Low coefficient NTC resistors?

 Sylvia Else wrote:
=================
>
> As I've indicated in my reply to Jan, I believe that the circuit in my 
> UPS is not providing anything like the required compensation. 

**As if a lying, autistic tenth wit like you would know . 


> the designer's remit was to provide some temperature compensation, so he 
> did, without caring whether it was the correct amount. 

**  Maybe he, she, it had brain tumor   -  big as yours. 



......  love,  Phil 

On Thu, 16 Dec 2021 01:20:00 -0500, Spehro Pefhany
<speffSNIP@interlogDOTyou.knowwhat> wrote:

>On Mon, 13 Dec 2021 22:40:09 -0800 (PST), Anthony William Sloman
><bill.sloman@ieee.org> wrote:
>
>>Back when I did one, back in 1979, you needed a lot of gain in that one op =
>>amp. I ended up going with four - it made the low pass filtering a lot easi=
>>er. My bosses got tetchy, so I had to draw the one, two and three op amp op=
>>tions. Electrotherm had a bulk deal on the uA715, so the one amplifier opti=
>>on wasn't all that expensive, but nothing else around at the time had that =
>>much gain. 
>
>Odd, RTDs have such high output even a single LM358 should be about
>good enough for *most* purposes unless you're running them at much
>lower than normal current or looking for off-label uK performance
>(which I remember  you were doing with thermistors). 
>
>At 0.5mA you get ~200uV/&#4294967295;C with a Pt100 ohm DIN RTD. Interchangability
>is maybe 1/3&#4294967295;C for inexpensive ones (at room temperature) and a degree
>or two at extremes, so for normal purposes- any modern op-amp assuming
>you don't mind trimming the offset. 
>
>Coincidentally, about the same output as you get measuring temperature
>with silicon BJT(s) operated at 10:1 current ratio. 
>
>>The linearising involved a smidgen of positive feedback which fr=
>>ightened the guy who took over the project, and nobody could make him see s=
>>ense.
>
>Shouldn't be too hard to show mathematically that the net feedback is
>strongly negative for all sensible RTD values. And that break
>protection is in the safe direction. 
>
>He probably would be terrified (with some justification) by an
>enhanced Howland current source. 
>
>The cool thing is that the same resistor that linearizes the RTD can
>be used to avoid having an active current source for excitation. Just
>a resistor from a reference voltage will do for excitation- there is
>no advantage to using a current source. 

I thought I invented that! One resistor to the + supply and one more
for positive feedback from the opamp output, RTD to ground.

I did invent a 3-wire version too. Can't remember that circuit now.
Nowadays we just use a mux'd differential-input delta-sigma ADC and
measure the RTD ratiometrically against a good resistor; don't need a
very good reference.


-- 

I yam what I yam - Popeye

On Friday, December 17, 2021 at 2:32:16 AM UTC+11, jla...@highlandsniptechnology.com wrote:
> On Thu, 16 Dec 2021 01:20:00 -0500, Spehro Pefhany 
> <spef...@interlogDOTyou.knowwhat> wrote: 
> 
> >On Mon, 13 Dec 2021 22:40:09 -0800 (PST), Anthony William Sloman 
> ><bill....@ieee.org> wrote: 
> > 
> >>Back when I did one, back in 1979, you needed a lot of gain in that one op = 
> >>amp. I ended up going with four - it made the low pass filtering a lot easi= 
> >>er. My bosses got tetchy, so I had to draw the one, two and three op amp op= 
> >>tions. Electrotherm had a bulk deal on the uA715, so the one amplifier opti= 
> >>on wasn't all that expensive, but nothing else around at the time had that = 
> >>much gain. 
> > 
> >Odd, RTDs have such high output even a single LM358 should be about 
> >good enough for *most* purposes unless you're running them at much 
> >lower than normal current or looking for off-label uK performance 
> >(which I remember you were doing with thermistors). 
> > 
> >At 0.5mA you get ~200uV/&deg;C with a Pt100 ohm DIN RTD. Interchangability 
> >is maybe 1/3&deg;C for inexpensive ones (at room temperature) and a degree 
> >or two at extremes, so for normal purposes- any modern op-amp assuming 
> >you don't mind trimming the offset. 
> > 
> >Coincidentally, about the same output as you get measuring temperature 
> >with silicon BJT(s) operated at 10:1 current ratio. 
> > 
> >>The linearising involved a smidgen of positive feedback which fr= 
> >>ightened the guy who took over the project, and nobody could make him see s= 
> >>ense. 
> > 
> >Shouldn't be too hard to show mathematically that the net feedback is 
> >strongly negative for all sensible RTD values. And that break 
> >protection is in the safe direction. 
> > 
> >He probably would be terrified (with some justification) by an 
> >enhanced Howland current source. 
> > 
> >The cool thing is that the same resistor that linearizes the RTD can 
> >be used to avoid having an active current source for excitation. Just 
> >a resistor from a reference voltage will do for excitation- there is 
> >no advantage to using a current source.
>
> I thought I invented that! One resistor to the + supply and one more 
> for positive feedback from the opamp output, RTD to ground. 

When? I didn't think that what I was doing in 1979 was any kind of invention, and I'm fairly sure that Honeywell Pt-resistance sensor option that came out within a year or two used the same trick - not that you can entireily rely on what you read in the trade literature.

> I did invent a 3-wire version too. Can't remember that circuit now. 
> Nowadays we just use a mux'd differential-input delta-sigma ADC and 
> measure the RTD ratiometrically against a good resistor; don't need a 
> very good reference.

I had to point out - to the software guy - that we were using the same voltage reference for the A/D converter and the bridge  excitation in the circuit that I publlshed in 1996. 

-- 
Bill Sloman, Sydney