Can I run a stand alone LV (from LV 7.10) under xp?

[TechTaleStream]

Dear all,

I have a problem. Since long we have used a LV .exe application under xp with no problem. Now this pc died. I could save the .exe program and also an installation of NI VISA run time 4.00.

I have come that far that I can start the application in xp compatibility mode (under w10) after having made the LV 7.10 run time library available in the same directory as the .exe.. I can also specify various parameters for the USB-rs232 communication from a subroutine in the actual .exe program.

And at some point it seems it works. BUT, the communication is to an AD converter, and some, say 10%, of the data is outright fake. And we cannot live with that error rate.

Something is utterly wrong.

If I try to start the program as a w10 program, it complains about missing semaphore files....

I need some in-depth input!

Would it be of any help if I post the .exe file? It is not too big 700k. Unfortunately, the source code is missing,

With the Best Regards from Stockholm

[altenbach]

---

@TechTaleStream wrote:

I need some in-depth input!

---

Us too.

What's the meaning of " after having made the LV 7.10 run time library available". Did you install the correct runtime engine or not?

(It could be a fundamental problem that the application was poorly written and now it just runs too fast for the hardware to keep up. Was it a real Serial port in XP, but a USB-Serial adapter in W10? Could it be you need 32bit windows 10? Can you define "fake"? (e.g. data missing, data there, but wrong value, etc.))

[TechTaleStream]

Altenbach,

Thank you for response. I am delighted to provide more info.

I downloaded the LV7.10 run time engine. This resulted in a number of files like lvrt.dll and resource files (*,rsc). I put these in the same directory where the .exe file is. With more of these files the program started. To the best of my knowledge the application was written with LV 7.10.

Yes, the original serial ports were in-built ports in a 'conventional' pc (not lap top), and now we use USB-rs232 adaptors. I have not yet investigated whether 32 vs 64 bit w makes any difference.

The 'fake' data: the rs232 communication reads values from an A/D converter. The origin of these values (from a magnetometer) makes it impossible that there are sudden jumps of order volts (FS is +-10V). I have tried many combinations of stop bits, baud rate, parity etc to no avail.

When I started this effort, there was already NI VISA run time 14.0 installed, which made it impossible to install NI VISA run time 4.00. Therefore I removed a lot of entries in a registry, and at some point I could in fact install 4.00. However, I am not that experienced so I could reliably tell exactly what I did.

The best I can achieve right now is a *working* program (it makes measurements, plot the data and the data can be saved), but with quite many data points which are outright wrong. This happens when I specify xp under compatibility mode.

Should I uninstall all software from NI, and carefully install LV7.10 and NI VISA run time 4.00, so I am sure that all registry and paths are ok?

Is it that I cannot expect to run the program directly with w10?

Does this info give a better view of the problem?

With The Best Regards

[Albert.Geven]

Hi

The only real problem left is the name and type of the magnetometer.

And except from logging, what does the program do?

[TechTaleStream]

Dear Albert,

the program helps you to specify maximum field, number of data points, waiting time between change of field (by sending set points to the power supply for the magnet) and measurement. You also set sensitivity and so forth. It also supports degauss of the electromagnet. The magnetometer is a EG&G model 155. The AD converter is a model 455 from the same manufacturer.

These instruments were commissioned ca 1983 and came with complete documentation(!). It means that if you have some background in electronics you can correct (almost) all electric *problems*.

This instrument is a work horse in our lab, and we are very impressed by the quality.

But now we are stuck :(.  If you can help, we would be more than delighted!

Valter

[altenbach]

---

@TechTaleStream wrote:

I downloaded the LV7.10 run time engine. This resulted in a number of files like lvrt.dll and resource files (*,rsc). I put these in the same directory where the .exe file is. With more of these files the program started.

---

I doubt this is a supported configuration. The correct way would be to actually install the runtime engine.

[Albert.Geven]

Just thought that maybe the source is still available in the .exe file.

If debugging was on during the build the exe could contain the vi's sourcecode.

If needed someone could look inside the .exe by renaming it to a .zip and check what is inside.

[TechTaleStream · Accepted solution]

Dear all,

Thank you for input.

Now it is time for some good news: IT IS WORKING FINE J.

I would now like to describe (to the best of my knowledge…, it is not easy when you try a myriad of things of which the interdependencies are not known..).

Soft ware. The first issue was the software. Yes, I downloaded and INSTALLED the LV 7.10 run-time engine. Then I had the *likely* issue with conflicting NI VISA run-time 14.0x vs 4.00. Then the VSM program properly identified various COM ports etc, time for testing….

However, after I migrated the folder with the actual .exe file plus various files from the LV 7.10 run-time engine to the lap top I intend to run the VSM instrument from –everything *seemed* to work…. Under w10 (!). But with the bug which anyway made it useless: out of ten points, ca 3- 4 are corrupt, ie wrong numbers.

An observation is that the MAX soft ware suggests that I only have NI VISA 14.0, not the 4.0 version(?). Is this important or not?

Hard ware.

The A/D converter is a neat construction (pics attached), dated Jan 1984. But a detailed inspection identified a couple of ‘errors’. The local power supply for the RS232 transmitter is supposed to be +-12V. But the positive ic (7812) was put in reversed(!), and consequently gave +15V instead. But, to my disappointment, this was not the problem…. After that I also saw that the strobe signal was never connected to +5V, as was indicated in the schematics. But that didn’t help….

Gender equality.

A classic thing with RS232 is whether the cable should be ‘straight’ or p2 changed to p3 and vv at one end of the cable. Historically, the cable was a fully populated 9p dsub to the COM port, and would not work with a cable with the minimum Rx, Tx and ground. But both these cables are ‘straight’. Strange!

So, this is the situation: the measurement program ‘works’, but with corrupted data and only with a full 9p cable….. and more strange things: it only functions at one of several programmable baud rates (4600).

However, after some more detailed inspection of the signals….., suddenly the COM port sends on the ‘other’ wire, meaning that now we need a cable with gender change!!

So, a question at this point: can a COM port change gender from the program?

Success.

Now with the gender changed cable it could handle all the various baud rates offered by the A/D converter, but the corrupted data bug persisted.. until I tried the xp compatibility mode…., but then the actual sequence during measurement caused the program to halt. However, if it helps to go from w10 to xp, it may improve even more…. If I use compatibility mode set to w98.

Yes, indeed! Now, ta-da. Everything works perfectly J J.

Sense Morales: never answer the question: ‘Can you tell me when you have fixed the problem?’