02-17-2015 05:12 PM
I use a bunch of GPIB-120A isolators. They are great for breaking ground loops and reducing CPU noise from a computer when making sensitive low temperature measurements. AFAIK the GPIB-120A has been discontinued by NI. We attempted to use the new GPIB-120B but it is a complete piece of crud and was discusses as a complete fail on these boards awhile ago. I got my hands on the last 5 GPIB-120A models that NI had. One of my units has a problem and we shipped it back to NI for repair. They offered a refurb or ***new***(???) unit of the same model.
Is there a replacement for the GPIB 120A that does not inject an unbeliveable amount of noise into a measurement system? Has NI recognized this failure and/or started up the production of the 120A again? Any alternative hardware to achieve ground isolation and noise reduction without adding high frequency noise to the system.
See http://forums.ni.com/t5/Instrument-Control-GPIB-Serial/GPIB-120B-causes-noises/td-p/878437 if there are any questions about how useless the GPiB-120B. I can also give a short rundown on the poor design choices with the GPIB-120B.
02-17-2015 11:51 PM
02-18-2015 07:22 AM
Albert, yes, it has been awhile since we addressed this issue! Only came up since NI emailed my technician and offered to replace a repaired unit with a "new" or refurbished unit.
The GPIB-120B has a power dongle that is like a cell phone charger. These chop the incomming high voltage AC at high frequency without a transformer to get the DC low voltage. This introduces a big signal at around a 100 kHz. It makes the charger lighter but cell phone chargers are very bad in a measurement lab. This was just a bad design choice.
Second, the unit itself is plastic and had no shielding. The two isolated sides should have independent copper shields around them to prevent crosstalk as much as possible. The entire unit should have a ground plane around it tied to the A side shield.
The power cord does not have a ground. There is no ground reference to the plastic case thus leaving all nosise to radiate wildly.
I don't know about the internal circuit layout but given the attention to the power handling I assume there was no attempt to minimize noise or crosstalk between the A&B sides. This project had bad design goals and bad testing.
It is much cheaper to manufacture but it doesn't accomplish the task. I got NI to refund my money for buying three GPIB-120B units!
02-18-2015 10:33 AM
02-18-2015 07:49 PM
Well, false alarm. NI has not restarted making the GPIB-120A, it was just an uninformed person in the repair section. Today NI clarified the status of the repair.
"ask if we could purchase another. The only ones they have are loaners and can not be purchased. They do not sell refurbished instruments and the 120A is no longer available new. The upgrade is the 120B"
Oh well, the 120B is really a stain on the reputation of NI.
02-19-2015 07:31 AM
02-19-2015 10:37 AM
That is a whole different discussion. USB is a bit tricky to isolate as well without causing unreliable connections and bizarre latencies. Ethernet is inherently isolated but has odd latencies. I give a whole lecture on different instrumentation buses and the pros and cons. GPIB has not been superseded in quality yet!