LMP91000: Oxygen measurement with the LMP 91000

Dear Esteban - 

Please post or attach your schematic, which will be very useful in assisting you further with constructive feedback. 

Esteban,

  My experience is that the USB 5V from a computer is typically an LDO and using a USB from a battery is typically a switching boost regulator running off of a single Lithium cell. This means that the switching noise from a boost circuit can effect the REF2933 output voltage due to high frequency noise.

It's possible that you are seeing a VREF shift. Add a 0.47uf capacitor to the input of the REF2933 and a 100nF capacitor to the output of the REF2933. This should stabilize the noise that is causing the VREF shift.

Thanks for your answer Gordon,

We have added both capacitors as you said but they had no impact. The signal is still way bigger when the supply comes from the external battery.

Any other idea?

Esteban,

  Have you measured the VREF voltage on each supply. I would be interested to know if the VREF is moving. The VREF is the bases for accuracy for this IC.

1) connect to PC, measure the VCC and VREF

2) connect to battery, measure the VCC and VREF

Please let me know the voltages for each that you are measuring.

The change you are seeing is 13uA, that means about 13% change since you are near 100uA. That is a lot of change. The VREF you selected is +/-2%. Based on what you report we need to determine if the change in VREF will change your current 13uA. It's possible that you need a 1% or even a 0.5% VREF.

Hi Gordon,

Here are the values I've measured in both cases:

1) Connected to PC: VREF = 3.2910, VCC= 4.794V

2) Connected to battery: VREF = 3.2913V, VCC = 4.960V

Esteban,

  Check your register settings. you should be running external reference set for VREF. The REF2933 looks like it's doing a good job. I wasn't expecting such a large change in VCC.  I will look at the VCC voltage on the bench and see if having the VCC change by this amount will effect the measurement value. I didn't think that it would, but I will check anyway.

Hi Gordon,

I've been using the LMP91000 library for programmig the device. Everything looks ok for me but anyway I will paste the code here so you can check it.

LMP configuration part of the firmware:

LMP.standby();
LMP.enable();
LMP.unlock();  // Enables changing the register
LMP.setMode(5); // 3 lead mode
LMP.setExtRefSource(); // External voltage reference
LMP.setRLoad(3); // Rload, 100 Ohm
LMP.setGain(3); // Rtia 14K Ohm
LMP.setBias(10, 0); //Bias a -600mV, 10 = 18%, 0 = negative
LMP.setIntZ(3); // Bypass = 3.3V (external ref)

On the other hand, I will also make some tests with other voltage values to supply my microcontroller with an DC power supply and I will send you the results.

Best regards, 

Asier.

Hi Josh,

Thanks for your contribution. I don't really think that is the problem since the configuration I want to use is the 3 lead mode,in which every lead is connected separately as you can see in the picture below.

Best regards.

Hi Gordon,

As I told I've been doing some tests with the DC supplier and as you suspected the change in the Vcc has direct influence in the value of the output singal. Here are some examples:

Vcc = 4.895 V  Output = 109.87 uA

Vcc = 4.933 V Output = 112.91 uA

Vcc = 4.945 V Output = 113.5 uA

Vcc = 4.955 V Output = 113.59 uA

Vcc = 4.975 V Output = 114.8 uA

Which could be the reason?

Best regards,

Asier.

Hello Josh,

The sensor in the schematic is not the same we are using, we are using the S+4OXLF oxygen sensor by DD Scientific. I used the one on the eschematic due to its' footprint is the same as S+4OXLF ones. Sorry if that has confused you.

Thank you,

Best regards.

Esteban,

  The data you provide definitely show a sensitivity to the VCC. Can you read back the data registers to verify that the systems is using the VREF. I know this sounds redundant, but we need to know that the code you are using is setting the part up correctly.

It should not change the current with VCC.

Gordon,

I'm pretty sure that my register is well configured. As I told you before, I'm using the LM91000 library, https://github.com/LinnesLab/LMP91000, and for configuring the reference source as external I use the command setExtReFSource().

//void LMP91000::setExtRefSource() const
//
//Unlocks the REFCN register for "write" mode. First reads the register to
//ensure that the other bits are not affected. Writes a "1" to the 7th bit of
//the REFCN register.
//
//Sets the reference source of the LMP91000 to an external reference provided at
//the Vref pin.
//
//Please consult page 22, "Section 7.6.4 REFCN -- Reference Control Register
//(Address 0x11)" of the datasheet for more information.
void LMP91000::setExtRefSource() const
{
unlock(); //unlocks the REFCN register for "write" mode
uint8_t data = read(LMP91000_REFCN_REG);
data |= (1 << 7); //writes a "1" to the 7th bit
write(LMP91000_REFCN_REG, data);

Esteban,

The data sheet shows that the PSRR and the output vs VDD has little to no impact on the measurement. I know that you feel comfortable with the code going into the device, but you really need to read the registers back to verify that it really is using external reference. I have not seen this except when the device has not been configured properly. Once you verify the registers are correct, then we can look else where.

Hi Gordon,

The data you have sent referes to the output of the temperature sensor instead of the oxygen sensor, my problem comes due to the change in the output value of the oxygen sensor. However, I understand the signal of the sensor shouldn't neither change with the change of VCC.

Regarding registers, I have been doing some test changing between external and internal reference source and also changing the % of them and everythig seems to be perfect so we might discard that option. 

Best regards,

Asier.

Asier,

  Because you have a large variation in VDD and we can't identify what is causing the current to shift. My recommendation is to run the LMP91000 VDD directly from the REF2933 and set the VREF to VDD. Then it wont matter what the USB voltage will be.

The REF2933 can supply 25mA and the average input current to the LMP91000 is 10uA. Make sure that you have the proper capacitance to supply inrush current to the LMP91000 and you will be good to go.

Hello Gordon,

I have been doing some test changing the values of IntZ, Rtia and Rload.

When IntZ = 3.3V (Vref), Rload = 100R and Rtia = 3500R, the output voltage is 4.6426V , which is higher than IntZ. In that case, the output current of the sensor must be negative following the equation Voutput = IntZ- (Rtia * Iwe). Furthermore, I have seen that the output voltage was limitited by the supply voltage of my circuit, in other words, the output voltage is always near the supply voltage (+/- 4.8V).

However, with other IntZ values the circuit works different. When IntZ = 0.658 (20% Vref), Rload = 100R and Rtia = 7000R, the output voltage is 0.03487V , which is lower than IntZ. This means that the current is positive following the equation aforementioned. When IntZ = 1.647 (50% Vref), Rload = 100R and Rtia = 7000R, the output voltage is 0.04144V. 

Hope you can give some light,

Best regards,

Asier.

Hi Josh,

I've been using the LMP91000 software trying to find the correct values for IntZero, Rload and Rtia introducing the different specifications of my sensor. Once I entered the values the software told me to short the RE and CE to properly configure it for Oxygen type sensors. So my question is if with every oxygen sensor this must be done or not, and if not in which cases this should be done. I'm also attaching the datasheet of the sensor so you can check the different parameters.

Best regards,

Asier.

s_4oxlf_datasheet_iss2.pdf

Thanks for your reply Gordon,

I saw that circuitry some time ago. However, that circuitry doesn't apply any bias on the sensor which in essential for the right performance of it. The company advises to use this circuitry for S+4OXLF but the functioning of it is very different since it doesent use any TIA. The output is measured through the 100R resistor in the potentiostat part of the circuit.

Best regards,

Asier.