MSP430i204x, MSP430i203x, MSP430i202x

Hello David,

I am inquiring members of our marketing team to find out about how to order the i20xx series devices and when that option will be available.

As for the design questions you pose, I have a few questions of my own:

• Both systems seem to be single-ended outputs, do you plan on grounding one of the inputs and operating the SD24 like a single ended ADC?
• Have you thought of using a thermistor conditioning circuit similar to figures 1 or 2 in this link: http://www.ni.com/white-paper/3643/en/? This would take advantage of the differential measurement inputs of the SD24, and may yield better results.
• For your thermopile design, will the contents in the red box be removed or still be there external to the MSP430?
• What accuracy over what range of temperature are you looking for?
• What voltage outputs do you expect across your thermistor/thermopile?

A large factor to getting the most accurate results will be ensuring that the output differential voltage of your signal conditioning circuit is well aligned with the 1.16V reference. You can use the internal PGA to help with this, but any use of the PGA will add noise to the system.

Answers to a few of these questions may help me figure out what you should expect in your results. Let me know if you have any other questions about the i20xx series.

Hi Tyler,

I don't have a very clear solution yet. I hope the measurement could meet the standard body temperature requirement, that is 0.2 degree Celsius accuracy between 35 to 42. I can use differential input if this can improve the measurement accurate. I hope the design to be simple, if a signal conditioner is required to achieve the measurement accuracy. I will use one.

for the red box, I don't know it is needed or not, I calculate it like this.
 the thermopile output signal is around 0V at 25 degree Celsius and 0.05mV/degree C. for 16 gain of the TI MCU. it will be 0.8mV/degree C.
ADC is 24 bits. assume I only use 20bits, for 1.16V reference, the resolution will be around 1uV. it is 1/800, should be good enough for my application. I am not sure my calculation working or not.

I don't want the red box in my design, but If my calculation couldn't provide the resolution, I will use it in my design.

for thermistor I can use a WHEATSTONE BRIDGE, or a signal condition circuit. so the voltage will be easy to control.

I am designing a consumer device, so the cost is sensitive. of cause for a measurement device, the accuracy is the most important.

if TI has a application note for a IR thermometer, that will be great for me to start my design.

i20xx may not be the best candidate, the reason I pick it because it is small,  has 24bit ADC and not expensive.

Sincerely.

David

David,

I have gotten word from our marketing team that once the distributors place orders for the devices, it will show up online. Usually this process takes 2-3 weeks. Our team has already sent out a stocking notification to the distributors, so this process should already be underway.

As for the design side, I can say that the i20xx series is a great selection for precision sensing and low cost point. You can expect 14 ENOB for AC signals, or close to 16 ENOB for DC signals from the SD24 module. This means you should easily meet the 0.2degC accuracy over your range.

Another option which you may consider for cutting costs would be to pick a MSP with integrated analog. The MSP430F2274 and its derivatives have 2 low-power integrated op amps which can be programmed for PGA implementations and internally connect to the 10-bit SAR ADC. 10 bits would still give you enough accuracy over the range you are looking for.

With this added information here are my thoughts:

• For the thermopile design, you can replace the red box with the internal PGA of the i20xx and gain the signal up to a level which should provide ample accuracy for your test case.
• For the thermistor design, I think you may want to implement a bridge/amplification circuit to achieve the best results. I have done some work with an RTD for an industrial sensor transmitter, which takes a bridge circuit and a differential amplifier to condition the signal. The circuit diagram is shown below. This design was done with the F2274 as stated previously, taking advantage one of the integrated op amps.

In the above diagram, R13 represents a PT-100 RTD, which has a nominal resistance of 100ohm at 0degC. R12 is an impedance matching resistor for the RTD, and R10/R11 are higher impedance matching resistors for a full bridge implementation. R8 and R9 can be scaled to adjust the sensitivity range of the sensor solution.

Let me know if this information helps, or if you have any questions.

A thermistor is a resistor with a large temperature coefficient.
If used as in your first image (I would swap thermistor and base resistor), it forms a simple voltage divider. The voltage on the thermistor changes (non-linearly) with the temperature.
If the thermistor is the base resistor, then the voltage goes from 0 (thermistor resistance 0) to VDO * Rt / (Rt+R5). You’ll need a pre-calculated and calibrated table (or a time-consuming formula) to transform the thermistor voltage into a temperature.
If instead a constant current is injected (there are e.g. some diodes with constant leakage current over a large voltage range), then the voltage on the thremistor changes linearly with its resistance.

A thermopile, however, is a voltage source. The voltage depends on temperature. In the second circuit, the resistor divider acts as offset voltage generator (so the OpAmp doesn’t have to operate near the negative rail, which would reduce precision, subtract offset voltage and increase operating current, if supported at all), the red box acts as amplifier and impedance changer. In theory, you can directly attach the thermopile to the two inputs of an SD16. However, you should still use the voltage divider to put the negative side on a defined voltage (or put it to GND directly and use the SD16 in single-ended mode)
SD16 input impedance is a problem, as the thermopile has 85K internal resistance. MSPs with input buffer on the SD have better (higher) input impedance. Or you have to live with it and count this in into your calculation (effectively a voltage divider, calibration needed)

Of course, Tyler's soluton with a bridge is the most elegant one, but calibrating the bridge is not trivial.

Hi Tyler

I prefer MSP430F2232IRHAT, because it is cheaper. I notice there is a chip ZSSC3218 from ZMDI, they call it signal conditioner. if TI has an application note or any kind of solution similar to ZSSC3218, that will be very helpful.

I don't not need it has supper fast speed, because it is for temperature measurement.

I hope I can find a economic solution from TI for consumer market.  it is very hard to compete with Chinese solution use local MCU.

thank you very much for your help.

David Sun

David,

I'm not familiar with all of TI's thousands of analog products, but I do know we have the TMP006 which is an IR Thermopile + 14-bit ADC which can then communicate to a generic MCU via I2C. The TMP006 will be accurate from 0 - 60C, so it should encapsulate your target range quite well with sufficient accuracy.

This part is quoted at $1.50/1ku, and would take the responsibility of the ZSSC3218 and your thermopile sensor all in one, which could save costs considerably.

From there, really any of our MSP430s would be a viable candidate for your overall controller in the system, especially the value line G2xx series MCUs.

Hi Tyler

thank you very much, I noticed TMP006 a long time ago, I was not very confident about the accuracy of TMP006. I will buy one evaluation kit and try to calibrate it, see how accurate it is. if it meet my design specification. that will be perfect match.

David Sun