Thermocouple input?

[puff]

Some time ago I've seen a project, where tiny85 was used for software compensation of temperature, with NIST look-up tables, etc. By that time they achieved pretty nice results, but for the operating temperature range limits (their board couldn't measure temperature below 0ºC). There are plenty of ways to measure cold junction temperature (dallas temperature sensors, LM35, thermistors, internal temperature sensor of the µc, etc). Their code used that ambient temperature as an input parameter, plus they read ADC from the thermocouple. That's it. I guess, the same could be achieved with STM, without any extra MAX chips?

How fast should we read EGT? One sample per second per channel? Hundreds of samples per second?

[AndreyB]

Here's a cheap eBay thermocouple, let's focus on the connector. Inner opening: 3mm. Outside width: 5mm

What kind of screw-in terminal do we need to use?

[puff]

I believe it actually doesn't matter. I believe you can just cut it and use wago connectors (preferably, gold-plated ). What is confusing is that the cost of any similar thermocouples here is ten times higher. The second weirdness is the temperature range: this one allows to make measurements within 0-800C (hopefully it is a mistake) - the locally available thermocouples work at the range of -40 - 1200 and higher. the third point is its thread and size. it could be something different - neither metric, nor imperial, - something chinese

[kb1gtt]

I've used this screw terminal before. it's reasonably rugged. It's not designed for thermocouple applications, but it would probably be good enough for our needs. http://octopart.com/7693-keystone-116638

Posted here as well
http://rusefi.com/forum/viewtopic.php?f=13&t=751&start=37

[AndreyB]

it would probably be good enough for our needs. http://octopart.com/7693-keystone-116638

Is it going to press the terminal against the PCB (and scratch it?)? Looks like this bolt would go right into the terminal opening and not hold it? Maybe I am not getting the way this connector works.

[kb1gtt]

Mounts something like the attached, but in reality, the screw is pushed up a bit as you typically have something under the screw head.

[AndreyB]

the screw is pushed up a bit as you typically have something under the screw head.

Would not the opening in the U-terminal be under the screw thus we are not really holding anything? Unless we would want to feed the terminal pass the U opening. Still confused.

[kb1gtt]

Terminal connects on top, not under the U. Does this clarify a bit? Shows blue ring terminal, or blue fork.

[AndreyB]

Terminal connects on top, not under the U.

Yes, this explains everything. I have just hit my new lowest. Let's go with this connector

[abecedarian]

How would CJT be accomplished, considering you're likely going from copper wire to zinc-plated copper in the forked & crimped-on connector, then who knows what the actual screw terminal is (stainless steel, zinc plated copper/aluminum), to the circuit board...?

[puff]

I think it doesn't really matter if we don't need accuracy of 2-3C and better. Just make those connections as close as possible to ads1118 or whatever we use to keep them at the same temperature (am i mistaken?) Besides, we could calibrate the whole thing with the ice bucket?

[kb1gtt]

Correct, if you want accurate, you really need Omega's PCC-SMP or equiv, if you want low cost and can tolerate less accurate, you do what they do in PLC's and industrial applications where they simply take your best stab at it. Such a connector should be less then +/- 5C, and you can typically remove much of that if you create a calibration factor. Basically if you generate a temperature, then you measure how far off you are, you then know your calibration factor. It's lame but there are a great many low cost industrial applications that do it this way.

[puff]

so now it comes to building the circuit...
ads1118 - i checked, aliexpresssells it for 45$ per 10. ebay sells 10$ per 5 - strange, isn'it?
as for psoc - if it is s cheap, why isn't it used by everyone? as far as i get it alan to uses it for his o2 controlers...

[kb1gtt]

About PSoC, some people get concerned with being single sourced. This is why ARM is king, as they allow multiple vendors for basically the same core of the chip. Being multi-sourced is a huge deal with many designers. As well it requires you to use their tools, so you don't have 3rd parties pushing their products. Many developers learn a tool they learned in school (C or C++) then they don't want to migrate to a new platform with a new way of thinking. These two items really limit PSoC's, the third key issue is scalability. PSoC's key feature is smaller PCB's, however as qty goes up cost of the PCB goes down, so they loose the competitive edge as qty increases. If you include issues harder to track issues like redesign due to obsoleted parts, or tracking your source materials for RoHS2, ect, I feel PSoC's are still very competitive at higher qty, but most people don't include those harder to track costs of a product life cycle. Basically it's a teach a old dog new tricks issue.

About product dev, why not just use the MAX chip? This code has already been developed in rusEFI and it's an easy path to implement. Should we really spend a pile of resources on saving a couple bucks? I'm tempted to think we should get something available and useful, then spend our resources on something more unique. I'll certainly support someone if they want to develop a board, I might even point them to a starting point, where I have trimmed down Frankenso. For now I have a pile of other items I'm going to bump up in priority, so It will be a while before I get back to continuing the EGT effort. If someone wants to finish the design, I'd be happy to help encourage the effort. It's probably a good opportunity for a young player of the KICAD game

[AndreyB]

why not just use the MAX chip? This code has already been developed in rusEFI and it's an easy path to implement.

It is the MAX chip as far as I am concerned - simply because I only have time to do the trivial stuff at this point

[puff]

you mean adding extra discovery or avoiding can?

[AndreyB]

you mean adding extra discovery or avoiding can?

I do not follow

[abecedarian]

you mean adding extra discovery or avoiding can?

I do not follow

Seriously, I don't think many understand the issues regarding temperatures, dissimilar metals and current flow, either.

[kb1gtt]

I agree, that's why so many industrial applications do it that way. The cheap ass industrial engineers have really pushed the limits of how far you can push the good enough envelope. My biff was when PLC MFG claimed +/- 1C, but I measured it at +/- 4C. This is why I suggest we consider it +/- 5C for 5C to 30C ambient. 5C to 30C was the limits of what I could do back in the day. Our suggest design guide spec's for hardware ambient between -30C to +50C (-22F to 122F). Based on my prior measurements and this extended range, I would expect a variation of +/- 10C. But I've only had the chance to measure between 5C and 30C.

[abecedarian]

so now it comes to building the circuit...
ads1118 - i checked, aliexpresssells it for 45$ per 10. ebay sells 10$ per 5 - strange, isn'it?
as for psoc - if it is s cheap, why isn't it used by everyone? as far as i get it alan to uses it for his o2 controlers...

ADS1118 on ebay are $3 for 1, $10 for 5, $18 for 10... et cetera. ADS1118 from TI is $5-$6 (depending on package- X2QFN or VSSOP) in singles, and do discount in accordance with quantity / breaks, however if you have an 'acceptable' email account on their site, i.e. verifiable corporate or EDU type, they offer samples for cost of shipping, IIRC. If there's sufficient interest in using this chip, I can talk to a few people to see what might be workable, as in some 'group buy' type thing.


Re: PSOC.... Cypress has been in the chip business for years. They are a second source for many things 'unlabeled' as well. They are used by nearly everyone, just not a lot in the maker / hobby community.

[AndreyB]

My attempt to find something similar on eBay or aliexpress has failed, I've placed an order for some of these Keystone 7693. Can we have a one-channel tiny board to use that connector?

[AndreyB]

the terminals are a bit bulky but still nice. now we need the footprint & a new revision of the board

[puff]

could you please measure the thickness of that braided cable?

[AndreyB]

D=3.0mm
maybe 3.2mm

[puff]

thx!