DIY Curve Tracer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I will try to provide some links to some of the posts within the thread.

Alfredo's AS-BUILT Schematics





START OF THREAD
*************************************************************
For some time i have been using my oscilloscope XY, camera and photoshop to create the curve family for some tubes. The process was time consuming as I had to layer all single curves into one.

I was playing around with some 555 ics i had on my bin and i was able to make a simple curve tracer driver.

For the plate sweep i use a variac into a HV transformer to I can limit the voltage excursion. I take the HV ac and put it to a full wave rectifier to get humps above ground.

See attached block diagram for the basic design. It works fairly well, see picture of 12ax7.

For now it only works with small signal tubes, I need to add a stair amplifier to test power tubes or anything that needs over 12V swing on the grid.

I also used a series resistor after the rectifier to limit the current for the 12ax7. I will need to build a series resistor box to handle different current rated tubes. I will also wire a permanent one on a Radio Shack project board.

Also a challenge is taking the current reading of a pentode's plate as my oscilloscope does not have differential probes. I may need to build a simple probe.

Any ideas?


alfredo
 

Attachments

  • IMG_0177.jpg
    IMG_0177.jpg
    278.9 KB · Views: 3,499
  • IMG_0176.jpg
    IMG_0176.jpg
    290.2 KB · Views: 3,306
  • diytracerdiagram.png
    diytracerdiagram.png
    44.7 KB · Views: 3,298
Last edited:
I was playing around with some 555 ics i had on my bin and i was able to make a simple curve tracer driver.

Nice work. Care to share the schematic of your breadboard circuit? I'd imagine there's at least a few of us that would like to try to replicate your efforts. I've always wanted a Tek 570, but I'll never afford one. It seems like you've got something that can produce very usable results for the common dual triodes.
 
I still have not got around to drawing a detailed schematic. I will post when it is complete. I want to run it for some time to make sure it prevails.

But btw, i use my other bench tools to complete the setup.. I use external power supplies for Vcc, tube bias, heater supply and HV ac.

Alfredo
 
You've got the curve tracer bug! (Been there!)

For a simple staircase waveform generator, I used half of a four-bit binary counter IC (HC393?) to get up to sixteen steps, with a passive resistor ladder on the outputs (but into an opamp buffer amplifier) to do the digital-to-analog conversion. I put that through a step attenuator and then a small power amplifier. A chipamp (e.g. LM1875, LM3875 or larger) might work well for the power amplifier, although I used an opamp with a BC139/BC140 push-pull booster amp inside the feedback loop, since I didn't need too much power. I configured the small power amplifier as a Howland-type voltage-controlled current source, but could switch in a power-resistor load to make it into a voltage source. That way I could get either calibrated current steps or calibrated voltage steps. I also had an inverting opamp amplifier that could be switched in, before the power amp. That way I could get either polarity for the staircase.

The clock pin signal for the four-bit binary counter IC was a pulse that was derived from the sweep signal. That way everything stayed synchronized very nicely. You want to switch to the next stairstep for each new sweep cycle.

I used "current sense" resistors in each of the three DUT (device under test) connections, with an opamp-based differential instrumentation amplifier's inputs across each one. I had switches that enabled selecting any of the three DUT currents for the vertical axis display and any of the three DUT voltages (i.e. across any two of the DUT connections) for the horizontal axis. And I had final inverting opamp amplifiers that could be switched in, for horizontal and for vertical, so I could flip the polarity of the display independently for each axis.

You might need switchable voltage dividers for everything, or something like that. I was only doing semiconductor DUTs, and only had a maximum of +/- 15V for the sweep signal amplitude, and only up to about 1.5A available for the sweep.

And yes, I had a whole series of switchable current-limiter resistances. I even had a z-axis output, for blanking the "retrace" when I used a sawtooth sweep waveform. (I also had selectable triangle and sine sweeps, and selectable sweep frequencies from 30 Hz to 22 kHz.)

Your biggest problem might be getting everything to stay calibrated when its temperature changes.

You can see an "older" (2nd-last) version of my curve tracer, and the specs, at Gootee Transistor Curve Tracer and Electronic Component and Device Tester . (Everyone please note that I do NOT sell these, any more.)

Let me know if there's any information I can provide that might be helpful.

Cheers,

Tom
 
A different approach

Still playing around with the circuit.

I changed the design to include a PLL locked a line frequency.

With the previous version, the # of steps per family jumped from 4 to 8 to 12 due to the way the 555 monostable works.

Now I put a Comparator IC referenced to the voltage of the largest step wanted for each family. So in theory it goes from 1 the the max number of steps the staircase gen can produce (limited by Vcc). The problem is that once you go past 10 the oscilloscope has trouble keeping up with the refresh rate.

I like this version better. Once I build it on a permanent board, I will post the final schematic.
 

Attachments

  • curve tracer v2.png
    curve tracer v2.png
    52.2 KB · Views: 3,061
  • IMG_0185[1].jpg
    IMG_0185[1].jpg
    240.3 KB · Views: 2,881
  • IMG_0181[1].jpg
    IMG_0181[1].jpg
    237.8 KB · Views: 224
  • IMG_0183[1].jpg
    IMG_0183[1].jpg
    238.2 KB · Views: 335
That seems like an awful lot of parts for the job? Since the frequency of the anode pulses and the signal to the Schmitt trigger is always the same (mains frequency), why do you need a PLL, or indeed, anything other than the opamp staircase generator?

My tracer currently uses only a comparator to generate a square wave from the AC, followed by a 4520 counter feeding a simple DAC using resistors and an opamp. Less than 20 components, not including the power supply.
 
That seems like an awful lot of parts for the job? Since the frequency of the anode pulses and the signal to the Schmitt trigger is always the same (mains frequency), why do you need a PLL, or indeed, anything other than the opamp staircase generator?

My tracer currently uses only a comparator to generate a square wave from the AC, followed by a 4520 counter feeding a simple DAC using resistors and an opamp. Less than 20 components, not including the power supply.

Hi Merlinb,

I used what I had in by parts bin. Since I've never done this before I studied the circuit of a Tek 570 that creates the stairs at 240hz rate. Since I am using a Miller integrator to charge up the staris, the PLL gives me a very even charge from step to step.If I want fixed steps per family it is easier. But I wanted to be able to adjust the number of steps per family so I can have enough traces equally spaced at whole step values for different tubes. With the comparator at the end i can adjust the number of stairs up to more than 20.

The less parts the better. Can you post your design? I am not sure how it works. Aren't the steps at 60 hz frequency? If this is the case the oscilloscope may have starting refresh problems past a family of 4 steps. Is it possible to change the number of steps per family?
 
The less parts the better. Can you post your design? I am not sure how it works. Aren't the steps at 60 hz frequency? If this is the case the oscilloscope may have starting refresh problems past a family of 4 steps. Is it possible to change the number of steps per family?
I'll try and post it later (the schem is drawn in Proteus and annoyingly takes up the whole screen since the symbols are so large)

My circuit produces eight steps at 3V per step. I added a potential divider so the volts-per-step can be reduced to 2V, 1V or 0.5V, but it always draws eight curves; you just select whichever option fits on the screen most conveniently.

I use full-wave rectified AC, so the steps are at 100Hz (UK), and I found that at least eight curves can be displayed without serious refresh problems, at this frequency.
 
I haven't fired one of mine up for quite a while. But I used 16 steps (4-bit counter) and don't remember there being any significant problem with the usability of the display at 60 Hz. It didn't all appear to be continuously on, at that frequency, of course. But it was usable. (I did definitely prefer the solid higher-frequencies' displays, though!)

I guess maybe you could just have the counter reset at whatever number you want, too, to vary the number of steps.

Best bet is probably to rig up more steps and see what it looks like.

Are you sure you wouldn't enjoy designing a high-power ramp-generator circuit, with a higher frequency? There are some very easy low-power ones with two opamps and then all you need is an audio-range power amplifier.

The classic two-opamp ramp generators usually also make a square wave as a byproduct, too, making it easy to then generate a short pulse to clock the digital counter. I, too, then used a simple resistor-ladder-and-opamp DAC, but followed it with a high-precision switched attenuator that had quite a few steps, and then a small power amplifier that could be switched to produce either current or voltage steps, both under voltage control by the output of the switched attenuator that the stairsteps had gone through.

And then you could add a rotary switch with capacitors to set the ramp generator to multiple frequencies.

(See how it starts? <grin!>)
 
Last edited:
LME49810 as step amplifier and grid current source

I am going to try to add a LME49810 as a step amplifier and grid current source so I can plot curves for tubes with large grid swing and also plot some A2 region curves.

I ordered 4 of the drivers but when I received them I realized that the pins are very close and do not fit on the breadboard. I searched all over and could not find an adapter for purchase. So I went to Radio Shack and purchase some parts.

My plan is that if the driver works I will also use it to make a A2 SE with some Jan1619 I have laying around.

I had to slightly bend some of the pins to make it fit the hole spacing on the perforated board.
An externally hosted image should be here but it was not working when we last tested it.



An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.


Below is the first one I made, after it worked I made the 2nd one in the photos above.

An externally hosted image should be here but it was not working when we last tested it.


I already ran some tests with it and it does what it is supposed to do. I followed the sample circuit depicted in the datasheet.

alfredo
 
Cool!

I did mine for low-voltage semiconductor stuff. So I wanted both voltage and current steps of either polarity, for gates and bases of transistors. After the staircase generator (4-bit counter, resistor-ladder DAC, opamp buffer amp, switched-in-or-out opamp inverting amp for negative polarity, and stepped attenuator), I used a Howland Current Pump (see AN-1515 at national.com), which is a voltage-controlled current source. But after its opamp, inside of its feedback loops, I added a small current-boosting power amplifier. I used push-pull transistors for that but today I would probably use an LM1875 chipamp. So then the staircase could go to ±15V max (1 V per step max) and ±1.5 Amps max (100 mA per step max). To change from current steps to voltage steps I basically just had a 100 Ohm power film resistor to ground, on the output, that could be switched in from the front panel. Worked great.
 
Today I hooked the chip to +-100v to check the maximun swing. Almost 190v pp. Next i am wiring up a 1619 se and se how well it cna drive the tube into A2. The swing for the 1619 will be around 70 vpp.

Love my iphone and the app Web albums. I can take a picure and upload it directly into picasaweb, then copy the link in my ipad to place in the post.

An externally hosted image should be here but it was not working when we last tested it.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.