HDC2010: Corrupted data readings

Dear Martin - 

Thanks for the post - I think here if you could share a capture of your I2C communication and your schematic, that would be appreciated.

The problems may be (along with an associated FAQ): 

1. in how long you are waiting after software reset (which is not really needed like it is with some other vendors devices) - when issuing the software reset, you need at least 300uSec delay time before attempting to communicate with the part.  

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/878720/faq-hdc2080-hdc2010---how-long-must-i-wait-after-powering-up-the-hdc20x0-family-sensor-for-good-data-to-be-returned-by-sensor-first-read?tisearch=e2e-sitesearch&keymatch=HDC2010#

2. that you don't wait long enough to begin with, to get valid measurements - and you might actually be always a measurement behind, and as you changed some trigger point, you may have uncovered the "bug" in your firmware. This would also be easily captured on logic analyzer or output log, as the first read, where it would show up as described in the FAQ. 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/878721/faq-hdc2080-hdc2010---i-am-getting-incorrect-values-for-example--40c-or-0-rh-or-both-from-hdc-device-for-my-first-read-or-for-any-read-how-to-troubleshoot-the-problem?tisearch=e2e-sitesearch&keymatch=HDC2010# 

Additionally, if you can read out the registers 0x04 through 0x0F on error, that would also be helpful - if the registers are corrupted, this would point us to the solution i outlined at the very end of this thread. 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/879641/hdc2080-taking-more-time-for-data-ready 

Dear Josh,

Thank you for the efforts. This is capture of part of our schematic:

Regarding the reset - I make HDC2010 reset only on invalid data detection. Тhat's innovation by me to recover from invalid data, still don't know whether this helps, I asked you about it.  Also wait the defined time after reset, in order to begin sensor communication.

I'm unable to attach the i2c communication because I don't have a tool for that but I'll explain the HDC2010 measurement process.

Briefly:

- The sensor is continuously powered on.

- Periodically I start the measurement by writing 0x61 to register 0x0F.

- Wait for 3ms and directly read the temperature and humidity registers.

Best regards,

Martin Peevski

Martin - 

Thanks for the reply and the drawing. 

I highly recommend that you get a logic analyzer if you are writing firmware or designing hardware that runs on a communications bus. 

Devices we use: https://usd.saleae.com/products/saleae-logic-pro-16 , https://digilent.com/shop/test-and-measurement-equipment/usb-oscilloscopes-analyzers-and-signal-generators/ 

At the very least, try to get access to an multi-channel o'scope to see what your timings are. 

I would also recommend you put the pull up resistors for the I2C in schematic near the parts.

Next - I am very curious why you set the part for 0x61. This is 11 bit temp and 9 bit %RH? Its a valid setting, but I would like to understand the use case for downgrading the measurement. The wait time after power up is 3mSec. The wait time after soft reset is 300uSec, the typical conversion time for the resolution you have chosen to use is 625uSec. This means that you are already waiting plenty of time for the conversion to complete. 

One technique you may try, as here we are a little blind still as to what is actually going on (without an analyzer) - would be to use a read of register 0x04 for status, before proceeding to read the temp and %RH registers out. This would basically block out all these errors you have and not return a value until that register was a 0x80. Here you will either get caught in an infinite loop or it will "fix" the issue. 

Looks like this (click on image to enlarge)

Here I am showing you issuing a 0x01 to 0x0F, waiting 1.3mSec (as I am using the part in default 14 bit mode), getting the IRQ, reading 0x04, getting a 0x80, then reading 0x00 through 0x03 for temp and %RH, then printing from UART. (this is using the Saleae LSA) - here you did not connect DRDY line from the part to your controller, but you can wait as you do, then read 0x04 in a loop until you get 0x80 (which should be first time), then read the measurement values out. 

Can you do any UART prints to help you troubleshoot this? Could you still read out 0x04 through 0x0F and print those out (on error)

Thank you for the reply, Josh,

I downgrade the measurement because this is enough for the device and consumes less energy. We are battery powered.

I made the software changes and now I'm waiting to see whether HDC2010 reset, when read temperature above 60 Celsius or below 0 Celsius, will fix the problem. I use UART prints to log everything important.

I don't like the idea to check 1st the 0x04 register, as I saw topics for endless loops here.

Dear Josh,

Thank you for the response. I'll wet you know if the HDC2010 software reset doesn't fix the invalid data readings.