What standards are there for connectors for 4-wire RTDs?
I have found 4 connectors so far, used by various manufacturers. For 3-wire systems there are two very well accepted standards, a larger “standard” version and a smaller “miniature” version that look like thermocouple plugs-and-sockets only with a third pin added. But I’m not finding a similar consensus for the superior 4-wire circuit.
I have no idea, Napier. Most of the 4-wire IPRTs I’ve dealt with were one of the following:
I think one of the reasons you find so many IPRTs hardwired to the signal conditioner/readout is because many users prefer to calibrate it as a system (IPRT + readout).
Depending on your level of accuracy, and the type of signal conditioner used, just about any connector could be used.
The ones I’ve used had bare wires.
What exactly is your situation? Are you trying to select a connector and define the pinouts?
My situation: I’m trying to promote 4-wire measurements for improved accuracy, around my organization. The 3-wire systems used to exploit a somewhat clever trick in constructing bridges which almost cancels lead resistance errors, but now they just use separate microprocessor operations to estimate lead resistance and subtract it from the total. I see them as an anachronism though they work pretty well IF you’re careful with wiring and the connectors all stay clean. The seeming elegant perfection of 4-wire systems seems well worth running a little more copper (it’s amazing to learn that ethernet wiring only uses half the conductors, so how bad can adding a 4th be?). Where I get stuck is that so many of the conveniences of 3 wire systems arise from standardisation on the connectors, which turn up all over the place in practical systems - but how big a job is it going to be to make that happen in 4-wire systems?
Omega sells several of their 4-wire RTD assemblies and meters with 4-pin “audio” connectors (I think their part number is something like F4A or TF4M or something). Maybe that’s a start. But they look a little fiddly for an industrial setting. I’ve thought about “dual thermocouple” connectors but they don’t appear on any premade RTD setups. Yes, PRTs come with 4 separate single-wire connectors, which is typically fine, but we wouldn’t do that on a big machine.
Napier:
I didn’t even know you could still buy 3-wire IPRTs. I was under the assumption that all currently-manufactured IPRTs were 4-wire. Which, as you have noted, is the best set up.
Have you checked out Lemo connectors? They’re pretty small, and of very high quality. And then there’s the old standby… military-styled Amphenol connectors. They’re big and ugly, but very high performance.
BTW… are you designing the signal-conditioning portion of the system? And what kind of accuracy are you looking for? The reason I’m asking… if you want to squeeze every last drop of accuracy out of your IPRT, you must design the IPRT’s signal-conditioning circuit to compensate for thermoelectric errors. There are three ways to do this:
Each technique has its pros and cons, naturally. If you want more info on these, just ask. If you *don’t * need to squeeze every last drop of accuracy out of your IPRT, then don’t worry about compensating for thermoelectric errors… just measure the resistance and call it a day. 
Of course, you’ll want to be careful to not introduce too much self-heating error, so keep the excitation current under 2 mA or so. (I’m assuming here that you’re using a 100 Ω IPRT.)
Oh, and one more thing: will the user be connecting/disconnecting the IPRT on a regular basis? Or is it likely the connector will stay mating for a long time? If the latter is the case, I would recommend gold-plated contacts. If the former is the case, regular 'ol nickel plated (or whatever) contacts are O.K., as the regular wiping action will keep the contacts free of oxide.
Crafter_Man, you never fail to please. In particular I think I recognize in your name choice a healthy cynicism about the inclusion of space characters in computer names.
So, whether you can buy 3-wire IPRTs depends on where you draw the line between IPRT and RTD. I find this important semantic issue somewhat inscrutable. You can’t buy highly accurate 3-wire sensors. I actually haven’t found manufacturers that label their products IPRTs though I’m not far down that particular road.
I think I’m trying to navigate from Omega RTD land, where equipment designers tend to congregate, towards PRT land but without losing the mechanically robust nature of RTDs in their wiring and connection hardware. Note that I understand the sensors themselves have robustness compromises with accuracy - hysteresis in particular - but the sensors are well protected and quiet; it’s the wiring and connectors I want to harden.
I think I also need better information about just how much accuracy we lose through 3-wire systems per se. I’m gonna guess it’s on the order of 0.05 or 0.1 C for clean new systems sitting on the benchtop, and maybe 0.3 or even sporadically 1 C or more for real systems with long wire bundles and infrequently remade connections that may go years without attention.
Yes, I know about thermoelectric effects. Premade products I buy use reversing DC. When I program my own systems I randomize the current over a range like 0.5 to 1 mA and reverse the voltage and do the appropriate renormalization and averaging. Standard RTD systems I see in use now do neither of these things.
I guess the battles I have are those of wanting to do better than run-of-the-mill RTD systems, to get total error budgets to 0.x C given months or a couple years between servicing. All in the 0 to 500 C range in otherwise pleasant and noncorrosive and quiet settings.
PS I use Lemo connectors on lab fluxmeters, and Jeeze they are small and delicate looking. Think they’d last long in a light industry factory (ie opposite of heavy industry) with maintenance men?
[QUOTE=Napier]
PS I use Lemo connectors on lab fluxmeters, and Jeeze they are small and delicate looking. Think they’d last long in a light industry factory (ie opposite of heavy industry) with maintenance men?
[/QUOTE]
Yea, they’re small, but impressively robust. I’ve had good luck with them. But there are a couple disadvantages to Lemo connectors:
They’re pretty expensive. Definitely not something you would use for a mass-produced product.
You need special tools to assemble them (unlike military-styled Amphenol connectors, which can be assembled using common tools).
Oops, I just saw your lengthy reply. A new reply is coming.
[QUOTE=Napier]
Crafter_Man, you never fail to please.
[/QUOTE]
That’s what the women say. 
[QUOTE=Napier]
So, whether you can buy 3-wire IPRTs depends on where you draw the line between IPRT and RTD. I find this important semantic issue somewhat inscrutable. You can’t buy highly accurate 3-wire sensors. I actually haven’t found manufacturers that label their products IPRTs though I’m not far down that particular road.
[/QUOTE]
A few things:
As can be seen, a 4-wire IPRTs is just one kind of RTD. But for whatever reason, everyone referes to a 4-wire IPRT simply as an RTD. While this is true, 4-wire IPRT is a more accurate description.
[QUOTE=Napier]
I think I’m trying to navigate from Omega RTD land, where equipment designers tend to congregate, towards PRT land but without losing the mechanically robust nature of RTDs in their wiring and connection hardware. Note that I understand the sensors themselves have robustness compromises with accuracy - hysteresis in particular - but the sensors are well protected and quiet; it’s the wiring and connectors I want to harden.
[/QUOTE]
Some RTDs are very mechanically robust (e.g. thermistors, film-type IPRTs), some are extremely delicate (e.g. SPRT), and some are in-between (e.g. IPRTs).
Here’s the deal: An IPRT is simply a platinum helix wrapped around a mica or ceramic insulator. Unlike SPRTs, an IPRT is not wrapped in a “strain-free” way. This allows the IPRT to survive a moderate amount of mechanical stress, but there is a price to pay: there is some hysteresis, and the repeatability is only so-so. But the uncertainty is still pretty impressive. If you thoroughly calibrate it, an IPRT can be assigned an uncertainty of around ± 0.05 °C over its entire operating range. But… the IPRT is *still * somewhat delicate. An IPRT *cannot * be smacked around or dropped on the floor, as doing so *will * throw R[sub]0[/sub] off.
We use SPRTs and IPRTs in our temperature metrology lab. Only two people are allowed to handle the SPRTs. We also have an IPRT that we take into the field for performing system temperature calibrations on certian things. To be safe, I instruct my field technician to handle the IPRT in much the same way she handles the SPRTs in the lab.
And you’re smart to think about wiring & connectors. They always seem to break. If you don’t care about size & aesthetics, look no further than military-style Amphenol connectors.
[QUOTE=Napier]
I think I also need better information about just how much accuracy we lose through 3-wire systems per se. I’m gonna guess it’s on the order of 0.05 or 0.1 C for clean new systems sitting on the benchtop, and maybe 0.3 or even sporadically 1 C or more for real systems with long wire bundles and infrequently remade connections that may go years without attention.
[/QUOTE]
It depends on a number of factors, but I would guess a 3-wire system adds an additional ± 0.3 °C or so. At any rate, if you want to use an IPRT, I would never even consider a 3-wire configuration. Go with a 4-wire configuration, and life will be much better.
[QUOTE=Napier]
Yes, I know about thermoelectric effects. Premade products I buy use reversing DC. When I program my own systems I randomize the current over a range like 0.5 to 1 mA and reverse the voltage and do the appropriate renormalization and averaging. Standard RTD systems I see in use now do neither of these things.
[/QUOTE]
Sounds good! And of course, you actively reverse the excitation current, and the voltage you’re measuring passively reverses. 
And if you’re going to the trouble of addressing thermoelectric effects, then you must use a 4-wire IPRT. Finally, if you want uber-accuracy, measure the resistance at two different currents, and then calculate the “zero current resistance” using a second-order curve-fit.
[QUOTE=Napier]
I guess the battles I have are those of wanting to do better than run-of-the-mill RTD systems, to get total error budgets to 0.x C given months or a couple years between servicing. All in the 0 to 500 C range in otherwise pleasant and noncorrosive and quiet settings.
[/QUOTE]
I have used IPRTs for years in our temperature metrology lab. Some things I have learned:
Treat the IPRT as if it were an SPRT. As an example, when you lay the IRPT on a table, there should not be a “tink” sound when its outer sheath touches the table. Yes, that is how I instruct my technicians to handle IPRTs! It is especially serious in the case of an SPRT. If one of my technicians lays an SPRT on a table, and I hear the sound of the SPRT’s stainless steel sheath touching the table, I immediately instruct them to stick the SPRT in the water triple point cell to see if the triple point resistance has shifted.
An IPRT must be calibrated at at least 5 points over its operating range at least once a year. The Callendar-Van Dusen equation is what I use to covert temperature to voltage. (I use the quartic solution to convert voltage to temperature for T < 0 °C.) In addition, I would suggest sticking the IPRT in an ice bath every now and then to see if its R[sub]0[/sub] has drifted.
An IRPT should be periodically annealed to relive molecular stresses in the platinum and to “boil off” impurities.
All very good, Crafter_Man.
I am using the terms RTD and _PRT in what I have picked up as a practical, colloquial sense. Both are metal resistance elements, typically Pt for RTD and always for _PRT. But there is usually an implied difference in the sense I’ve picked up. RTDs have their platinum rigidly bonded or held while _PRTs have theirs free, so RTDs have better impact resistance but worse hysteresis. RTDs all match a standard curve and are interchangeable (within tolerances), while _PRTs are always individually calibrated. RTDs always use Calander-vanDusen (and I always use them above 0 C) while _PRTs usually use ITS90. RTDs are referenced to 0C while _PRTs are referenced to TPW=0.01 C. RTDs come in 100 ohm and sometimes higher, even 1000 ohm. _PRTs come in 100 ohm and sometimes lower like 25 or 2.5 or even 0.25 ohm. RTDs are sometimes made with thin films and _PRTs never are. You buy RTDs from Omega or Watlow or Tempco and _PRTs from Hart or Tinsley. RTDs are measured with direct current without any treatment of the thermal emfs, while _PRTs always use various debated methods of dealing with it. RTDs cost 20 to 100 dollars while _PRTs cost 300 to 5000. And most RTDs are 3-wire while all _PRTs are 4-wire.
Or, at least, this is what I gather. But then this is what I’m trying to grow beyond. The current border of my experience does not extend into the transitional regime between the Omega and the Hart regimes. I guess I’ll go do a web search on “IPRT”, or maybe I’ll call Burns Engineering, because it looks like they have products there. I see little diagrams in the books that show, for instance, a coiled Pt wire that is pretty closely confined in holes through, or a spiral groove around, an alumina slug - this is supposed to be a transitional design between RTD and SPRT. But I don’t actually know anybody that makes it.
So. I know how to buy Standard or Secondary Reference PRTs and have a vendor calibrate them, I have a TPW cell and a chiller for using it, I have some nice PRTs that I send out for calibration, I can see a path toward being able to calibrate them in-house with tin and indium and zinc cells, I tinker with Vishay foil resistors and mercury-wetted reeds and computer programs. And I see people build a machine - they order RTDs with a 1 C tolerance and order controllers based on their features list and build the machine and move on. And I’m trying to picture how to build machines 3X or 10 X better. So where do I go to learn about a better grade of RTDs, and find people who supply 4-wire controllers?