Data Acquisition Systems

[PedroDaGr8]

It's time to get some advice from you guys. I have been tasked by my boss to plan out some new data acquisistion hardware. For most instruments in the lab I know what to look for and what will fit my needs. On the other hand, I know many of you have been involved in data acquisition so here we go.:

Data being acquired:
1) Primary signal;  "dumb" 280nm UV absorbance detector. By dumb I mean it outputs a simple voltage via a two-wire differential pair. Full range: 0-1V but most of the time we set the range on the chart recorder to 100mV, 200mV or rarely 500mV range; in order to see more fine detail on the paper.
2) Signal from a fraction collector that shorts, what I think is a floating input, to ground for 100ms each time it moves to a new tube.

These signals CAN be combined (that's how we do it now) but it would be nice to record them separately and overlay them. As sometimes when the signal is at or near zero it can be very difficult (or impossible) to tell what the tubes are. This isn't critical because we don't collect at zero and we typially mark a tube point during collection and count from there.

The UV signal is slow (recording is typically over 30min-2h depending on the procedure). On the other hand, the pulse from the fraction collector is very short.

Current Device (and why it works but sucks):
The current device is a Kipp&Zonen BD11E. We have 1 at each station with 2-5 stations per bench.

Benefits:
It works and makes hard copy of the absorption profile that we want.
Can write on it, for tube counting.
It's easy to use for everyone (including non-tech-inclined people).
With one at each station we can use them for small column rapid collections.

Downsides:
Changing ranges mid-run makes an UGLY spectrum with a notch in it (very minor problem)
Hard copy only means no online storage or retrieval.
HORRIBLY GODAWFULLLY POS UNRELIABLE!!!!!!!!!!!!!!!!!!!!!!!!!  They break, the motor burns out, etc. Sometimes even when you barely use them.

This last point is a HUGE problem. We have FIVE out for repair at about $1k in repairs per device.

In the past we had been content and not upgraded because there was no financial justification for scrapping working systems and upgrading to digital. Basically, if it ain't broke then don't fix it. Well its broke and broke bad.

Goal and Budget:
An easy to use system which can record these charts in digital format (for printing and storage) or both formats. At a reasonable cost:P if done individually $<2000 per unit. If I can incorporate multiple stations into one (ideal) I can go above $2500. I just have to justify why its better. Each bench has a computer and between 2-5 stations at it as I mentioned before.


Basically an all digital or hybrid system. Right now, I am leaning towards an all digital system because the hybrid systems seem unreasonably costly for potentially no improvement in reliability. That being said, the software that the end-user uses has to be pants-on-head retarded simple to use. Either out of the box or something that I can design a simple interface myself. Additionally, if the input modules have a degree of splash resistance that would be a DEFINITE bonus that I could sell to my boss as these have lots of buffers (basically water with lots of salt) around. While we don't in general have an issue with liquids, better safe than sorry.

As for the data, absolute accuracy and precision are ENTIRELY unimportant. We are doing entirely qualitative work here (basically establishing where the product is based upon visual inspection). Resolution is much more important, the ability to resolve the peaks is a must. Though the peaks slowly change over time and we need to see the peak shape. Whether the peak is 1.134mV or 102.345mV is completely irrelevant. Hopefully this makes sense.

I have looked at NI stuff (gotten a bit confused on what we need). The NI stuff seems like I might be able to make it work but I am worried about cost and in particular all-at-once cost. For example, it's easier to replace a bench at a time than it is to replace them all. The cost of LabView license etc. might kill thigns as well.

Similarly I looked at Yokogawa and a few others. With this stuff it seems like either its prohibitively small, slow or costly.


Any advice and recommendations you could give is great.

[Rerouter]

while i am not sure about the totalising counter, there is an easy way to record your UV, make a scaled voltage to current amplifier, most of the cheaper digital recorders will happily accept 4-20mA in atleast 1024 steps, this would also make your system completely future-proof as they will always be around, (and are stupidly cheap), only issue would be aggreing on an open data format so that you can go and data mine it,

[PedroDaGr8]

while i am not sure about the totalising counter, there is an easy way to record your UV, make a scaled voltage to current amplifier, most of the cheaper digital recorders will happily accept 4-20mA in atleast 1024 steps, this would also make your system completely future-proof as they will always be around, (and are stupidly cheap), only issue would be aggreing on an open data format so that you can go and data mine it,

Unfortunately, I can't build anything. The hardware has to be off the shelf. In case something breaks and I'm no longer employed there.

By the way an example image of a spectrum.



You can see my numbering of the tubes.

Sent from my EVO using Tapatalk

[Rerouter]

same difference,..

something like this http://www.acromag.com/catalog/872
a millivolt to current converter, (sometimes called transmitters in the trade)

the reason why i am trying to keep it current based is that millivolt recorders cost more and are rarer,

still i am not sure how you could go about including a totalizing counter to this,

[jeremy]

Operating system preferences?

NI DAQ sounds perfect for what you want. I personally don't like labview, but I know many who are less code oriented and really like it. The PCI cards are fairly straight forward to program in C and various .NET languages too and give you lots of IO.

From the sounds of it your signal is actually quite slow; 100ms is 10Hz? Even the cheapest option from NI does 10kHz.

You could probably even get away using the cheapie NI USB DAQs for $200ish with a couple of differential input channels on a 14 bit ADC.

[PedroDaGr8]

Just an update. We moved forward with LabView. Purcahsing a single bay USB CompactDAQ and one module. After a bunch of learning, banging my head and work (as well as some great help from the labview forums) I was able to get a program to do EXACTLY what I wanted. Thanks guys for steering me in the right direction.