Anyone have an estimated emissivity for the IVAC 9000 radiators?

[Fraser]

_Wim_

Great work, thank you 

Fraser

[globoy]

Thanks for all that, _Wim_.  An easier start than trying to replace the controller.  As today was a very snowy day I hacked on my unit.  Mostly I followed your lead but since I didn't have an 8V PSU, I modified the IVAC unit and I played a little with the cold side.

I adjusted the PSU to 6V out by including a 10k resistor in parallel with a 3.3k part (R21) in the feedback path and adjusting the trim pot.  Above about 6.1V the PSU crowbar fires.  I suspect that could also be adjusted up but didn't pursue that today.  I put a 1N4001 diode in series with the branch going to the logic circuits (in series with FB4) to drop the voltage to a legal level.

I was able to achieve 50C on the hot side with no problem.

I also put the largest value multi-turn trim pot I had (20k) in parallel with the NTC for the low side.  Unfortunately it was too low of a resistance value and the unit could not achieve a temperature lock.  I then tried some discrete resistors and had success with a 47k resistor, although that took a long time to reach stability.  I think the temperature was around 19C (measured 19.2C with my Agilent temperature sensor stuffed in a screw hole on the cold-side block).

I'll try again with a 100k trim-pot and try to post some plots.  I'm still trying to figure out the best way for me to have a precise temperature measurement as the goal of this project is a calibrator my the camera I designed.

[globoy]

And pics of the display board (with wire extensions) and cold side measurement.

[_Wim_]

Thanks for all that, _Wim_.  An easier start than trying to replace the controller.  As today was a very snowy day I hacked on my unit.  Mostly I followed your lead but since I didn't have an 8V PSU, I modified the IVAC unit and I played a little with the cold side.

I adjusted the PSU to 6V out by including a 10k resistor in parallel with a 3.3k part (R21) in the feedback path and adjusting the trim pot.  Above about 6.1V the PSU crowbar fires.  I suspect that could also be adjusted up but didn't pursue that today.  I put a 1N4001 diode in series with the branch going to the logic circuits (in series with FB4) to drop the voltage to a legal level.

I was able to achieve 50C on the hot side with no problem.

I also put the largest value multi-turn trim pot I had (20k) in parallel with the NTC for the low side.  Unfortunately it was too low of a resistance value and the unit could not achieve a temperature lock.  I then tried some discrete resistors and had success with a 47k resistor, although that took a long time to reach stability.  I think the temperature was around 19C (measured 19.2C with my Agilent temperature sensor stuffed in a screw hole on the cold-side block).

Thanks for posting your tests also. I wondered how good cooling would be, but I kinda expected cooling down would only be minimal. I think cooling was only provided to achieve 26°C in hot environments, and potentially also to regulate out small overshoots in temperature during heating up to achieve a faster lock.

I'll try again with a 100k trim-pot and try to post some plots.  I'm still trying to figure out the best way for me to have a precise temperature measurement as the goal of this project is a calibrator my the camera I designed.

Accurate temperature calibration is hard. For this purpose I would expect a small pt100 connected to a high res dmm would be the most accurate (because they have a "large" contact surface and do not need a reference junction). I bought some small ones on Ali-express (https://www.aliexpress.com/item/32689032460.html?spm=a2g0s.9042311.0.0.27424c4dYv6HSk) to build a laser power meter, but have not done anything with them yet. These would fit the side entry of the black body normally. 

I would fill the the hole on the black body with some thermal compound so the pt100 is completely submerged in paste. This should ensure a good coupling (there will be a lag in measured temperature, but once stable, the results should be fairly accurate). The hard part is the accuracy of the PT100 itself. Maybe this weekend I will put one in boiling water and melting ice, and then compare it with the results of the flir. This should hopefully be +-1°C accurate.

[Fraser]

Remember that Black Body units are calibrated either as ‘Actual’ or ‘Measured IR’ temperature. ‘Actual’ uses a surface contact sensor on the emission plate and ‘Measured IR’ uses an accurate IR Radiometer that is viewing the emission plate.

‘Actual’ calibration requires knowledge of the emission plate emissivity

‘Measured IR’ calibration takes the emissivity of the emission plate into account as part of the calibration process as the Radiometer views the emission plate. ‘Measured IR’ calibration is liable to the measurement tolerance of the Radiometer so ‘Actual’ calibration using a very accurate contact sensor can be more accurate if the emissivity of the emission plate is well documented and tested.

Fraser

[Fraser]

If you work with a true surface emissivity of 0.95, you will not be far off but tests would reveal the true emissivity of the black paint coating on the emission plate.

Fraser

[_Wim_]

Remember that Black Body units are calibrated either as ‘Actual’ or ‘Measured IR’ temperature. ‘Actual’ uses a surface contact sensor on the emission plate and ‘Measured IR’ uses an accurate IR Radiometer that is viewing the emission plate.

‘Actual’ calibration requires knowledge of the emission plate emissivity

‘Measured IR’ calibration takes the emissivity of the emission plate into account as part of the calibration process as the Radiometer views the emission plate. ‘Measured IR’ calibration is liable to the measurement tolerance of the Radiometer so ‘Actual’ calibration using a very accurate contact sensor can be more accurate if the emissivity of the emission plate is well documented and tested.

Fraser

Good point!

These were indeed adjusted to give a reading of 38°C to simulate an in ear temperature which has emissivity of almost 1 (0.9988 according to the paper I posted in the 3rd post of this thread), and this was confirmed by the initial measurements with my flir Exx (38.1°C with its emissivity set to 0.99), and also when I tested our infrared fever thermometer we have for the kids (showed exactly 38.0°C  )

But the paint probably on the black body will have a lower emissivity, so the actual temperature of black body must be higher to achieve this result. According to the same paper the emissivity of such a calibrator was 0.972, so the actual black body temperature should be 38°C/0.972 =  39.1°C. This small difference is already difficult to measure accurately, but it is indeed best to know what the theoretical value should be.




 

[_Wim_]

Good point!

These were indeed adjusted to give a reading of 38°C to simulate an in ear temperature which has emissivity of almost 1 (0.9988 according to the paper I posted in the 3rd post of this thread), and this was confirmed by the initial measurements with my flir Exx (38.1°C with its emissivity set to 0.99), and also when I tested our infrared fever thermometer we have for the kids (showed exactly 38.0°C  )

But the paint probably on the black body will have a lower emissivity, so the actual temperature of black body must be higher to achieve this result. According to the same paper the emissivity of such a calibrator was 0.972, so the actual black body temperature should be 38°C/0.972 =  39.1°C. This small difference is already difficult to measure accurately, but it is indeed best to know what the theoretical value should be.

From the above, we could also do the reverse calculation. These are probably still perfectly in spec at 38°C, so when measuring the actual temperature, we could calculate the emissivity!

[_Wim_]

I just reread the first post from globoy, he measured an actual temperature that was lower than 38°C. This I would not expect, or do I have some flaw in my thinking above? 

[_Wim_]

If you work with a true surface emissivity of 0.95, you will not be far off but tests would reveal the true emissivity of the black paint coating on the emission plate.

Fraser

I should of read that post of you too 

Thanks for already confirming my logic BEFORE I asked the question!

[globoy]

I wouldn't trust my temperature measurements to date.  It's also possible my unit is not accurate anymore.

I have taken your advice and will try to incorporate two precision RTDs into the radiators (with thermal grease).  I hope to try with one tonight.

From the instructions on the top of the unit it seems that the in-ear IR thermometers were meant to see the 26/38°C values so assuming the paint could not match the 0.9988 emissivity of the inner ear then your assumption must be right.  The block temperatures must be slightly higher.  I guess measuring the radiator temperature precisely would allow deducing the actual emissivity of the paint.

[_Wim_]

Accurate temperature calibration is hard. For this purpose I would expect a small pt100 connected to a high res dmm would be the most accurate (because they have a "large" contact surface and do not need a reference junction). I bought some small ones on Ali-express (https://www.aliexpress.com/item/32689032460.html?spm=a2g0s.9042311.0.0.27424c4dYv6HSk) to build a laser power meter, but have not done anything with them yet. These would fit the side entry of the black body normally. 

I would fill the the hole on the black body with some thermal compound so the pt100 is completely submerged in paste. This should ensure a good coupling (there will be a lag in measured temperature, but once stable, the results should be fairly accurate). The hard part is the accuracy of the PT100 itself. Maybe this weekend I will put one in boiling water and melting ice, and then compare it with the results of the flir. This should hopefully be +-1°C accurate.

First step, test the above PT100's for accuracy with boiling water and ice. Result are quite accurate! 

The boiling water fluctuated between 138.50 and 138.55 ohm, the melting ice was quite stable between 100.03 and 100.04 ohm. Meter was zero checked with parallel short banana cable.

 

[_Wim_]

Ok, and now for the actual test...

FLIR indicates 38.1 °C with emissivity set to 0.99.PT100 Resistance is 114.72 ohm and very stable (very occasionally going to 114.71, but 99% of the time at 114.72 ohm).

According to this PT100 calculator (https://www.peaksensors.co.uk/resources/rtd-calculator-temperature-resistance/) this equates to 37.9°C

Verified also again with our in ear thermometer, was again bang on 38.0°C. If I put the FLIR emissivity to 1.00, it also indicates 37.9°C

This would mean the emissivity of this black body must be quite high, possibly >0.98, although I find that hard to believe looking at available high emissivity paints and materials. The only thing that comes close is vacuum deposited black coating (https://www.acktar.com/products-services/high-emissivity-materials/)

I am on the limit here of what I can accurately measure, and also have no intention to invest in "temperature nut" equipment 

[Fraser]

Very interesting results 

Emissivity of 0.98 is very achievable so it could be a high performance coating 

These were for medical use and, as we know, the cost of such kit is often high. This may have justified using high performance emission plates and coatings. These are certainly very nice bits of kit 

No need to go down the ‘temperature nut’ path  The temperature measurement tolerance of a typical thermal camera is enough to make that path unnecessary.

Thank you for spending the time doing the tests  Much appreciated as my three units are sat doing nothing under my desk. It may be time to make them useful.

Fraser

[_Wim_]

Same test at above, but now with the series resistor installed again so the black body radiator regulates to 60°C. FLIR emissivity set to 0.99.

Resistance reading is 123.38ohm which equates to 60.3°C. Flir reads 60.1°C

[_Wim_]

These were for medical use and, as we know, the cost of such kit is often high. This may have justified using high performance emission plates and coatings. These are certainly very nice bits of kit 

These are indeed very nicely made. And for prices between 15 and 20$ on the US Ebay, this is an extremely good deal if you can use it.

 Glad I bought one to play around with. I also had lot's of fun doing these tests.

[globoy]

Great, _Wim_!  Thanks for posting your results.  My order for PT100s was held up but I am looking forward to again following in your footsteps. 

[globoy]

I finally got my temperature sensors and had a chance to instrument my unit.

I measure the low side to be 26°C as well.  The low side is very stable and I see a little drift after the unit has dialed in the temperature.  Interestingly while the high side heats up I've seen it overshoot on occasion.  But then it settles back down.

At 60°C my high side series resistor was about 3303 ohms (2-wire measurement). Slightly more drift on the high side that I attribute to the resistor's temperature coefficient (for example from night to day in my lab).  At 6V in I couldn't quite get the high side to 65°C but could get it to 64°C.

I think this unit is now good enough to run some experiments.  Thank you again, Fraser and _Wim_, for all your work.  This device can become a useful piece of kit.

On the chart below, the X-axis is seconds.

[_Wim_]

I finally got my temperature sensors and had a chance to instrument my unit.

I measure the low side to be 26°C as well.  The low side is very stable and I see a little drift after the unit has dialed in the temperature.  Interestingly while the high side heats up I've seen it overshoot on occasion.  But then it settles back down.

At 60°C my high side series resistor was about 3303 ohms (2-wire measurement). Slightly more drift on the high side that I attribute to the resistor's temperature coefficient (for example from night to day in my lab).  At 6V in I couldn't quite get the high side to 65°C but could get it to 64°C.

I think this unit is now good enough to run some experiments.  Thank you again, Fraser and _Wim_, for all your work.  This device can become a useful piece of kit.

On the chart below, the X-axis is seconds.

Looks very good! 

[Fraser]

Globoy,

Nice work 

That is a neat dual RTD thermometer you have produced. Any chance of sharing the design ?

Fraser

[globoy]

Thank you Fraser, and sure.   It's a pretty simple build around an ESP32, a pair of MAX31865 ICs and PT100 RTDs in a 4-wire configuration and a 2x8 character LCD module.  I used a random ESP32 board I had laying around but I think almost any ESP32 dev board will work fine (see notes below).  Code was done in Arduino with the ESP32 add-on.  It has a very simple serial interface available for logging temperature data on both the built-in HW serial and a Bluetooth SPP serial.  See below for zip files with the sketch and the version of the Adafruit MAX31865 library I used.  Also see below for a quick hand drawn schematic.

Parts
1. I used a ESP32 WROOM module on a breadboard from circuitsetup.us (https://circuitsetup.us/index.php/product/solderable-breadboard-esp32-esp8266-esp01-module-breakout-with-3-94-x-2-375-x-1-875-100x60x25mm-project-box-2-pack/).   This requires an external FTDI cable pinout serial interface (and the corresponding manual pressing of IO0, press/release of RST, release of IO0 to enter bootloader mode).  But any ESP32 board with the appropriate IO (see the sketch) will work.

2. A pair of Adafruit MAX31865 boards I got from Mouser (https://www.mouser.com/ProductDetail/485-3328).  They're ready for 4-wire connections.  Just have to add the connectors and pin-header.

3. A pair of Hareaus 32208550 RTD sensors I also got from Mouser (https://www.mouser.com/ProductDetail/956-32208550).  They are connected up with 4 wires (2 wires per lead) - heat shrink tubing around the solder joints and covered with thermal paste when inserted into the extra screw holes on top of the IVAC 9000 forward-facing aluminum blocks (next to its own temperature sensor).

4. A 2 line, 8 character LCD module I had laying around (Hantronix HDM08216L-3). But any HD44780 type display should work fine.  One might have to adjust the calls to lcd.begin() and lcd.setCursor methods in the sketch to match their display.

5. A couple of LDO regulators (whole system takes less than 100 mA).  Either two 5V units or one 5V unit and 3.3V unit (see below).
 I connected the input to the IVAC power supply on the main PCB (outputs of the ferrite beads going to the digital logic) and my IVAC power supply has been adjusted up to 6V.

6. Misc parts.  Some 100 uF electrolytic, 1 and 0.1 uF ceramic caps, 10K pot, 10K resistor, resistor selected for the LCD module backlight, buttons as necessary, etc.

Power Supply Considerations
My circuit.us board requires a 3.3V LDO for input->ESP32 conversion.  I included a 5V LDO to provide power for the two Adafuit boards and the LCD module.  The curcuit.us board has a terminal header for the FTDI cable for programming but only connects RX, TX and GND so the system must be powered externally for programming. (Plus the ugly manual entry and exit from bootloader mode).

Many ESP32 dev boards have a built-in USB Serial chip and 5V -> 3.3V LDO plus auto-bootloader circuitry making programming much easier.  If you use one of these then you can replace the 3.3V LDO with a 5V LDO followed by a schottky diode to power the board's VIN terminal.  The diode prevents conflict between USB supplied power and board supplied power (well ideally you'd like a schottky on the USB power in as well).

Some LCD modules are backlit and some are not.  Some backlit models have a built-in current-limiting resistor and some don't so you'll have to match driving the backlight LED with an appropriate external resistor connected to 5V as necessary.

Firmware Command Interface
Super simple command interface available over serial and Bluetooth serial.   Type 'H' followed by carriage return (or enter 'H' into the Arduino serial monitor and click Send) to get it printed out.

Temperature of sensor 0 (low-side)

Code: [Select]

T0<CR>
Temperature of sensor 1 (high-side)

Code: [Select]

T1<CR>
Log both temperatures at the log interval specified in mSec (minimum 500 mSec and 500 mSec increments, L=0 to stop)

Code: [Select]

L=<msec><CR>
Zero timestamp (useful when starting a log you want to import into something that will graph it since most of them want the X-axis to start at 0).

Code: [Select]

Z<CR>
Typical logging looks like

Code: [Select]

31726.07 26.04 64.05
31731.19 26.04 64.05
31736.31 26.04 64.05
31741.43 26.04 64.02
31746.55 26.04 64.05
31751.67 26.04 64.05
31756.79 26.04 64.05
31761.91 26.04 64.05
31767.03 26.00 64.05
Where the first column is seconds, the second column is temp 0 and the third column is temp 1.

This code could also be stripped of the bluetooth serial stuff and probably run just fine on any old Arduino as well with appropriate pin remapping.

[Fraser]

Globoy,

Thank you for sharing your design 

Fraser

[Userli]

I also purchased an IVAC 9000 but unfortunately it doesn't heat. The display shows R0.03 NOT READY, blinking at roughly 4 Hz. The unit draws 50mA and the 4 test points all show 5V. I exchanged the MosFETS, which seemed the most obvious possible problem to me but still no luck.
Does anybody know what R0.03 means? I assume it's the error description?

[_Wim_]

I also purchased an IVAC 9000 but unfortunately it doesn't heat. The display shows R0.03 NOT READY, blinking at roughly 4 Hz. The unit draws 50mA and the 4 test points all show 5V. I exchanged the MosFETS, which seemed the most obvious possible problem to me but still no luck.
Does anybody know what R0.03 means? I assume it's the error description?

How are you powering it? It seems like only the control section gets power, and not the H-bridge section. Have a look at this post and the datasheet attached: https://www.eevblog.com/forum/thermal-imaging/anyone-have-an-estimated-emissivity-for-the-ivac-9000-radiators/msg3504564/#msg3504564

The power supply connector has 3 pins connected to ground, and 2 pins connected to 5V. All of them are not shorted in the unit, so to make is work you do need to supply all of them, or add a solder bridge as shown in my post above.

[Userli]

Thanks for the fast reply Wim.
I followed in fact your post, put the bridge and powered it even at the same points.
I tested the peltiers and the temperature probes. The uP works since the display works and the ADC also shows activity. The ANDs driving the MFETs seem to work but I didn't yet re-engineer the logic, which I assume ensures the right combination of MFETs of the H bridge to be driven or switch between cooling and heating?