Thermo couple input conditioning

From DIY contraptions to sophisticated FreeEMS-specific designs! Plus general hardware development!
MotoFab
1N4001 - Signed up
Posts: 307
Joined: Thu May 29, 2008 1:23 am
Location: Long Beach CA

Re: Thermo couple input conditioning

Post by MotoFab »

jharvey wrote:The slow sensor is the one that costs more. I think the real expence is the PCB. You can always leave off the chips if you don't want them. Or populate them down the road if it's a feature you do want. I'm tempted to draft it up with both, and expect for blank pads.
Have you priced the Maxim and AD parts in quantity? And are you looking at the low or high accuracy AD595 part? You're right about the other costs though. Board real estate, larger enclosure, panel cutouts, interconnects, dual-tech assembly.
mk e
1N4001 - Signed up
Posts: 230
Joined: Mon Aug 17, 2009 3:20 pm

Re: Thermo couple input conditioning

Post by mk e »

MotoFab wrote:
mk e wrote:At $100 to read 12 TCs as well or better than any other option on the market it’s already priced at less than 10% of what any other option would cost and a great value.
A 12 TC conditioner using the AD595 will likely add $300 all in.
Still a very good value I think for people who need TCs and a nice feature to have available.
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Re: Thermo couple input conditioning

Post by jharvey »

AD chip is $13 qty 1

http://octopart.com/info/Analog+Devices/AD597AHZ

Maxim chip is $13 qty

http://octopart.com/info/Maxim/MAX6675ISA%2B

Looks like I buggered my thoughts, the non alarm version of the AD597 is $6, so both chips cost the same if we want to compare apples. If we get rid of the alarm feature we could save a couple bucks, but that's not the title of this design. So $13 x 12 is $156 for the thermocouple chips.
MotoFab
1N4001 - Signed up
Posts: 307
Joined: Thu May 29, 2008 1:23 am
Location: Long Beach CA

Re: Thermo couple input conditioning

Post by MotoFab »

jharvey wrote:AD chip is $13 qty 1

http://octopart.com/info/Analog+Devices/AD597AHZ

Maxim chip is $13 qty

http://octopart.com/info/Maxim/MAX6675ISA%2B

Looks like I buggered my thoughts, the non alarm version of the AD597 is $6, so both chips cost the same if we want to compare apples. If we get rid of the alarm feature we could save a couple bucks, but that's not the title of this design. So $13 x 12 is $156 for the thermocouple chips.
The page also shows a link to AvnetExpress (Avnet's 1sey 2sey online store), and shows stock of the Maxim part at $8.44 qty1. A roll or tubes of the Maxim part is nearer to $3 from first-tier distribution.

Octopart is a great referral source for a few of something. Octopart is a for-profit link provider and they do a good job. For production quantities, check out the first tier distributors listed on part manufacturers websites.

Also, if you use the 595, the AD595A is the lower accuracy part. Check out the pricing on the C part.

The higher accuracy part is a better choice as the high voltage output signal needs compressing which will add more offset and gain errors. These Analog Devices parts have a 10mV/°C transfer function because they were originally designed to drive needle gauges.

I'm not pushing for the Maxim part, we're just talking here. I'm surprised the 'open thermocouple' alarm would be forfeit.
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Re: Thermo couple input conditioning

Post by jharvey »

I didn't see the avnet price listed on octopart, I didn't dig into it that much. You are right, they offer it in qty 1 for a bit over $8, so the digital could be lower cost if purchased from this vendor. Seems the typical price is $13, I wonder why theirs is so low.

We were thinking the the 597 instead of the 595 because of the extended ambient temperature range. The 595 does 0C to 50C while the 597 is 25C to 100C. Both can work in a much broader temperature range, but the error brackets will increase if you are out side this ambient.
toalan
Wideband Wizard
Posts: 241
Joined: Tue Jun 10, 2008 2:53 am
Location: Toronto Canada
Contact:

Re: Thermo couple input conditioning

Post by toalan »

at 10 bits the resolution is quite poor, I have used 11 bits without amplification and the Temperature resolution and accuracy is too bad without doing special stuff.

If your ADC has an offset of 1 LSB the introduced error is quite large, and if you operate near the rails you have to deal with non linearities of the ADC.

I ran my EGTs off the internal bandgap voltage buffered by an opamp, so that I am not operating near the rails.

I did a calibration of my ADCs to remove offset and linear gain errors.

Also I added in an IIR filter, depending on the strength of the filter you can pickup alot of resolution and reject out alot of noise. For a slow thing like EGT, it is a very good approach. With the IIR, I could achieve ok resolution. Oversampling + decimation would work too, but IIR nets you more resolution for a given # of samples and lowpass filters the samples to boot.
User avatar
KW1252
LQFP112 - Up with the play
Posts: 166
Joined: Tue Jan 15, 2008 5:31 pm

Re: Thermo couple input conditioning

Post by KW1252 »

The Maxim chip is in my favor, given that SPI is already in use. The chip needs no external circuitry, and will not eat up an AD channel. It's true that 4MHz does seem low, but for actual road-use thermocouple response time that's rather quite good. The figures I've seen for EGT thermocouple response times are few hundred milliseconds, which is rather good. The fastest I've found is 50ms, but that's for testing and tuning only, not for extended use. Industrial TC probes have time constants from several seconds to several minutes for comparison.

For the AD chip, I think there's a great difference whether the ECU is in the engine bay or inside the car. For me, the interior is usually below 25 degrees, save the hottest summer days.
arsenix
TO220 - Visibile
Posts: 10
Joined: Thu Jan 08, 2009 3:17 am

Re: Thermo couple input conditioning

Post by arsenix »

This topic is a bit old at this point... but hey it is still on the first page of threads :)

One thing that wasn't mentioned previously is that the Analog devices chips can be MUCH cheaper if you need more than about 4 channels. Since the speed is so high, you can run the chip through an analog multiplexer. So for eight channels you can use one of the high accuracy analog devices chips, and one 8 channel multiplexer. I think both of those chips are under $15/ea. The Maxim chip is very convenient, but at 4Hz you can't really multiplex it so you are forced to use one for every channel.

I have a schematic I designed a while back for an 8 (or was it 16?) channel thermocouple board that used this strategy. If anyone is interested I'll see if I can dredge it up and post it.


James
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Re: Thermo couple input conditioning

Post by jharvey »

I'm interested, please post. Most importantly, what specific chips do you recommend looking into.
arsenix
TO220 - Visibile
Posts: 10
Joined: Thu Jan 08, 2009 3:17 am

Re: Thermo couple input conditioning

Post by arsenix »

My design was basically:
2x AD595's
2x ADG507A - A 8 channel, two input multiplexer. This allows both + and - pins of the TC to be multiplexed. I believe this chip was recommended on an analog devices app note.
This allowed 16 channels of TC. Design has a 15V switching supply to feed the 595's. This will allow up to ~1300C to be measured. 2 op amps are used to scale 0-15 down to 0-5 to feed into the micro.

Looking back this now I remember what I was doing. I designed the board to allow ganging three PCBs together with two of the boards having unpopulated microcontrollers, for up to 48 channels of TC. The intent was to hook this up to my ECU (MS2) over CAN.

I'll post the whole schematic although you will mostly be interested in page 3. I guess I'm not cool enough for attachments, so here is a link:
http://peverill.org/files/designs/megat ... 7-2010.pdf


James
Post Reply