bizarre temp controller behavior

I've been working on a large C-41 film developing machine that has vats of chemicals which must be held at exact temperates of 100F.
Each tank had 100 ohm platinum junction probes and the controllers are the digiwheel style Omron E5CS-somethingsomething units.
Most have failed and other methods are used to maintain chemistry temperature, but the critical one for developer is acting strange.
We got a new probe for it, a two lead Pt probe and wired it up into the cable harness going back to the controller of the machine using the original three leads.
When exposed to air the proble and Omron unit work fine. If you warm it with your hand, it works fine. If you dip it in a jug of any liquid it measures the temp corrected.
If the probe goes into any part of the developer, the temp reads as out of range (39.9C or around 106F).
Ok, maybe it's electrical leakage. Putting the probe in an electrically insulated sheath doesn't work either. Temps go out of limits or read really high.
Ok, maybe the probe isn't grounded, or there's a weird electrical leakage issue, so I ground the probe to the original thermistor shield. No dice, temps our of range.
I even attached a lead to the metal probe and dumped it into the developer tank to see if it's electrical noise or leakage. No problems with temp readings in air or liquid if the probe housing is wired to the developer tank.
It just never works when immersed, which is baffling.
I tried the other "spare" controllers and they all seem to behave the same way, or are just dead. Tried other new probes, they all behave the same way too. They all measure 107ish ohms at room temp and the controllers are the correct ones for platinum probes.
The only and next move is just replace the temp controller. The omron stuff has crappy docs (good luck finding old data sheets) and is overly complex, so I'm steering towards something from Panasonic. No crazy hysteresis loops or fuzzy logic are needed. If the heater has to cycle every 10 seconds, that's fine as the racks going in an out of the tanks can shock the temperatures, and fast recovery is needed.
The machine itself has no docs, is no longer made and there are no original parts available. Having custom parts made for these things is becoming the norm, and pretty expensive. The original probes were sealed in plastic tubes, but the reason why isn't known. There are no spares of even duds left to take apart. I tried replacing the cable of about 10 feet between the controller and probe, but it behaved about the same way but still registered a temp too high.
Has anybody come across something this strange, or know what may be causing it?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

The heating element might emit a lot of IR in some bands that are not strongly absorbed by the solution. If the tank walls are nice and shiny it might bounce around until absorbed by the probe. Take a sample out of the vat and immediately measure it. Wrap the probe with insulator and then aluminum foil and see if the reading is more accurate. The tank walls might even focus the IR at some particular point.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

The temperature of developer if immediately removed reads correctly. The tank itself is some sort of grey plastic. There's a compartment off to the side where the probe usually sits, which is connected to a filter and recirculator pump which is always on. The heater is one of those flow through units attached with hoses to the main tank, pump and well which has the filter cartridge.
The probe reads high temps anywhere inside the main tank or the small one off to the side, so it doesn't seem like it's getting blasted with the output of the heater, I'm not dunking my arm in there to feel where the currents are though.
I'll try the insulator/foil experiment. The probe itself is some sort of stainless, I have no idea how it reacts to IR. The plastic sheath I tried before was just a piece of black PVC with one end heat sealed.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload


Then again, it might be a no-no to put alumium in a tank of reactive chemicals like developer.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 15 Apr 2013 20:02:01 +0000 (UTC), Cydrome Leader

Have you simply considered putting in a balancing resistor/pot?
Sometimes screwing the unscrutable is a pain in the ass..so find a work around to make it work....is much simpler
Gunner

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 15 Apr 2013 20:02:01 +0000 (UTC), the renowned Cydrome Leader

107 ohms is about 65F on a Pt100 alpha = 0.00385 RTD. A bit cool for room temperature, but okay..
There are other types of RTDs.. but the old "US Standard" alpha 0.00392 is not very much different from the DIN standard at 40C (less than 1 degree C). The old J Pt100 Japanese standard was similar to the obsolete US standard. Everyone uses the Euro standard now, pretty much.     
What you are saying does not make a lot of sense. What is the temperature of the developer? 100F is 37.8C, so very close to what you say is the upper limit of the controller (but usually the standard range is to 49.9C).
Is this thing sitting all apart? Are the sensor leads grounded at any point? (Unplug the temperature controller, remove the sensor from the bath and measure the leads to earth with an ohmmeter - power off).
It is possible they have grounded one of the sensor leads to reduce bobble in the controller- the E5CS is/was an early switchmode supply controller. Usually Pt100 probes are floating wrt to the shell.. is the shell floating on the new probes? Any vigorous EMI sources kicking around? Is the developer tank earthed?
Does it make a difference if the controller is calling for heat or not?
One possiblity is that there is not much wrong, but the setup is not properly wired so the controllers appear to be malfunctioning. Maybe a heater is leaking AC current into the tank. If you take an AC voltmeter and hold one lead in your hand and touch the other in the tank do you read any voltage?
If you replace, be sure to use a decent auto-tune PID controller- AFAIR, the C41 process is pretty fussy .. IIRC (frm a long time ago) the spec is +/-0.1F, which is not easy.
Best regards, Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
snipped-for-privacy@interlog.com Info for manufacturers: http://www.trexon.com
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

The installed omron controllers max out at 39.9C, which really doesn't give much wiggle room. The operating procedure for the machine is ignore what the controllers say, but trim the reset pot on the front panel and fiddle with the numbers until the the temp measured in the tank with a normal thermometer is correct.

The machine is in use, so and can only be played with at slow times.

The original probes would have been what was cool in West Germany the mid to late 1980s. It's not clear if the Omron controllers are for the international Pt100 or Japanese version. They are all fairly old so it's not out of the question they're all bad.
I've not tested if the new probes are floating or attached to the red or white lead. Flipping polarity doesn't seem to have any effect. I've not examined how and if the tank is directly grounded. There is a chain driven mechanism that raises and lowers rack of film into the tanks. The machine is mostly single phase 208. Motors and pumps are run with contactors and heaters and fans with "advanced" solid state relays.
The highest tech thing in those machines are the omron controllers. There's no know EMI sources, just other similar machines.

It does not. None of this machine makes sense. The rest are straight forward- replace/rebuild the bad part and they work again.

Will have to try that. Leakage makes sense as a cause, but again, a wire wrapped around the probe's metal sheath dunked into the developer tank doesn't cause weird readings of air temp, hand temp or even temp of a container of developer. Just thought about it now, but the test of is this an electrolytic reaction problem is bridge the container of developer with probe to the main tank with a wire. Maybe the wire itself isn't reactive.
The cable running from the operator panel to the back of the machine is three conductor, white, brown and green with a braided shield. I believe the shield is chassis grounded, but I'd have to look at it again. The last working but not OEM probe was somehow lost while trying to source replacements, so it's not clear how or what was even connected internally.

Somehow the size of the machine and all the adjacent tanks and the constant pumping somehow make it possible.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Cydrome Leader wrote:

I haven't read all the posts, but beyond leakage from the heating elements, is it possible that the original probes were insulated could relate to their housing acting as an electrode in a chemical battery with the developer solution? Your test with a dunked wire might not have had enough surface area or been the correct metal to reproduce the effect?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

That might be. It doesn't explain how the probe in a plastic sheath also read high temps, but we'll see.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I'd check simple things first. Like a thermal open circuit in the new RTD sensor. I've had that problem. Use an ordinary ohmmeter to read the RTD resistance as it warms up in a bowl of hot water from the tap.
Joe Gwinn
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 15 Apr 2013 19:04:54 -0400, Spehro Pefhany

If the spec is +/- 0.1F, then a 2-wire RTD ain't going to get it. 4-wire preferred, and that's pushing it, but I seem to recall the Omrons taking 3. I always specified 4-wire and put two leads from one end under the + terminal, the other two under the measurement and - terminals. I also always used ungrounded sheathed elements. Don't recall ever seeing a grounded RTD, but you could get t/c's that way. A/C heater leakage played havoc with grounded t/c's.
Pete Keillor
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 16 Apr 2013 09:23:06 -0500, Pete Keillor

He's got a 3-wire cable going to two-wire very near the end. Probably okay provided the end is reasonably well controlled.
Perhaps worth stating that 3-wire RTD compensation requires that the three wires (well, usually it's only two of the three, but you may not know which ones) be as close to identical as possible, since one wire is used to compensate for the resistance of another.
Four wire is theoretically independent of that kind of matching requirement, but nobody outside of laboratories uses it much.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

f
e
f
r
e
e
y
t
Having worked with photochemicals, the developer is VERY reactive(or it wouldn't work). You no doubt have an electrochemical reaction going on between contents, sensor and some fastener or plumbing somewhere in the tank. If I were speccing sensors for such an item, they would be glass-encased if immersed. It wouldn't have to be much of reaction to get a few millivolts out on the leads, enough to mess up any temperature signal. It wouldn't take a lot to seal the end of a suitable glass tube, drop the temp sensor in that and see if the problem goes away when dunked in the tank. It would take a pretty sensitive VOM to measure the voltage at that level, they DO exist, but usually aren't a shelf item at Lowe's. We generally used a mirror galvanometer and a Wheatstone bridge when messing with such things in the lab.
Stan
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
In article

The sensor appears to be a 100-ohm RTD, not a thermocouple, so millivolts should not have that big an effect. And the RTD is probably encased in but galvanically isolated from a stainless-steel probe.
I'm betting on a bad RTD.
Joe Gwinn
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

ts of

e the

he

it

t
ut of

ly

kage

ce,

loper

p
per

same

me

s are

ly

e
ks

s
led

of

feet

ut

Have you checked that the housing IS isolated? I've been surprised more than one time that way.
Stan
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I'll be back tomorrow and an Omron rep is dropping by with some controllers to play with. Panasonic still hasn't contacted me back, so they lose, which is too bad, I liked the timers and stuff back when it was Aromat.
As for the probes, I took one for testing. I can't get it to read wrong from 50 to 120F. It jives with my analog TelTru thermometer, which is actually more accurate than the thermocouple that came with the last Fluke meter. The fluke has more digits, but they're all wrong.
With a standard meter, there's no connection between the stainless sheath and the leads. I don't know what the internal construction is, so I'm not going to try any 500 volt hipot-like tests, and I doubt it's rated to that anyways.
milivolts should not have an effect on a RTD, but how electrically isolated the current temp controllers are is unknown, and again, I can't do a 500 volt hipot test on them, and I can't locate data on what the isolation ratings are to start with.
So the plan is down to
- try the probe I tested - try a probe inside something glass. There's never glass tubing around when I actually need it for some reason. - try a new controller from the sales guy
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.