Products > Test Equipment
So what do you do when you need a modern transistor curve tracer?
exe:
Why do you need a "trusted calibration"? Semiconductors drift quite a lot with a slight change in temperature (that's why SMU do pulsed current test). Anyway, how important it really is for *you*? Do you really care about every PPM on your curves? I remember if you design a circuit that depends on particular bjt beta then you are doin' it wrong).
As for me, I'm more curious about leakage currents and frequency response.
switchabl:
Even if your design depends on device matching (or selection for some parameter) you may not strictly need tracable calibration as long as you can indepently verify the performance of the final product in a tracable manner. But you may want it to be safe and to reduce the chance of rejects.
If, on the other hand, you are manufacturing the BJTs, etc., then you absolutely care. Even more so if you are creating a PDK that others will need to depend on to design complex ASICs. It is true that parameters can change greatly with temperature, and also due to process variation, but that is exactly what you need to characterise. Most people never need to do this, however this is exactly the space that these expensive semiconductor analyzers live in. As for frequency response, they usually come with built-in impedance analysis these days.
Lastly, if you are doing research and are planning on publishing your measurements, IMHO you should care as well. Though I know too well that money may well be tighter in that case...
Martin72:
--- Quote from: dobsonr741 on February 19, 2023, 10:20:13 pm ---Why would one want to use a curve tracer in 2023?
--- End quote ---
What should have changed that it no longer requires a Curve Tracer ?
I am currently building one based on Arduino.
This is first for learning purposes, because the data is not different from most other tracers.
Because 12Vdc and a few mA are not really suitable to test transistors/FETs for correctness regarding data sheet.
chilternview:
--- Quote from: alm on February 20, 2023, 10:35:10 pm ---
* An accurate measurement function (5.5-6.5 digit), which some bench supplies could do an okay job on.
--- End quote ---
A 3.5 digit measurement capability would be enough, that's a 12 bit ADC and in fact what the HP4145 has (e.g. on +/- 2V range measure with 1mV resolution).
Then if your display has (say) a 0.1mm minimum pixel size, then you could have a screen of up to 40cm across.
dobsonr741:
--- Quote ---What should have changed that it no longer requires a Curve Tracer ?
--- End quote ---
Few changes: analog building blocks are readily available and cheap as integrated ones with very tight tolerances. Like a voltage controlled amplifier, with accurate logarithmic input. Or true RMS converters. 10x better than discrete parts can ever do.
Then in high power you have specialized parts, again with tight tolerances and guaranteed specs. No longer using a vanilla part, then selecting into tighter specs of finding matching pairs like in the 70’s.
And there is the safe operating area, more important than the curve itself. How would you measure that with curve tracer?
In practical terms, if a discrete curve tracer has a meaningful use in today’s engineering than you’d be able to buy one new.
Btw, I’m not discounting semiconductor measurement as a new chip rolls out of the production line or when it’s developed. But that’s not curve tracer, rather a much more complicated semiconductor test equipment, going into measuring dynamic behavior up to the GHz range.
Oh, if you are into retro instrumentation, please go for it. I saw a wonderful curve tracer in the mid eighties, a vacuum tube based transistor curve tracer. It was already vintage 40 years ago.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version