"XP Mode" under Win7 - Catastrophic 6008OEM failures?

[SnowMule]

Here's a strange one. 

Been developing a test system to blast through and verify all the 6008's inputs/outputs for ISO certification.  Programmable power supply to source voltages, DMM to measure them and provide traceability.  The usual safety stuff applies here, power supply's output is only turned on when I'm using it, then turned off again after the measurement.  Current is limited to 100mA.  The 6008 is reset before the tests so it's in a known state.  Operator is prompted to make the connections then click "Continue", then a task is created, value is applied or read, and the equipment is reset to a "safe" state. 

I developed the system on my native WinXP machine without any sort of problems.  Everything worked as expected...

Brought a new tech on board, her new computer is a Win7 machine.  :ugh

No serial port?  Really?  Hmm... open up a USB-Serial adapter and install the driver... sorta works in 7, real flaky.  Tried a NI USB-232 adapter with similar results.  I've not had much luck dealing with serial devices in 7 (The handful of 5680 RF power meters on the manufacturing line gave me days of headaches when those boxes were upgraded to 7).

Hunt around on teh interwebz and find this "windows xp mode", downloaded that.  It's a virtual machine hypervisor that XP runs inside of.  Sweet... I can do VM's.  I like VM's.  Install LabVIEW, DAQmx, and the project in the XP VM.  Install USB-Serial drivers in the VM.  Everything is working better now than it did in 7 (no surprise to me).

Plug in a 6008 and attach to it in the VM.  XP sees it, installs the device, shows up in MAX... everything is happy.  Run the test... it'll take a few measurements, then the DAQ dies.  Light stops flashing, doesn't show up in the VM, doesn't show up in 7, doesn't show up on my computer anymore, doesn't even show up in UVCView.  Nothing visibly damaged on the board, magic smoke is still inside all the IC's, and the LED stays off.

One board was known-bad, started failing measurements on the line and I replaced it.  The reference voltage had drifted far enough that it wasn't outputting the right voltages.  Who knows what else was going on with it too, it's one of our oldest 6008s.  Plugged in another board and it fried one test further in than the first one did.  Fuse tests good on both boards.  The voltages on U1 (appears to be a voltage regulator of some sort) and the J3 pins are drastically different between the two boards.

Only differences between our two setups are the host PC/OpSystem and the USB adapter on the PSU.

Any suggestions for further troubleshooting/repair of the hardware?  Anyone experienced this phenomenon before?  I'd like to not fry any more 6008s...

[NIthan]

Hey SnowMule,

When you are verifying all the inputs and outputs, what are the tests you are running? Also, is it failing during the input tests or output tests? If you are overloading the inputs and frying the card, then I assume it would be failing on the input tests, but it may be a different issue altogether if it's failing on the output tests. Refer to this manual: http://www.ni.com/pdf/manuals/371303l.pdf. Page 21 and 22 have sinking and sourcing info that implies the DIO can only handle ±8.5mA, so make sure the 100mA is not going to DIO. Also, page 23 gives you a list of considerations for I/O protection. 

I would also trying checking the voltages and current from your PSU with a DMM before sending it into your card, just to double check that the signals are not overloading the card. You have a analog input voltage protection of up to ±35V, so just make sure you don't go higher than that (really you shouldn't be going higher that ±10V since that's the spec on the card). I hope this helps.

-Nathan H

[SnowMule]

Pretty straightforward.  Analog Input test prompts the user to connect the power supply and DMM to the analog input on the tester, when the "continue" button is clicked then the power supply is configured and the output is enabled, short delay for everything to stabilize, the DMM takes 100 readings at 1ks/s and averages them, DAQ does the same thing, the two readings are subtracted and the error is spec'd according to the documentation.  Apply and read 10V from input to ground on the AI-RSE configuration, then 20V on AI-Diff.  Anything on the digital side is either 0V or 5V.  Been through the manual and everything is within the limits of the equipment.  And it all worked flawlessly on my machine.

As I said in my previous post, the power supply's output is only enabled during a test.  When the measurements are complete, the output is shut off.  When the DAQ is sourcing voltage, the power supply is disconnected from the test setup. 

One DAQ fried just after the analog input test, the other one fried a few readings into the digital output test (power supply wasn't connected). 

The strangest part is I'd never fried a DAQ during all my development of this on my native XP machine.  Same power supply, same DMM, same DAQ... then I move the software over to a win7 box, fight with it to get it to sort of work, install microsoft's virtual-XP hypervisor, install LV, USB-RS232 drivers for the power supply, and test software in that VM, _then_ DAQs start dropping dead.....

I'm starting to think this is worth a call to NI.

[NIthan]

Hey SnowMule,

It could be that these 6008's are bad. In which case, if they are still under warrenty, you can call NI and request an RMA, and they should be able to fix these for you. However, our support team doesn't actually support DAQ on virtual machines. This is mainly because each virtual machine software uses different methods for communicating with external hardware and troubleshooting approaches vary depending on what virtual machine you are using. If the problem is that the VM is causing the boards to fry, then I would verify this by measuring the output of your board and testing equipment with a DMM or scope and making sure your current and voltage levels are within normal ranges.

-Nathan H 

[SnowMule]

The boards are dead-dead.  As in, don't show up in MAX or Device Manager.  Nothing comes up in UVCView either.  They're just... dead. 

Pretty sure these are all well past their warranty period.

I'm a little hesitant to plug another 6k8 into the win7/virtualXP machine without knowing what exactly causes the boards to fry.