ST25R95 Tag detection issue on temperature change

[Jhern.1]

Hello,

I am testing my reader (based on ST25R95 ) across temperature and I have detected the device wakes up by itself (with no tag on field) when introduced into the fridge (about 5 ºC). The device which had been previously calibrated on ambient temperature (20-25 ºC).

I am currently using a guard of +/- 8 as recommended by ST.

I am having calRef values about 0x20 and 0x1C on different boards (same board model and antenna tunning).

Questions:

1. Is this a normal behaviour ? I am using NP0 capacitors, so I do not understand why the temperature change is affecting my system. Could you provide some advice about this?
2. I guess if I increase the guard I will reduce this effect, but I am afraid this will lower the detection range, right?
3. High Q antenna tunnings are more sensitive to temperature changes than low Q antennas?
4. How do you recommend to lower the Q?
5. Is the temperature provding higher or lower antenna current?
6. When a tag enters on field the antenna current increases or decreases?

Thanks in advance.

Regards, Ramon.

[Ulysses HERNIOSUS]

Hi Ramon,

I am not a full expert here but I have some experience from ST25R3911B/3916 in this area. @Brian TIDAL_O​  may be able to add here more information.

ad 1) With change of temperature I assume some of the components may drift. Even NP0 caps have a slight drift but you have more components like inductors and resistors in the matching network. From my experience such effects cannot completely be avoided.

ad 2) Yes, the alternative would be to re-calibrate (at least partially to adapt to the changed environment).

ad 3) + 4) don't know. Not an expert here.

ad 5) + 6) I think it can go in both directions. I assume it to depend on the type of matching (smith chart) and also the type of card/phone which enters the field.

Regards, Ulysses

[Brian TIDAL]

Hi Ramon,

I've done some experiments on my side with an X-NUCLEO-NFC03A1 board: the CalRef at low temperature is between 2 and 3 Dac steps (i.e. between 8 and 12) from the CalRef at ambiant temperature. Therefore, this may cause unwanted wake up from Idle.

I would propose the following workaround:

if several wake up with no response to polling are detected, re-calibrate the Tag Detector (st25r95CalibrateTagDetector()) and assign the CalRef returned value to gRFAL.wum.CalTagDet.

Rgds

BT

[Jhern.1]

Hi Bryan,

Thanks so much for your answer.

Yes, I know the re-calibration can tackle this problem at the end, but I wanted to know if this is a normal behaviour or there is sth wrong related to my antenna tunning or antenna Q, or any component.

1. High Q antenna tunnings are more sensitive to temperature changes than low Q antennas?
2. How do you recommend to lower the Q?
3. Is the temperature providing higher or lower antenna current?
4. When a tag enters on field the antenna current increases or decreases?

From my experiment this morning the calRef value goes lower at higher temperatures (from 0x2C to 0x20) so my guess:

• Lower temperature -> Higher antenna curent
• Higher temperature -> lower antenna current

Does this match your experiment at low temperature?

Do you think a lower Q antenna can minimize this effect?

Regards, Ramon.

[Brian TIDAL]

Hi Ramon,

as far as I know, resistor variation due to temperature is a normal effect (with negative temperature coefficient for composition resistors and positive temperature coefficient for metallic resistors). I will ask to an HW and RF expert to provide you answers to your precise questions.

Rgds

BT

[Jhern.1]

Hi Brian,

I am using 330 ohm (1% tol, 100ppm/ºC) for RX resistors, which can swing from :

• 330,8 @ 0ºC to
• 329,2 @ 50ºC

Resistance variation is tiny compared to current. I am detecting problems with at 5-10 ºC. when calibrated at 20-25 ºC.

Is the reception side (where rx resistors are) involved in tag detection? or should I care only for the componentes in the TX side?

I am using same guard for high and low threshold (8).

I am using the calibration algortihm found in RFAL.

My design is pretty similar to design on M24LR kit transceiver board (find it attached). This base design is for the CR95HF and I am using the ST25R95, but I think both chips are the same from the RF perspective.

Do you think the ferrite beads L01, L02 are causing the drift? In the nucleo-nfca3 board there are 560 nH inductors instead, but the BOM for this board is not available.

Regards, Ramon.

[Brian TIDAL]

Hi Ramon,

here is the BOM for the X-NUCLEO-NFC03A1.

I have forwarded your question to our HW/RF experts.

Rgds

BT

[Jhern.1]

Thanks for the BOM.

I see you are using here a different inductor on hte EMI filter. Why is this?

For the former inductor -> Not easy to get a precise inductance value from Wurth datasheet for calculating the EMI filter.

I have been reviweing the components and tuning my system. Here are some points I want to let you know.

1 - VDC on ST_R0 when fine tuned is about 4 V, which is ok (below 7V). I could lower R_rx so as to increase this voltage, but not sure if thes impact on detection.

2 - Would an increase on V_ST_R0 benefit the tag detection?

3 - I have measured impedance of M24LR kit and gave me 66 -36j, which surprise me. I expected sth around 27-30 ohm (purely real impedance, minimum reactance). Anyway, matching seems to be ok at target frequency with this impedance.

 VNA is calibrated and measured at TX pins

Would I improve detection If I move this tuning to 27-30 ohm ?

Looking forward to hearing answer from yout RF expert.

Note we are launching this new product these days and need to solve these issues asap, before going to mass production.

Regards, Ramon,

[Brian TIDAL]

Hi Ramon,

I am trying to find answer with RF/HW experts.

As I reproduce your issue with a freezer spray directly on the ST25R95 on an X-NUCLEO-NFC03A1 board, I tend to believe that the variation is most likely due to the IC itself. Can you confirm that on your side, you can also reproduce your issue with a freezer spray directly on the ST25R95 IC?

If it confirmed, I would suggest to implement the recalibration workaround proposed in previous posts.

Rgds

BT

[Travis Palmer]

​Hello Ramon,

A temperature change is affecting any component on the PCB - even the PCB itself. Some components have more, some have less influence. Some of them you have already identified.

To my experiance the most critical one is the PCB antenna itself. Lets look at a normal ST25R3916-DISCO antenna at 25°C, +95°C and -40°C - same is valid for any other PCB antenna with different values.

25°C:

-40°C

+95°C

As you can see on those pictures, we see quite some change in the series resistance at low frequency and Parallel resistance at resonance frequency.

This is due to the copper traces getting hot and cold and changing its resistance.

As mentioned also other components will face similar changes. The EMI coils for example, will be dependent on the type of material they are made of. Here i recommend you browse through the vendors offering.

The question is: Can those effects be minimized? The answer would be: Yes

Can those effects be removed? The answer would be: No

I think the before mentioned recalibration is a good workaround to overcome this challenge. For future products you should have a look at our other products. The ST25R3911B and ST25R3916 offer an automatic averaging filter which is slowly adapting the reference value. Slow changing environmental affects would not cause a wake-up event here.

BR Travis

[Jhern.1]

Hi Travis,

Thanks for your thorough answer. I really apprecciate your assistance. I was ready for mass production and found this issue. I need to solve to solve it asap.

I do not see series and parallel resistance on your pictures. I just see impedance at 2 and 78 MHz. I see the real part increases with the temperature, which is normal due to copper conductance variation.

Are you sure this small variation can really impact on detection? Taking these Z values at 2MHz into the excel spreadsheet I see rx Voltage goes from 3,150 V to 2,919V. not sure how this can be translated to antenna current.

I was also wondering how the PCB could affect this drift, but not easy to quantify. I did some experiment in my end:

• On CR95HF eval kit (MB1054B). Calibrated at 25ºC. I applied selective air heat (50-60ºC) to the PCB antenna and the system did not wake up. The DAC Ref changed when calibrated at heat, but no wakeup happened. I also applied heat to components and same result.

Since my board is more sensitive to the temperature change I guess I can change sth in my board in order to minimize this effect, but I would like to know how. ¿Could you provide some clues?

I am waititng new inductors for the EMI filter. With regards to the EMI inductors. Which is the key parameter? a very low Rdc value?

With regards to EMI filter. I think it is best to tune it at a higher frequency tahn 14,6MHz so as to avoid getting closer to 13,56MHz due to tolerances. I chose 20 Mhz. Is this ok?

My board is providing lower CalRef values (about 0x40) than the CR95HF kit (which shows values of 0x68 at ambient temperature). I also see dfference in the voltage at ST_R0, but I think both data are related, right?

Based on these experimetns I guess the higher the antenna current (CalRef Value) the more tolerant to temperature changes. Does it make sense?

If so, How can I rise the antenna current ? by lowering rx resistors?

FYI: I am using custom board with ST25R95. Is there any difference between ST25R95 and CR95HF that may impact?

Is there any tweak I could do in the tuning so as to minimize the temeprature effect?

You have seen differences on a very wide temperature span (95º to -40º). I need to have stability, at least, between 5º and 45 ºC. I have already implemented the re-calibration workaround, but the system should be tolerant, at least, to nigh/day thermal variation.

Regards, Ramon.

[Travis Palmer]

Hello Ramon,

Every component on the PCB (and the PCB itself) will change its behavior when being exposed to temperature. This effects the PCB antenna, EMI inductors as well as the CR95 itself and will therefore change the voltage across the NFC antenna and measured current.

The above mentioned behavior of your experiment could be linked to heating of the EMI inductors. For example, if the field has been turned on (e.g. due to polling for a tag for several 100ms) before the calcRef is executed. This will heat up the inductors (depended on their current rating) and produce a calcRef value which is too high - or equivalent to higher ambient temperature. Your higher level and lower level might be shifted. As a result the device would wake up faster when being cooled down, but slower when being heated up. It would be interesting to know what the current consumption during continuous Tx of your board is and what type of inductors you are currently using. Maybe you can measure the temperature using an IR camera? It would also be interesting if you could share the refCal values for 25°C, hot and cold.

In general less antenna windings will cause a higher current flowing through the antenna. Be aware that any change which might be favorable for the wake-up performance might effect other modes of the IC - which could for example reduce the sensitivity and therefore read range.

Increasing the offset between lower and upper limit could be one other solution. But it will reduce for sure the detection range.

Looking at our ST25R3916 or ST25R3911B the principle is quite similar to the CR95HF. There it is a common procedure to obtain the reference value every time before going to enter wake-up mode. This is a common practice since these devices are even more sensitive and therefore more sensitive to such environmental effects.

Criteria for good EMI inductors are:

• Q-Factor => crucial for performance ( I guess not so important for your application)
• low DC R => very interesting for you - less losses and heating
• High saturation current => same as above
• temperature stability

I am not sure, if changing the EMI cut-off will influence the temperature behavior. More likely playing the inductance value (while staying at the same cut-off) could have a positive effect => lower inductance value = lower DC resistance. (for example if it is related to inductor heating).

The ST25R95 and CR95HF should behave the same in this regard.

BR Travis