FXOS8700 - Bad Z Accel Axis

[ERussell]

I am using the FXOS8700 as an e-compass and of the first 10 prototypes, we have found that on one the accelerometer z axis never changes. On several others the accelerometer z axis is intermittent. I did not experience this with the 5 sample chips I tested last spring. I'm using the same code as before. I tried setting the self-test bit in CTRL_REG_2 and again, the z axis never changes. The value read is always 0x1FFF (raw value of 0x7FFC). I am using hybrid mode, normal read, and am reading out 12 bytes (I2C) for each reading starting with register OUT_X_MSG. When the accel z axis is working, the compass works very well.

Has anyone else experienced this issue?

Thanks,

Elizabeth

[sanchezf]

Hi Benjamin,

The same problem with Z-axis appears at one of my customers. In this case the read value at Z-axis registers ( 0x05 and 0x06 ) is 0x3FFF. Sometimes the system is recovered after several movements and sometimes they need to do a HW reset.

The failing devices are marked as follows :

263

8700

I2ZG

The customer is using your sensor to detect the position of a electro valve.

Do you have any application note explaining how to read the marking ?

Thanks,

Francisco

[oferfreilich]

Hi

I have the same problem on one chip, but on the z-axis of the magnetometer.

I can read only 0x8000 value from this axis from only one board.

The other axes of the magnetometer (and of the accelerometer) are OK.

We replaced the FXOS8700CQ chip with a new one and the problem disappears.

[benjaminbush]

Hey Ofer,

        Did you see this problem consistently or just on one chip? I'm just concerned that when we try to ramp up we will find this problem consistently. Is this a soldering issue or a manufacturing defect?

[TomasVaverka]

Hi Benjamin,

Could you please tell us more about your application? Are you using the FXOS8700CQ just for static orientation measurements or are you exposing it to mechanical shocks?

However, if the Z-axis output is reporting exactly 0x7FFC or 0x8000 (+/-full scale) whereas the unit is only exposed to earth gravity, the probability that the device is suffering Z-axis stiction is high. 

Customers that experience an issue similar to this can contact our quality team and customer service team. The NXP Quality team will enter the information into the NXP CQC system to route to the correct quality team and business line contacts for support.

Best regards,

Tomas

[benjaminbush]

Hey Tomas,

        We will be exposing it to shock, I'm not sure I can say more than that.

I'm pretty sure that we are experiencing Z-axis stiction at this point. What is the procedure for getting replacements? We purchased the chips through mouser so do we reach out to them?

Thanks for your help.

Benjamin

[TomasVaverka]

Hi Benjamin,

Yes, I would recommend contacting Mouser Electronics directly.

Best regards,

Tomas

[benjaminbush]

Hey Tomas,

        Mouser responded with this message

My technical team confirmed that we haven’t any reported issues with this part in the past year, so the rest of our stock should be fine. I found that the date of manufacturing on your order was 1541 (the second week of October in 2015).

Has this batch had reported problems from z-axis stiction? I'm concerned that it might be more of an issue with noise in the system or some other problem. Thanks for your help.

Benjamin

[benjaminbush]

I have encountered the same problem in  2016. Instead of 0x1FFF I am reading 0x8000. My chip is 

263

8700

DNJ

I'm using i2c, and hybrid mode. I've used 2 chips, that had the z-axis start with incorrect values, then correct values, then consistently send out 0x8000.

Thanks for your help

Benjamin

[TomasVaverka]

Hello Elizabeth,

Could you please send me a complete part marking found on the top side of affected parts?  

According to your description, it sounds like the Z-axis stiction failure that rarely occured on the MMA845xQ accelerometers.

There has been made some design improvements to eliminate this failure as described here. 
 
However, I have not heard of this issue on the FXOS8700CQ yet. I have requested the responsible team for more information and I will get back to you as soon as I know more details.

Regards,
Tomas

[ERussell]

Tomas,

The marking on the chips I sent you was from a 10 piece purchase from Digikey in September by our Contract Manufacturer. They purchased production reels from Arrow and Future which have the following marking:

263

8700

I4ZF

The CM is planning to run 500 boards later today or tomorrow with the production chips. I would really appreciate hearing as quickly as possible whether Freescale has confidence that these chips don't have the problem.

Thanks,

Elizabeth

[TomasVaverka]

Elizabeth,

As your question requires more detailed communication, I have just created the Service Request #1-1213293747 for you. You will receive an email from Freescale acknowledging the creation of this Service Request. All additional communication will be worked via this Service Request.

Regards,

Tomas

[dimitrov]

We've experienced something similar. See this.

The FXOS8700 used as magnetometer overflows start sending -32768 or 32767.

Sometimes we fix it just by reflowing the chip, sometimes we had to replace the chip, but it happens quite often. We are blaming the HASL finish so far. The next batch of boards will be with different finish, so we'll see.

What finish do you have on your boards? When you put a magnet next to the compass does the Z axis change?

[ERussell]

We are using an immersion gold finish rather than HASL. As expected, a magnet affects the magnetometer readings but not the accelerometer readings.

[dimitrov]

Somehow I got the impression that you are having problem with the magnetometer. My mistake.