymore]

With the schematic shown, probe A1 will measure voltage and in the scope you'll have to invert the signal, since the common is at higher voltage. The probe A2 will be connected to the sense resistor to measure current. In XY mode the curve I vs V will show on scope.

The signal generator will do fine to test the passive components or a single junction, but to actualy trace all the curves for the MOSFET, you will have to raise the auxiliary voltage Vgs, as explained on step 10 of your experiment guide, and for instance, draw on "millimetric paper" (if that's what you call it) each of the observed resulting curves (Id vs Vds) to achieve something like you see on figure 2 of your guide, by superimposing all the curves registered.

I built a component tester from an Elektor article some time ago, similar to the 'octopus curve tracer'

EDIT Link to the FR Elektor magazine with the article: https://archives.doctsf.com/documents/feuilleter_document.php?num_doc=71521&num_serie=211



But for transistors, would be better something  like this if you are into Arduino. This will draw all the transistor curves at once for you, as the "real" transistor curve tracers do.



https://www.instructables.com/id/Transistor-Curve-Tracer/

[k8943]

Gosh, thanks for all the fantastic replies.

Did swap A1 and GND around just to get some kind of result and play with XY. However not only does the 1054Z not have XY from Math, nor (looking in the online manual) does even the RTB2004 so guess that's hard to come by.

SLab looks very elegant. Regarding the octopus understand it may be possible to put a little transformer on the sig gen output to have something isolated - at least for sine like waves. PWM could use an optocoupler...

Partly got into this because made a list of things wanted to do with a sig gen before getting one, so trying to avoid building a board (analog or arduino) to generate the appropriate waves. That being said plan to watch the teardown video for the SDG5000 series which DOES have isolated outputs. Would certainly be prepared to make a board / mod to make this available to the SDG1000. Thing is, where does the problem start? At the DAC - did they wire AGND to DGND? Or at the power stage that increases the voltage? Hmmm

Have made a couple of copies of the Bob Pease differential probe and could maybe use one for the scope. It's just that having that floating sig gen allows for a beautifully simple WYSIWYG type situation so the brain cells can focus on trying to interpret the XY instead of worrying about what's going wrong. Was looking forward to using a staircase function from the second channel on the mosfet gate. Grrr.

[rhb]

The *only* DSO I know of that gives XY from math is the LeCroy. That was a major reason for buying a DDA-125.  It's damned ridiculous. 

They all put in a trace integration feature, but you can't plot a hysteresis curve without building an outboard analog integrator.  Positively stupid.  Just one of many reasons I'm working on a FOSS DSO FW stack.  Even $20K Keysight and R&S DSOs can't do it.

[Mortymore]

Casually found another video from w2aew that I think is of interest on this subject:
#232: More Circuit Fun: Simple transistor curve tracer using Stairstep generator circuit



And I also tried something today, just for proof of concept:

Since my Siglent SDG1032X generator has 2 outputs, I set CH1 to generate the step signal (8 steps, signal available in ARB built in waves) for Vbe, and CH2 generates the ramp signal for Vce.

CH1: 8 steps, 1KHz, 2Vpp, 1Vdc offset
CH2: ramp, 8KHz, 8Vpp, 4Vdc offset

The ramp has to last the same time as each of the 8 steps, hence 8KHz for ramp, and 1KHz for the 8 step signal.

On the breadboard placed a BC547B with 1KOhm base resistor and a 470 Ohm at the emitter. Just some random values, with the only care not go over the current specified for the generator.
Since the generator and scope are tied to GND (any of them floating), I couldn't place the GND scope probes as the experiment guide suggests, but I was not tyring to measure really anything, just to see some curves on the scope. And in the attachment they are.

[rhb]

That looks quite good.

[Mortymore]

I know that this post is a bit old but for future reference, here's some curve tracing with 2 different AWGs.
Apart from the AWG and the Oscilloscope, all that's needed is a couple resistors and some care on the values chosen, so the AWG will not be overload.

So here it goes, something like this: "Forgive me the crudity of the schematic, I didn't had the time to put it on scale or to paint it"




Option 1) Results with the more recent test gear, AWG: Siglent SDG-1032X (Max.Iout: 200mA) and Scope: ISO-Tech IDS-2074E

EDIT2:  Something  posted on an GW-instek GDS2000E scopes that adds to the subject.
To draw a PNP curve in the same way as a PNP, invert ch.1 and ch.2 on scope




Option 2) Results with some old/cheap gear, AWG KKmoon MHS-5225A and Scope: Tektronix 2445

For NPN: BC547B




For PNP: BD140




Setting the signals needed on the SDG-1032X is fairly easy. He has built in the ramp and 8 step signals.
But the MHS-5225A only has the ramp, so the 8step signal had to be built (rampup_8steps.csv attached). For the PNP transistor is just a matter of inverting the signals.

MHS-5225A settings for NPN




MHS-5225A settings for PNP (ch.2 on the scope is inverted)




EDIT: In all cases, don't forget to set the proper signal offset so they start at 0Volt

Steps vs value to be set for the 8 steps ramp-up to set a csv file for the KKmoon Generator
built on Excel, setting the same over the specified rows

----- Rows ------
first -------- last --------- value (on column A)
0001 ------ 0256 -------- 0
0257 ------ 0512 -------- 0.142857
0513 ------ 0768 -------- 0.285714
0769 ------ 1024 -------- 0.428571
1025 ------ 1280 -------- 0.571429
1281 ------ 1536  ------- 0.714286
1537 ------ 1792 -------- 0.857143
1793 ------ 2048 -------- 1

[EDIT2] to (re)place the pictures along the post

[rhb]

Thanks for posting that.  While it is obvious how to do it if you read the Tek curve tracer documentation, lots of people overlook how simple it is once you have a good AWG and a scope.

Have fun!
Reg

[Calvin]

Hi,

I´m still rolling the idea to increase collector signal and -power by simply adding a dc-amplifier.
One of those cheap chinese ebay offers should suffice, eyh?
Collector drive signal could be a rectified sine, same as in the Tek576

regards
Calvin