Choosing equipment for automated PSU testing in LabVIEW

[Betonmischer]

I'm looking to build a small lab for testing computer ATX power supplies, which includes voltage regulation, ripple, efficiency and power factor. For that I need a power analyzer, a four-channel scope and a rack of electronic loads. All controlled with NI LabVIEW to set up an automated test sequence that would measure everything with multiple load combinations on different PSU rails.

I think about buying PEL-2000A/B loads from GW Instek for their fast slew rate, option to drive loads synchronously in dynamic mode and reasonable pricing, GDS‐1054B scope and GPM-8213 power meter of the same brand. Would you recommend those or something else for a similar price? Programmability is the main concern for me, as I'm afraid of potential issues when trying to tie it all together in LabVIEW.

Also, I've been recommended to choose equipment with GPIB over USB, and that raises a lot of questions for me. Is there any advantage to use GPIB ? Would mixing GPIB and USB devices work (the scope is USB-only)? If GPIB is indeed the way to go, would a cheaper clone GPIB to USB adapter work in LabVIEW instead of the original one from NI?

Any piece of advice is welcome.

[nctnico]

I'd go for LAN (ethernet) instead of GPIB nowadays. With LAN you can connect equipment to a network more easely and use off-the-shelve cables and equipment (like ethernet switches). USB can be finicky to keep going reliably as the USB bus can be upset quickly / easely in a lab environment. USB just isn't as robust as GPIB or LAN.

Not sure whether using LabVIEW is the best choice. If your requirements are not that high, then using the Python programming language + Pyvisa (together with the VISA layer written in Python) is likely easier to get going compared to getting all the NI layers to play nice. With ethernet connected equipment, there is no benefit from using NI's obfustigated Visa layer as you know the IP addresses. In my own lab I have all my equipment on fixed IP addresses and can access these directly over the network. Python has all the libraries needed to do analysis and visualisation.

[Betonmischer]

Thanks, Ethernet it is then. Speaking of Pyvisa, that seems promising, but I couldn't find any VISA libraries on GW Instek's website, only LabVIEW drivers. Although, I'm not familiar with either of those, so I could be missing something.

[nctnico]

I'm always using SCPI commands as described in the remote programming manuals to control instruments. A Labview driver probably hides that from the user for easier integration in a visual environment like Labview. I have used Labview over 30 years ago but I recall it had a learning curve. If you have some programming experience, the Python way may be easier. If you are more geared towards using a visual 'click it together' environment, then Labview may be more suitable.

[Betonmischer]

I'll try it both ways then when I get the hardware. SCPI commands don't look very complicated. Thanks for the advice.

[rfclown]

You can mix GPIB, USB, LAN. I use NI-VISA. I also use LabVIEW, but it just makes VISA calls, so as long as NI-VISA is installed (free download) I can send SCPI commands with LabVIEW, C, shell scripts, whatever.

GPIB cables are bulky. It's what I used for everything many years ago, but now it's mostly LAN. You just need a switch with enough ports for your gear. I'm not a fan of USB. As someone else said, seems more finicky. Hubs and enumeration and stuff. With LAN I set IP addresses to a dedicated subnet. For some reason someone at work settled on 172.9.33.x and I use that.

I would shy away from clone USB GPIB adapters. Google and you'll find all the time people put in getting things to work with this or that. I don't like the HP adapters either. The NI GPIB-USB-HS ones just work. You can get them off eBay used for <$200.

You're right, the SCPI commands aren't that complicated. You look up the commands you need from the manual, and just send them. Everything is sent and received as a string. For the reception you often have to parse the result to get what you want.