how to calibrate usb2533 and strange behavior with differential readings

[surfboard]

I have a usb2533 (serial #391549, not sure how old it is) that started acting up yesterday. It isn’t exactly “not working” but it is giving pretty unreliable readings compared to the other usb-2533’s I have on hand. Is there a guide somewhere to validate calibrations or perform field calibrations on this device? I saw the magic “calibrate” button in InstaCal but it didn’t have any effect on the results I was seeing. I was using the universal library AInScan function to scan all channels and double-checked with the multi-channel scan in InstaCal.

 

Given a slightly noisy source (~60mV peak-peak) on one of the channels that was around 3.0V, I was seeing very noisy readings (~200mV pp) at either -2V (yes, negative) when measuring differentially, or +2.6V when tying it to earth ground with a settling time of >2 us, or +2.9V when using 1us. I tried 2 other usb2533’s I had on hand, and they read the exact same source at +3.0V correctly, differentially or when tying to earth ground, regardless of settling time. The weird behavior has me thinking the board may be defective. This would be the 3rd or 4th board that’s died on us, perhaps there is something we can do to make them last longer.

[DAQman]

Here's a simple test to run. First Disconnect everything so that nothing is attached to the inputs terminals. Start up InstaCal, select the board and from the menus up top, choose Test->Analog. Click the Scan Test tab. It will allow you to take some snapshots of data. Press Scan Options and select Eng(Volts). Press OK, press Start then View Data. The USB-2500 series is unique in that the open channel voltage reading hovers around zero. Below are the results from my board. If your channel are much higher, like over a 0.5v, then I suspect the inputs got zapped by ESD or experienced continuous over voltage situation.

The InstaCal calibration for the USB-2533 adjusts the field calibration table not the default factory table. In order to take advantage of it use SetCalTableType. The two types are FIELD & FACTORY. 

BoardConfig.SetCalTableType()
Specifies the coefficients table to use for calibration.

Member of the cBoardConfig class. Accessible from the MccBoard.BoardConfig property.

Function Prototype
VB .NET

Public Function SetCalTableType(ByRef calTableType As MccDaq.CalTableType) As MccDaq.ErrorInfo

C# .NET

public MccDaq.ErrorInfo SetCalTableType(CalTableType calTableType)

Parameters
calTableType

Specifies the coefficients table to use. All edge settings are MccDaq.CalTableType enumerated constants. Valid values are Factory and Field.

Returns
An ErrorInfo object that indicates the status of the operation.

From https://www.mccdaq.com/pdfs/manuals/Mcculw_WebHelp/ULStart.htm

Best regards,
John

[surfboard]

I will try to read the floating input values next time I'm in the office. Should I read them single ended or differentially? I assume you want a single ended scan.

- What is the correct procedure to perform a field calibration of the A/D? There are no instructions in InstaCal. Should all the inputs be disconnected, should they be connected to one of the analog outs as a loopback, etc? Does it not matter?

- Is there a way to run the field calibration procedure from universal library functions?

- Where is the field calibration stored? Is it on board in eeprom or is it a local file somewhere?

[DAQman]

Single-ended or differential, it shouldn't matter. 

The USB-2533 calibration is done by the board itself. It has an onboard voltage reference and performs a two point cal using the reference and ground. During the calibration the board internally switches the inputs to the reference and to ground. So there is no need to hook anything up. 

The calibration table is stored on the board with the factory table and is kept in an eeprom. The only way to run the field calibration is with InstaCal. 

 

 

 

[surfboard]

The open multi-channel scan is quite different from yours. This is with nothing connected to it.

If I scan only a few channels, I get near 0 V:

If I scan ch0 - ch20, I get within 0.5V:

If I scan all 32 channels, I get -5, settling to -10 V

 

As a side question, is it possible to read the coefficients directly? I would like a way to compare the factory vs field calibration coeffs.

Edited January 17, 2023 by surfboard
more details

[surfboard]

I also did some individual channel scans, the weird behavior starts at ch23. ch0 - ch22 reads around 0V, but ch23 and up give all kinds of strange values (many around -1 to -3V, some at -5V, some at -10V). It seems like one of the mux ics is fried.

[DAQman]

I will send you a quote to order the repair service. It will be delivered by way of a private forum message.

[surfboard]

Are there any components on this board that you think would benefit from a heatsink? I doubt this was due to overvoltage since all the instruments the board is connected to are powered by a 24V power supply (although normal signal range is -2 to 10V) so my only guess is that they're wearing out due to heat. These boards are often running 32-channel differential scans 24/7 or close to it so our application may be more stressful than most. The datasheet says the max input range is +/-30V.

[DAQman]

To my knowledge the USB-2500 series boards do not run hot and I don't think heat sinks will help, but they definitely can't hurt. However, if they are installed into a restricted area, ventilation needs to be provided. This particular board is from January of 2018 and I couldn't find any record of it or any of your USB-2533 boards being in previous for repairs or calibration.  

[surfboard]

Yes, most were not attached to my name, although the previous board I sent in was damaged beyond repair so it was trashed (s/n 391536 I think). I'm surprised this board is that old, I know of at least 3 failures that we've had since 2018. This must have been an older spare that we swapped in. I will follow up on the private message for the repair service.

 

So is there a way to read the calibration curve coefficients directly to compare factory vs field? I'd like some sort of way to set up a "canary" for when channels are acting up without stumbling across it randomly. I was thinking a periodic "field calibration" in InstaCal and then just see if the curves are "substantially different". It's a shame there's no pass/fail testing built into InstaCal for the A/D module.