Why a Type K thermocouple?

Hey, I was just wondering why we are limiting ourselves to a type K thermocouple with the new circuit board. THats if we are limiting ourselves to that at all really.

Personally, I would think that for the moment we would want to use a type T thermocouple. It is much more accurate over the current temperature ranges than the type K that is being used.

---

Jay

the board isnt a part of any sort of 'recommended' reprap electronics right now. i was looking into thermocouples, found a simple circuit, and made a board so that people can easily start building and researching reprap machines using thermocouples.

nobody is limiting anyone here. before, no thermocouples were supported, and today you can easily use a type K thermocouple.

type K is the most common, and the AD595 chip is intended to be used for type K thermocouples. the datasheet (my guide in making the circuit) showed how to use it for type K, so thats what i did. it can be used for others, but it requires extra work.

if you think type T is better, then feel free to modify it. its open source after all, not to mention experimental at this point. i'll even go one step further: if you modify it to work wtih *both* types of thermocouples, i'll incorporate your design into v1.1 and you'll have boards manufactured and i'll send you some free ones.

The data-sheet (http://www.analog.com/UploadedFiles/Data_Sheets/AD594_595.pdf) for the AD595 also covers the AD594. Between then they can handle E,J,K and T thermocouples.

Using a Type J thermocouple just involves using an AD594 instead.

Using an Type E thermocouple looks like the hardest to me, as it involves calibration, but step by step instructions are given in the data-sheet.

Using a Type T with the AD595 requires no hardware changes. In fact from 0C -> 50C it involves no changes at all. Beyond 50C or below 0C you will need to calculate a new lookup table to use in the micro-controller as Type K and Type T thermocouples have different non-linearities. (I assume we are already using a look up table of some kind as even with a Type K the voltage out is not directly proportional to the temperature).

So the existing board will work for type J,K and T thermocouples.

How's that, I got it working with both type K and type T without changing anything ;-}>

I leave the calculating a lookup table and modifying the firmware as an exercise to the reader. ;-}>

Edited 1 time(s). Last edit at 05/06/2008 11:15AM by peteredworthy.

Hi, I wana use ad595 with mux 4052B(4 channel dula mux). can any one help me regarding this circuit i.e how to calibrate Ad595 and C code for selecting mux channel. also converting the out put back to temperature. thanks . need urgent reply

Hey ayesha.

I suspect that using a MUX chip to switch between thermocouple inputs like that will not work too well. If you look up your standard K or T type thermocouples, they will generate 41 microVolts per Kelvin (type K) or 43 microVolts per Kelvin (type T). With a temperature differential of around 200 Kelvin, you're looking at somewhere around 8.5 millivolts of signal that you want to measure.

I think that your suggested MUX chip will probably have some non ideal characteristics that stuff up this very small signal. If you want to use the AD595 I would suggest that you go for some nice small DPDT relays to switch the signals. If you want to get really fancy, you can get latching ones (i.e. you don't need to supply current all the time) .Keep your wiring quite small and tight, and try to keep all of your wiring in pairs to reduce noise / thermal issues.

Alternatively, you could go for a slightly more difficult approach which could offer significantly better performance and use a high accuracy ADC (say a 24 bit one to get good temperature resolution) with SPI and a built-in input multiplexer on it. Add an external digital temperature sensor (again you can get SPI based ones) and you can do the same processing as the AD595 does inside your micro.

This would probably have a similar total cost as a relay setup but would involve surface mount chips (sorry).

The advantage is that it will be smaller, more accurate and not tie up so many microcontroller pins.

Unfortunately I can't be much help with the coding side because I'm not familiar with programming PICs or Arduinos. Maybe someone else would like a go?

jbb

hi...i wanna test thermocouple type k with amplifier circuit...but i dont know how..please help me..

There is a circuit and some software information here: [reprap.org]

---

[www.hydraraptor.blogspot.com]

please tell me what will be the output of the circuit if 50 thermocouples in series are connected to AD595 amplifier? what about the temperature difference? what is the minimum temperature difference to generate voltage across two terminals?

Thermocouples are based on a slight difference in electronegativity between dissimilar metals. The thermocoiple is made by taking wires of two different metals or alloys and having them connect together at only one point. That point of connection is where the temperature sensitivity is. Lets simplify the metallurgy and just call the low side "A" material and the high side "B" material. So the thermocouple is measured where wire A meets wire B.

If you connected 2 Type K thermocouples together (or two of any type) then you would add third junction, where you wired the high side of the first one to the low side of the second one. In effect, you have an A-B-A-B thermocouple. Which means that the voltage goes up where the first A meets the first B, then you have a reversed junction where the first B meets the second A, so the voltage goes back down to near zero, and then another A to B junction where the second A meets the second B and the voltage goes back up. So you have a +, followed by a -, followed by a +, or about the same voltage you had with just one thermocouple. Worse, the minus depends on the temperature where you joined the other ends, probably room temperature. As that varies, the output voltage changes even if the spot you are trying to measure stays exactly the same.

This means that putting 50 thermocouples together in series wakes the final voltage no larger, but completely confused by all the pluses and minuses. To get a higher voltage, you need an instrumentation amplifier, a special op-amp that amplifies the difference between the two inputs. Normally this is done with several normal op-amps, one for each input to buffer and amplify them individually, then another to combine the two and convert the difference into a larger voltage, often zero based.

But if you look at the specs, the AD595 is the perfect device to connect a thermocouple too, and they are cheaper than any other kind of instrumentation amplifier you can buy, so why try to work around it? Use 4 AD595s, and then a 4 to 1 analog mux to switch between the thermocouples. Otherwise, you will have to use several op-amps to buffer (convert very low voltage low current signal into a high current, possibly still low voltage signal) before each thermocouple reaches the analog multiplexer inputs.

Mike

Edited 1 time(s). Last edit at 08/13/2010 02:18PM by rocket_scientist.

I was mistaken as to how inexpensive the AD595 is. Most sites like Digikey and Newark list it for about $30 US, with some vendors with lower prices showing out of stock. However, doing a little googling around, I found it in stock for $18 at SparkFun, who are nice people to work with anyway.

Mike

the expense of the AD595 is one reason that I went with the MAX6675 instead

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

hai there, well regarding the thermocouple, previously some student had made a pro-type on how to measure temperature, he used AD595 amplifier and a microchip. now the task is now, i need to improved the pro-type for better performance. so here i'm looking for some advice and guide how to build a pro-type and and how improve it, hope to hear from you all and i hope everyone can help on this.

About ad595 I tried using it and in my experience i got it running but it ended up with having "noise" giving some 2-3C swings at room temp. I used EC2.2 made by myself, board has its own 7805 with caps but sadly does not have that L-C filter for Vcc, so maybe it was related to that too. Tested with no actual thermocouple (short ends read room temp), with RC filter on transmission (diff values), tried diff approaches even with ad595 compensation capacitors, etc. I had alot of tries but couldnt get rid of this 2-3C swings which appeared like after 4-5 correct readings. I ended up suspecting some internal adc noise or some comparators issues, or some picked up noise, something like that just couldnt put my finger on it.

I ended up using thermistors for both bed and extruder, and these seems to give very stable readings and i am satisfied with them. IF they are calibrated properly, i find the system very good in practice.

About thermocouples i would say MAX6675 is probably better in the terms that it uses its own adc and just reports the value, this process being probably more resilient than ad595 or other options which only reports voltage. That being said i havent used MAX6675 myself.

About why type-K thermocouple, i think there is no reason, just i think its the most common type around DIY environment.

AFAIK, I think most thermocouple chips can either be made to work with Type-J, or otherwise, they have chipset variants for Type-J or others.

Edited 1 time(s). Last edit at 09/03/2011 01:28PM by NoobMan.

So, kind of an old post...but i have been wondering why, in all the posts about thermal couples, people are using such complicated chips? Not that im against it, just wondering my not simply wire in a opamp circuit with a LM358 [NTE928M]? For example: [www.bristolwatch.com]

This could easily be tied into the Servo ports on a Ramps and added to a Reprap machine easily. Just a thought i've been having.

Thanks!
Cid

Thermocouple ICs have more than calibrated amplifier, among other things they got some absolute temperature reference built-in. Some kind of temperature sensor to which readings are compared and adjusted, for high accuracy. AD8495 is replacement to AD595, can read both datasheets and look for "ice point reference" / "cold junction compensation". I cant find in that link anything about where that reference would be. And probably there is more in case of thermocouples, e.g. errors, accuracy, etc, that wont be attainable in such a diy strip-down fashion.

Scale of mV voltages, sensitive reference point, high degree of accuracy required, etc, all these are not a good mix to have on a plate. And if somebody does this good with individual / discrete parts, then probably will cost much more than the IC itself.