Fool'n around with DSO & AWG

[mawyatt]

Grabbed a 2N3904 from the junk pile and wondered if it was any good? Could have used a DMM or T7 Component Tester but that's no fun, so decided to see if the SDS2000X+ with an AWG could be easily "coached" into a Component Curve Tracer.

Set the AWG (SDG2042X) Channel 1 to provide the Sweep Voltage using a Trapezoidal Waveform with a DC Offset so the waveform is ~unipolar, this goes to DSO Ch1. AWG Channel 2 was setup with a Rising Stair-step waveform for the bias stepping. Used a 100 ohm current sense resistor for DSO Ch2 in the DUT emitter and a 1Meg base bias resistor to the AWG Ch2. AWG Ch1 was connected to the DUT collector. Here's the results with scope set to XY mode, it's noisy and made no attempt to reduce noise. Result was 2N3904 is fine

#88 is 2N3904 with Vertical scale of 20mv/100 ohms, or 200uV/div

#91 is with higher collector sweep range

#90 was using a custom HV amplifier (we developed awhile back to augment the AWG, details somewhere here on EEVblog) connected to AWG to sweep the voltage higher to show beginning 2N3904 collector breakdown, note horizontal (DSO Ch1) scale factor.

Then we got a 3A 400V NPN and used a Juntek DPA1698 AWG Buffer Amp for more DUT current.

#92 is 3A NPN with 10 ohm emitter current sense R, so scale is 200mv/10 ohms or 20ma/div

#93 is with 1 ohm emitter resistor, scale is 200mv/1 or 200ma/div

Anyway, this was a little fun play'n around this afternoon.

Best,

[RoGeorge]

He, he, nice weekend proj:  "Ad-hoc Curve Tracer!" 

To see the reverse breakdown, I've used once a step-up transformer connected straight to the AWG's output (all passive, no amplifier).

P.S.
Please always add a hand-scribbled schematic with the setup, or else only those who already know will understand.

[mawyatt]

Good name!!

Attached is the sketch and setup images.

Best,

[tautech]

Thanks for sharing this with us Mike. 

Can I ask if you have done this before with other DSO's or did that little voice inside your head tell ask you what happens if I do this ?

[mawyatt]

Never tried this before with any DSO, back in our lab when I was working we had a proper Tektronix Curve Tracer, so no need to try.

Here's another trace with a Power NMOS device (18A 200V), works equally well as it should

Edit: And another of an IRF740 showing the NMOS Triode Region.

Gotta love just how versatile these DSOs are, they aren't just an oscilloscope but a very good Analog Data Acquisition system with almost unlimited flexibility. As 2N3055 mentioned over on the new Rigol HD thread, the ADC is only part of the overall System, the rest of the Data Acquisition needs to been done well and Siglent has certainly demonstrated such to us with the SDS2000X+

Best,

[mawyatt]

Now here's a very interesting set of curves from way way back!! This is a 2N3904 operating upside down or reverse, with the emitter and collector exchanged. So Vce is now from the emitter which is positive to the collector, and Ic now flows into the emitter and out the collector toward ground thru the current sense resistor which is 100 ohms.

Back in the very old days (60s) we utilized certain transistors upside down or reverse as analog switches because they had such a low saturation voltage when operated reverse.

The current scale factor is 50mv/100 ohms, or 500ua/div, and note the first image the low breakdown ~5 volts of the emitter junction and the low saturation voltage is the second (scale is 100mv/div).

Best,

[RoGeorge]

Here's another trace with a Power NMOS device (18A 200V)

Modern instruments are indeed great, not to say nowadays everything can be interfaced with a PC, then there's no limit for what they can do.  For example, for MOSFETs and couple of amps range current, did once a curve tracer from the power supply alone  , no generator and no oscilloscope, just that the RIGOL DP832 power supply was controlled by SCPI over LAN (LXI), then cobbled a few Python lines in a Jupyter Lab Notebook, and got this:


https://www.eevblog.com/forum/testgear/rigol-dp832-power-supply-as-automated-curve-tracer/

Gotta love just how versatile these DSOs are, they aren't just an oscilloscope but a very good Analog Data Acquisition system with almost unlimited flexibility.

Thinking about this as we speak, a good preamp+ADC, and DAC+amplifier, FPGA+RAM, all multichannel, high speed and attached to a PC, would make quite a universal EE instrument.  Those blocks could be configured as DSO, DAQ, AWG, DMM, power supply, controlled load, whatever.  Wonder why nobody does such a box. 

[RoGeorge]

Back in the very old days (60s) we utilized certain transistors upside down or reverse as analog switches because they had such a low saturation voltage when operated reverse.

What!?! 
Gotta try that myself tomorrow (or today, 'cause it's 2AM here ).

[mawyatt]

Here's another trace with a Power NMOS device (18A 200V)

Modern instruments are indeed great, not to say nowadays everything can be interfaced with a PC, then there's no limit for what they can do.  For example, for MOSFETs and couple of amps range current, did once a curve tracer from the power supply alone  , no generator and no oscilloscope, just that the RIGOL DP832 power supply was controlled by SCPI over LAN (LXI), then cobbled a few Python lines in a Jupyter Lab Notebook, and got this:


https://www.eevblog.com/forum/testgear/rigol-dp832-power-supply-as-automated-curve-tracer/

Gotta love just how versatile these DSOs are, they aren't just an oscilloscope but a very good Analog Data Acquisition system with almost unlimited flexibility.

Thinking about this as we speak, a good preamp+ADC, and DAC+amplifier, FPGA+RAM, all multichannel, high speed and attached to a PC, would make quite a universal EE instrument.  Those blocks could be configured as DSO, DAQ, AWG, DMM, power supply, controlled load, whatever.  Wonder why nobody does such a box. 

That's great looking plot!!! A good PS or SMU is certainly an excellent method for creating a nice looking curve tracer. Very nice cobbling

A while back we had thought of developing much of what you describe as a Component Curve Tracer, since good accurate ADCs and DACs are available, however not very good at coding and such, so abandoned the idea. Here's a list of the specs we were after.

Fully self contained with no need for PC or anything else. would be designed around RPi with +12V DC input power.
3 or 2 Terminal Devices
N and P Types
+-30V LV Range @ 2A
+-175V HV Range @ 0.1A
+-15V Gate Drive
+- 0.1A Base Drive
XY DSO outputs
Pulsed Capable to limit heat build up in DUT
DUT Thermal Sensor

The device current would be sensed with a precision Transimpedance Type Amp which forces the test terminal to virtual zero volts. This is similar to high quality Lab type LCR meters. Also various DUT current ranges from 100na to 10 amps full scale.

Anyway, with your skills this type of device should be achievable.

Best,

[ledtester]

Thinking about this as we speak, a good preamp+ADC, and DAC+amplifier, FPGA+RAM, all multichannel, high speed and attached to a PC, would make quite a universal EE instrument.  Those blocks could be configured as DSO, DAQ, AWG, DMM, power supply, controlled load, whatever.  Wonder why nobody does such a box. 

Given your extensive posting history I can't tell if you're being facetious or if you're referring to an Digilent Analog Discovery.

https://digilent.com/reference/test-and-measurement/analog-discovery-2/start

[MikeP]

 For the purity of the experiment, current feedback should be eliminated. I'm talking about the emitter resistor. The entire test setup must be grounded at the emitter. Accordingly, the current signal from the resistor will appear in an inverted form.

[HighVoltage]

#90 was using a custom HV amplifier (we developed awhile back to augment the AWG, details somewhere here on EEVblog) connected to AWG to sweep the voltage higher to show beginning 2N3904 collector breakdown, note horizontal (DSO Ch1) scale factor.

Interesting setup, thanks for sharing!

Can you point to the HV amplifier thread, somehow I can not find it.

[mawyatt]

For the purity of the experiment, current feedback should be eliminated. I'm talking about the emitter resistor. The entire test setup must be grounded at the emitter. Accordingly, the current signal from the resistor will appear in an inverted form.

Yes this would be the proper method, and why it's mentioned in the above (#8) if we were going to design and build a proper Curve Tracing Fixture using a Transimpedance amp which moves the current sense resistor into the feedback loop and forces the emitter node to virtual ground. Of course one also could move the current sense resistor above the collector and use a differential voltage probe across said resistor.

Anyway, the kludge setup shown here was just for some fun playing around 

Best,

[mawyatt]

#90 was using a custom HV amplifier (we developed awhile back to augment the AWG, details somewhere here on EEVblog) connected to AWG to sweep the voltage higher to show beginning 2N3904 collector breakdown, note horizontal (DSO Ch1) scale factor.

Interesting setup, thanks for sharing!

Can you point to the HV amplifier thread, somehow I can not find it.

Sure, here's a couple that you might find interesting. The one we used is based upon the OPA462 HV Op-Amp.

https://www.eevblog.com/forum/testgear/hv-buffer-amp-for-awg/msg3623953/#msg3623953

https://www.eevblog.com/forum/testgear/cheap-diy-high-powervoltage-awg-buffer-amp/msg3449784/#msg3449784

https://www.eevblog.com/forum/testgear/awg-buffer-amp/msg3368650/#msg3368650

Best,

[mawyatt]

Here's a few more interesting plots using this setup with just voltage sweeping the device.

#99 is a 6.2V Zener at 1ma/div with 100 ohm sense R
#100 is a 1N4733 Zener (5.1V)
#101 is a 1N4799 Zener (24V)
#102 is a 1N5817 SBD, note the low froward conduction voltage SBD are noted for

#103 is a very interesting plot, note the zero axises are moved to the lower left corner for a better display. This is a reversed (exchanged collector and emitter) 2N3904 with base connected to ground thur a 100K. A 1K is used from the collector to ground for current sensing (500mv/1K or 500ua per div) for CH2 and 10K is to the emitter from the + sweeping voltage with the AWG feeding the mentioned HV amplifier. Scope CH1 is across the emitter to ground to show the DUT voltage.

Note the negative incremental resistance (dV/dI) exhibited by the voltage across the reversed 2N3904 folding back occurring ~11V while the current is increasing, thus dV/dI is negative!! As the current continues to increase the negative resistance subsides and goes thru zero ~700ua and then become positive.

Just to proof to ourselves we were actually "seeing" a negative resistance, we attached a 1uF Mylar capacitor from the emitter to ground and set the AWG sweep to produce ~13.3VDC across the DUT, which produces about 250ua current, right in the middle of the negative resistance range as shown in plot #103.

#104 shows the classic relaxation oscillations created by the negative resistance reversed 2N3904.

Anyway, hope some folks find this stuff interesting!!

Best,

[RoGeorge]

This is too much fun already!
You sure it's not illegal? 

A switch with zero AC resistance seems a very interesting component, is that exploited in practice?

[mawyatt]

Don't know if it's useful as a switch since the zero ohms would only be over a narrow bias current range. Just did a quick test where we injected a signal into the emitter of the 2N3904 at 1KHz 2Vpp thru a 100K resistor and 1uF decoupling capacitor. You can see the signal collapse as you slowly sweep thru the zero ohm region at ~11.9 VDC bias, and then expand as you pass thur. It's like sweeping a notch filter only in the bias domain!!

Best,

[RoGeorge]

we utilized certain transistors upside down or reverse as analog switches because they had such a low saturation voltage when operated reverse

Is the BJT model critical for the reversed Vsat trick to work?

Asking because in simulation it worked great, seeing reversed Vsat of less than 2mV, while in practice the reversed Vsat was about the same as the normal Vsat.  Wanted to use the reversed transistor's low Vsat to switch on a discharging resistor for a 1.2V Ni-MH cell (discharge current in the range of 100mA) only to find out Vsat is about the same for either normal or reversed Vsat (70mV Vsat reversed, instead of 80mV Vsat normal).   

Tried with two transistor, both with B=430 normal, and with CE reversed one was showing B=3, the other B=12.  The transistors type is SF128 D, video/RF NPN 600mW, TO-39 metal can.  Will look for other medium power transistors to try, and will use soldered wires instead of alligators next time.

Attaching the simulation for a generic NPN (not SF128) and the only datasheet I could find for SF128.

How to choose the type of the transistors better for reversed mode?

[mawyatt]

A lot will depend on what parasitic resistances the transistor has, and how well the model takes care of these as well if it's even capable of accurate reverse behavior. When we utilized this way back, recall the switching type transistors behaved the best (have a higher doping level in the base/emitter) while reversed but can't remember the exact device type tho.

Best,

[RoGeorge]

Will test some switching transistors then, thank you. 

Just for the docs, found a brief class notes for the reversed transistors, has formulas for reversed Beta:
http://web.mit.edu/6.012/www/SP07-L18.pdf

[nctnico]

we utilized certain transistors upside down or reverse as analog switches because they had such a low saturation voltage when operated reverse

Is the BJT model critical for the reversed Vsat trick to work?

Asking because in simulation it worked great, seeing reversed Vsat of less than 2mV, while in practice the reversed Vsat was about the same as the normal Vsat.  Wanted to use the reversed transistor's low Vsat to switch on a discharging resistor for a 1.2V Ni-MH cell (discharge current in the range of 100mA) only to find out Vsat is about the same for either normal or reversed Vsat (70mV Vsat reversed, instead of 80mV Vsat normal).   

Tried with two transistor, both with B=430 normal, and with CE reversed one was showing B=3, the other B=12.  The transistors type is SF128 D, video/RF NPN 600mW, TO-39 metal can.  Will look for other medium power transistors to try, and will use soldered wires instead of alligators next time.

How to choose the type of the transistors better for reversed mode?

Likely the older transistors work better. A long time ago I played with an avalanche pulse generator. I found out that the modern transistors don't work well or not at all. The good old 2N3904 worked like a charm though. Seems to have to do with different production methods.

[tautech]

Yep as Nico says the old bipolar types particularly 2N2369 was famously used for the Jim Williams avalanche pulser.
Dave did a vid on it years ago and for a while member free_electron was making kits however IMO todays Leo Bodnar pulser is a superior product.
Some reading:
https://www.eevblog.com/forum/blog/eevblog-306-jim-williams-pulse-generator/

IIRC it was based on App note AN47 from Linear

[HighVoltage]

Yep as Nico says the old bipolar types particularly 2N2369 was famously used for the Jim Williams avalanche pulser.
Dave did a vid on it years ago and for a while member free_electron was making kits however IMO todays Leo Bodnar pulser is a superior product.
Some reading:
https://www.eevblog.com/forum/blog/eevblog-306-jim-williams-pulse-generator/

IIRC it was based on App note AN47 from Linear

Yes, the fastest rise time I ever got out of the Jim Williams 2N2369 avalanche pulser
was around 400ps. The Leo Bodnar pulser gives faster than 50 ps rise time.
The concept of the Jim Williams avalanche pulser is just amazing!

[RoGeorge]

The Vec saturation voltage in mV range is expected for reversed mode, reversed meaning here that the CE terminals were swapped between them.  It all works at voltages lower than 5V (lower than Vbe reverse breakdown).  I don't see how the reversed mode BJT, and its expected very low Vsat, is related with the controlled avalanche breakdown from that pulse generator.

If the pulse generator mention was only as a side note while recommending older transistors, then these SF128D tested in reversed mode happens to be old indeed.  They are NOS found this weekend while digging in one of the scrap boxes I have from 20+ years ago, found together with some other NOS transistors and ICs from the same era, fabrication date probably during the 80s-90s era.

[mawyatt]

The ones we used reversed way back in 60~70 were a fast pulse transistor, but can't remember the number!! Don't think it was the 2N2369 tho.

BTW thanks for the MIT find, that's a great reference

Best,