Keysight software rant

[taydin]

So my new keysight awg arrived today, and the first thing I wanted to try is to capture a signal using my agilent scope and give it to the awg to generate the same signal.

So I go ahead and install the software that comes on the DVD. But right off the bat, the installed gets stuck and makes absolutely no progress. Can't find any information about it on the net, so I go ahead and do a windows update session, which takes a few hours. Then I try the installer again and this time it worked. I'm thinking, well, with windows this kind of crap happens, so no worries ...

So the software installation finishes, lo and behold, it has installed about 20 software components on my system !!! And when looking at the task manager, about 10 keysight related software tasks are activately running and in memory! two license manager services, I guess one wasn't enough and they are running two for reduncancy's sake. I'm a little pissed off now, but ok, let me just continue with what I want to do.

So I run the benchvue thing, and it doesn't find my scope. Trying to understand what's going on, I started to run utility programs on the scope. I run something called LXI discovery or what not, and then benchvue discovers the scope. I click on the scope icon, and a window pops up, saying that this application requires a license, which costs $700 or something similar, and offers me 30 day trial  I take the offer and I'm seeing the scope screen now.

The awg was discovered already, I click on the awg icon, and again, it says this requires a license, about $600 or something similar, don't remember now, and again 30 days trial. I take it again.

Now I have both scope screen and awg screen open and scope has the PWM signal i want to test. But for the life of me, I can't find out how to transfer the signal to the awg! The awg wants ARB files, the scope cannot save ARB format. I use keysight's WaveBuilder software (yup $800 license and again, i took the 30 days trial ) and even that damn thing can't save ARB files. benchvue cannot use any other awg file to give to the awg device, it definitely wants ARB. After one hours of cursing and trying random things, somehow I got a new button to appear on the screen which transfers the scope's waveform to the awg.

Even this most basic thing is so hard to do with this software, it's unbelievable!!! 20 odd software components, and it still doesn't make the most basic task easy. And on top of that, it will ask me to pay close to $2000 after 30 days. I am now uninstalling all of this stuff, and I don't want it even if it was free.

[nctnico]

The right way to accomplish wha you want is probably to save as CSV and then import as CSV.

BTW I have given up on software which comes with test equipment a long time ago. It is always written by interns which took a semester of 'the programming language du jour'. Writing your own software or hack some scripts together is always more time efficient.

[lundmar]

I share your frustration. Here we are paying big bucks for expensive instruments only to find out we need to pay even more to do even the simplest of things 

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi-tools.github.io

That is, simple tools that just work with LXI compatible instruments for discovery of instruments, sending SCPI commands, and retrieving screenshots - no need for stuff like NI VISA drivers or other bloated and often inferior software with costly licenses. I think we can do better. After all, LXI is an open standard so why not make some proper open source tools to support it.

Armed with the SCPI command manual of your LXI compatible instrument and lxi-tools you should be able to accomplish most operations.

Now, unfortunately it is not going to work for you on Windows as it currently only supports GNU/Linux-like systems. Maybe it will be ported to Windows in the future, but for now the focus is on new features and testing with more instruments.

For those interested in helping improving the tools there is an active discussion thread here:
https://www.eevblog.com/forum/testgear/open-source-lxi-tools-and-liblxi-v1-0-released-for-gnulinux/?all

[taydin]

The right way to accomplish wha you want is probably to save as CSV and then import as CSV.

BTW I have given up on software which comes with test equipment a long time ago. It is always written by interns which took a semester of 'the programming language du jour'. Writing your own software or hack some scripts together is always more time efficient.

Yes, the scope can save CSV. If I saved the waveform to a flash stick and plugged it into the awg, it probably would have worked. But the bench woe thing only looks for ARB files, so it won't even offer to take the CSV. And yes, you are right, these softwares are a waste of time. Good thing the device itself doesn't depend on these and works independently.

[taydin]

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi.github.io

I will definitely look into this and help if I can. I'm developing system software for GNU/Linux, so I'm quite well versed in it.

[lundmar]

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi.github.io

I will definitely look into this and help if I can. I'm developing system software for GNU/Linux, so I'm quite well versed in it.

Great. I invite anyone to contribute in any way they can (code, ideas, doc, testing, etc.) so that we can steadily improve lxi-tools and make it a good alternative so that we can escape the shackles of those silly and bloated proprietary software offerings.

[HalFET]

I've also pretty much given up on using the standard software. At work I just always wing it using Matlab scripts, the standard Keysight software couldn't even do something as simple as "write value to excel file on key press". You know, the most basic thing you need when you have to verify 50 boards.

[popp3r]

That brings up the question ... are there any test equipment company's out there that have good software integration with their hardware and at a reasonable cost?  Understanding that cost is a relative term.

I know this could be the irresistible force paradox. Just thought I would ask.

...

[Dubbie]

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi.github.io

Do you have any plans to port this to windows?

[bd139]

This is quite funny because I just got a Keysight DMM and didn’t even bother to play with the software because (a) it doesn’t work on a Mac and (b) it is going to be shit anyway. I fired up python and poked the serial port until I worked out how to scrape a reading out of it.

lxi-tools looks awesome. Can it talk to serial devices at all? I’d like to add support to it for my equipment. Assume you accept suitably non poopy PRs?

[lundmar]

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi.github.io

Do you have any plans to port this to windows?

Yes, in the future, maybe a port to Windows. However, current focus is to stabilize and add features to make sure it works with the most popular instruments. One of the new features I'm working on is HiSlip/TCP support as it is used by next generation LXI instruments.

[rhb]

Abandon all hope, ye who enter here.

I returned a Chinese SA because of crappy UI software thinking that if I spent more money I could get better.  Sadly the disease has spread throughout the world.  What is the point of an AWG if you can't load a waveform from a DSO?

So far as I can tell, the only hope is to convince an OEM to open source the UI portion of the FW.  I don't really care about the details of driving the hardware.  Whether using a FW API or the HW API, it's pretty much the same.  Ideally the  FW API is simpler than the HW version. But I would like to escape from stupid.

[lundmar]

lxi-tools looks awesome. Can it talk to serial devices at all? I’d like to add support to it for my equipment. Assume you accept suitably non poopy PRs?

Thanks. However, I'm sorry to say that serial communication is not included in the LXI (LAN eXtensions for Instrumentation) standard. So unfortunately no, no support for serial nor USB. lxi-tools is strictly for modern instruments.

That being said, there are ways to support older or serial-only instruments by using LAN-to-serial VXI11 server devices which can work with lxi-tools. I believe you can buy those online.

Actually, once I'm finished writing the HiSlip/TCP library you will be able to e.g. implement a modern LAN/LXI server on a device like the Raspberry Pi connected to your instruments via serial so you can use lxi-tools with them this way.

[thermistor-guy]

...
BTW I have given up on software which comes with test equipment a long time ago. It is always written by interns which took a semester of 'the programming language du jour'. Writing your own software or hack some scripts together is always more time efficient.

Same here. I save my money for high-quality hardware. The tradeoff is I spend time writing software, which is fine by me.

For test equipment, I write GNU gawk scripts which handle SCPI commands and responses. The scripts run on Linux and on Windows (with Cygwin). Easy peasy.

And when it comes to troubleshooting, there's nothing like having a system which you designed and you control.

[HalFET]

That brings up the question ... are there any test equipment company's out there that have good software integration with their hardware and at a reasonable cost?  Understanding that cost is a relative term.

I know this could be the irresistible force paradox. Just thought I would ask.

...

National Instruments and PicoTech instruments are usually hassle free and work out of the box without additional expenses.

Now the main thing Keysight has going is quite simple, Tektronix is horrendous at software. Ever tried to interface an instrument of theirs? Took me a week last time to get it talking.

But don't forget, for an average company it's cheaper to just buy it if you consider the cost of an engineer. One day of work will cost as much as some of those software bundles...

[nctnico]

I don't get why KeySight software needs a license manager at all. After all you have to connect their hardware anyway. They could sell the licenses as an option on their equipment if they really want to so the equipment becomes a dongle.

[Someone]

I don't get why KeySight software needs a license manager at all. After all you have to connect their hardware anyway. They could sell the licenses as an option on their equipment if they really want to so the equipment becomes a dongle.

Some of their other software is licensed that way, but benchvue has been a bit of a failure and changing direction with the licensing several times doesn't instil any confidence in the stability or support for the product long term. I get they want a unified environment and interface but the ah-hoc software from the existing instruments was quite good (and often free), its the automation and co-ordination that I'm willing to pay for. Its a shame the interfacing and replication of features otherwise available on the devices isnt simply free, then charge for the other things which "add" to the product and extend its capabilities.

[Messtechniker]

As much as I like my 34465A, I dislike Keysight's Benchvue, much for the reasons stated above.
I've tried really hard to like it, but now I have uninstalled it. I would recommend not to waste any more time with Benchvue if you dislike it already. In my case I rolled my own for the few specific functions - long term logging mostly - I needed.

[xani]

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi.github.io

That is, simple tools that just work with LXI compatible instruments for discovery of instruments, sending SCPI commands, and retrieving screenshots - no need for stuff like NI VISA drivers or other bloated and often inferior software with costly licenses. I think we can do better. After all, LXI is an open standard so why not make some proper open source tools to support it.

Out of curiosity, what does it do what sigrok (https://sigrok.org/wiki/Main_Page) can't ?

[lundmar]

I've recently started an open source project/effort to create better and simpler tools for controlling LXI compatible instruments. See http://lxi.github.io

That is, simple tools that just work with LXI compatible instruments for discovery of instruments, sending SCPI commands, and retrieving screenshots - no need for stuff like NI VISA drivers or other bloated and often inferior software with costly licenses. I think we can do better. After all, LXI is an open standard so why not make some proper open source tools to support it.

Out of curiosity, what does it do what sigrok (https://sigrok.org/wiki/Main_Page) can't ?

While sigrok is good software and can do a lot of things (like protocol decoding, gui data representation, etc.), lxi-tools is simpler and strictly focused on modern LAN/LXI instruments. If you simply need to discover the IP address of your instrument, communicate SCPI commands, or grab a screenshot from your instrument, then lxi-tools might be a better choice.

[HalFET]

Out of curiosity, what does it do what sigrok (https://sigrok.org/wiki/Main_Page) can't ?

Sigrok pretty much only supports hobbyist toys, no decent instruments except for a few multimeters maybe and the Lecroy scopes. It's a gimmick, I know of no one that actually uses it.

[TD-Linux]

sigrok is also basically an abstraction layer over many different pieces of hardware, it could use lxi-tools underneath (though it currently doesn't). lxi-tools gives you access to everything your scope does, but many commands are instrument specific. sigrok needs to be higher level in order for its protocol decoders to have a common ground to work on.

Sigrok pretty much only supports hobbyist toys, no decent instruments except for a few multimeters maybe and the Lecroy scopes. It's a gimmick, I know of no one that actually uses it.

I know this is probably bait, but sigrok is primarily a logic analyzer tool. It can certainly capture mixed signals from a scope or DMM too (and in the next version, can use analog channels as digital ones). It has quite a few users in this space, especially with the pulseview gui (along with its primary competition, the saleae software). I guess if you measure "professionalism" as "number of Windows CE splashes shown during boot", then it doesn't measure too high.

[HalFET]

I know this is probably bait, but sigrok is primarily a logic analyzer tool. It can certainly capture mixed signals from a scope or DMM too (and in the next version, can use analog channels as digital ones). It has quite a few users in this space, especially with the pulseview gui (along with its primary competition, the saleae software). I guess if you measure "professionalism" as "number of Windows CE splashes shown during boot", then it doesn't measure too high.

It isn't bait, it's simply on observation in terms of hardware support, even for logic analysers: https://sigrok.org/wiki/Supported_hardware#Logic_analyzers I simply see no decent high speed platform with a good memory supported at this time, so it's quite useless once you start dealing with larger FPGAs or DRAM. And let that be the main use for logic analysers, for debugging MCU applications you rarely need a LA if you're a bit handy with JTAG debugging.

On the multimeters they're missing the prevalent Keithley and Solartron meters. For LCR meters they don't support the classic HP ones, which are pretty much the gold standard, or that newer USB based one from that little Dutch company. They're also missing the common counter and signal generator variants. (e.g. you can pave the street with the old R&S SMx signal generator variants, they're incredibly common.) So yeah, I just find it quite disappointing, I mean it's nice for hobbyist use, but this is not even close to competing with the Keysight and Tek software in my humble opinion. And I'm not up for the "then add it yourself" argument, I'd much rather invest my time in just scripting what I exactly need using the NI APIs.

[Mr. Scram]

National Instruments and PicoTech instruments are usually hassle free and work out of the box without additional expenses.

Now the main thing Keysight has going is quite simple, Tektronix is horrendous at software. Ever tried to interface an instrument of theirs? Took me a week last time to get it talking.

But don't forget, for an average company it's cheaper to just buy it if you consider the cost of an engineer. One day of work will cost as much as some of those software bundles...

It's not cheaper if it doesn't work.

[nctnico]

I know this is probably bait, but sigrok is primarily a logic analyzer tool. It can certainly capture mixed signals from a scope or DMM too (and in the next version, can use analog channels as digital ones). It has quite a few users in this space, especially with the pulseview gui (along with its primary competition, the saleae software). I guess if you measure "professionalism" as "number of Windows CE splashes shown during boot", then it doesn't measure too high.

So yeah, I just find it quite disappointing, I mean it's nice for hobbyist use, but this is not even close to competing with the Keysight and Tek software in my humble opinion. And I'm not up for the "then add it yourself" argument, I'd much rather invest my time in just scripting what I exactly need using the NI APIs.

I agree but it would make sense for Keysight, Tektronix, etc to use their resources to improve an open source project by adding support for their equipment rather than coming up with yet another failed attempt at instrumentation software.

[Mr. Scram]

I agree but it would make sense for Keysight, Tektronix, etc to use their resources to improve an open source project by adding support for their equipment rather than coming up with yet another failed attempt at instrumentation software.

If you watch the John Kenny interview, you'll see that Keysight dreads open-source development. I guess the Rigol affair didn't help either.

[Dr.Trinity]

In our labs we tried the Keysight BenchVue software really hard but came to the conclusion, that it sucks.
It was a horrible experience with the License Manager running in the background and causing even more problems.

A friend in the lab came across this post:
Keysight BenchVue is a Great Example of an Unfavorable Approach to Hardware-Supporting Software
http://toolguyd.com/keysight-benchvue-is-a-great-example-of-an-unfavorable-approach-to-hardware-supporting-software/
So, we knew that we are not alone in this experience.

Why would a billion $ company that firstly makes really good hardware, not offer a simple and good working software
that goes along with their equipment, is hard to understand. BenchVue was on a good way to revolutionize the industry.
But now, I don't think people will buy it anymore.

Does it really have to be a profit center in each department?

Well, we stepped away from BenchVue and I don't think it will be installed in our labs again.

[JohnG]

Python + PyVisa + KS Visa solved a lot of my problems. I will never look back...

Now, if only I could do without KS Visa or NI Visa 

John

[xani]

I know this is probably bait, but sigrok is primarily a logic analyzer tool. It can certainly capture mixed signals from a scope or DMM too (and in the next version, can use analog channels as digital ones). It has quite a few users in this space, especially with the pulseview gui (along with its primary competition, the saleae software). I guess if you measure "professionalism" as "number of Windows CE splashes shown during boot", then it doesn't measure too high.

It isn't bait, it's simply on observation in terms of hardware support, even for logic analysers: https://sigrok.org/wiki/Supported_hardware#Logic_analyzers I simply see no decent high speed platform with a good memory supported at this time, so it's quite useless once you start dealing with larger FPGAs or DRAM. And let that be the main use for logic analysers, for debugging MCU applications you rarely need a LA if you're a bit handy with JTAG debugging.

On the multimeters they're missing the prevalent Keithley and Solartron meters. For LCR meters they don't support the classic HP ones, which are pretty much the gold standard, or that newer USB based one from that little Dutch company. They're also missing the common counter and signal generator variants. (e.g. you can pave the street with the old R&S SMx signal generator variants, they're incredibly common.)
[/qoute]

That's the problem with non-corporate-supported open source, even if someone *wanted* to write the driver for free, they need to own the hardware in the first place.
And the company doesn't want to invest money in OS because having their own is "competitive advantage", even when it isn't because every other company reinvents the wheel too.

So yeah, I just find it quite disappointing, I mean it's nice for hobbyist use, but this is not even close to competing with the Keysight and Tek software in my humble opinion. And I'm not up for the "then add it yourself" argument, I'd much rather invest my time in just scripting what I exactly need using the NI APIs.

Chicken and egg problem, you need software to get to the critical point where most people's needs are fulfilled so it makes more sense to hack on that project rather than set up something on your own. But until software gets to that point, everyone reinvents the wheel and wastes a ton of time.

[HalFET]

For keysight it's simple economics really. It costs less to make it than they earn on it, plus their direct competitor is way worse at it.

Additionally this isn't chicken and the egg problem. NI GPIB/VISA/LXI is universally supported. So you pretty much have to install it. It has reasonably easy APIs, is well documented, and you could find LabVIEW drivers for a turkey sandwich if it comes with a LAN or GPIB port.

While if we use Sigrok we'd have to install a massive toolchain, and the accompanying slew of dependencies, as well most likely. Then we'd have to learn the Sigrok coding standards and how to extend it. It just isn't worth it, especially if I can get along with Matlab or LabVIEW.

And loads of us have the hardware at home, or have access to it at work after hours. It's a matter of asking really, happy to try things out -- as long as I don't have to mess with 20 installers.

[nctnico]

IMHO the NI package is way overbloated, some vendor software depends on a specific version and the latest versions won't even install on older Windows versions (like Windows 7).

[HighVoltage]

I am not sure all this fuss about bad Keysight software is really appropriate.

For one, Keysight started to offer easy to use software to get some data out of instruments and it is far ahead of its competition. I remember well, how frustrating it was 10 or 15 years ago, to get any data acquisition going, even on a 34401A and I was really happy when the first (FREE) logging software was released by Keysight, long before BenchVue.

I have been using BenchVue at some customer sites for simple tasks and it worked really well for this purpose. Installed on a stand alone PC that is not used for anything else, the license manager does not matter. So, for simple business applications it is OK. And usually a business does not care about the cost of a few hundred dollars.

But, I also understand that Keysight closed the door to the hobby electronics world. The older and license free BenchVue software is still available for download and works OK. I am using a dedicated laptop just for BenchVue and it works without major problems.

I wish Keysight would have kept a free and simple license for entry level connections and data collection and control of the instruments. Who knows, may be one day Keysight we will realize this and we will see a U-Turn in their policies.

Or, may be one day an independent company will write cool software for all kind of instruments. But may be that market is too small.

 

[bd139]

Or, may be one day an independent company will write cool software for all kind of instruments. But may be that market is too small.

Working on it

[HalFET]

IMHO the NI package is way overbloated, some vendor software depends on a specific version and the latest versions won't even install on older Windows versions (like Windows 7).

Sure it's bloated, but all VISA packages are bloated. I remember Tek's one was already in the multi-gigabyte territory in the mid 2000s.

[nctnico]

I have been using BenchVue at some customer sites for simple tasks and it worked really well for this purpose. Installed on a stand alone PC that is not used for anything else, the license manager does not matter. So, for simple business applications it is OK. And usually a business does not care about the cost of a few hundred dollars.

My problem is not so much the purchase costs but the maintenance costs caused by the dependence on a license manager and third party libraries. It is very easy to waste many hours to solve issues with drivers and incompatible versions because the symptoms and error messages are clear as mud. In many cases the application just crashes or doesn't work without any indication on why it is not working. Ofcourse things can be mitigated somewhat by using seperate PCs but these also don't come for free and need extra maintenance. Creating software which just works and doesn't need an expert to install and/or diagnose it is an art very few master.

[D3f1ant]

As much as I detest the new licence only model for benchvue, the older free versions still work.

I don't have it installed any more so can't confirm but from memory one of nice features of benchvue is you can drag and drop traces from the oscilloscope to the signal generator. Haven't done it for a while, but don't remember it requiring any great effort to get it to work.