Ni1000 Accuracy

[ScumPunk]

Hi, All,
I'm after some background info regarding HVAC applications.
We're working on an analogue input circuit for Ni1k sensors, and I'd like some background info such as:
1. Typical applications - building heating & air con, ovens, etc.
2. Typical temperature ranges - ties in with the above
3. Expected accuracy - I've been given +/-0.5K, which seems to be the sensor accuracy, but given the application, what sort of accuracy would be required?
Thanks in advance,
Mort.

 

[itsmoked]

+/-0.5K

Does not seem like any sensor accuracy to me.

Typical accuracy for all the stuff you mention is 1/2[º]F.

Typically you would aim for 1/4[º]F and be happy with 1/2[º]F.

Most of the listed devices are only user adjustable to 1[º]F. Most refrigeration and gas based heating systems cannot or should not cycle very frequently so they don't like temp control much finer anyway. Cooking food does not require better than 5[º]F and this makes the relays cycle much less and so live longer.

As for Ni1000.. please be more specific.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ozmosis]

Mort
Applications: HVAC typically measuring temperature from -(minus)50 to +150degC. Used in return temperature measurement for cooling towers, HW and CW pump systems and room temperature feedback.

Temperature at 1000ohms is 0DegC.
In a VFD, Siemens have it in their SED2, Danfoss have it in their FC102 as a feedback taking the actual resistance input and auto scaling it as a temperature feedback.
Don't worry too much about accuracy, this is HVAC.

 

[Skogsgurra]

+/-0.5 K is very close to +/-0.25 F why don't you like that?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ykee]

I think you mean +/-0.5K is close to +/-1F?

 

[Skogsgurra]

Yes. Not easy to get that right. :-(

But a decent temperature tolerance nonetheless.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ykee]

I would ahve to agree with you, HVAC is usually pretty cost sensitive, and since most settings are hidden from occupants, or are relative, this tolerance seems adequate.

 

[ScumPunk]

Hi, all,
thanks for the information, it mostly confirms what we thought. But, given that we are in a lab and rarely (if ever) get out in to the field, we wanted to get some more opinions.
ykee: your comment:
"and since most settings are hidden from occupants, or are relative, this tolerance seems adequate."
hits the nail on the head, I feel. Given that your other comment regarding cost is a major factor in this design, this is the main problem. 1 count on our 12-bit ADC = 0.8mV = 0.098K, so trying to hit +/-0.5K is proving difficult given the following other conditions:
1. It's an electrically noisy environment;
2. The circuit has an operating range of -10 to +80 degC
Personally, I think the requirement is unrealistic without using more precise (expensive) hardware, the best we've achieved so far is +/-0.7K :-(
Keith: When you say "be more specific", are you referring to the different versions of Ni1k (class A, B) etc? From the documents I've seen, they all have the same response, just different tolerances. If you are referring to something else, please explain, this is the kind of information I'm trying to find, but no-one here has this kind of experience.
Thanks,
Mort

 

[Skogsgurra]

Are other measurement techniques ruled out?

I have been using thermistors in HVAC applications. They are cheap. They don't even need an AD (we used RC time constant measurement in the MCU, the cheapest PIC we could find) and they are small.

This solution is used in a very successful range of ventilators and no one seems to bother about accuracy. If it is hot, they just turn it down a bit. And vice versa.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ScumPunk]

Gunnar, Your description of the sensor use in these applications (If its hot, turn it down), is exactly how I see it being used. Unfortunately, we've been tied down to this spec (it seems for now), so we have to try and meet it. Unfortunately there isn't another measuring method we can use, as it's to go in to a new HVAC drive, and this is one of the features the marketing people want in it.
Thanks for the input,
Mort

 

[Skogsgurra]

Marketing is always right ;-)

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[roydm]

Both Johnson Controls & Siemens rate their accuracy as +/- 0.1%
The resistance chance per is 3 Ohm/ Deg F
So that works out +/- 0.33 Deg F
I think they are often wired with just 2 wires. The error due to change in lead resistance will be quite insignificant relative to the change in sensor resistance.
Roy

 

[itsmoked]

Yikes!cYou are entering the most competitive field in industry. There are literally thousands of temp controllers.

The commercial refrigeration field almost never uses three/four wire devices. They really go for two wire.

osmosis' -50 to +150degC is pretty much it.

Your Ni has a higher temp coeff which helps.

The the most coveted of industrial temp controllers was the GoldLine made by a company that also made swimming pool controls and actually grew from that field outward. Unfortunately that company stumbled over digital pool chlorination and doing it by just adding salt to their gizmo which cracks it into chlorine. They can then creep the Cl into the pool 24/7 instead of the normal, big blasts of it, causing wild chemistry swings. In their infinite wisdom they dropped their comprehensive temp control line completely, and nearly instantly.

Replacing it as the simple, most inexpensive biggest bang for the buck digital controller for temp, in the industry is the 'Ranco'.

http://www.grainger.com/Grainger/wwg/catalogPageView.shtml?CatPage=3801

It's nicely done. (I just hacked 4 of them for a job) You should pick one up and dissect it. It uses the industry favorite - the thermistor.

Thermistors limit the maximum temp and are the main reason designers head to RTDs.

With your RTD you can easily do 1/4F but no one will care and you immediately add display costs doing it, as you then need a decimal and another place.

Most people in industry/commercial realize that any space bigger than a domestic refrigerator will not be one uniform temperature. Corners may be warmer, there may be some stratification, warmer areas, colder where the evaporator outlet is, etc. So sub degree control is pretty pointless. It's also is much harder on all the equipment.

Hence the 1F or 1C most commonly seen.

Whatever you do, you'll probably want a factory calibration, as the Ni1000 are out of the box about +/-2 degrees. You will probably want to allow calibration by the installer too. The typical situation is the unit is installed, run to setpoint, then a hand held thermometer is put in the most rational location. The thermostat display is then adjusted match the thermometer. This takes into account everything previously mentioned about cooled space inhomogeneity.

If you want a more accurate assessment you'd need to put your proposed controller into a more specific field for us. Like wine cooling, or blast freezing, etc.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ozmosis]

Mort
If you are working on the project that myself and Michael Uhlig put together 3 yrs ago, then the definition of the NI1000 should be clearly outlined in the SBT Signal norm spec attached in the lahstenheft for the 230 CU.

 

[Skogsgurra]

Nicht Pflichtenheft?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...