[PSIG]

Random stuff, as I don't know your test conditions — but I don't think I'd want to see the sensor glow under normal warmup conditions. Is the MOSFET getting hot, possibly indicating it's having issues with the heater inductance? I haven't looked at the code; but is it intended to drive the MOSFET with an inverted signal (E9) or non-inverted (E10)? Probably, as it would likely go out-of-control, but just asking. Unless time-to-temp is excessive; it's working. Yay!

David

[Groopan]

Random stuff, as I don't know your test conditions — but I don't think I'd want to see the sensor glow under normal warmup conditions. Is the MOSFET getting hot, possibly indicating it's having issues with the heater inductance? I haven't looked at the code; but is it intended to drive the MOSFET with an inverted signal (E9) or non-inverted (E10)? Probably, as it would likely go out-of-control, but just asking. Unless time-to-temp is excessive; it's working. Yay!

David

Basically I am using blyunds shield and his code . I just used a different mosfet. The spirit commands to calibrate responds well the values are as expected.once I start measuring it starts to read lower (2%). I then notices that if I reset the controller a few times succesively after the heat up phase I get a normal reading 20%. So I changed the setpoint manually until I got the 20%in air . I the proceeded to use a lighter and flooded the sensor and it responded as expected. With that said I did some mods to the code to display reg1 and does not match settings for normal measuring mode at 17v. I will do some more tinkering and see what I can find .

[plazomat]

I haven't looked at the code; but is it intended to drive the MOSFET with an inverted signal (E9) or non-inverted (E10)
David

@PSIG - according to schema E10-non inverted.

Plaz

[JHolland]

If you are using PWM then you may need a logic level FET. Many Arduinos don't have a de-coupling cap on the analogue reference as per the datasheet.. You have 100k pull-downs on the output of Ua and Ur, should they be fitted?
You really do need good grounding when you have sensitive analogue signals and relatively high power switching signals.

[JHolland]

Which sensor are you using? the schematic shows the circuit configured for a 4.9.
Also it looks like there is a resistor missing, the datasheet shows a resistor in series with the Un output.

Edit: The app, note shows that resistor as 0 Ohm so that can be ignored.

[Groopan]

Which sensor are you using? the schematic shows the circuit configured for a 4.9.
Also it looks like there is a resistor missing, the datasheet shows a resistor in series with the Un output.

Edit: The app, note shows that resistor as 0 Ohm so that can be ignored.

Lsu 4.9

[JHolland]

Most of the code has been taken here: https://github.com/Bylund/Lambda-Shield-Example
Chris

I had a quick look at that code, the PID is not properly implemented, the PID parameters haven't been converted to the Z-domain. You may be better off using the Arduino library code, the transform is implemented within that code although its rather crude. There is an explanation of one method of implementing the transform in this application note:

http://ww1.microchip.com/downloads/en/A ... 03136A.pdf

The Microchip application doesn't include an integral anti-wind up term so it may not be ideal. Also the paragraph that states:


In a continuous-time system, the PID controller can be transferred from time domain to s-domain by using a Laplace transform.

should read:


In a continuous-time system, the PID controller can be transferred from the s-domain to the z-domain by using a Laplace transform.

[Groopan]

Most of the code has been taken here: https://github.com/Bylund/Lambda-Shield-Example
Chris

I had a quick look at that code, the PID is not properly implemented, the PID parameters haven't been converted to the Z-domain. You may be better off using the Arduino library code, the transform is implemented within that code although its rather crude. There is an explanation of one method of implementing the transform in this application note:

http://ww1.microchip.com/downloads/en/A ... 03136A.pdf

The Microchip application doesn't include an integral anti-wind up term so it may not be ideal. Also the paragraph that states:


In a continuous-time system, the PID controller can be transferred from time domain to s-domain by using a Laplace transform.

should read:


In a continuous-time system, the PID controller can be transferred from the s-domain to the z-domain by using a Laplace transform.

I don't think there is a problem with the pid . The setpoint for temperature is pretty much met and held .

[Groopan]

Can any one confirm that they receive 0x2889 when they check the status of REG1 when the controller is set into normal measurement mode at 17V.

[JHolland]

I don't think there is a problem with the pid . The setpoint for temperature is pretty much met and held .

There is a big difference between something that works properly and something that appears to work properly. Whoever wrote that doesn't understand how to implement PID on a microcontroller - PID is an analog algorithm, it must be converted to a digital algorithm. If you are going to try and produce something then you should try to ensure that its done right, how much confidence do you have that that will always be stable?

Study the Arduino library code here:
https://github.com/br3ttb/Arduino-PID-L ... PID_v1.cpp

In the PID::SetTunings function the PID values are rescaled relative to time, this is a very simple implementation of a conversion to the z-domain.

Code: Select all

111     double SampleTimeInSec = ((double)SampleTime)/1000; 
112     kp = Kp; 
113     ki = Ki * SampleTimeInSec; 
114     kd = Kd / SampleTimeInSec;