Measurement deviations when comparing to other instrument

Hi.

I just got the otii and hooked it up for some measurements on a PCB with DC/DC step up regulator.
The otii i set to output 3V. Current set to auto range.
The readings show an average of 70 uA with a lot of spikes with up to 1 mA peaks.

When comparing this capture to another one from my Keysight 34465A, here showing a histogram view of a 5 seconds sample with 50kHz sampling speed, what puzzles me is why I can’t see these spikes on my Keysight. Also the average consumption is different, around 25 uA.

Here’s another view of same capture:

The keysight has a sampling speed of 50 kHz vs the otii 4 kHz, so I should be able to compare the results.
Any ideas, anyone?

Hi,

There’s definitely something triggering the Otii Arc to switch to high-range. We have solved one instance of this happening recently (after the 1.0.7 firmware), but the firmware release is still pending. If you create a case here: https://www.qoitech.com/customer/my/cases/ we can attach the release candidate firmware for you to try out.

I don’t have the high-speed digitize + deep memory options on my 34465A so I can only go to 3.33kHz, and in that setup the Arc and the Keysight have definitely agreed.

/wj

I had a test setup running on my bench that consumes less current, but just wanted to show how I expect the Otii recordings to look like when not in high-range. The steps can easily be seen when in high-range at these loads (as the steps are ~80uA), but they are masked by noise when the hardware is in auto-range and has selected low range. The first readout is done at full speed (~3k3sps on the Keysight, 4ksps in Otii), the readout top left in the monitor DMM window is just a snapshot, not an average:

The second reading is done at 10sps (Otii records at full speed, then downsampled by factor 400 after the recording is done):

These recordings are performed with the Arc running the release candidate firmware.

/wj

I unchecked the auto-range to double-check:

Average current is around 112 uA on the otii while it’s 25 uA on the Keysight.
I think we need to get to the bottom of this since it seems it’s not related to auto-ranging.
The two instruments actually report different result measuring the same circuit board.
I mentioned the DC/DC converter earlier since I thought it might produce current peaks from the switching, that could interfere with the measurement. Another possibility is that the Keysight has a bandwitdh limitation that “filters” the signal in a way it looks better than it is.

I hooked up my logic analyzer in analog mode with sampling rate set to 50 Ms/s and got this (1mV = 1uA).
I used an adapter for converting measured current to a voltage (http://www.eevblog.com/product/ucurrentgold/)

Looking at the image below I can confirm that the peaks of 500 uA are actually there so I guess the Keysight has a filter that truncates these spikes which have a duration of around 2-3 us.

image

That being said, There is still an issue when using auto-ranging since it, when active, reports avg current 72uA and when not active 112 uA.

Hi,

I sent you a PM with the RC firmware for the Arc.

The reason for the additional spikes in your first recording is most likely an artifact of range switching (something that has been solved in later Arc firmware). Both recordings definitely looks like the hardware is in high-range to me (but the first one tries to go back to low-range repeatedly).

Please upgrade to the new firmware (and also perform a zero calibration from the Device menu after the upgrade is done). I believe this will improve the measurements.

/wj

FW upgrade completed. Looks much better now and also measures same average as the Keysight.
Thanks for the quick and accurate support!

BTW, I compared using autorange for same mesaurements.
Readings are different, less detail.
Is that due to a different sampling rate in this mode?

That’s great! But I believe the recording was done in high-range, did you re-select Auto-range in the Current tab? You should see a major improvement in the peak-to-peak noise when in low-range (496uA peak-to-peak in high-range in this screenshot).

Just noticed the second post, the second recording looks more like correct auto-range operation. The high-range recording is not more detailed but it does contain a lot more noise. Auto-range will attempt to stay in low-range as much as possible (until current consumption exceeds 19mA when it transitions into high-range and then automatically switch back to low when current consumption drops again).

When you select a second or so worth of data, you are getting the expected average from auto-range I assume?

/wj

Yes indeed. It’s same reading as with the Keysight.