02-15-2013 12:12 PM
Hi,
I have issue that a continuous current offset of 8uA is at each of the inputs of my USB-4431. The IEPE excitation is turned off but the current is still there. The offset current creates a maximum voltage of 2.27V when the input is in open cricuit mode. Although 8 uA doesn't sounds much, it does mess up with all my results.
Am I the only one that is experiencing that issue? I have that issue since the first day I got the device, and I just been looking around trying to find what is the issue ever since.
02-15-2013 12:48 PM
The specifications for the device indicate a typical input offset voltage of 750 uV when driven by a 1 ohm source impedance. That could result in as much as 750 uA of input current. The manual also mentions that wwhen DC coupled an open input will float to 2 to 2.5 V. The input circuit diagrams do not show any explicit sources, but what you are seeing appears to me to be within specifications.
What kind of problems is this causing for your measurements? What is your signal source?
Lynn
02-15-2013 02:25 PM
Thank you for your replie,
The attribution of the offset voltage to the offset current seems a bit deceiving espacially when the test assumes a impedance of less than one omhs (which normally will be done to represent the offset volatge excluding the offset current i.e. shorted input offset voltage). The 2.5 V is for open circuit, but even with a 10komhs (1/20 of the input of 200komh) I can read of offset of about 120mV (this is from the offset current only, I'm measuring with a picoscope the voltage input).
As for the issue it can cause, for one, that current alone is a source of rms noise. For second, it does affect other stage of my system. What is the point of having an input impedance of 200komhs with DC coupling at 24 bit if you have an offset current capable of offsetting an inpout source as low 1 ohm?
I'm taking measurement at various stage points of a anlog filter system. Even with the buffers it does still affect the signal.
Ultimately the datasheet are vague enough to allow just about anything, but I would think that there is some commun sense.
Ben
02-15-2013 03:04 PM
I agree that the datasheet is quite vague on the DC input behavior. My suspicion is that this device was designed for applications, particularly the IEPE enabled ones, where a little DC wandering around the input was insignificant. I wonder if the IEPE current sources are disabled but not disconnected and have microampere-level leakage?
You may need to put low output impedance buffers between your measurement points and the inputs.
Lynn
05-11-2013 09:32 PM
Hi all,
I have a shared frustration with Benure and would like to urge NI to consider a re-design of this product. Not only does the high input bias current put a large offset on the input, but the input impedence is only 200KOhm. That makes this box almost useless for measuring passive circuits or to use it as a scope with probes. It would have been trivial for NI to add appropriate buffer amplifiers to the circuit. Neither of these issues is listed in the 'spec sheet' provided on the web page for the product. Like Benure, I assumed that the 4431 would meet generally accepted standards for test measurement equipment. In my opinion it does not. Now I must spend signficant time and money building a preamplifier.
Very dissapointed.
05-12-2013 09:33 AM
05-12-2013 11:24 AM
Hi Benure,
I purchased the 4431 to use as a spectrum analyzer, which is its advertised purpose. I've used analyzers from HP, SRS, etc. for years and was looking for a compact alternative. These boxes always have input impedances ranging from 10MOhm to many GOhm. And essentially no input bias current. The 4431 may well be in NI specs, but if so there is a real lack of understanding of test equipment at NI and how it is used. I was also looking for something to use as a portable scope, and the 200K input impedance really makes that impractical.
Your comment seems to suggest that you can't have an ADC input impedance higher than 1 Ohm or you become dominated by the noise of the ADC. Or did I miss-read that? In any case, that's not true. A FET op amp as an input buffer would only have ~10nV/rtHz of noise but an input impedance of ~10^13 Ohms. And a bias current of ~1pA. With this input buffer, the noise would be limited by the *source* impedence if it is above about 6KOhm (thermal noise is 130nV/rtHz*sqrt[source impedance in MOhms]). But even that's not really relevant since with 24 bits the LSB is still only a ~microVolt, so you are still limited by digitization.
I've worked in precision electronics for decades and I think NI just totally missed the boat on this one. The input stage on the 4431 is just a very poor design and I was also expecting more for a 2+K$ price tag. So I'm building a preamp. Bummer.
SG
05-12-2013 10:12 PM
Hi SG,
Yeah, I just wanted to spare you time in arguing with their "Tech help". Mainly, they would just try to prove you that that device is within the specs while you try to explain that the given specs are just non-sense, they are defined within absurde measurment parameters.The example of the 1-ohm input is that I was telling them that the noise level I'm measuring is higher than the specs and then they tell me that because the input isn't shorten, the noise can come from anywhere. Summary is that I totally understand your frustration, it's baffelling that NI, a producer of instrumentation(software and hardware) used by other manufacturers to do their data sheets, did a very poor job in defining their product and also did a poor product. I work in the precision too and I'm fully aware of the wide variety of amps and switches that would have corrected the problem at minimum cost. I really don't undestand why/how their design are like that.