At the start of the year when I first received the ARTIQ crate (CU3), I did a test on the TTL pulses, and they were all above 3 V (high impedance set on both the on-board switch and the oscilloscope) as you can see from the following figure.

After we installed the DAC crate, we noticed that the output voltages on the TTL pulses dropped. The same 1 µs pulses only go up to (at most) 3 V.

What is interesting is that if we lengthen the pulses, we see that the output voltages grow and saturate around 4 V. The following picture shows two channels superimposed on each other. Channel 1 is just a 15 ms pulse, while Channel 2 consists of three pulses with a 1 ms delay between pulses (pulse 4 ms - delay 1 ms - pulse 4 ms - delay 1 ms - pulse 5 ms).

The output from the TTLs go directly to the oscilloscope, with short coaxial cables. I also played around with switching the termination switches on the boards ON and OFF, and terminating my scope at 50 Ω and 1 MΩ. If the termination switch on the board is set to ON, there is no RC-like growth in the output voltage after the initial jump to 2.5 V.

I understand that this is within specs (Output channels can supply >2V into 50 Ohm loads), and we do see that if our load is set to 50 Ω the initial jump goes to 2 V before growing to 2.5 V. But it is just a little surprising that the output voltage has changed since March.

Many of our equipment require a triggering voltage of at least 3 V. Am I doing something silly here? Do you know what could have gone wrong?

Thank you in advance!

The RC time constant in https://github.com/sinara-hw/DIO_SMA/issues/13 looks like it is on the scale of <1 µs (consistent with ~ 10 pF gate capacitance), but we are seeing an effect that is on the scale of ~ 1 ms (from last plot in post above).

I could try adding a 1 nF cap across gate and source of the FET to see if this goes away, but I feel like these two cases feel different because of the different time constants. Do you suggest that I proceed?

4 days later

sb10q We have the DIO-BNC v1.3. It looks that there are no 10 kΩ resistors across G and S.

Will we void warranty if we add the 10 kΩ resistors across all 64 channels of G and S? Or do you recommend us sending this back for the upgrade?

      ngkiaboon Will we void warranty if we add the 10 kΩ resistors across all 64 channels of G and S?

      No - but make sure to take the usual precautions, e.g.

      • ESD mat
      • ESD bracelet
      • never let cards touch conductive objects while powered up
      • never hotplug EEM cables
        17 days later

        Adding the 10 kΩ resistors solves the problem (Ch 1 has the 10 kΩ resistor across G and S, the rest do not; scope at 1 MΩ input impedance).

        We shall proceed to add 10 kΩ resistors for all channels, including those for the input TTLs.