DIY calibration source

[Edwin G. Pettis]

The 50? terminator is for use with the pulse generator, high speed/rf generators have 50? output impedance so you must match it at the other end of the coax or the signal will be anything but accurate.  Audio generators usually have a 600? output impedance and need to be correctly terminated unless you are using an accurate DVM/scope as the monitor. 

[MisterDiodes]

Just another general "Head's Up" and as David L. Jones would put it:  "A Trap for Young Players" if you're building a DIY 'scope calibrator box:

If you're using something like an LTC2057 amp - or any precision amp in general for your calibrator circuit - and the datasheet gives you just the "indefinite" short-circuit output current, usually around 20 or 30mA.  As Edwin pointed out that is a useless condition, since at 20mA or so you're driving the amp's output signal into a zero volts short.  Normally when you design your circuit you will keep your normal operating current no more than say 10% of the spec'd "short circuit" current if that's what's given on the datasheet.  That is around ~ 2mA max normal operating output current for '2057, and any more than than that and the quiet, soft output stage of a precision amp starts acting like it has noisy, distorting resistor drivers; not transistor drivers - and it becomes obvious they are bouncing along on the same die as that 100kHz chopper system.  Oops.

Another possible hidden issue here is if you're using that '2057 (look for this for any op-amp really) as some sort of output buffer that goes to the calibrator box output terminals.  Maybe the op-amp is used as a unity-gain or boost buffer:  So you accidentally short out the output terminals to zero volts.  Now trace back your feedback path to the input of that op-amp and for example:  If you're using a 10V Vref on the non-inverting input to the amp, and your output is being fed back to the inverting terminal (which is suddenly now zero volts) you just created a 10V differential voltage across the amp's input terminals.  In the case of the '2057, absolute max differential input voltage is only 6V.  Oops. So you've just blown out the amp's -input- stage, even though you thought you were safe with that 20~30mA "indefinite output short circuit current" spec.  Now you have to solder in a new chip, and you realized why you should've used a better op-amp in an 8-pin DIP socket <Laughing>.

So when you design your "simple cheapskate calibrator box with op-amp" <Grin>, keep those things in mind as you trace everything out when those output terminals are shorted - typically you might want to have your Vref at say 5V (use an off-the-shelf chip) and boost the output to 10V if you need that - of course that needs decent feedback resistors and maybe a trimmer pot.  If you can calibrate often you don't need exotic resistors. 

I would also look at the specs of what you -really- need for DC calibrator accuracy for an old, low-res 'scope (What are those TDS300's - 2% vertical accuracy, 8-bit?).  I'll bet something like an '6655-5 could be an adequate 5V source, and gain that X 2 via a quiet bipolar amp, and you'd be fine.  Or just put a 10V Vref chip in a box and be done (with perhaps a low-value limiter resistor for a small bit of short-circuit protection) - maybe that's all you need?  Just an idea if you want something as cheap as possible.

[MisterDiodes]

james_s:  A simple "Ghetto Fix" idea that sometimes works to clean up '8013B output pulses for better risetime - you might try feeding the 8013B pulse output signal thru about three or more 'LVC14 logic gates in series, then the output of the last gate goes to 'scope input.  You should get pulses with risetimes < 1nSec...maybe even down to .8 or .7 nSec if you got a fast chip.    Sometimes that's all you need.  Whatever pulse timing inaccuracy the generator is spitting out will still be there, but you'll have about 10X better risetime on the pulses.

This is a handy trick for adjusting higher freq passive probes (>300MHz) when the 'scope's own calibrator output doesn't have a fast enough risetime to adjust its own probe's HF compensator section (esp. LeCroy).

The real fix is to use a better pulse gen, but a 20 cent chip might be worth a try at least.

[james_s]

I was really surprised how picky the scope is about the calibration signal. The standard I have is a Time Electronics DC calibration standard with 0.02% accuracy, it does the job but everything has to be just right or the scopes will fail calibration. I too figured something far less precise would probably do the trick but I was wrong.


The rise time of the 8013 looks really good actually, what I suspect is the problem (but not yet verified) is either excessive jitter in the pulse triggering, or perhaps the pulse length. I'm not really sure what the scope is looking for in that particular case since the failure is the oh-so-helpful "check signal source" but I have not succeeded in calibrating the timing with any equipment I have. That said, the actual accuracy of the scopes seem to be bang on. I sent one home with my UK friend and part of the reason for this endeavor is that I'd like to provide him a way of running through the calibration routine should it ever become necessary. I did dump the DS1644 and verify that now I can load it with the existing data and it still reports pass so this may not even be needed but it seemed like an interesting project to take on. I mean it *should* be easy, on the face of things.