Help sought: Panel temp sensor dead in 13-year old flat panel water heater system.

[bcroe]

Corrected LM135 CKT

This one might work better. Bruce Roe

[bcroe]

fix

Of course, I reversed the + and - pins on the OP AMP, so change 2 & 3 before you build it.
Bruce Roe

[MikeSolar]

The earlier comment about us guys just giving up and buying new controller is, in this case, not a bad idea. I love fixing things but sometime the extra work just isn't worth it. Besides, with a new Resol controller you can get a lot of new features.

[bcroe]

Ckt

According to my calculations, this attachment ought to let LM135s do your job.
Bruce Roe

[windbag]

But I've learned a lot & getting my first quote on Monday

At least I now know what most systems in the UK use, what the transducers are and what they cost. I know what to ask anyone proposing a new control system and I've got some idea of how much controllers cost (plus some of the features that might be of use).

So at least my knowledge is further on, and next time there's a problem, I'll have a much better idea of what I'll be facing.

Better : I found a UK site UK which can take details of what solar panel (both main types) work you want, plus your location, then farms the "job" to local solar specialists with the aim of getting three interested ones to contact you for a quotation/site visit.

The first one has already contacted me, we'll see how that pans out on Monday.

[windbag]

A simple OP AMP and several resistors can convert the LM135 output to any other linear slope, even reversed if needed.

You're right of course, but as I pointed out my partner after making the measurements, we've been working with a system that not only is outside what has become a de-facto standard, it's doesn't even fit that that sub-standard either.

To use an analogy which shows my age:
I could get another betamax tape to keep my video player going for another 10 years, then I'd be happy to.
But if I've got to buy a open reel of 2" pro video tape and THEN have to trim its width down to 4cm to use it in my machine, perversity is taking over.

[bcroe]

Well, I spent ages trying to extract the transducer from the fritzed top tank sensor.
No use. when it got fritzed, it got well & truly melted into position - so much so that it glued to the tube better than itself, crumbled into bits - no m'facturer lettering remained.

Seeing your reply Bruce, and the datasheets, I calibrated the response of my remaining sensor properly.
It gave 2.198 V swing between ice/water & boiling water.
So not an LM135. Or even a degree per Fahrenheit equivalent!

There's the rub. There are NO calibration parts (eg trimpots) on the control board so it must be achieved by fixed components.
So unless anyone knows of an equivalent to the LM135 that has 22 mV/degreeC, this controller is no longer viable.

What is worse is that the panel sensor is somewhere in the body of the thing, so the entire top will have to be removed to replace it.
I'm NOT going clambering on my roof, so it looks like I'll have to get in a pro - which goes against the grain

A simple OP AMP and several resistors can convert the LM135 output to any other
linear slope, even reversed if needed. I'd use a rail-to rail (LM6482?) to keep it
simple. Sure noise has to be removed, but with high impedance and extremely
low frequency operation, its easy to filter out. Leave the other working sensor as
is. Don't try to remove the bad, just attach another. I'll try to attach a picture of
my hot water circulator control, which uses LM135s and gives a Deg F display.
Bruce Roe

[russ]

**** runs downhill and sometimes it seems like one is stuck standing at the bottom of that hill.

[windbag]

I wish...

Well, I spent ages trying to extract the transducer from the fritzed top tank sensor.
No use. when it got fritzed, it got well & truly melted into position - so much so that it glued to the tube better than itself, crumbled into bits - no m'facturer lettering remained.

The LM135 can solve your problem;

Ah, if life turned out to be that easy...
Seeing your reply Bruce, and the datasheets, I calibrated the response of my remaining sensor properly.
It gave 2.198 V swing between ice/water & boiling water.
So not an LM135. Or even a degree per Fahrenheit equivalent!
Rats.

Your controller might need a calibration to match. Bruce Roe

There's the rub. There are NO calibration parts (eg trimpots) on the control board so it must be achieved by fixed components.
So unless anyone knows of an equivalent to the LM135 that has 22 mV/degreeC, this controller is no longer viable.

So I guess I'm cream-crackered.

What is worse is that the panel sensor is somewhere in the body of the thing, so the entire top will have to be removed to replace it.
I'm NOT going clambering on my roof, so it looks like I'll have to get in a pro - which goes against the grain

[bcroe]

The LM135 can solve your problem; I have been using them forever.
The output voltage is quite linear, referenced to absolute zero. No need
to calibrate them, simple .01V /deg K. Your controller might need a
calibration to match. Bruce Roe

[windbag]

True, and some devices allows you to alter the program to accept different curves for the thermisters or TCs but not typically simple solar controllers. We use a lot of DDC controls and dataloggers where we have to calibrate device to the curve of the sensor for proper operation.

The bottom line here is that unless the OP can find the same sensor (assuming there are some markings on it) and there is nothing wrong in the control (such as damage from lightning), he is better off picking up a RESOL or similar control.

I 98% sure the controller is AOK. I "just" need replacement sensors.

I almost wish they WERE a thermocouples. There are only a few types, and monitoring microvolts Vs temp would tell me which.
On the other hand, I'm glad they aren't as thermocouples can have a real problem with long connections, and you need proper compensating cables to ensure temperaure variations at connections along the path don't mess up the results. Plus the issue of very small signals being susceptible to RF etc. I used to work with thermocouples in large scale field trials, so I have experience here.

But as I said, they are NOT thermocouples, AFAI can tell electrically.

I have the dead sensor in my hands - you guessed it, no markings - so I'm going to have to try and get the transducer out of the tube.

Wish me luck!

[MikeSolar]

Calibrating a TC is nothing more than comparing it to a known temperature.

There is nothing to calibrate - only two wires and a cold junction.

True, and some devices allows you to alter the program to accept different curves for the thermisters or TCs but not typically simple solar controllers. We use a lot of DDC controls and dataloggers where we have to calibrate device to the curve of the sensor for proper operation.

The bottom line here is that unless the OP can find the same sensor (assuming there are some markings on it) and there is nothing wrong in the control (such as damage from lightning), he is better off picking up a RESOL or similar control.

[russ]

Calibrating a TC is nothing more than comparing it to a known temperature.

There is nothing to calibrate - only two wires and a cold junction.

[MikeSolar]

[QUOTE=windbag;46553]Gosh, folk here are really quick to throw up their hands and "buy a new one". Not my approach!



I don't understand. All the thermocouples I have ever used (including the one I have to hand for a multimeter) have two wires and require no power - but do need very careful connection methodologies, which is why I'd have thought they wouldn't be used in a domestic fitted unit.

Testing the thermocouple I have, it measures ~ 15 Ω whichever way I connect across it - so the behaviour doesn't correspond in any way to working sensor I have.

It DOES correspond with a thermistor/resister in series with a centre-tap, but I can't see why you'd put power and the balancing resistor in the sensor like this, so I think it is unlikely that it really is that either.

I'm hoping someone on this forum is familiar with the precision temperature sensors such as the LM135 which is available as a three-wire TO92 package which could easily be mounted in the head I've described, and which I've certainly seen cited as used in domestic solar installations. Thomson and National Semiconductor (& probably many others) make these, for example.

From the Thomson data sheet:

"The LM135, LM235, LM335 are precision temperature sensors which can be easily calibrated. They operate as a 2-terminal Zener and the breakdown voltage is directly proportional to the absolute temperature at 10mV/

[windbag]

Not a thermocouple - but might be a Zener with temp-variable breakdown voltage

... think you need a new control. Sorry

Gosh, folk here are really quick to throw up their hands and "buy a new one". Not my approach!

Right so you have a control that uses thermocouples.

I don't understand. All the thermocouples I have ever used (including the one I have to hand for a multimeter) have two wires and require no power - but do need very careful connection methodologies, which is why I'd have thought they wouldn't be used in a domestic fitted unit.

Testing the thermocouple I have, it measures ~ 15 Ω whichever way I connect across it - so the behaviour doesn't correspond in any way to working sensor I have.

It DOES correspond with a thermistor/resister in series with a centre-tap, but I can't see why you'd put power and the balancing resistor in the sensor like this, so I think it is unlikely that it really is that either.

I'm hoping someone on this forum is familiar with the precision temperature sensors such as the LM135 which is available as a three-wire TO92 package which could easily be mounted in the head I've described, and which I've certainly seen cited as used in domestic solar installations. Thomson and National Semiconductor (& probably many others) make these, for example.

From the Thomson data sheet:

"The LM135, LM235, LM335 are precision temperature sensors which can be easily calibrated. They operate as a 2-terminal Zener and the breakdown voltage is directly proportional to the absolute temperature at 10mV/