Bug: Channels get mixed after some time

[Martin Lintzgy]

We can confirm that opening Edge on widows 10 does indeed provoke channel swapping.
From the technician who carried out the test.

"I went out to the lab where the piclog is running, monitoring UUT's , and Edge is on this Laptop, but doesn’t seem to be running. I had a look on Task Manager.
There is one unit with high current, 4mA. approx.. it stands out visually, and as an experiment, when I opened up the Edge app…..the 4mA. unit jumps channel. I did it 3/4 times and it jumped each time.
However, Edge doesn’t seem to be on in the background"

[Martyn]

This is not specifically due to Edge, but more that launching Edge is a CPU intensive process, and can also be seen when the O/S decides to do updates.

What is happening is that the driver is not given sufficient time resources by the O/S to read values from the device and so the devices on board memory fills.

What would be the correct course of action when this occurs, as there will be missing data. Would it be better to acknowledge that there has been a problem and leave gaps in the data, or use some form of interpolation to fill in the gaps.

I can't guarantee how easy any solution will be to implement in the codebase, but can pass on your thoughts to the developers.

[Martin Lintzgy]

Sounding like a buffer overrun problem in the driver.
Your comment
"I can't guarantee how easy any solution will be to implement in the codebase, but can pass on your thoughts to the developers"
Says to me that you are not taking this problem seriously. All very well shouting from the hilltops about your new Cloud service, rather like selling a car with a faulty engine but, hey, it features exiting new cupholders.

We want a logger that will reliably log 16 channels over a few weeks, where the channels logged dont get randomly mixed up!!
If there are data readings skipped because of insufficient resources, then so be it-- Random Channel swapping makes the product 100% useless!.

[mchamley]

I agree with Martin, far better to have missing data than swapped channels, and missing data is also preferable to interpolated data which risks being misleading. And the comments about advertising features versus supplying something reliable are spot on - I'd even rather fiddle about logging the data in a terminal session with some clunky solution over RS232 than have something fancy that doesn't get the fundamentals of reliably logging data right.

[mchamley]

It would be nice to know whether there's any update on this. We've given up on Picolog 6 now and have written our own Python scripts that just write the data to csv files to reduce the CPU and memory usage, in order to reduce the probability of running into this data corruption issue. Of course, knowing that the fundamental cause of the problem is still present doesn't fill us with confidence, so we still want this to be fixed, if Pico ever plans to give it any priority. At this point it doesn't look sensible for us as a business to remain loyal to Pico's products.

[Calo0]

I am currently facing the same issue.
The channel mixing is clearly due to a high CPU usage (not sure about memory) as it happens each time a CPU peak occurs, when I open a browser or launch a programm with visual studio for eg.



Reducing the CPU usage by all means could be a solution, but it feels uncomfortable knowing anything could go wrong if the CPU usage goes up for any reason...