Advantages/Disadvantages NI, LabView, etc.

[Coliban]

It´s a little bit surprising for me that this professional test equipment gains such a low interest if i compare the posts about test equipment. Once in a while in my business there come the question to get National Instruments and, for example PXI chassis and modules because of something "standardized" testing for example with LabView or LabWindows.

Therefore i reconsidered my private test equipment with several multimeter, oscilloscopes, signal generators, etc. and when i think about it, it would be much more compact and concise to have one chassis with several modules and a software in favour of several test devices and a zoo of test programs, equipment and starting from scratch every time i think of a new project. And therefore, the ramp up for every new project is cumbersome if there is some amount of time between them, because OS changes, computer changes, and i have to search and collect often from scratch interfaces, probes, devices and start from zero with testing or developing.

I am aware that such devices like from NI are expensive, but they have also small chassis for an reasonable amount of money with slots for 5 or 6 modules. In my mind (it is only a theory) the devices would be in place, the software would be in place and if i use the NI software, LabView for example, i always have a defined hook where to start from. And, before all, a functioning one.

Ok, maybe that's only ideal (or wishful) thinking from me, a theory, but what do you think, is it a good way (maybe a little bit more expensive) to set up an NI testing environment? 

Ok, when i look at the prices i am convinced that the PXI prices makes the difference: a PXI chassis with 5 slots costs more costs 1250€ and the modules are not very cheap so an oscilloscope (modul) to operate on an connected  computer would already costs. For companies where time is money (but for whom not) this would be a useful way.

But the compactRio modules are low priced and would be a good alternative for standardized and repeatable measurements in the private sector (?). The advantage is supposedly the high number of modules and the fast integration into the test environment.

Nevertheless, when i see the many test equipment here on EEVBLOG, which is sometimes very high priced, i wonder why such environments like LabView is not seen more often.

I wonder what the pro and cons may be.

Or do i bark up the wrong tree?

regards

[Coliban]

For private use I found already an obstacle: for the software, in the cheapest configuration, LabView costs a fee of 400€ per year. I would not pay for software subscription (i refused that already for photoshop).

[JPortici]

My three personal ramblings at NI
- Labview is one of the slowest pieces of software i've ever had to use. No matter how fast is the PC
- We have a bunch of NI interfaces of different ages and guess what, each supported from to and up to a different version of labview so if we want to use them all we need multiple installations and multiple OSs as some of the older one require XP to work, other 7.
- Each installation would obviously require a different license

[Coliban]

@JPortici,

and what do you use instead?

[JPortici]

It would have been used for automated test. Luckily one of our boards has the appropriate number of input/outputs (and then some) so i decided to write a dedicated firmware and app in one day

[wn1fju]

At my job, I've probably bought well over one million dollars worth of test equipment hardware/software over the years.  LabView is the ONLY thing I
ever returned for a full refund.  Simply horrible to learn, horrible to use (if and when it works with NI hardware) and very slow.  Take a look at the NI
forums - hundreds and hundreds of posts complaining about errors.  Yet it does have a cult following - some people swear by it (don't know why!).

I have been somewhat successful using NI's Measurement Studio.  I wrote my own application software using Microsoft Visual Studio (.net) calling
the Measurement Studio which calls the specific NI drivers for the hardware.

My preference is Matlab.  However, this is pretty expensive once you add a few of their toolboxes, and is probably impractical for most home users.

[daveyk]

I used VBA (with Access for results databasing) and VBA with GpIB between the instruments and recently added support for some Siglent equipment via network cable.

Of course developing and perfecting your apps while still getting work done takes months, if not years.

Then if you talk to an instrument under test via RS232, and use mscomm32, you find you can not get it to work under windows 10, but everything else does, even VB6!!


Sent from my iPhone using Tapatalk

[joeqsmith]

I wonder what the pro and cons may be.

Or do i bark up the wrong tree?

regards

I've used Labview since version 4.   We looked at LabWindows back in the 80s I think.  This was a DOS based system.   
There was a person who ran a website that was titled Labview Sucks.  Or Why Labview Sucks.  Something like that.   It was several pages long with hundreds of complaints.   I  was not able to find it but there are several smaller similar lists out there.  Even a cartoon:



I have used very few of NI's hardware.  I have their first version of Ethernet GPIB controllers but that's it.  Speed wise, its subjective.   I can tell you that I used to do a lot of low level code in C and would call the routines from Labview.  A couple of years ago, I bought a new PC and at that point stopped using C all together.  Everything I do now on the PC is Labview.   Maybe the best example I could provide you is a home made motorcycle simulator I came up with.   I'm far from an expert with it but there is a pretty decent forum with people who should be able to answer any questions you have about it.

Here you can see me using it with a handheld meter over BLE.
https://youtu.be/e_YzwO62feQ?t=1240

Here I am using it to collect data from a spectrum analyzer and then decoding the FSK data from another meter.
https://youtu.be/Mkec-Eyhg7c?list=PLZSS2ajxhiQBcHhIaGpmm9GyZQfrCzqkv&t=356

[JPortici]

Yet it does have a cult following - some people swear by it (don't know why!).

bet they all are Uni Lab Teachers/TAs (cough cough)

Then if you talk to an instrument under test via RS232, and use mscomm32, you find you can not get it to work under windows 10, but everything else does, even VB6!!

Actually it's working for us.. No idea why or why you have problems. We communicate with the units with a usb->uart TTL cable and the application (written in vb6) also works on windows 10 with no modifications

[Mr Nutts]

For private use I found already an obstacle: for the software, in the cheapest configuration, LabView costs a fee of 400€ per year. I would not pay for software subscription (i refused that already for photoshop).

More like $49.95:
https://store.digilentinc.com/labview-home-bundle/

[Mr Nutts]

In college I used both Labview and VEE (something like Labview but from Keysight) but after a while I dropped Labview and went with VEE because it was much easier.

What a shame that Keysight has clearly no interest in VEE

We also used PXI systems but most of the modules were from Aeroflex and Adlink. I looked at PXI for my home lab but regular instruments are still a lot cheaper

[JPortici]

For private use I found already an obstacle: for the software, in the cheapest configuration, LabView costs a fee of 400€ per year. I would not pay for software subscription (i refused that already for photoshop).

More like $49.95:
https://store.digilentinc.com/labview-home-bundle/

Cool!

however, i'll add another reason to hate labview that i just remembered..
The last time i tried to install it (2016-2017) they asked for a lot of personal informations and about 10 minutes later i filled the form, along the email with the download code and license entry i received another spam email and a spam call. I don't remember if it was NI or one of their associates but they had emailed and called me at the same time ask if i wanted support (paid obviously) and i didn't even have started downloading the software yet. It happened twice over the same month, as i tried to download the thing twice and the multi GB download wouldn't come to finish, or would finish corrupted

[Berni]

I had to use LabView in school and i never really liked it.

The GUI designer is quite good (Even if it does look a kida ugly) for quickly throwing together a nice functional user interface. But the actual programing using the flow diagram never really worked well for me. Sequencing things in that language is annoying and any big program just turns into a huge tangled mess of spaghetti that makes it hard to even see what is going on.

Most of my test automation ended up being in C# trough the standard VISA DLL. I did some with VBA in Excel too, but that was not that great.

Still i have not found the perfect test automation tool. I want something where i can quickly throw together a test without any significant programing effort. For example hook up a few multimeters and PSU to it, just drag drop the desired values into a graph and have it chart my experiment in real time. Then if i want to do more then i would want some user definable "test cycle" feature where a seqence of actions is fired off each iteration and the result of each iteration is plonked into a chart. Then in the end i could just save the chart in any common format, analyze and measure it with cursors,  perhaps even feed that graph into another test cycle sequence to grab some information from it and put it into a new graph etc.

So far i have never found any piece of software that does this in such a quick and easy way. I'm even considering making it myself and open sourcing it.

[SilverSolder]

The big disadvantage with any software suite is the work in getting it up and running, and working with all the instruments.

For hobby use, just using the instrument's own GPIB capabilities with a serial to GPIB converter is easy, cheap, and fast enough for basic instrument control and getting readings on to your hard disk.

The downside is that you have to use Excel etc. to make graphs,  or use your scope as an XY recorder and only use GPIB for control.

[joeqsmith]

This is hardware in the loop motorcycle simulator, currently all written in Labview.   The basic idea is I have the real engine controller plugged into the simulator.  This allows me to try things without having the actual motorcycle.   Originally, I was doing the engine model in C but as I mentioned, the new PC is so fast it's now 100% Labview.  The hardware is all custom.  Labview communicates with this through the PCI bus.   

https://youtu.be/q_89qoFMivg?list=PLZSS2ajxhiQBvWvqMVLdRQMjGofKpQUJr&t=2194

[joeqsmith]

In this video, I am measuring a several shunts out various meters I have looked at.   

https://youtu.be/ne4Ls2mPF-s?t=2173

[AlanS]

Has anybody used EEZStudio? They have done great work in other areas with their programmeable power supply.

https://www.envox.hr/eez/studio/create-an-instrument-extension/introduction/11-eez-studio.html

[joeqsmith]

Having used Labview for so long now, the main reason I continue to use it is because it saves me time.    I can pretty much automate any sort of test I come up with.  This includes collecting the data and post processing it.  I have been burned by LabView a few times.  The most recent, they dropped support for my Ethernet GPIB controllers after I had upgraded.  I had to spend time sniffing the traffic and then wrote my own interface for it, in Labview of course.  It actually works much better now! 

Dave ran a test about a year ago where he life cycle tested the 121GW multimeter rotary switch.    If you watch their video, you get an idea about their test setup.  Pay attention to the yellow sticky notes as I understand that's how you  decode the data.   I was never smart enough to translate it.   

https://youtu.be/uEfzy3xmReI

I had ran a similar test on a few meters but used Labview to control mine.  24 or so minutes in you can see how I am post processing the data.   Setting up this little fixture wasn't days of effort.  Maybe I spent a few hours on the main program and after some feedback from viewers, I spent a bit more time changing it.    It's not uncommon for me to run a test of the course of weeks using Labview.  It's been a very stable platform. 

https://youtu.be/bs5n3a__Yq0?t=449

[Plasmateur]

We started out using Matlab and LabView in my lab about 5 years ago. And, for the most part is was great for introducing me to SCPI codes and how to automate stepper moters, oscilliscopes, NI-DAQ's, etc.

But then the codes just grew too large to manage.... And I'm sure some NI guy could come to the lab and show us how to better manage it.

I really love my job, and I want to be able to take work home. But that would require having the licenses and the equipment.

So we had an intern from some top university come into our lab. He's an EE guy. We're all physicists. We just get the machines working the best we can with duct tape and WD-40.

He shows us python, but we really don't pay much attention to it. He takes whatever code we had and just made it better. Then I download Anaconda python just to putz around with it a bit. Best thing that's ever happened for our lab.

Saved us alot of money and we vowed to rid ourselves of Labview and Matlab. It's just the best thing ever. Get yourself a copy and just connect to one of your instruments. If you love this stuff. You'll be hooked.

I've been around the PXI chassis. They're nice, but too expensive to justify for our type of research.

Edit:I'm really bad at spelling.

[0culus]

Python is awesome for that sort of thing. There's an enormous ecosystem of powerful scientific packages available, to do practically anything imaginable. True, it's probably almost never the fastest performance-wise, but once you get a feel for the Python way of doing things, it's incredibly fast to get code up and running. JetBrains Pycharm Community Edition (it's free) is a great environment working with all things Python as well. Although I'm very much a "vim+shell" guy, PyCharm CE won me over for Python work. It handles all the virtualenv stuff for you, provides an excellent debugger, and overall just shows nice level of polish.

[jeremy]

Another vote for python. When I was doing my PhD, I would have scripts which could acquire the data from experiments (basically some lasers and mechanical stages interfaced to a bunch of GPIB, serial and LAN devices from many different vendors, as well as some NI DAQ cards over PCIe), post process it and generate the eps files for publication, all in one go. It certainly makes reproducible experiments much easier. I even had one experiment where I was “unit testing” my simulation software (also written in python/C++) by actually doing a real experiment, and checking that the simulation matched the real results.

MATLAB is good for maths, but it sucks as a general programming language, and a lot of non-trivial problems end up needing the features of a general language. Just about the only thing MATLAB has going for it imo is that everyone else historically uses MATLAB (and that is changing, thankfully!)

I also use python and the unittest module to do flashing and production testing of assemblies. Works great, and I can ssh into the test system halfway across the world to check on things or examine an unusual fault.

I also vote for pycharm (although the license cost for the pro version is easily worth it for me).

Edit: also, if you have any problems with performance in python, just write the optimised code in C or C++ and plug it in using Cython. I wrote a GPU accelerated ray tracer which loaded geometry from disk and set up the scene in python, precalculated some stuff in C++ and then pushed it to the GPU to calculate using CUDA C and some PTX assembly.

[TheUnnamedNewbie]

At my Uni lab we have constant arguments about Python vs Matlab. Nobody uses NI LabView.

We make quite decent sized automated setups because we do a lot of performance sweeps vs bias voltages etc. We have discrete standalone equipment, PXI stuff, VXI stuff (parbert for example), MXI stuff. Right now, on of our setups uses 3 PSGs, 1 BERT, 3 scopes, 7 supplies, 2 Spectrum Analyzer, automatic attenuators, and a few switch devices. This is stuff from Rohde&Schwarz, Keysight, Anritsu, Virginia Diodes, Keithley.

The people who use of Python say they prefer it because it is free and easier to use. The supporters of Matlab prefer it because the manufacturers tend to have more examples for it or provide code for us (EG, our 65 GS/s AWGs come with a set of matlab tools to configure it and integrate it more easily into other code) and tend to be able to support us better. I haven't used python much, but what I like about a lot of the stuff is that the software takes care of most of the setup and interfacing.

[joeqsmith]

This one is pretty old.  Video showing the Labview to control my old VNA.  The VNA doesn't perform any calculations.  The SOLT is all done in Labview.  The graphs in this program are all custom.  I don't think they supported Smith charts back when I wrote this. 

https://youtu.be/GJd8niwbic8?list=PLZSS2ajxhiQBcHhIaGpmm9GyZQfrCzqkv&t=185

[JohnG]

Another vote for Python.

It has the following benefits:
1. No license to purchase, maintain, or renew
2. Can do everything from simple text scripts to full-blown web driven GUI
3. Can interface with many other programs
4. Easy to find solutions to programming problems
5. Python skills transferable to other tasks besides instrument control
6. Many more people know Python than Labview
7. Python doesn't crap all over your Windows installations (although VISA still does ).

Number 1 has proven to be a really big benefit for my company, since we have several sites with common test setups. Purchasing multiple licenses, dealing with license servers, renewals, etc., has not only direct costs, but all the indirect costs as well. We already do this with Altium, and you can sink a surprising amount of time into dealing with them. And, every time you need another license, their salepeople use it as leverage against you. Don't want to repeat that with another company.

John

[joeqsmith]

I bought a Labview license for home with version 5.0.  I had been using Labview at work for a few years by then and knew that the time it was going to save me would pay off.   I don't stay current with the license and won't upgrade unless they have something I feel will be of benefit to me.   

I have used Labview to write programs for friends and no they don't need a license.  Making an executable is simple.  I provide them an installer like any other program.  That's worked out really well.     

One program I put together is for an engine dyno we built.  I made a model dyno that uses a toy electric motor that I used to develop the code on.  Sorry but I don't have a good video showing the toy or real dyno in operation.  Like the motorcycle HIL, it runs closed loop all in Labview.     

I think NI allowed web based control in Labview 5.  They had some demos setup onsite that you could log into and run.  One was a light bulb and temperature sensor.  I think it was basically you could tune the PID and watch how it controlled.  They had a camera set up with live feed.   Trivial today but I yet have a need for it.   

Recently I wanted to measure the battery life of a few handheld meters.  Many companies will provide Labview examples for their hardware.  In this case, the DAS is custom made.   
https://youtu.be/IoCGMRi536Y?t=235