Awful 7.2 and 7.3 Drivers DAQ 6052E firewire 1394

[Tony Roberts]

The supplied NI-DAQ 7.2 and 7.3 drivers do not work with analogue output and analogue input waveform operations that occur within the same program. Leading to rapid and severe termination of Windows with firewire 6052E, but perfect operation with PCMCIA 6062E.

The only drivers that work with 6052E firewire are 6.9.3.

However, deinstalling 7.2 / 7.3 drivers and then installing 6.9.3 does not work.

Hope this saves some people several wasted days. Fortunately, we had excellent support from the local field service engineer in Stuttgart. Unfortunately, this problem was not expected, or known, nor was it in the NI knowledge base. Please investigate and make others aware of this problem.

Cheers,

Tony.

[Sebastian N]

Dear Tony,

could you give me more informations please?
I want to write a knowledgebase entry, but I need more Informations from you.
What for a software and SP you use, are installed?
What was the application, could you poste the VI par example. What Error you get?
Could you work with Max and the Daqpad6052E (1394)?

What are your steps to find a workaround?

This will help to write a sheet.

Thank you for your help and advanced,

best regards.

SebastianN
NI Germany

[Tony Roberts]

Hi Sebastian,

This is a really horrible problem.

The solution to the problem is to use NIDAQ 6.9.3, and to avoid ever installing NIDAQ 7.2 or 7.3.

The steps to find a workaround involve copious amounts of coffee, junk food, and working until 4am in the morning. I wouldn't recommend it. I would, however, recommend a heartfelt apology from whoever wrote NIDAQ 7.2 or 7.3.

The problem lies with NI_DAQ 7.2 & 7.3 and thus affects NI-MAX as well as LabView 7(?) and Measurement Studio 6.0 (easy.io function calls). It is independent of platform (Windows 2000, Windows XP) and computer (3 Sony VAIO notebooks, 1 Dell(?) notebook and an IBM desktop). It only affected DAQPAD6052E 1394, and not a DAQCard 6062E (PCMCIA). It affected two different DAQPads 6052E in exactly the same way.

(The "(?)" refer to equipment and attempts made by the National Instruments engineer to get the DAQPAD6052E to work properly)

I wasted 6 long evenings on this, 2 days, and 2 afternoons. Your excellent and helpful local Stuttgart service engineer kindly stayed with us until late last evening trying to solve it. Eventually I solved it last night at 3 in the morning. So I'd be glad to help you write a sheet.

We initially thought it was a hardware error. What was happening was that when capturing a waveform, above a sample rate of 1000 samp/sec, (using any method) a random amount of data at the end of the waveform was missing / replaced with a odd selection of numbers. Win2000 or XP, desktops or laptops.

There was no error. The card passed all the tests. The returned value of the ScanFunctions was 0, indicating a perfect scan success. The actual scan rate returned by the functions (triggered or simple capture waveform or intermediate I/O) match the desired scan rate (100kS/s to 500S/s). The error could be replicated in NI-MAX.

The workaround is to grab twice the data you need and ignore the last half of the waveform (not exactly elegant, but it worked).

There is another problem, at high frequencies there is a ~1% cross talk from the 1st sampled channel to the second sampled channel. Best case is 1% with FS & differential, worst case is GS & single ended. The workaround is to grab a dummy middle channel, but this reduced the overall maximum sample rate, because the 333kS/s is shared between more channels. The error could be replicated in NI-MAX.

The incorrect data problem was cured by upgrading from 7.2 to 7.3, but the crosstalk problem remained.

However, with NIDAQ 7.3 the real nasty problem occurred.

If a waveform output was looped from the card, and then the instant a waveform input operation was undertaken (the "read" section in intermediate I/O, the aquirewaveform in easy I/O), the computer crashed badly (hanging, not allowing you to "End task" with task manager, or simply rebooting instantly). There was no error. Perhaps a -10800 or similar (cannot complete within the time limit) if we were very lucky.

Putting a Delay (1), Delay (0.05) etc between the output and input functions, had no effect.

We bought a multifunction input output card because we needed multifunction input output. One without the other is pretty useless.

This problem affected the standard sample programs supplied with CVI. As well as the standard LabView programs on the engineers computer.

After the engineer had left, I tried with an earlier NI-DAQ drivers (before the "mx" functions were added.

The best ones in the past seemed to be the 6.9.3, and so I deinstalled NIDAQ 7.3 (reboot) & Measurement Studio 6.0 (reboot). Then I reinstalled Measurement Studio (reboot), and then installed NIDAQ 6.9.3 (reboot), then connected the card. Exactly the same problem occurred.

So I installed NIDAQ 7.0.1. Same problem.

So I started with a virgin machine, installed MeasStudio 6.0, then NIDAQ 6.9.3, then connected the card.

Perfect results. No missing / incorrect data, simultaneous input & output, crosstalk of <0.3%.

So, this morning I deinstalled the machine I'd been trying to get working of every National Instruments program. YOUR DEINSTALLERS ARE AWFUL. They do NOT remove everything. They leave the registry in a horrible mess of references. The \windows\system directory still has NI programs. I quickly lost count of the number of NI programs remaining after deinstalling the programs. THIS NEEDS TO BE FIXED. Please pass on this information to your programmers and managers.

I tried to remove as much as I could find, then I reinstalled Measurement Studio. Then NI-DAQ 6.9.3. Then the 6052E firewire DAQ-Pad. Initially the pad could not be used by windows (this device is not working properly). So I deleted the device from the Windows System Device Manager. Then rebooted, then the card was properly installed. I imagine I missed some of the files that were left on the machine after deinstalling NI software. PLEASE MAKE YOUR DEINSTALLATION PROGRAMS BETTER. MUCH BETTER.

The card now behaves perfectly with the software we tested it with Measurement Studio 6.0 & NI MAX 2.3(?) or 2.2(?).

Hope this helps,

I'd certainly hope this is entered as a knowledgebase sheet, and I'd would very much appreciate an apology from whoever writes the NIDAQ 7.2 & 7.3 drivers and the deinstallation programs.

Cheers, with thanks for your rapid response to this forum.

Tony.