interpolating ribbon temperature by resistance

I am currently building a small high temperature heatsealer for fluoropolymer plastics like Teflon. I have used off the shelf controllers with thermocouple inputs and have even designed one (with the help of an electronics friend/genius) that is in use today in our business (using an 877 and picbasic pro).

I would like to do away with reading a thermocouple (too slow) and read the resistance of the ribbon directly with an A/D pin, go to a lookup table and display the temperature based on imperical data taken during the initial setup.

The ribbon is pulsed with current from an alternistor driven transformer. The transformer is only turned on a few milliseconds every eight miliseconds. While the current is off, I want to read the resistance of the ribbon.

I have maybe a four millisecond window at most to read this resistance and act on the A/D value. Timing is everything.

I do not have a good handle on how long it takes to convert analog to digital in a pic running at 20mhz using PicbasicPro. I've also seen mention of reading the pic directly for faster access.

Can someone tell me how fast the A/D channel works, and what issues you see in how I'd like to do this? It has been done before in a similar controller (Ropex) but they measure by current and voltage during an on-cycle. I just want to know what others might have tried and find out if this can be done.

Don't beat me up too badly... I'm mostly made up of crazy ideas... not a genious in the electronics department.

Many thanks!

Ross

unless we have a little bit more details it will be hard to answer this. But look in the datasheet about acquisition time, it will depend at least of the external impedance.

Read/Write to the A/D register instead of ADCIN will also increase speed.

You may find some good infos and trick bellow
http://www.sfcompiler.co.uk/wiki/pmw...User.MinADtime

Quote:

---

I have maybe a four millisecond window at most to read this resistance and act on the A/D value. Timing is everything.

---

As Steve noted, more details are needed...but as a rule of thumb, a 10 bit ADC acquisition takes about 20uS or less to complete (provided you do not violate the input impedance specification) Further, the lower the input impedance, the lower the acquisition time. If you do not need full 10 bit accuracy, you can cut the acquisition short to further save time - see the Microchip MidRange Manual for details

An ADC example using the registers directly (and not the ADCIN command) is here
http://www.melabs.com/resources/samples/pbp/a2d10.bas
(if you use this latter approach, drop the Pause 5 in the loop)

Hi guys, thank you for your reply.

I'm sorry I didn't get back to you quicker... I did write a reply a couple of days ago.. when I hit the send button it asked me to log in. When I did... something messed up and I lost the message I was going to send. Bother.

So this will be a quick one.

The ribbon is stainless steel, .005 inches thick by 3/4" wide by 17" long. I did the math at work and the resistance at room temperature is something less than one ohm. (.34?) I don't remember. Sigh**

The ribbon is mashed between two parallel bars with silicone and mica and a few other materials in between. I want to pulse the ribbon with enough current to get it up to about 700 degrees F in 10 seconds or less. Then hold it there for a few dwell seconds, and then turn off the power and allow it to cool back to 150 degrees F.

Of course I want to control this sealer for all other temperatures as well. The typical maximum is 700 degrees F. Fluropolymers such as Teflon FEP, PFA and Modified PTFE seal between 550 and 650 degrees F.

The sealer I built currently has a Dallas thermocouple device that reads a thermocouple input and interpolates it. The pic polls the device and retrieves temperature serially every 21(?) seconds or so. This is pretty good but not nearly as fast as reading the ADC channel directly. So if I could tap in to that speed, I could more precisely control current and thus temperature.

I know this can be done because it is being done in a Ropex controller. I believe they are using a current sensor to measure the current during an on cycle. They are also either reading the voltage drop across the bar during an on cycle or the resistance inbetween cycles.

Well I have to run for now.

Thank you for your feedback! Any other comments, criticisims, wise cracks or suggestions are eagerly sought!

Ross

Hi, Ross

I do not exactly know the 'professionnal level" you want to build your control, but, what exists :

1) the programmable temp regulator ... I used it to control Cast iron heat treatments. costs some $$$ ... but most reliable !!!

2) The "el chapo" device ... have a look to Parallax "industrial applications" booklet ... there's how to realize a simple PID temp regulator ( w/soft offered !!! ) . an AD595 or MAX 6675 will be a good interfacing to Thermocouple.

With Darrel's instant interrupts, il wil be easy to drive the system clock , now.

On this basis, I've realized a vent regulator to control with a R/C servo the cowling apertures for a model plane engine ...

Here, I output the command signal @ 40 Hz ... so it might be much enough for your application !!!

Alain

Testing Testing...

Ok...

It worked.

I've tried twice now to submit a reply to this thread from an email link over the last couple of days and both times lost all of what I had written. Bother.

Thank you all for getting back. All good stuff.

I am familiar of course with the temperature controllers out there. They do work great... no question. I guess I just want to re-invent the wheel. There... I said it! :-)

There are a lot of off the shelf modules out there. Every one of them will do MOST of what I want... but never quite all of what I want. And then to do what I want requires that the operator be able to go through multiple layers of menus. Not good for the business I am in... heatsealing high temperature fluoropolymers. The only controller that would appear to come the closest is the ROPEX controller and it's expensive... about $2k.

What makes my heat control dream somewhat unique is that I want it to be easily configured to different material heater ribbons, of different shapes, of different thickness and lengths. Then I want the controller to control the current to the heater ribbon, bringing it's temperature from room to 700 degrees F (or any temperature in between) in 8 to 10 seconds and hold it there for an operator chosen dwell time. When the cycle is over, the current is shut off, the temperature drops and the mechanical jaws open at 150 degrees F to allow the sealed materials to be removed from the machine.

All this control, in a small package, built on the cheap (say under $200 US) with no more than two-layer-deep operator configurable control over heat time and pressure. And of course... I only need two or three of them. HA HA HA HA HA!

Ahhh... but those small problems don't quench my insatiable desire to stick my neck out and say "LET HER GO!"

The ribbon I am currently designing around is 303 stainless... .005" thick x .75" wide x 17" long. That works out to about .12 ohms of resistance at room temperature. I honestly haven't worked out what the resistance is at 700 degrees F. I keep burning my fingers (kidding!).

My current controller is using two pic processors. One is dedicated to doing only one thing... waiting for a zero-cross pulse, checking an eight bit port and then outputing a pulse to turn on an alternistor after a pauseus (based on the port data input. It repeats this every half cycle at 60hz (120 times a second). And it is rock solid. No complaints there.

The other processor polls a max 6675 for a thermocouple reading. Based on this reading, it raises or lowers the pauseus data for the second controller. But the 6675 is slow at 220ms between conversions. So I can poll it about 4 times a second.

I want it to be faster! :-)

I'm banging a lot of amps (75 to 100) or so at 1-2 volts ac for microseconds through this ribbon. It heats up quick and cools down in roughly the same time.

That's the story... I'll fill in more details as you guys thrash me behind the dumpster.

Thanks for the great input. So... how would you read the ribbon resistance between current pulses? Any thoughts??????

Ross

This is just to correct an oops in an earlier post. I didn't even realize the post went through. Ah well...

Anyway... I posted that I polled the thermocouple device every 21 seconds. What I meant was every .210 miliseconds (aprx). It looked stupid the way I wrote it... needed to correct that. :-)

Thanks again for all the comments!

Ross

Hi rossfree,
I was just wondering if you had tried sandwitching 2 of these SS strips together with some heat sink compound and (mica ?) insulators so you could heat one with current and measure the hot resistance of the other one ? It seems to me you could then allow adequate time for heat saturation to get an accurate idea of the hot resistance of the strip. Other methods might be to insert strip into a heat treat oven while wired to an ohmmeter. If you do not have one, I'll bet your local junior college does and pizza seems to be their official currency :) .
JS

Hi, Ross

Depending on the voltage applied, the trick could be simple:

for DC supply, just apply Heating power by PWM ... and allow a continuous current flowing into your resistor.

So simple ... just measure the voltage across your heating element during the "0v" part of the PWM ...

For AC supply, Low voltage ... the PWM will be replaced by burst mode ...and constant DC current only applied when no AC provided.

AC, High voltage ... a probe really seems the simplest way !!!

Alain

Thank you for the feedback...

Right now I am pulsing a 2KVA 48 Volt CT transformer using 110VAC line and an alternistor. One pic does the timing and sends the pulse to the alternistor to turn it on at the right time in the half cycle. The other pic runs the program, picking up the temperature and passing timing information to the first pic.

Depending on the ribbon size and length (it changes all the time) I may require 50 or so amps of current pulsed 120 times / second for very short duration to achieve the ramp and dwell time I require to bring the temperature of the ribbon up to the required temperature and sit there soaking heat into the film to melt it.

The voltage across the ribbon would never exceed 48VAC.

I like the idea of applying a constant current across the ribbon in series with an external resistor. Then reading the voltage across the resistor and interpolating the ribbon temperature with a look-up table or something like that. But I'm not to good at figuring out how to decouple the 48VAC transformer such that it does not affect the results. And forgive my ignorance... but I also don't know how the pulses would affect the constant current circuit. Boy... I wish I had a mind for this stuff!

My electronics buddy suggested injecting a 100khz signal across the ribbon and external series resistor and using a capacitor to decouple from the 48vac transformer. But he said that would likely get into some parts and require some tweaking.

And lastly... I have again revisited using a thermocouple... but instead of the MAX6675 Thermocouple to Digital converter (I am currently using)... I would use a thermocouple amplifier/conditioner like the AD597 which ouputs 10mv/degree C analog in real time. I could feed that into the A/D pin of the pic and convert it to Farenheit for display. This would be much faster than the MAX6675.

So I have choices... but still... I would really like to know the simplest... most precise way of reading the ribbon resistance. That would be the cats meow!

Do you smell it???

That's my brain.

Ross