Handling High Number Of Temp Readings

[ZedIDave 0]

We have a machine in the field where the customer now wants to install a thermocouple on every ceramic heater. The oven currently consists of 40 Zones. They want to install a thermocouple on the 300 heaters that are not sensed by the controller for temperature feedback to a display. They want to know when an individual heater has failed. The thermcouples are Type K. I have looked but cannot find a suitable, cost effective solution and there may not be one. Does anyone have any suggestions on implementing this that they have done? Thanks

[TimWilborne 108]

There are many ways to do this...depends on what your calling economical... Any particular brand components you want to use? How spread out are the thermocouples? Do you want to connected it to the existing controls or will they be completely new?

[ZedIDave 0]

The heaters are spread out across an area 20ft and 3ft. The added thermocouples will be added to a stand-alone system. They want a monitor to view the temperatures of the elements. We are not inclined to use any particlar controls except for AB PLC's. The heater is currently being controlled using RKC temperature controller modules that communicate back to an HMI. We have thought about using a plc with thermocouple cards connected to a monitor. This would at least allow us to get 8 pts per card but if we use a SLC, that would need 3 racks. (They currently have AB in house and prefer AB). Noise can become an issue too. Ceramic heaters are 460V with the thermocouple built-in. What I would prefer to do is monitor the current on the heater zones. If the current for a particular zone drops, you know you have a defective heater. The defective heater will be isolated to a particular zone. The operator then looks at that zone (heaters change color when heating) to find the defective heater. But, this is what the customer wants. Thanks

[TimWilborne 108]

I would go with a distributed I/O such as Flex I/O and put it on Devicenet. You could get 8 modules per node, 8 points per module. So you could space 5 panels down through it. Then you could connect them to a SLC through a SDN module if that's your preference of PLC...or do you need a PLC? Or do you even need a PLC. Could you connect the Flex I/O modules through Devicenet or Ethernet directly to a display since your just monitoring?

[kaiser_will 65]

The customer wants to measure ~ 300 Type K thermocouples and you would prefer Allen-Bradley controls and to keep the cost low. When it comes to temperature measurement, I usually turn to Omega Engineering first (www.omega.com). Their product specification people are fast to get in touch with over the phone and are pretty good at guiding you to the catalog cut sheets. From there you can weigh your options between Omega and their competitors. Have you started building a specification for the project for accuracy? It would be good to know up front what the customer expects from this system so as to insure they will be happy with the results (or not). Will the application have thermocouple probes inserted into the zones and wired into a patch panel (or signal conditioners) or will there be thermocouple heads (such as a probe)? If the customer will go with probes, there are many signal transmitters that are integrated into the heads (converts thermocouple to analog signal). Have you considered comparing wired temp measurement to a non-contact temperature scanner? Raytek makes a rugged line of temperature scanners that have very decent resolution and accuracy.

[ZedIDave 0]

Will, Customer requested AB PLC. I would not use AB if I was trying to keep the cost down. The thermocouples are imbedded in the ceramic heater. All the heaters that do not have a thermocouple will be replaced. I have looked at Omega but did not see any suitable solutions. For the money that we are going to charge the customer to implement this just so they can see the exact heater to be replaced seems ridiculous. We could easily narrow it down to a zone for 1/3 cost and then they could use a non-contact transmitter to find the defective heater. I don't see the payback

[Veganic 8]

Monitor the heater current instead?

[robh 12]

What about using some sort of thermocouple switch? Say each zone has 10 TC, you could have one input from the switch to TC card, then have multiple TCs connected to the other side of the switch. You could sample the temp every few seconds? I have never seen such a device other than manual hand switch. Like this! I just found this automatic switch. Edited 21 Nov 2008 by robh

[paulengr 44]

If you switch to RTD's, since they are resistors and you don't need the special thermocouple wire (made out of the same metals), then you can get creative with relays and switching and such to multiplex everything. RTD's are also more accurate although there are temperature limits (don't expect to go to 2000 degrees F).

[BobLfoot 302]

Allow me a little liberty while I crunch thru some math. 320 TK's in 40 zones thats 8 per zone. 20 ft long for 40 zones that 1/2 foot or 6 inches per zone. So you have 8 TK's every 6 inches if we run averages. using TW's example 8 modules per node and 8 TK's per module means 1 zone per module and 8 zones per node. This means 5 nodes or a node every 4 feet. Devicenet is one nice way, I'd choose ControlNet just because it has a ton more bandwith than Devicnet. 40 1794-IE8's and 5 1794-acn15's runs $41K almost $42K on RA's website and that doesn't include the PLC the HMI, the cable and the Power Supplies. Delivery in two weeks. Hope this is a 100K job at least if you want to make a margin. I think that the OMEGA solution http://www.omega.com/pptst/OME-PCI-1800.html would be more economical.

[kaiser_will 65]

Is your proposal budgetary or is the project a done deal once you find the hardware solution? What is the business purpose for the project? If the customer is adamant that you must use Allen-Bradley, I suggest you offer up a couple of alternative solutions for them. Quote up the Allen-Bradley solution, such as what BobLFoot specifies. Then also get on the horn with Omega Engineering and National Instruments and work with them to spec out a solution with their products. Omega and NI both have considerable hardware for this task. Go into any research lab where high-density temperature measurement is going on and I bet you will find Omega or NI products in use.

[ZedIDave 0]

Last night I was looking at Red Lion controls. They looked like they had may solution. Using the CSM controller and CST modules I can get the hardware for around $16K. The controller has numerous serial and ethernet drivers. Looks like my answer but I want to investigate the other suggestions. I am going to quote our customer with the AB PLC option and what other solution I may find, but the Red Lion solution right now appears to be the most cost effective. All this to allow the maint. people find the defective heater a little faster. Thanks for the help.

[Clay B. 1]

2 cents worth...... Your customer may find that this is not going to give them the answers they are looking for. If the reason they want this is to find a broken heater temp it may not work. If one heater is broken then it can still be heated by the other heaters around it. I would strongly suggest a system using current detection If heater is broken then current goes to zero. Problem is obvious. If current drops IE...element has parcial short to ground this can be detected. If element is has dead short to ground current meter can shut circuit off and you do not end up playing find the short game. In injection molding this is used all the time. Some of the mold machines I have worked on had well over 100 heating elements in thiem and this system let you find the problem in just seconds. Usually it took you lopnger to walk over to the machine than it did to trouble shoot. The other plus side here is that the zone did not have to go "cold" before you detected the problem.

[ZedIDave 0]

We told them they would need to allow the heaters to cool and then turn the heaters back on to even detect the failed heater due to heat absorption. I would much rather use current detection as I stated in an earlier reply, but the customer wants to "see" the defective heater. We may end up doing both, since the customer may run numerous parts before noticing the part is out of spec due to heat. The application is on an inline thermoformer. Thanks

[BobLfoot 302]

Red Lion Website pricing was CSTC8000 @ $536 each handles 8 TKs so 40 * $536 == $21,440 CSMSTRV2 @ 308 each handles 16 CSTC8000 so 3 * $308 == $924 So $22,364 from the web beats the $50K at Rockwell.

[Veganic 8]

Missed that, sorry. But while on the subject, how would you monitor the current. Just wondering if there is a better, cheaper way than I'm currently (!) doing it.

[Mickey 71]

You didn't say how you are doing it , but below are a couple of options. http://www.loadcontrols.com/products/products.html https://www.ohiosemitronics.com/products_current.asp

[ZedIDave 0]

I thought about using the old current doughnut with a digital output and wiring it and the ssr trigger signal to inputs that would then be monitored by a stand alone PLC. The temperature control is on a standalone system with the setpoint and current value being used on a comm line. Since then we have finalized on what the customer wants. We are going to monitor each heaters temp through the red lion control. The Red Lion control will then send the current temperature over to a PLC. This PLC will have the setpoint and high and low alarms that are set by the operator transmitted by the main PLC. If the heater goes above or below the setpoint, an alarm will be triggered. We will have logic to prevent the alarms from being set until the heater has reached initial temp and an overall timeout if the heater never reached set temp. Thanks for everyones input

[TimWilborne 108]

Please come back and let us know how it goes. I will be very excited to hear how you like Red Lion's modular controller

[ZedIDave 0]

TW, I hope our customer accepts the quote. I will let you know how the Red Lion controls work out. I am also looking at them to do some protocol conversion on a different job using the Data Station Plus. They have some really nice hardware and the software is free. Hard to beat these days.

[paulengr 44]

RTD's these days have higher accuracy and cost roughly the same as thermocouples...probably about $40 per unit. You do not have to buy the special thermocouple wire...any wire will do, and junctions are not as critical. This being the case, you can think in terms of a matrix arrangement which means you can do multiplexing of your inputs and drastically decrease costs. So, going to Omega, the D2PF relays have 4 poles each for $24, for a cost of $6 per RTD and $2000 for all of them. Now, let's take a 1794-OB32P, for 32 total outputs. Now imagine your RTD's in a matrix arrangement (rows and columns). When you trigger an output, fire the appropriate relays to multiplex only 10 of the 320 possible RTD's onto 10 inputs (each input shares 32 different RTD's). Then you need just two 1794-IR8's (remember...you can't use your proposed IE8's because you need thermocouple readers) to read those inputs, at a list price of $3000 for the input cards. Cycling through all 32 bits gives you all 320 RTD's, 10 at a time.