Thermocouple rectifier?

[TConnolly]

In this particular app the customer has a heated/cooled mold with a TC probe in the mold. Electrical contact is made with probe pads when the mold is clamped. Sometimes the operator can put the probe in upside down, and the part will be scrap. The customer needs a way to rectify the voltage from the TC so that it does not matter which way the probe is placed and he would prefer to not have to take time to align the positon of the probe and mold. He wants a "slap it in and push go" solution for max throughput.

I considered bringing the mV signal into the PLC and doing the polarity adjustments in software and linearizing it myself, but I'm looking for a better way.

Rectifiers usually cause a slight voltage drop and I think would be unsuitable for use with thermocouples, or am I up in the night? Has anyone heard of such a device? Any ideas?

[jimbo3123]

In this particular app the customer has a heated/cooled mold with a TC probe in the mold. Electrical contact is made with probe pads when the mold is clamped. Sometimes the operator can put the probe in upside down, and the part will be scrap. The customer needs a way to rectify the voltage from the TC so that it does not matter which way the probe is placed and he would prefer to not have to take time to align the positon of the probe and mold. He wants a "slap it in and push go" solution for max throughput.

I considered bringing the mV signal into the PLC and doing the polarity adjustments in software and linearizing it myself, but I'm looking for a better way.

Rectifiers usually cause a slight voltage drop and I think would be unsuitable for use with thermocouples, or am I up in the night? Has anyone heard of such a device? Any ideas?

Thermocouples rely on having special wiring so that there aren't other bi-metalic junctions to throw-off the reading. (That's why thermocouple connectors and wires have a specific polarity/alignment and can't be connected backwards) If the sensor is connected backward, you will have a negative voltage that is ALSO affected by this.

I'm afraid that this problem can't be corrected at the panel. You need to ensure that the operator aligns the mold so that the device is connected properly. I would look into modifying the mold to make it asymetrical (only able to go one way) if possible, and clearly labled if that isn't an option.

[CroCop]

Can you go to a non contact thermocouple?
Is it the + - leads that are getting flipped, or is it the probe being put in completely upside down?

[rdrast]

Maybe you should switch to RTD's.

[CroCop]

Maybe you should switch to RTD's.

Not with the temps he runs.

[randylud]

Thermocouple rectifier? Sorry, no such animal. Rectification is the process of changing AC to DC. The milivolt signal from a TC is already DC and therefore can not be rectified, no need. Perhaps you are thinking of some other polarity sensing situation, but rectifier is the totally wrong idea.

[TConnolly]

I guess I should have been more specific. A bridge rectifier is what I was thinking of. A bridge rectifier can not only used to change AC to DC, but it can also assure that the polarity of a DC signal is always the same. Its done all of the time. The problem I see here is the voltage drop across the rectifier, not that its already a DC signal. Thanks for the input though.

Jimbo,
This is an S type TC probe (expen$ive) so the extension leads are both copper (platinum's seebeck coefficient is 0) so it shouldn't matter which way its connected except at the instrument itself. Thats one reason a S type probe was selected. I really wish it was as simple as chaning the molds, but in this app that will cost several hundred thousand dollars, and that is out of the question.

Ron's suggestion for an RTD is very clever since RTDs don't care about polarity and it wouldn't matter which pad the third lead was on, or better yet, 4 leads could be used. Unfortunately I don't know of any RTD that has a wide enough temperature range.


:banghead: :banghead: :banghead: :banghead: :banghead:

[randylud]

I figured you meant something else, that is why I posted my reply. The diodes in the bridge or any other for that matter will require about .6 volts to put them into forward biased condition. The TC will not produce that kind of voltage and as you said, the drop would kill your accuracy.

[danw]

I'm still confusted about the real problem, but it seems the T/C gets installed somehow or oterh so that when it is connected it's reverse polarity on the input.

I have connected a single T/C to more than one input dozens of times, sometimes on two different devices (controller and high limit).

Years ago there could be problems with cold junction compensation and T/C break detection. Improvements in analog input technology have made those issures largely disappear.

Why not take whatever leads you have coming out and conect two sets of extension leads to those two T/C wires.

Run one set of extension leads to input A and wire it for standard polarity.

Run the other set of extension leads to input B and wire it for reverse polarity.

One input will read OK, the other won't. Pick the right one.

All it costs you is a econd AI input.

Dan

[Terry Woods]

One method that I can think of is to replace the T/C with a Thermistor-type detector, and of course, the appropriate controller.

Of course... that might not get much money for you.

Alternatively... you could talk them into modifying the mold and the T/C so that they have a keyway which would insure correct installation when assembled... thus forcing correct installation.

That won't get much money for you either. That might be a deal-killer... however, a little bit of clever thinking goes a long way! How about simply inserting a pin into the T/C connector and a matching hole in the mold? Do they have a macine shop?

Another option would be to modify the T/C connector in some way so that there is an additional set of contacts. That additional set of contacts, on the Mold side, would be connected to a diode (internal in the connector). The diode would be oriented to indicate proper installation. An external source would be applied, through a load, to those additional contacts. The source could be a simple 5V-DC. That circuit could then be run through a 5V PLC Input... or, through an opto-isolator which controls a 24V-DC source, or a 120V-AC source, to a PLC Input.

If the PLC "sees the light" from the opto-isolator... then the PLC sends a signal to the temperature controller indicating that the probe is correctly oriented. Until the PLC "sees the light"... the controller is prevented from operating.

You could make a few bucks on that one... however... modifying the Mold and/or the T/C might be a deal-killer. Although... if thought through... it could be a minor modification.

I've got yet another idea... one that doesn't involve any modification to the T/C or the mold... but I'll have to do a bit of thinkin' (MGD) to realize it!

Hint: Milli-Voltage Controlled Opto-Isolator (as opposed to current controlled). This would function, or not function, in a polarity sensitive manner.

[rsdoran]

I believe Alaric is actually thinking of a Wheatstone bridge system like a Thermistor would use. A thermistor nor RTD will usually go to the 2500 deg F range which is what I assume Alaric needs.

I am cofused how a probe could be put in upside down, the connector should only be able to connect one way using proper polarity, at least those I am familiar with.

Is there a special reason the probe has to be put in a certain way physically??

May just want to go to www.omega.com (http://www.omega.com) and use their configurator to see if you can find a different type probe that will work in the application.

If all of what I said is bullcrap then it may be easier to take it to the plc. Using a signal conditioner to change the signal to 4-20ma may offer a cheaper option than purchasing a T/C card.

[geniusintraining]

You may want to contact these guys

http://www.statesvilleprocess.com/profile.htm

I have all of my RTD's PT100, T/C's and other stuff made by them, they are up by Randy in NC, They may be able to come up with a 'well style' that may be robust enough for your applacation, their prices are good also.

[TConnolly]

I am cofused how a probe could be put in upside down, the connector should only be able to connect one way using proper polarity, at least those I am familiar with.

This is a custom TC with contact pads rather than connector pins. When the mold is clamped contact is made with the pads. Normally + is to the right, - is to the left. If its in upside down, then the pads will be reversed, + on left, - on right.

[keithkyll]

Dan's suggestion of 2 Conditioner Modules is probably the cheapest way out of this. Parallel the 0-10 volt outputs with two Schottky diodes (.2 volt drop). The highest voltage output will prevail. Compensate for the .2V drop with an offset in the PLC. Keep the diodes at a stable temp to avoid drift.

1) A full-wave bridge with zero voltage drop is done with 4 op-amps. This would get too complicated unless you can find a commercial unit. I don't know a source.

2) Detect a low temp. If low, latch in a polarity reversing relay or a shutdown alarm.

3) Read temp. Fire heater for a few seconds. Read temp again. It will be lower if T/C is reversed. Energize reversing relay or alarm.

[randylud]

Dan's suggestion of 2 Conditioner Modules is probably the cheapest way out of this. Parallel the 0-10 volt outputs with two Schottky diodes (.2 volt drop). The highest voltage output will prevail. Compensate for the .2V drop with an offset in the PLC. Keep the diodes at a stable temp to avoid drift.

1) A full-wave bridge with zero voltage drop is done with 4 op-amps. This would get too complicated unless you can find a commercial unit. I don't know a source.

2) Detect a low temp. If low, latch in a polarity reversing relay or a shutdown alarm.

3) Read temp. Fire heater for a few seconds. Read temp again. It will be lower if T/C is reversed. Energize reversing relay or alarm.

Just a point here about voltage drops. The .2 volt drop mentioned above is 200mV. Depending on the TC, the available voltage from a typical "J" type is in the range of 70mV maximum at the highest temperature it can measure. That is far less than the .2V needed to forward bias a diode. The problem here is not going to be solved by using diodes. Check out this link on TC output voltages. http://www.omega.com/temperature/Z/pdf/z203.pdf

[rdrast]

Aside from 'Detect the error' and signal an alarm, I'd say the only option here seems to be have the machine and the thermocouple's redesigned so that they are keyed.

A short-term solution might be to BOLDLY color code the connector and it's recepticle, but maybe your problem with with the people installing them... No help for people being stupid or malicious.

[danw]

Detect and fix.

Apparently, when the mold is put in backwards, the pads don't match up, so the T/C has reversed polarity to the indicator/controller.

T/Cs that are wired reverse polarity show the temperature going in the wrong direction. When heated, the temperature goes down, when cooled, the temperature goes up.

It is not uncommon to not be able to detect reverse polarity at ambient temperature, because at ambient, both ends of the T/C are close in temperature to one another, the difference is negligible so the T/C doesn't have a substantial difference between hot end and cold end, it doesn't generate much signal, so the huge majority of the indicator's reading is the cold junction value. I suspect there are too many operating conditions to account for to be able to now whether the mold is always cooler or hotter than the cold junction temp.

But, as soon as the mold starts to heat, the temp will indicate in the wrong direction, if it has reverse polarity.

I like Keith's idea of a reversing relay feeding the AI on the PLC.

You test when the heat cycle starts. Are successive values each lower or higher than the previous? (time the test, so once you've established correct polarity, you don't lose it)

If successive values are lower, energize the reversing relay, if higher, leave the reversing relay coil as is.

Temp values for the duration of the test would be missing, if reversed. Would that matter? Are they looking for a max temp or for a profile, start to finish?

Dan

[rdrast]

I like Keith's idea of a reversing relay feeding the AI on the PLC.

You test when the heat cycle starts. Are successive values each lower or higher than the previous? (time the test, so once you've established correct polarity, you don't lose it)

If successive values are lower, energize the reversing relay, if higher, leave the reversing relay coil as is.

Dan

That is a possible solution, BUT quite possibly not practical. A Thermocouple depends on the junction of dissimilar metals. That is why it is important to use TC extension wire if needed to connect to the actual TC Transmitter/Analog input. Even the 'Universal Terminal Blocks' used for extending TC leads are designed so that the leads themselves must be in physical contact, and a compression screw or clamp merely forces contact between the actual wires. Any break in the signal path, that does not use the proper cabling will create additional junctions of dissimilar metals, each of which can (and probably will) severly degrade the signal accuracy.

If the 'Detect' method shows an error, ideally, the process should be stopped, the actual problem corrected, and the process restarted.

In the case of connecting a junction backwards, it isn't just that the polarity is reversed, but now you are adding two additional junctions, which are in series with the desired junction. Typically this results not only in a 'reverse temperature coefficient', but also a strong non-linearity.

[rsdoran]

Everyone onct in a while my ignorance shines through and I sound nasty when I do not mean too.

None of this makes sense in many ways.

First you have an expensive special configure S type T/C that appears to be removed on a regular basis.

Second it appears that this expensive special configured S type T/C can be inserted 180 degrees out of position which in turn reverses the polarity.

To me this appears to be a bad engineering physical design. I can not see the application therefore I may be wrong. I have not seen it mentioned that the Operator has a way to know the T/C can be upside down; which is what I am basing this reply on.

I would think the simplest thing to do, since the T/C is special made, is have it made with an alignment notch or something similar. This could be as simple as a small line on the probe or connector, any kind of marking. If its a handle probe then make the handle with a marked point to indicate the correct position.

I may be wrong but it does not appear the Operator(s) are being given the ability to do their job correctly. In other words this is not a situation where you are foolproofing it against Operator error, it was designed where the Operator could make errors 50% of the time.

If I am way off base, my apologizes, but I doubt I am far off.

[danw]

In the case of connecting a junction backwards, it isn't just that the polarity is reversed, but now you are adding two additional junctions, which are in series with the desired junction. Typically this results not only in a 'reverse temperature coefficient', but also a strong non-linearity.

The application uses an S type (Platinum Rhodium) thermocouple with which they've used copper extension wire. I assumed from one of the comments that the pads referred to are not the T/C elements or extensions of the T/C elements themselves, rather, that the pads are the extension wire pads, with the T/C being buried somewhere in the mold, connected to the pads with copper extension wire.

If the pads are T/C elements, and not copper extension wire, rdrast is correct. Attempting to correct for reverse polarity will result in a bogus signal that cannot be corrected.

Dissimilar metals used in junction blocks (like nickel plated screws in ordinary junction blocks) do create an addtional junction, but the assumption used with junction blocks is that the junction block remains isothermal, same temperature at both connections, in which case the error on one leg cancels the error on the other (one is plus, the other minus). If the connections are not isothermal and at different temperatures then you've got error.

If the pads are copper extension wire pads, then the mold is a junction block, with whatever error is present in non-isothermal conditions. But they have lived with that as part of the design. And as RsDoran points out, it is a somewhat somewhat lacking design?

Dan

[Terry Woods]

+====================+ +====================+
| | | |
+--( + )------( - )--+ +--( - )------( + )--+
| | | |
+====================+ +====================+


Ron, what is the difference between the two figures above?

If not for the "signs", the two figures are identical! And yet, in seeing the signs we can see that one figure is upside-down relative to the other. Without the signs there is no way to know if they are identical or if one is inverted.

+====================+ +====================+
| | | |
+--( )------( )--+ +--( )------( )--+
| | | |
+====================+ +====================+

.
.
If not "pinned", then, at the very least, apply color to the connector body!

The lower side should be RED, and the upper side GREEN!

GREEN is GO!

RED IS STOP!

+====================+ +====================+
| | | |
+--( )------( )--+ +--( )------( )--+
| | | |
+====================+ +====================+

.
OMG!!! I've turned this into a Traffic Light Problem!!!

[CroCop]

OMG!!! I've turned this into a Traffic Light Problem!!!


Bwahahahaha!!!!
I think the reason Alaric is being asked to do this is he regularly makes miracles happen. The solution is to put the mold in right, (duh) but it's far easier for a mid level manager to point to the automation guru & say fix it............

[rsdoran]

By now everyone is aware of my background so I will use it as an example. The picture included is of an electrical box similar to what has been used for years. The whole purpose of these boxes was to eliminate connection issues. If a ride got improperly conneted you could burn out lights, ballasts, and single phase the motor.

Note the round colored connectors. Black, red, and blue are "hot" phases. White neutral and Green ground. Makes it easy for the electrican to make the connection, the problem was that the machine (ride) operator also had to connect his end which sometimes was not always marked properly, for whatever reason, and was connected improperly.

The whole point of my last post was that the design appeared to allow the machine operator, or anyone, to make an improper connection. I understand what thermocouples are, what they do, but not to the depth of others here so cannot offer anything concerning the T/C.

It just appears that the issue is try to eliminate the possibility of an improper connection.

http://www.plctalk.net/qanda/uploads/gflex-1.jpg

[Terry Woods]

Ron...

The impression I get is that the "connection" is made by the mold-clamping device... NOT the operator! The contacts on the T/C connector are oriented as the operator (or whomever...) layed it in while assembling the mold.

The contacts on the clamp-side are already wired into the controller... I don't think the operator is making any electrical connections.

When the clamp closes on the mold, the contacts on the clamp-side simply come in contact with the contacts on the T/C in the mold. And, since the connector on the T/C is symmetrical... and I would expect that it is visually the same in either orientation... it can be installed upside-down.

Apparently they have a reason for installing the T/C while the mold is assembled rather than simply inserting the hard-wired probe into the mold just prior to clamping. Maybe the T/C is anything but straight... maybe it turns a few corners as it goes through the mold...

[TConnolly]

That right Terry. The connection is made by the clamp, but the operator assembles the mold. CroCop and Ron are also right, its a situation where the customer wants the controls guru to fix it rather than fix his operator/assembly problems. Modifying the molds will be a very expensive propositon for the customer (hundreds of thousands of dollars) so I can understand not wanting to do that and perhaps only 1 in 20 times, maybe less, is the mold assembled incorrectly, mostly by novice operators. So the customer could train the operators better and he could do something about turnover, but instead he wants magic. You know the saying, "any sufficiently advanced technology is indistinguishable from magic." Well to too many managers, we automation folks are "magic."

The customer has a history with Dan's two channel suggestion on an older system, but was getting some kind of strange interference that disappeared when going back to only one channel - so there is some customer resistance to that - despite having a more sophisticated PLC with better inputs now. Again, its that "magic" factor that is probably coming into play - this time in regards to a problem that was not understood before a solution was found. I'm leaning in the direction of a two channel system, it seems to be the simplest. By saying that however I don't want to stiffle ideas, there have been some good suggestions so far.

[Terry Woods]

An FET Op-Amp has a very high input impedance and can operate at very, very low voltages (millivolts).

So... how about making a connection between the Positive leg of the T/C and the Positive input of the FET. Then use the amplified output to drive an opto isolator. The PLC reads the opto-isolator. If the pretty light goes ON then the connection is correct. If not... then perform a lead-switch via relay... it must be a very, very good relay... gold contacts?

The normally closed contacts should be used when the connector is properly oriented... this assumes that they are at least trying to install the connector correctly. But then... maybe they install the connector wrong more often than not...

Some adjustments might need to be made to accomodate the T/C going through the relay contacts.

Take a look at this...
http://www.aldinc.com/pdf/AN17.pdf