Data acquisition and logging

[Swede]

Thank you, Frank! Yesterday I breadboarded a differential amplifier to measure voltages across current shunts, and it worked perfectly... a little trimmer pot sets the gain, so it'll amplify 0 to 100 mV --> 0 to 5V. So that's another option. As I mentioned to Tentacles in a PM, I don't know why I'm getting into this particular project so hard, other than that it's fun, and something new... turning a cheap notebook PC into an effective chart recorder.

[Swede]

Quick update on the project... three of the four signal conditioners I picked up off eBay work and are rock solid, signal vs. noise. The fourth was shorted internally, but fortunately the guy is going to exchange it. I refer to these 5B modules:

 

http://www.5bears.com/chem1/data010.jpg

 

The RTD module is a better choice, accuracy-wise, for most chemistry, rather than the thermocouple device. So I now have multiple options for the channels. It turned out that there was no way to wire these modules without a true socketed backplane, so I was forced to order those new. I need to rig up some sort of physical support structure - there's not enough room on the PCB's to use simple spacers.

 

The power supply is going to be either simple TO-220 devices, or a surplus rig.

 

All in all, it has been more complicated (hardware-wise) than I thought it would be. There are always questions in my mind about appropriate grounding. The ADC (The Dataq device) has four differential channels, meaning the unit measures the difference between the two leads. Each channel has a "+" and "-", and further, there is an "Analog Ground" terminal that is not well documented. Dataq's support forum has not been much help. I just hope I don't let the magic smoke out of this thing. I don't know if I should let the signal conditioners' power supply float vs. the Dataq ADC, or tie the power supply's GND to the ADC's Analog Ground.

[Richtee]

I don't know if I should let the signal conditioners' power supply float vs. the Dataq ADC, or tie the power supply's GND to the ADC's Analog Ground.

In instrumentation it is usually set up so there is a "signal" ground and a "system" ground. If you have the option, try the grounds seperate. This is usually a noise reduction thing. I am assuming your power supply input to the units in question have a plus and minus and a "ground" terminal. THis ground is for the internal functioning of the unit, not necc. for it's inputs. Try an ohmmeter check from the power ground to the signal ground. Also, in instrumentation apps, the SIGNAL ground, or cable shield, is usually grounded at the transducer end if at all.

[Swede]

In instrumentation it is usually set up so there is a "signal" ground and a "system" ground. If you have the option, try the grounds seperate. This is usually a noise reduction thing. I am assuming your power supply input to the units in question have a plus and minus and a "ground" terminal. THis ground is for the internal functioning of the unit, not necc. for it's inputs. Try an ohmmeter check from the power ground to the signal ground. Also, in instrumentation apps, the SIGNAL ground, or cable shield, is usually grounded at the transducer end if at all.

Thanks, Richtee, this is helpful. I'm going to do another blog soon, but what I have so far is this: I've got my raw "signals", 3 or 4 in number. These go to a bank of 5B signal conditioning modules. The 5B modules have completely isolated inputs in the form of screw terminal blocks, each with a + and a - , NEITHER of them continuous with the 5B's power supply common.

 

The entire 5B bank is powered with a single beefy 5V supply. On the output side of the bank, where the amplified signals are tapped for sampling, there are 4 terminals, +5V, PWR Ground, Signal Ground, and Signal + . The signal ground and PWR ground ARE tied together, via a cuttable jumper. The Dataq data acquisition unit is self powered by the USB bus. My plan right now is to let everything float, as the Dataq inputs are differential inputs, simply measuring the voltage difference between the two wires. All of the 5B outputs are 0 to 10V, and the Dataq can handle +/- 64V.

 

Here is the homemade 5B setup holding commercial boards:

http://www.5bears.com/chem1/5bt01.jpg

 

And on the output side; you can see the jumper hiding behind the terminals. The right two terminals are power, the left two are output. I grounded the entire aluminum structure to power common:

http://www.5bears.com/chem1/5bt02.jpg

 

That damned jumper... I'm not sure if I should cut it or not. Here's what the data sheet says.

 

Jumper W1 connects the 5V power common to input/output common (backplane measurement ground). A connection between power common and i/o is important for the 5B series modules to function properly; however, if this connection is made elsewhere in your system (the best place is usually near the A/D converter), the W1 should be cut since a ground loop could result.

I've got some excellent two-conductor cable with a braided shield. What I could do is cut the jumper, and use the shield to re-connect PWR gnd and I/O gnd at the Dataq "-" input, but this seems fussy.

 

Your thoughts are appreciated... I thought I knew what I was doing, but this is slightly over my head. So far, though, as each module is tested individually, they seem to be working well. The suite of signals will be a PT100 RTD for temp, 0 to 100 mV for current via shunt, and 0 to 10V for the electrodes. pH later.

[Richtee]

That damned jumper... I'm not sure if I should cut it or not. Here's what the data sheet says.

 

Well, best I can offer: Leave it for now. When you get multiple inputs/outputs set up and you notice any "flakiness" in the circuitry or responses, make that your first tweak.

 

Grounding is a black art, I swear... Key thing to keep in mind is keep the resistance <Both DC AND AC...> between all common connected grounds...be they signal or power or whatever..to an absolute minimum, thereby avoiding the dreaded "ground loop".