Data Acquisition, Part I
Entry posted by Swede
496 views
First, a postscript from the previous blog entry, where I attempted to clean perc with Sodium Bisulfite, a reducing agent... following some suggestions from Tentacles and others, I proceeded a bit more methodically and was rewarded with success - perc that tests as clean as my commercial stuff! It simply required an acidic environment to liberate the SO3 gas, which dissolved immediately into solution, and reduced the stray chlorate. Future cleanups will use potassium metabisulfite so as to avoid annoying potential contamination with Sodium.
I am really going nuts in the electrochemistry department. I've got multiple cells for both chlorate and perchlorate, supporting hardware such as power supplies, and further, I've easily doubled the MMO mesh on hand with some judicious eBay purchases. The eBay MMO is especially thick when compared to the previous, commercial-grade MMO, and makes chlorate like crazy. The only hassle I've noticed with it is that the eBay MMO must be scraped off the spot-weld points, down to bare Ti, or the weld quality sucks or is non-existent. No big deal... I just need to avoid the resultant Ruthenium dust! MMmmm, Ruthenium!
Quickie plug for my Agora ad. I've really got a lot of this stuff, so if you want to make chlorate or mess with lead dioxide plating, give me a shout.
When I first started my basic experiments, I was religiously logging temperature, voltage, and current by hand, with pencil and paper. At the end of each run, I attempted to do a bit of graphing, but I was lacking all of the data points generated between bedtime and sunrise, and as the experiments ground on, I found myself doing less and less logging, out of pure laziness.
I am convinced that valuable information on the state of a (per)chlorate cell can be obtained with voltage and current graphs; enough so, that I decided some time ago to automate and permanently log data acquisition. If I can find identifiable points relating to power that are distinct and reproducible, I might be able to deduce chloride, chlorate, and perchlorate concentrations based solely on the data, rather than tedious or expensive titrations. A bit of research revealed that dataloggers range from thumb-drive sized temperature loggers and similar, and from there, they then take a giant leap into the stratosphere, cost-wise, as you examine true lab-capable equipment. A bit bummed, I kept plugging away until I found Dataq. Take a look... these guys not only have inexpensive hardware, they also have powerful software that is well-written, and makes using their hardware a breeze. At least, that is my initial impression! 
Hopefully there will be no changes to that favorable outlook.
I ordered one of their higher-end units, the DI-158U. Two reasons: First, the unit can measure more than 10 volts, and it is conceivable that the cell voltage in a perchlorate cell can exceed that number. But the primary reason was the extensive programmable gain available only in that particular model, available in binary increments from 1 to 512. With a 12 bit resolution, this means that the unit can possibly detect and display voltages in the microvolt range.
I installed the software on my tiny Asus Eee notebook with no problems. The DI-158 uses USB for both data and power. I have only scratched the surface with the software, but it appears exceptionally powerful for the price (free!). Everything, and I mean everything, is scalable, and custom "labels" can be affixed to the axes. The software samples at a rate of less than 1 Hertz up to 14,400. For this process (electrochemistry) anything faster than 1 Hertz is overkill, and will consume disk space quickly, I'm sure.
Just to be sure everything worked, I attached my homebrew thermocouple amplifier to the unit, powered everything up, and began to play a bit with the thermocouple and the software.
This next picture is representative of what you see on one of the channels, channel 1. I have temprarily lablede it "Voltage" because in the final setup, channel 1 will, in fact, be cell voltage.
The blue line at the bottom is the signal, and the displayed 0.00934V is noise. No doubt if this was a $1800 unit instead of a $199 job, there'd probably be less noise, but any system is prone to stray voltages on this scale. Simply moving a copper wire through the air generates microvolts. Ultimately you filter it, bias it, or ignore it. The latter is probably easiest, given a signal that is normally (or amplified with a signal conditioner) several volts in strength.
A vertical bar moves from left to right, as in an oscilloscope sweep, leaving behind it the reading on all of the active channels.
I powered up the thermocouple amplifier, and was immediately rewarded with a jump in the signal, indicating it was detecting ambient temperature, and ouputting approx. 200 mv in a 20C room, as the amp is set up to deliver 10 mV per degree C. Time for the big test! Since a type K can handle serious heat, I applied a heat shrink gun to the tip, and was rewarded with a perfect trace on the PC screen. It worked!
Note that I changed the Y-axis to Temp as it should be.
With the basic system ready, and having browsed eBay thoroughly for any bargains, I began to accumulate transducers and integrated signal conditioners at fire-sale prices. It turns out that Analog Devices makes dedicated signal conditioners, and they are fairly common on eBay. Quick summary: A signal confitioner is a device that takes a "signal", usually voltage, from a device or transducer of interest. For example, a simple thermocouple, due to the Seebeck effect, produces a voltage all by itself as a result of two dissimilar metals that are welded together at the tip, but the voltage is miniscule, and not linear. A signal conditioner takes this microvolt signal, rejects RF noise and hash, and then amplifies and linearizes the signal, outputting 0 to +5VDC, rather than 0 to 0.003V over the entire range of the thermocouple. Importatly, it isolates the fragile data acquisition device from the nasty, outside world, where human error can do stupid things, like put 120VAC into the input, rather than 0.001V DC.
Here is one of Analog Devices RTD modules:
The pinouts for these modules are all the same, making a multi-channel device, fully signal conditioned, a reality rather than a fantasy, given that these can be obtained at 10% to 20% original cost.
The final setup will use this RTD module, a voltage module, and for current, one last surprise gadget that I did not even know existed until a few days ago...
A DC current trnsducer! It uses the Hall Effect to detect the current, and ultimately, potted electronics inside of it amplifies yet another very weak signal and makes it usable by us. By its very nature, it is already isolated.
I have some instrument amplifiers, and also some isolation IC's, but ultimately, I am going to build this rig out of surplus signal conditioners, and fall back on the scratch-building concept only if the conditioners cannot do the job.
Why go to all the trouble? I can't think of many chemical processes where you are not tracking something, and I expect the Dataq unit will receive plenty of use. The DI-158 also has digital and analog outputs, making it possible to automate tasks that were previously done by dedicated controllers. I am not 100% positive, but I am pretty sure it can be set up to act as a controller. Let's say your pH probe detects alkaline conditions, and you want your project acidic... a digital output can be automatically turned on, which in turn, activates a solenoid valve, admitting acid into the experiment. Besides, it is both cool and fun to work with such hardware, and the satisfaction (for me, at least) is very high.
Part II will consist of further integration and testing.
2 Comments
Recommended Comments