Precise and automatic sample temperature control The key for biomedical and material science applications

Size: px

Start display at page:

Download "Precise and automatic sample temperature control The key for biomedical and material science applications"

May Carr
6 years ago
Views:

The key for biomedical and material science applications Application Note #02 Many mes, the goal of a magne c hea ng experiment is not to study the hea ng capaci es of the sample material or the

1 The key for biomedical and material science applications Application Note #02 Many mes, the goal of a magne c hea ng experiment is not to study the hea ng capaci es of the sample material or the specimen under test, but to reach a certain temperature and maintain or modify this temperature according to previously designed requirements. That would be the case of an experiment with cells loaded with MNPs in which the researcher is interested in obtaining a stable hypethermia temperature (i.e. 42ºC) for a certain me, in order to assess, for example, the cell death rate in that condi on. Another example would be an experiment in which some chemical reac on or physical transi on is expected to happen in a certain temperature and that temperature is expected to be obtained by magne c hea ng. The temperature control func onality (embedded in DM100 Series s so ware, MaNIaC 2.0) allows the se ng of one or more temperature set-points during the experiment, and through an advanced feedback algorithm it controls the applied field for that goal. This feature is based on a PI (propor onal-integral) algorithm that allows precise control of the output temperature. A fine tuning of the control response (velocity, overshoo ng and steady state error) can be easily achieved by simple manipula on of the propor onal (P) and the integral (I) terms. This Applica on Note shows you how to quickly, safely and efec vely take advantage of this tool. Made in collabora on with The Ins tute of Nanscience of Aragón 1/5

2 Temperature Control adjustment. MANIAC V2.0 features a propor onal-integral (PI) controller in order to drive the Temperature Control func onality. It is mainly a feedback control mechanism that figures how intense the applied field should be in order to reach the temperature setpoint, by making a calcula on that starts from the value of the error, here defined as the difference between the temperature measured at the sample and the temperature setpoint programmed by the user. The generic PI controller algorithm is defined by the following func on : In the following example, the experiment was set to reach and maintain a temperature of 35 ⁰C. A er a short and small overshoot of <0.2 ºC (<0,1 %), full stabiliza on is achieved within the next 120 seconds. Where: u(t): controller otput (B(t)) e(t): error Kp : propor onal gain Ki : integral gain Le axis (red line): sample temperature vs me. Right axis ( blue line): applied field vs. me MANIAC V2.0 introduces a varia on respecting to the general expression: the Ki parameter is replaced by the Ti parameter, which is related to the scale of the integral s me variable, being Ki = Kp/Ti. If you want to use the Temperature Control funcionality, you just have to choose one of the temperature sensors in the Control window. This op on is also available when the Infrared Imaging System (IR1) is installed (DM2/3), so the infrared data can be also used as a control variable: You could remotely sense the surface temperature of your sample and use that informa on to control the experiment. In a single run, your sample can be fully evaluated at different target temperatures. In the example above you can see a one hour experiment, with three different setpoints, which mimics the typical culture cell temperature (37 ⁰C), the hyperthermia cell death temperature (42ºC), and finally thermoabla on temperature (45º). The stabiliza on me was always be er than 3 min for each pulse. In order to obtain the best performance of from the PI controller, some mes the propor onal and integral parameters have to be manually adjusted. In the next sec ons you will find some ps to get a stable control ac on and a smooth temperature curve. 2/5

looking for a fast transient from the ini al 25º to the 50º setpoint. This led to an important overshoo ng and later oscilla on.

3 Influence of the propor onal term. The propor onal term basically tells the algorithm how fast it has to react to the current value of the error. In MaNIaC, this term is represented by the nondimensional variable Kc and can be set within a range of 50 to 300 in the Kc configura on screen: In the first case, a high value of Kc was set (290), looking for a fast transient from the ini al 25º to the 50º setpoint. This led to an important overshoo ng and later oscilla on. In order to avoid that, in the second example Kc has ben reduced to 100. obtaining a be er behavior: Ideally, we would like our control to achive the setpoint as fast as possible, so we are induced to set high values for Kc. But choosing a value that is too large can lead to big overshoo ng and/or no ceable oscilla on, yielding undesired overhea ng of the samples. In the other hand, a value of Kc that is too small will lead to very long transients. When op mal Kc values are chosen, short stabiliza on ( 1) mes (t < 3 min) and small overshoo ng are obtained. In the following examples, the experiments were performed by fixing the frequency at 580 khz and modifying Kc. The temperature setpoint is 50º. Three different Kc have been switched to check for the different responses. A sample with a high hea ng rate was used. Overshoo ng could be reduced and even eliminated just by reducing Kc even more, but that would lead to a no ceable steady state error. When the error gets too small, the propor onal term does not act anymore, so when the temperature curve is so ly aproaching the setpoint value, it se les in a final value that is less than the setpoint. To avoid this error and get fast transients at the same me, there is the Integral term. (1): It is absolutely necessary to keep in mind that the actual response me of any magne ng hea ng experiment is fundamentally dependent on how responsive the sample itself is. These experiments have been done using Magno, nb s standard colloid and, of course, the results obtained by the user will vary according to what sample is under study, and how stable and how linear its response is. 3/5

4 Influence of the integral term. In the examples shown in the previous sec on, the experiments were run while keeping a small constant value for the Ti parameter in order to enhance and isolate the effects of the varia ons of the Kc parameter. But something that is not evident from those examples is that when an automa c control is based only in a propor onal ac on, it is not possible to adjust the feedback loop to avoid both oscilla on and steady state error at the same me. For that, the integral control is included. This term will contribute with an ac on that is propor onal to the integral of the error from the beggining of the experiment. The u lity of the integral term is evident when trying to avoid the steady state error: even when the propor onal term is no longer sensi ve to the small error in the final part of the experiment, the integral term will keep increasing its contribu on and then helping the system to reach its setpoint with a negligible steady state error. In MaNIaC 2.0, the integral ac on is controlled by the Ti parameter, defined as Ti = Kp/Ki. From this defini on, the integral ac on will be equal to the propor onal ac on but trimmed by Ti. The following examples were run with a less responsive sample and a constant Kc value in order to let Ti s effect show. In the first example, a small Ti value allows the integral term to greatly contribute to the global control ac on, and then, even when the propor onal ac on stops, the integral ac on drives the curve to an overshoo ng period before stabilizing on the final value. In the second example, a smooth temperature curve is obtained by adjus ng Ti to a value of 250, that is limi ng the integral ac on down to a adequate level. In MaNIaC, Ti can be set within a range of 50 to 300 in the Ti configura on screen: Another great advantage of the integral ac on is that it helps the experiment to run with much smoother field varia ons, avoiding the turn on-turn on effect that is typical of a propor onal only control algorithm, which can be appreciated in the blue lines of the examples in the previous sec on. 4/5

... summarizing MaNIaC running on any DM100 system enables accurated control of the temperature of the sample or specimen in a

The Temperature Control func onality is available for the three DM100 applicators (DM1, DM2, and DM3), and can be run with op

modes standard fiber op c probe DMC1 Controller Standard 1,5 kwa DM100 Series controller So ware MaNIaC V2.0. For a closer look to DM1 go to h p://www.

5 ... summarizing MaNIaC running on any DM100 system enables accurated control of the temperature of the sample or specimen in a wide range of experimental setups, while reducing temperature transients at the same me. The Temperature Control func onality is available for the three DM100 applicators (DM1, DM2, and DM3), and can be run with op onal temperatures sensors like the local invivo sensors or the Infrared Imaging System (IR1) designed for DM2 and DM3....see it on Applica on Note Nº3. Gear The following devices were employed for the experimental procedures described in this document: DM1 Applicator 8 frequency modes standard fiber op c probe DMC1 Controller Standard 1,5 kwa DM100 Series controller So ware MaNIaC V2.0. For a closer look to DM1 go to h p:// For a closer look on MaNIaC, go to h p:// ware.html Wri en by Lic Beatriz Sanz Sague (INA) Eng Antonio Bermejo (nb) Prof Gerardo F Goya Rosse (INA) Eng Nicolás Cassinelli (nb) 5/5

CSE 3215 Embedded Systems Laboratory Lab 5 Digital Control System

Introduction CSE 3215 Embedded Systems Laboratory Lab 5 Digital Control System The purpose of this lab is to introduce you to digital control systems. The most basic function of a control system is to