Smoothing a 6221 sweep

[Joshua S.]

I'm using a Keithley 6221 current source to supply 10-500 microamps to an experimental sample. My experiment is extremely sensitive to transients, so I want to turn the current on smoothly with a sigmoidal (error function) ramp over the course of about 300 milliseconds. I've managed to get this working in principle, using a custom sweep loaded via serial and triggered externally by a digital pulse on the trigger link line.

However, I'm still seeing the effects of the staircase nature of the sweep. The minimum delay per step is set by the device at 1 millisecond, and practically I find that triggering any faster than 800 Hz causes the 6221 to miss a pulse or two intermittently. The resulting steps are still large enough to affect my experiment.

My question is:
-- Is there a way to bypass the minimum delay time per step on a custom sweep, so that I could trigger more like 10 kHz?
-- OR can you suggest any low-leakage-current low-pass filtering solutions to remove the 800 Hz steps without compromising the current accuracy?

Thank you,
Josh

[brian.d.smith]

A customer of mine tried the low pass filter idea and the instrument oscillated. That is no guarantee that you would experience the same problem, just a data point of someone who tried.

If you look at the datasheet, the programming resolution for the 20uA range is 1nA. That is the size of your stairstep and is a fixed behavior.

For the 200uA range, 10nA is your step size.

One idea is to use a voltage source with a large inductor in series with your device. V = L* di/dt and di/dt = V/L
You will get an exponential ramp with a discontinuity only at the beginning.

[Joshua S.]

Brian,

The steps I'm seeing are much larger than the programming resolution, and are limited by the delay time. For instance, consider a linear ramp from 0 to 100 uA in 300 ms. If I trigger at the fastest possible rate of 800 Hz (and I can't go any faster since each step requires a delay of at least 1 ms) that means I'll have 240 steps in that time frame, each of which will have to be ~417 nA high. That means the sweep step delay limit is more than 40 times the resolution limit, and this only gets worse for nonlinear sweeps.

I'll try the inductor idea, though, and see how that works.

Thanks,
Josh

[brian.d.smith]

I see your point. Is 300 ms a requirement for the sweep?

[dbellisario]

I am having the same problem - a sensitive device, on top of significant voltage overshoot in a differential conductance measurement. Any success story?

[brian.d.smith]

I am having the same problem - a sensitive device, on top of significant voltage overshoot in a differential conductance measurement. Any success story?

The current source has nearly infinite output impedance. This is one reason that the 6221 is such a good current source. One disadvantage is that there is no built-in voltage overshoot damping. A voltage source damps voltage overshoot by design wheres a current source damps the current but not voltage.

If you are experiencing voltage overshoot, you may want to consider some form of R/L/C to counteract the R/L/C of your cables, connectors and/or your device. You could begin by trying to characterize the cabling and connectors from the 6221 to just before your device.