It has been a while since the last blog entry, but it has not been wasted time. Much of it was spent researching the type of data collection I wanted to attempt, and the remainder of that was gathering the necessary components, much of them from eBay, although I was forced to buy a few items new. One of the things I learned was that with data collection, it is not the analog to digital conversion that is difficult or expensive; rather, it is the art known as "Signal Conditioning" that was the greatest challenge.

Signal conditioning (to repeat a bit from the previous blog) is the act of taking what can be an incredibly minute signal, isolating it, amplifying it to a more reasonable value, and then feeding that signal into the ADC (Analog to Digital Converter). It, in turn, feeds the now amplified data, which has been digitized, into the micro notebook PC that I will be using.

My ADC from Dataq is the DI-158U. The DI-158 is a four channel differential analog signal device, meaning it can accept signals below 0 volts. It is reasonably robust, capable of +/- 64V per channel. In retrospect, one of the other units may have been a better choice, such as the DI-148, which has 8 single-ended channels, measuring from 0 to some positive voltage, since all of my signals of interest are DC and one-rail. But the DI-158 has some excellent, and very flexible, software that allows me to truly manipulate and display the data for best analysis.

As mentioned, I want to measure (per)chlorate cell voltage (0 to 10 VDC), Cell Current (via shunt or hall effect transducer), typically 0 to 100 millivolts, and temperature, using a PT-100 RTD, a resistance device capable of excellent accuracy.

Each of these channels needs massaging and amplification using some method of signal conditioning. My original plan was to make amplifiers for those channels that need them, and hooking things like cell voltage directly to the Dataq unit. Then I found out about isolation, specifically the very real need for it. Isolation simply means the signal is electrically isolated from the Dataq ADC. Analog to digital converters are fragile units. In a complex system, stray voltages can be generated that can wreak havok on the ADC, releasing the magic smoke contained inside that makes it work.

I won't go into HOW isolation is done, as it is not important to understand the theory behind it, but generally they'll use either a transformer, or some sort of optical setup. I found that Analog Devices makes a series of modules they call "5B" that are potted units that both isolate and amplify, and are perfect for data acquisition tasks. But at $150 to $200 each, they were out of range, price-wise. But eBay once again played an important role. I found several 5B modules that would do what I wanted, and slowly gathered them up at good prices.

My original plan was to mount them on perf board or otherwise permanently solder the pins, but I found that backplanes are available for 5B modules, ranging from single prototype boards like the AC1360, simpler boards that will hold one or two of the modules (5B03 and 5B04), and all the way up to 8 and 16 channel backplane boards. I ended up buying new 2 ea. 5B03 and one 5B04, and they are NOT cheap. Then I found a guy on eBay selling an 8 channel board, NIB. Sigh!

With the boards on hand, I needed to figure out a way to mount them, and power them as well, with correct fusing. I probably could have gotten away with a 5VDC wall transformer, but instead modified a really nice little 5V power supply:

On the back, poking out, you can see a fuse holder, with a second one not visible. The hidden one on the left has a 1 amp slo-blow fuse for the supply, while the output I rewired for a 500 mA quick-blow fuse for the 5B bank.

The next step was to mount the boards. I went the overkill route and used two pieces of aluminum angle, slotted lightly for the board edges, and mounted on a heavy piece of Al plate, which acts as a common power ground for the system.

In the second picture, you can see a fuse that has since been remounted on the power supply, as I mentioned. Each module requires 5 volts, so I daisy-chained the 5V output through each board, and in addition, each board is grounded to the Al base. Also, in the second picture, you can see that the right-hand Al angle has a piece of phenolic board glued to it, as there were PCB traces very close to that edge, and mounting directly into the Al angle would have shorted some of them out.

I tested first the voltage unit, which accepts +/- 10VDC, and outputs +/- 5VDC; essentially just a voltage follower that divides the voltage in half. It worked perfectly. On the Asus PC, I could see the output voltage directly, set high and low limits for the display, and otherwise make the presentation meaningful. Resolution is excellent, down to the 100th of a volt.

The next installed and tested 5B module was the PT100 RTD (Temperature) unit in the very first picture. It too performed perfectly, and I calibrated it on the low side with ice water. For the high-side, I immersed both the RTD and a good lab thermometer in a beaker of near-boiling water. The Dataq displays the voltage generated; I enter 85 degrees, the temp. of the water, and the system now has two data points to work with and create an accurate linear response.

The next was the trickiest - current. There are two ways to measure high current. The traditional way is with a shunt, seen on the far right of this board:

The current-carrying line is simply connected in series with the shunt, and the shunt itself is tapped for a voltage measurement. As the current travels through a known, very low resistance, it generates a voltage drop across it. This voltage drop is measured, and is linear with the amount of current traveling though the shunt. Shunts themselves are carefully-made and calibrated devices. This one in particular delivers 100 mV at 60 amps.

The mV signal is amplified by a 5B module into a 0 to 5V signal, and that in turn is fed into the Dataq unit. It worked. But I also wanted to try the previously mentioned hall effect transducer, so I made this board of PVC plastic, left over from the T-Cell, to do both.

The transducer has its own small power supply (8VDC) mounted on the perf board, and outputs, at zero current, exactly 1/2 of Vcc, meaning with no current flowing, the signal is 4.00VDC. As current flows, the signal is linearly increased at some modest value per amp, and at the rated 72 amps, it should deliver approx 7V. The 3 volt spread is more than adequate for the Dataq. Tested, it too performed perfectly, and by its very nature is isolated, meaning I don't need to use a 5B module. But since I have one, and it works fine, I will probably just use the 5B module rather than the hall transducer. It was a cool experiment, regardless.

The overall system, as it stands - calibrated, and ready to go to work:

The wiring is twisted pair PTFE-coated aircraft wire; good stuff. So far, tests have shown excellent performance and little noise or interference, even when using a switching supply. The leads to the cell are to the left, while the right side of the bank has connections for 5B power, and for the 5B outputs, fed to the Dataq ADC.

When I get a bit more used to the software, I will try and show some cool traces. I also need to determine what sort of sample rate is adequate for runs that can last for days. Disk space and file size may be an issue. One final benefit - the PC display itself becomes "cell central." At a glance, I can see the condition of the cell. The fourth channel is still open, and I am hoping for pH there, if the probe can survive the immersion.

I can use this setup both for oxidizer production, and for lead dioxide plating. I'm sure too that there will be future uses for it in other chemistry processes. I realize it is not everyone's cup of tea, but I have hopes for it in the sense that the data it gathers can be analyzed and used to determine end-of-run conditions based upon how the voltage and current interact. Given that, one wouldn't need a similar rig; simply keep an eye on V and I and from there, know when to pull the plug on a process. Finally, it can be turned ito a crude controller by tapping a byte of digital output. By using MOSFETS and a simple optoisolator, I can turn on and control just about anything from the PC. I am hoping the software can be set to recognize certain conditions, like excess temperture, high pH, etc, and respond with digital outputs.

After this, I plan on setting up a strain transducer on my hydraulic press for rocketry. Then it's back to making Pyro stuff rather than electronics!