Page 1 of 5

Thermo couple input conditioning

Posted: Thu Aug 27, 2009 2:03 am
by jharvey
The 5554 project has brought up the use of thermocouple. I think the use of the AD597 is a good choice. That's the chip used on the relow oven I'm working on. Here's a link to the reflow.

http://www.thesiliconhorizon.com/reflow.htm

The reflow has a good response, and they claim is handles more than 1000 cycles before the probes they use break. I think this is a good reference design. Are there any thoughts, other alternatives, ect?

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 4:22 pm
by jharvey
I see there are some variants of the above mentioned chip. The AS594, AS595, AS596 and AS597. There are some questions that should be answered before choosing a chip. Perhaps answered with a feature spec of some sort.

What is the temperature range we are looking to measure? Do we care about - temps? These thermo couplers are intended to be in the exhaust gasses right?

Can the brain ADC convert negative voltages? I suspect it can not. I'm assuming the ADC will range from near 0V up to near 5V. Is 10mV/C a good accuracy? I suspect it will suffice.

What do we expect for an ambient temp of the FreeEMS unit? The variants of these chips have different tolerances based on the chips ambient temperature. The 94 and 95 prefer 0°C to +50°C while the 96 and 97 prefer +25°C to +100°C. All will work from -55°C to +125°C, but are most accurate in the preferred ranges. I think ultimately the question is, will this be in the engine bay, or in the cab.

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 4:54 pm
by mk e
On the 5554 I/O doc I listed
“K type, 41microV/dC” – these are common, cheap and work from -200C to 1350C
I will add a range of 0-1000C to the spec, 600C-700C should be about normal.

I don’t know what is being considered for FreeEMS.

Exhaust temp gets measured for a couple reasons. The turbo guys don’t want to fry the turbine. Roadracers want to know about cylinder to cylinder differences mostly as it indicates problems. The exact number is not all that important, being of 10C at 600C is a non-issue but being off 100c probably is a problem.

The case temp question is a good one. It is normally not considered acceptable to put an ECU in the engine bay and I don’t know of any aftermarket manufactured that does not specifically recommend against it. Even still I would expect 50C is not high enough for the in-case temp and suggest you go with the higher temp 96 or 97.

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 5:30 pm
by jharvey
If we sample that directly, the output voltage will be from about 0V at 0C and about 10V at 1000C. I believe the ADC can only handle about 5Vmax, so we would need to scale that temp range down with a resistor divider or something similar. What's the resolution of the ADC? 10 bit better worse? I'm ignorant about the brain board at this point.

Assuming a 10 bit DAC, ADC rail to rail linearity, and 5V = 1000C, the resolution would be around 5V/1024 = .005mV per bit, so call it about +/- 1C. Therefore a linear scale should be just fine. No need to warp the scale for a more accurate reading at 600 to 700 degrees, right?

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 5:38 pm
by mk e
jharvey wrote:If we sample that directly, the output voltage will be from about 0V at 0C and about 10V at 1000C. I believe the ADC can only handle about 5Vmax, so we would need to scale that temp range down with a resistor divider or something similar. What's the resolution of the ADC? 10 bit better worse? I'm ignorant about the brain board at this point.

Assuming a 10 bit DAC, ADC rail to rail linearity, and 5V = 1000C, the resolution would be around 5V/1024 = .005mV per bit, so call it about +/- 1C. Therefore a linear scale should be just fine. No need to warp the scale for a more accurate reading at 600 to 700 degrees, right?
no, it's about .041V and we need to amp it.
0.000041V/c x 1000C = 0.041V right?

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 6:02 pm
by jharvey
Looks like your thinking uV not mV. The data sheet has a table that helps simplify it, or perhaps you're thinking of the thermocouple itself. This chip is an amplifier.

The table is nice because you can look at it's variations over what we are assuming is a linear scale. I'd say it's fine to consider it linear, for this range, however it does wonder a little bit. Down under 1%

Take a look at table I.

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 6:13 pm
by mk e
I'm giving you the SYSTEM INPUT spec, which is the output from the TC.

The processor wants its input to be 0-5V, which would the be the output from the I/O board.

So the spec for the I/O board TC circuit is:
0v-.041V input /0-5V output.

You want a 100:1 amp right?

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 7:20 pm
by jharvey
I see what your saying. The type K thermo couplers products the 41 µV/°C, and is the input spec. It will need to be amplified to fit the processors voltage range.

I choose to look at the AD597 for this amplifier because it is specifically tuned for K type thermo couplers. The AD597 includes some features like coupler failure detection, and ICE compensation.

The output of the circuit in our case wants to be 0-5. Working from the sample schematic found on pg 4 of this pdf, http://sigma.octopart.com/54925/datashe ... 597ARZ.pdf I see a gain of 245.46 which should produce the voltage shown in table I. The voltages noted on table I are too high for our processor, so I have to attenuate them before they are good for the processor. My plan is a simple resistor divider, allowing us to pull the scale down to an good range for the processor. After I attenuate the signal, the overall system gain will be closer to 100, instead of the 245 provided by the AD597 chip.

Does this sound like we are on the same page?

Re: Thermo couple input conditioning

Posted: Wed Sep 02, 2009 7:34 pm
by mk e
That sounds right to me and I like the added features on the AD597

Re: Thermo couple input conditioning

Posted: Thu Sep 03, 2009 6:23 pm
by Fred
mk e wrote:I'm giving you the SYSTEM INPUT spec, which is the output from the TC.

The processor wants its input to be 0-5V, which would the be the output from the I/O board.
That is great, but there is no point ignoring the fact that he is obviously talking about designing the circuit to read the otherwise useless sensor and continuing to talk about the sensor voltage. Knowingly talking at cross purposes does no one any good, wouldn't you say?

Fred.