RTD response sloooow


Thread Starter

Bruce Cincotta

We are having a lot of trouble getting good fast temperature response from a SS thermowell/RTD combination. We are using a rosemount Thermowell and transmitter with a spring loaded RTD. The process goes from ambient to 240 degrees quickly, but the RTD takes over a minute to respond to the change. The thermowell is well placed in a tee with adequate flow, and the liquid is just water.

Any thoughts on a solution would be appreciated.
Dear Bruce:
Before i make any conclusion, i think the best way for you is to list out what type of RTD you're using, whether it is pt100 or others because i believe there is selectable range for you to select. You can ask the Fisher Rosemount rep. for help too.
The one minute response time sounds a good guestimate. One thing mislead me in that one minute.
If you wait until the reading is 100% stabilised: this is not the response time. The graph is exponential. In that case , the response time is the time it takes to reach 63% of the full value.
However, you are concerned with faster response.
XTR has noting to do. Your problem is one of proper design in the implementation of the measurement itself.
If the loop is closed, put substantial amount of derivative, very litlle integral and like normal proportional. It looks a perfect case for using
Lead/Lag; you can cut the overshoot by adding
an override on the output. If the valve is equiped with a positionner, you may also replace it by just a 1/1 booster. Positionner delays the action. If the loop is a cascade, the overall tuning is of great influence ( in that forum, somewhere, you will find an excellent tutorial on tuning cascade).
Further question: what is the accuracy figure ?
If a ± 1 deg.C would be satisfatory, then you may consider using open type T/C. The type 'T' is just the best one from absolute zero up to 400 deg.C
Note: tables do not show it as low as absolute, but it is in fact (look NBS Monograph 125).
The 'Marlin Manufacturing Corporation;
12404 Triskett Road, Cleveland ,Ohio 44111'
specialised in temperature measuement for Research & Industry.
They were supplying standard grade T/C and special grade, as well as many other laboratory
The special grade Marlin is simply a selection.
For the type 'T' they select within ± .5 deg.C
An open type 'T' will respond in few seconds.
Certainly you will have to redesign the installation (I can help you).

Roughly, with the off the shelf special grade, the
overall accuracy of the istallation will be around
± 1 deg.C

Now: if you require better accuracy. I have a cat in my bag. This is personal but it might interest you and others in that forum.

I dedicated several thousands of hours in improving two aspects of NIST (formerly NBS) in the matter of thermocouples.
It consists in personalising the accuracy of thermocouples down to the International Temperature Scale.
The principle: first install the T/C (any grade)
the T/C c/w extension line. Second, laboratory portable bath for reproducing ITS 90 reference points.
From there: I calculate, instead of polynomial, a fractional approximation of my own. The fractional approximation saves several of these expensive bath. Then your accuracy is witin ± .01
In that way, the temperature point is personalised. It has its own set of coefficients. If no one can supply XTR with modifiable coefficients then you would have to consider reading directly into an analog/digital input and the software.
For more: [email protected]

Bruce Cincotta

We are using at 100 ohm Platinum RTD. The one minute lag I mentioned was before any significant change was seen. I would guess the time constant was closer to two minutes.

Brian P. Romano

Although Stainless Steel is great for corrosion resistance, it is a lousy conductor of heat. If you have a "beefy" thermowell, the response of the
RTD tip is delayed by the temperature lag of the thermowell heating up. Is it possible for you to change thermowell material top something like brass or bronze? We once had to make a custom brass thermowell for an oil heating process we worked on for just this reason.

Good Luck!

Steven Landau

1) Remove TEE and use compression fitting directly on the RTD.
2) Use special thin-wall Tee's, you can find them listed under "REDUCED TIP"

In the past I have used 1/8" SS RTD's and make my own thermowells out of closed end SS thin wall tubing, and had them brazed into a spool piece or elbow.

Steve Landau
VP Controls & Automation
92 Montvale Ave
Stoneham MA 01890

Office: 781-438-3337
Fax: 781-438-5297
Cell: 617-908-9232
e-mail: [email protected]
e-mail to pager: [email protected] (short messages

This could be due to a high integration time on the part of the signal conditioner, or excessive digital filtering further down the line.

Your process is going at 4 degrees per second, whilst many RTD signal conditioners will
happily integrate this down to less than 1 degree per second.
No mention if the well is tapered or stepped. Stepped is better for response, tapered is mechanically more robust.

Ron Sewell
Sigmatic Controls
Kelowna, B.C.

Curt Wuollet

Hi Bruce;

Assuming that the electronics are not that heavily damped, I would seek to increase the rate of heat transfer. A dab of thermally conductive grease ( heatsink compound from Radio Shack )in the right place to increase the transfer area might help a lot. This would be where the RTD element contacts the well.
Also, the change is exponential and this may be seen as slow when only a narrow range close to the ultimate temperature is examined. It actually
never gets there, achieveing thermal equalibrium where there is a drop in temperature corresponding to the thermal resistance between the process and the RTD at the small but finite heat flow out of the leads.
The job then, is to reduce that thermal resistance, hence the thermal grease. There's not much to be done to reduce thermal resistance between the process and the well except keeping it clean and the fluid turbulent around the well. If that doesn't fix it, perhaps a different sensor might have better coupling. I think the grease would equalize that.

Hope this helps


Curt Wuollet, Owner
Wide Open Technologies.
Hello Bruce
You may check up the following. Are you using some barrier in the circuit / This will give you around 50 ms delay . What is the
scan of your DCS / receiving system ? What is the screen update rate of the MMI ? Check also the really installation of the RTD insert within the well . Compare the response with a local
temperature Gauge and the RTD signal at receiving end . Best of luck !
Nilanjan , India

Thermowell is acting as heatsink, directly coupled to piping. Is it possible to use a much longer thermowell, (with corresponding longer "T")? Also is RTD coupled to end of thermowell with a good thermal compound?

Just ideas,

Tony Firth, Electrical Eng.,
Quester Technology Inc.,Fremont,CA
for hvac use i have had the same problem so i use "thermowells" which have the probe sticking into the water stream. the probe is glued with epoxy into the well so it is a one shot deal but the
response looks ok even when sampling at one second. this avoids the thermal inertia of the thermowell and also the thermofluid and is more accurate.


Bruce Durdle


1. Have you got good contact between the thermoprobe and the inside of the thermowell? I trust the dimensions of the probe and the overqll length of any extensions are matched? (A well-intentioned purchasing officer of my
acquaintance decided that we didn't really need 10" thermocouples when the wells were only 7", so changed the records so we got 7" replacement probes for 7" wells.) (For "well-intentioned" read "interfering", "I-can-do-your-job-better-than-you", or just plain "interfering idiot")

2. You might also try some thermally-conducting paste in the bottom of the well.

Having said that, I think you need to use a very ewell-designed well to get better than 1 minute time constant - I seem to remember reading a few years ago about some tests to optimise the well performance which got down to 20 seconds by turning down the well so it was as thin as possible in the vicinity of the tip to minimise the effective thermal inertia.



extended surface (finned) thermowells can be used to reduce the static and dynamic measurment errors.

Dave Bartran