SCXI-1001 Chassis causing bad cold junction measurement?

[dabretty]

Hello -

I just spent half a day doing some work on our equipment in another building, where it appears that a possibly bad SCXI-1001 chassis was causing bad cold junction measurement in thermocouple measurement.  I'm curious as to if anyone knows why?

The details of this setup:  In the PC we have a PCI-6052E DAQ board, cabled to a SCXI-1001 chassis, which is full of SCXI-1102 thermocouple amps all with SCXI-1303s (terminal blocks).

The symptoms were that the software I (unfortunately!) wrote was displaying bad, extremely-wrong temperatures (oh, 700 degC or so).  I at first thought that this was some sort of open-thermocouple issue, but when looking at channels individually in MAX I could see that they appeared to be reading correctly.  After a little bit of research, I discovered that the default setting for MAX was to use a CJC source of "Constant" (at 25 degC) while my software was hard-coded in VB to use the built-in sensors.  In MAX, switching this to use built-in yielded the same results (bad readings).  Unfortunately for us, however, we have some temperature swing at the terminal blocks so using a set temperature is highly undesirable.

Thinking that it could be the terminal block (where the cold junction sensor is, right?) we moved our thermocouples to another SCXI-1303 block (and SCXI-1102 amp), but still in the same chassis.  This produced an identical result (apparently bad cold junction readings).  We even tried swapping in a spare 1102 amp to see if the 1102s had all been fried, with no difference.  Suspecting the chassis (we were out of options!), we swapped it out with a spare ... and, voila, it worked!  The cold junction source for our TCs was once again usable.

So my question is, since all data here points to a bad chassis, could this really be the source of bad cold junction readings no matter what 1102/1303s are plugged into it?  I thought the chassis was really nothing more than a backplane, but I guess I must be wrong.

Thanks for any insight,

Brett

[PBear]

Hi Brett,

I want to make sure this isn’t a problem in software first before we deem that this hardware fault. You are right, the chassis does not really have anything to do with CJC so I tend to think there is a configuration or driver miscommunication. First off, check to make sure that in MAX, your chassis that is not working properly is set in the correct daisy chain index. If you have no other chassis connected to it, make sure that it is set to daisy chain index 0. This can change if you have set up another chassis in MAX but it is not running. Also, double check that the address is set properly.

If this does not solve the issue, delete all chassis under devices and interfaces. Then change the address on your chassis’ front and reconnect it to your DAQ board. Set up your modules as always and make sure the address and daisy chain index are correct.

Finally, if none of the above work, try using one 1102/1303 in different slots other than one and see if you are able to get good readings with CJC enabled. If you cannot and you have the correct daisy chain and address set up, then I’m inclined to think this is a strange chassis defect and would call National Instruments.

[dabretty]

Patrick -

Thanks for the reply.  It is a single chassis configuration, with the the index set to 0 (both via the dip switches and the software setting).  Unfortunately I cannot change the address on the chassis right now and reconnect, as we are in the middle of a several-day test (using a spare).  Once this is done, I will try forcing it to another address and reconfiguring.  In my troubleshooting the other day, I did delete it in MAX and add it in again (and all modules were successfully detected), but this was all using the same address again (0).

We also did try different 1102/1303 combinations in different slots of the same chassis.  We hauled over a SCXI-1000 (six slot) as well and directly moved one 1102/1303 combo over, only to see it work there.  To double check that we didn't inadvertantly change something, we immediately moved the pair back over to the SCXI-1001 and it still was in this failure mode.  It was at this point that we dug out a SCXI-1001 spare and began our test.

Also, we checked the three fuses of the chassis and all were operational.  In doing so, we did actually see some poor soldering work, but our techs here indicate this is par for the course with NI chassis they've seen in the past.  I don't mean to knock you guys (you have lots of great hardware and a great software suite), but when we're tracing strange hardware behaviors and discover shoddy (albeit functioning) soldering it doesn't exactly make us feel like we're barking up the wrong tree.

Anyways, when our test finishes I'll play around with it a bit more and try different addresses for the chassis.

Thanks,

Brett

[PBear]

Hi Brett,

When you checked the backplane fuses, did you use a DMM to check the resistance? If the fuse reads something other than approx. 0 then it needs to be replaced. A fuse on the backplane blowing will cause strange readings in a multiple of different ways, sometimes hard to predict. If these indeed are still good, then we'll have to continue debugging when you get a chance to use the test system. Again, make sure the "Daisy Chain Index" (not the address) is set to zero. This is a different index than the address and there is very strange things that can happen if the address is correct but the daisy chain index is set to something > 0 when infact that is the only chassis connected to the system.

[dabretty]

Patrick -

Sorry for the slow response, and thanks for your suggestions.  I have been away from this project for a bit but took a look at it today.  We are not seeing any funny resistances on the backplane fuses, nor is the daisy chain index set incorrectly.  I think we are looking at some weird issue with the chassis.

Where to from here?  We fortunately had a spare that works like a champ ... it is also a newer revision of the SCXI-1001.  I understand our defective chassis is probably not under warranty, is there any way we can get this thing repaired, or is a new chassis the most economical way of dealing with this?

Thanks,

Brett

[PBear]

Hi Brett,

It does appear that the chassis is functioning improperly. The CJC channel should be multiplexed to the DAQ device the same way that the rest of the channels are so perhaps the analog bus on the backplane is bad but we can't notice. Have you tried just scanning the CJC channel ("SC1Mod1/_cjTemp") or tried putting known voltage into one of the channels from an external source and confirming that it reads back correctly?

Assuming you have used all the exact same hardware with a different chassis and got it to work, it looks like pursuing an RMA from National Instruments is the next step. Go to ni.com/support to contact NI and refer to this forum thread for RMA approval. I'm sorry about your hardware problems but this will allow NI to repair your hardware.

[dabretty]

Patrick -

I have been monitoring the CJC channel now separately, but it is misleading as the chassis is now in its functioning state.  Looks like we have an intermittent issue, unfortunately.  I'm going to let this thing run for a few days, and see if it drops out again.

Looking back at all of the evidence, I'm certain that I can point at the chassis as the bad spot, or possibly its communication (1349) adapter.  With an electrical tech, about an hour ago we pulled apart the chassis to look at things but did not see anything obvious.  The power supply slice did appear to have some re-work (and by hand) solder jobs done to it, and our tech resoldered one of the transistor connections as it looked iffy.  But unfortunately we were already in the apparently "working" state of this intermittent problem, so I suppose only time will tell.

Thanks,

Brett