Signal conditioning for pH & ORP probes

[David MC Lee]

Can anyone point me to a site or explain how to design the input signal conditioning for pH & ORP probes. What do the probes really output? Are they all the same?
Thanks

[bigcat39]

Heh, Finally something on this board I know the answer to! pH probes output a millivolt signal, but it's not that simple..... the output of these probes is governed by the Nernst formulas, which describes mathematically the output of the probes @ pH & temperature. Basically, they output 0 mV at a pH of 7, about .4 volts @ pH 1, and -.4 V @ pH 14. BUT! The output is highly dependent on temperature, and also on the probe in question... that's why pH meters are calibrated very regularly with buffer solutions. ORP sensors work pretty much the same, just use a different formula.
Hmmmm, you could design a pH circuit I suppose, but why buy a cow when milk's so cheap? Look on ebay in Buisness & Industrial, pH meters with automatic temperature compensation and a signal conditioned mV output can be had for, oh, $30.
HTH,

Bill
BTW, I'm a metrologist and Instrument/Controls engineer by trade...

[David MC Lee]

Thanks for the info. I did find a schematic for the pH probe at
www.national.com under the LMC6001 op amp data sheet.

I’ve just been kicking around the idea of building my own controller for my tank. Yea if I can get a box that will just output mV directly from the probes that would be great. I’m just not satisfied with the features on the consumer grade controllers that are available. I really want to be able to write my own software for the controller.

[BeanAnimal]

Okay lets take this a step further, or simpler maybe.

I am going to use my PC to control various aspects of my tank. I would also like to buy probes and interface them with my PC.

If what you are saying registered correctly with me, I could take the direct ouput of the probes, plugged into a ADC, and use software to adjust and normalize the data, then do something useful with it. That would mean writing the code to "fix" the output (which others have already done numerous times?)

*** OR ***

I can buy a "black box" that outputs an already normalized signal, and use that to drive the ADC, allowing the software to just do something "useful" with the data?

Thanks in advance,
Bill

[bigcat39]

Beananimal,
Oh sure! I'd be happy to send you the formula, in fact I am thinking of doing just that with Softwire and a measurement computing USB box. Here check this out... http://www.measurementcomputing.com/usb.html
Softwire can be a real PITA, it's a Visualbasic frontend. I like National Insrument's Labview much better,

[tschopp]

David,

[tschopp]

David,
I am replying to this because of your email. I have looked into signal conditioning for ph/ orp / echem cell probes. The biggest chalange is the output impedance of the probe. My understanding is that they can be as high as 100M Ohm. I think it gets worse over time. If we want to be able to measure accurate to the mV level this puts a limit on the leakage current of 1pA. 10^-3 V / 10^8 Ohm = 10^-11 A, then divide by 10 so its not a factor. This is a non-trivial leakage, but it can be done. The pinpoint monitors get around the problem by lifting the ADC pin on the microcontroller off the board into the air and wiring the probe directly to it. This eliminates leakage on the printed circuit board. The leakage is then determined by the input bias current of the cmos input. This will be OK if the temperature is stable, cmos input bias increases rapidly with temperature. I would say the leakage current problem is not the biggest problem with the pinpoints. I have noticed a large change in the reading due to changes in the input voltage, i.e. poor PSRR. I do not mean to critisice the pinpoints, I think for the money they are great and I have seen much more expensive "scientific" equipment designed much worse. To measure the signal correctly you need to use a printed circuit board with a guard trace around the high impedence line. A guard trace is simply a trace that surrounds the trace it is protecting. The guard trace has its voltage set the same as the high impedence trace. This reduces the electric field and the high impedence trace has no reason to leak current any more. The signal then needs to go into an amp that is designed for low leakage applications. The AD549 from analog devices pops to mind. The AD549J (cheapest version $12) list an input bias of 250 fA at room temp with an offset voltage of 1mV. You could use other amps, check input bias and offset voltage. Set the amp up in a unity gain configuration. The signal is now low output impedance and can be measured by an ADC. use a good voltage reference and ADC if you want accurate results. I like the AD7708 for a 16bit ADC or the AD7718 for a 24 bit ADC (you get 18 bits of accuracy out of the AD7718). I like references from Thaler. They make the best references I have seen and the cost is often better than lower quality from other places. The ADC connects to the microcontroller with a SPI bus. With the chips listed the accuracy would be limited to 1mV by the offset voltage of the OP-amp.

[BeanAnimal]

Question come to mind then.

Probes:

What is the cheapest probe available for our use:

I know that I can buy a "Kit" that I can calibrate and "dunk" into the tank. However, I want something I can setup and forget.

Is this possible? Place the probe in the sump and hook it up to the computer. How long can it go wihout being re-calibrated, damged or worn out?

In your opinions will it be cheaper to
A) build a drive circuit AND ADC and Controler to interpret the output

B) build a drive circuit AND ADC using PC software to interpret the output.

C) Buy a controller and use the supplied DLLs or ActiveX to funnel the results into the desired application (home grown VB package)

Bill

[BeanAnimal]

Originally posted by bigcat39
Softwire can be a real PITA, it's a Visualbasic frontend. I like National Insrument's Labview much better,


I am a VB programmer, so this is most likely e benefit for me. I see that the dev tools also have components useable in C++ .net and delphi.

I am very interested in something I can incorperate into MY CODE, not a standalone unit and interface. To me that is no better than the handeld with a digital readout, or test strips and inicator solutions.

I simply want to use the data in my software (for simple monitoring at first).

Eventually I would like to use the realtime readings to operate PH and other buffering or adjusting components of the system.

Bill

[barebottoms]

From my bookmarks:

http://www.maxim-ic.com/appnotes.cfm/appnote_number/34/ln/en

http://www.weissresearch.com/

[bigcat39]

Originally posted by BeanAnimal
I am a VB programmer, so this is most likely e benefit for me. I see that the dev tools also have components useable in C++ .net and delphi.
Ahhh, sorry didn't know your expertise level. The components that come with Computerboards stuff are very nice.
I am very interested in something I can incorperate into MY CODE, not a standalone unit and interface. To me that is no better than the handeld with a digital readout, or test strips and inicator solutions.
COLOR=blue]Ahhhhh, but most of these instruments come with a mV output, linear and proportional.[/COLOR]
I simply want to use the data in my software (for simple monitoring at first).

Eventually I would like to use the realtime readings to operate PH and other buffering or adjusting components of the system.

Bill
[

[capescuba]

Just tagging along also - I'm very interested in doing some of my own PC automation too. Maybe we can pool some guys together and share source? I know VB, VB.NET, ASP.NET and c# .. I already have the bases for some X10 automation using c# and the firecracker RF transmitter, was just looking for the monitoring side of the equation ...

[BeanAnimal]

I have an interface alsmost done in VB6. The light and pump timers and wavemakers are pretty much done. The moon phase calculator works and I am working on the tons of other crap that I palnned for it. I even have planned x10 compatible output, though I am not a fan of X10 stuff.

I may just scrap it and move on to .net ... I am not sure yet. .net seems to have problems with a lot of older activex controls so I may be shooting myself in the foot.

In any case I need to figure out if it is even possible to dunk the probes in the sump and leave them there indefinatly.

I would like to monitor pH ORP Nitrate and Amonia. I need to find advice on which probes are suitable, acceptable or ideal.

I have found many USB based IO controllers, and own 2 96 bit ISA based IO cards. I just don't know a damned thing about probes!

Bill

[capescuba]

Bill - do you already have an ADC?? I have just done some looking around am waiting for a MAX187 to come in, then I'll look at figuring out what I can do with it.

If you know VB (I was a VB developer for ~10 years) then you will LOVE .Net. COM interroperability is pretty good, I don't know of any problems. And communications to ports/networks etc. is all native so you don't need ActiveX/COM controls etc to help ..

I am also planning to make this a web app so I can monitor remotely, something that is a no brainer in .Net.

Dave

[BeanAnimal]

I have purchased a few ADC chips and but can not construct the circuits (of which I am just learning) until I understand the voltages that will be input to them.

My inputs so far will be the water quality probes, several temperature probes, and simple logic from switches and similar 2 state devices.

The temp sensors are serial...not ideal but it will work. So this leaves the probes for ADC.

WE should definatly try and work together on this one!

Bill

[David MC Lee]

Thanks for all the info.

Here is the inside of my pinpoint meter. Same PCB for orp and ph. I like the idea in the maxim data sheet of putting the buffer in the probe. Does anyone know of anyone how is really doing this in their probes?

[David MC Lee]

One more pic of the inside.

[tschopp]

David, I do not know about the maxim chip, but it seems unlikely that it would become mainstream. National is trying to do a similar thing with a pre-amp for microphones. Microphones already use a FET preamp and have bias voltages in the "standard" wiring, but the national thing is not catching on even though it would be a significant improvment and requires no changes to the rest of the micrphone citcuits. The maxim idea assumes the rest of the circuits would be different, or batteries would be put in the probe. I have used microphone cable on high impedance circuits like this with good results. I have also used teflon wire on other circuits (for high temp stuff). I do not think teflon is really needed for pA stuff. If we were talking 10fA then teflon and air wires would be a must. I looked at the national circuit for the ph probe. I think compensation / calibration is typically better done in the digital domain. I would recomend digitizing the raw siganal and appling a digital calibration and temp compensation. On one of my sensors I had to use a compensation that used the time derivative of temperature. I had no clue ahead of time that would be required and would not want to think about how to do it in analog. You could get a good temp compensation by calibrating with standard solutions at different temps. The thermistor in the national circuit would just be an approximation of the real temp compensation. Also pots tend to be poor electically and can introduce many problems. I try very hard to avoid them, and my boss gets mad if I use them. I have used a couple of digital pots recently, they are better than mechanical, but still have some capacitive problems. David, do you do pcb layout or circuit design? Or do you mainly do the programming? I will have to look at my pinpoint again, it has been a few years since I opened it, but I do not remember it looking as nice as yours (FR4, soldermask, etc).