Thermocouple signal conditioners 1

[asimpson]

I am using PLC to monitor temperature 500-900 c using K type thermocouple. PLC only takes 0-10V analogue signal.

I found a thermocouple transducer to convert thermocouple 0-1000C to 0-10V however I would like to maximize accuracy over range I am looking at 500-900C

Can anybody recommend signal conditioning thermocouple transducer that might do this without having to do too much electronics building?


Thanks

 

[VEBill]

The solution you have found so far is pretty close. How many bits is the analog input? Is it worth worrying about what might be (at most) only one additional bit of resolution? How does the PLC's A/D resolution compare to the overall system accuracy?

It's a good principle that the measuring system should have plenty of head room at the top, and same thing at the bottom. One should beware of what happens to the control system and/or user displays if the variable happens to get out of range.

 

[IRstuff]

Gaining up the signal is not necessarily an accuracy improvement. Since no gain processor is perfect, additional gain implies additional error. The typical transducer appears to have resolution smaller than the accuracy already, so there is nothing more to be gained:

http://www.intertechnology.com/Doric/pdfs/600_Series.pdf

TTFN
faq731-376

 

[asimpson]

PLC has 8 bit res. on A/D conversion however it states a +/- 5% accuracy. which is 100C over 1000.

VE1BLL
What do you mean room at top and bottom? Does that mean if I am interested in a range I should have capability be reading 20% above and below in my PLC?

 

[BrianG]

+/- 5% accuracy is not good - can this be calibrated by the user with a reference voltage, i.e. so that 10.000V input gives exactly 8 bits?

8 bits can give a resolution of approx. 0.4% once calibrated, so if you have a signal conditioning unit that can have an offset voltage applied to it of -5V then add a gain of 2 your input to the PLC could give you 500-1000 deg C range for 0-10V input. The only downside is that you can't then tell any difference between 500C and a lower temperature. If this never occurs in your process it may not be a problem, otherwise you may have to programme you PLC to recognise a low A-D count near zero reading as being = 500C and anything from there down to all bits = 0 as out of range or sensor failure.

 

[VEBill]

What do you mean room at top and bottom?

Adding to what BrianG already provided:

In my line of work, we pay very close attention to safety as a science unto itself. There have been accidents (in many fields) where having a sensor/display system bottom-out, or max-out, has made a bad situation much, much worse. It should be standard policy that the sensors and displays cover a range that provides useful information even in the event of failure modes.

Of course, there may be situations where one needs to make a trade-off. The point is to not allow oneself to instinctively narrow the sensor and display range, without at least taking a moment to also consider failure modes.


8-bits is pretty low-end in this day-and-age.


One old trick is to convert the signal-of-interest to the frequency domain (V/F converter), and then measure the frequency. There's PLENTY of resolution (accuracy is another topic) in the frequency domain, even in the low kHz range. This assumes that your PLC has a discrete input and enough uC horsepower to measure frequency. Also, think about the counting time.

 

[MacGyverS2000]

8-bits is pretty low-end in this day-and-age.

Not to mention 5% accuracy... that just sucks.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

Seems odd that you don't want to muck with circuits, yet choose a rather pitiful one. The transducer cited above:

http://www.intertechnology.com/Doric/pdfs/600_Series.pdf

has RS-232 digital outputs. No fuss.

I'm sure there are other modules with the same capabilities that might be cheaper.

TTFN
faq731-376

 

[asimpson]

The truth is I only have access to PLC programming (ladder logic) capability at the moment and wish to do a proof of concept with a simple PLC.

When I get my hands on a good electronics engineer I will probably use a PIC based controller whish can take in 14 bit temperature values. Much smaller, cheaper, and capable.

 

[IRstuff]

14 bits seems a bit light; that'll get you 0.61 mV resolution, but 10V/1000°C = 1mV/°C. If you want to be able to average down the noise, you might consider going to a 16-bit ADC.

You might consider:

http://www.sparkfun.com/products/10294

While it has only a 10-bit ADC, it comes with a bunch of other devices that might prove interesting, and the tools are either cheap or free.

TTFN
faq731-376

 

[danw2]

There are hundreds of 2 wire loop powered temperature transmitters whose output can be scaled as needed, like 500-900°C over the 4-20mA output range.

When you drive 4-20mA through 500 ohms and you get 2.0 -10.0Vdc suitable for your 0-10Vdc analog input.

Yes, you lose 20% of the scale by having a live zero at 4mA, but with a 400°C span (900-500) you lose 600°C of the 0-1000° range that's not of interest.

No, 2 wire loop powered transmitters can not be scaled for 0-20mA because the transmitter uses about 3.5mA run its own electronics.

Google loop powered temperature transmitter for 1.3 million hits

Be sure to inquire of the vendor about getting the range done for you, so you don't have to buy software or a special cable to configure the range. Either the manufacturer or the distributor should be able to range it for you.

 

[sreid]

Does the PLC Manufacturer offer a temperature extension block?