TMP102 takes 5-10 seconds before giving valid 13-bits temperature reading

Hello Ren,

The sequence is:

- cold boot the PC (the TMP102 is on a card in a PC).

- once the driver is loaded, it start the micro controller.

- The micro-controller program the cfg register

- Then the micro controller read the temperature register (right after the cfg register programming)

- after that, it will read again the temp reg at every second.

So right now, there no delay between the cfg register write and temp register read.

The patch I am using right now is to wait N seconds between the cfg write and temp reg read.

The micro controller we are using is made by our team. There is no read buffer on it.

It drives the I2C pins (sda/scl) to write and read register, So everything is in order.

I couldn't find anything in the TMP102 spec that say we need to wait after a cfg register write, before reading the temperature register...

Jay

I have done additional testing and here are the results:

Sequence 1 (13-bit):

- Cold boot

- write CFG (EM=1)

- (NO WAIT)

- read TEMP, then wait 1 second (repeat this step)

I get 9 invalid temp (seems to use temp12 format, but lsb is 1), then all subsequent read are fine.

example:

0x2021 0x2021 0x2021 0x2021  0x2021 0x2021 0x2021 0x2021

0x2021 0x1021 0x1039 0x1051  0x1069 0x1081 0x10A1 0x10B9

Sequence 2 (13-bit):

- Cold boot

- write CFG (EM=1)

- wait 1s

- read TEMP, then wait 1 second (repeat this step)

I get all valid values, but the first 8 values are always the same.

example:

0x1219 0x1219 0x1219 0x1219  0x1219 0x1219 0x1219 0x1219

0x1221 0x1229 0x1239 0x1249  0x1259 0x1269 0x1281 0x1291

Sequence 3 (13-bit):

- Cold boot

- write CFG (EM=1)

- wait 16s

- read TEMP, then wait 1 second (repeat this step)

I get all valid values, and they are all incrementing as expected ... (I'm cooling the chip before power up to see the temp rise).

example:

0x10B9 0x10D9 0x10F1 0x1109  0x1129 0x1149 0x1159 0x1179

0x1189 0x11A1 0x11B9 0x11C9  0x11E1 0x11F9 0x1209 0x1219

Running the same test without programming the CFG register (ie, remain in the default 12-bit mode) gives me similar results.

I get the same readings for the first 9 seconds... (the temperature is always valid in this case, since it is always in 12-bit mode)

So it seems the TMP102 needs:

- a delay between the CFG write and TEMP read.

- a delay before being able to read different values.

I have a work around, where I simply wait 16 seconds after the CFG write, that works fine.

However, I don't understand why this is needed ???

Jay

This is what I have read and this is what I would expect.

It could explain why the first reading is wrong, because it would only update the LSB bit (12-bit or 13-bit), but not the actual temperature value.

However, in my testing, it takes 9 full seconds for the value to be updated after the configuration register write.

Here is the sequence I am doing:

Sequence:

READ CONFIG REG = 0x60A0 (which is the reset value)
WRITE CONFIG REG = 0x60B0 (we simply set EM=1)
READ TEMP = 0x2911 (1st reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (2nd reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (3rd reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (4th reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (5th reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (6th reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (7th reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (8th reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x2911 (9th reading : wrong)
WAIT 1 SECOND
READ TEMP = 0x1489 (10th reading : ok)
WAIT 1 SECOND
READ TEMP = 0x1489 (11th reading : ok)
...

I have modified my microcontroller code to store information at specific locations in memory:

The HMC store in memory the CONFIG reg value read at address 0x2F42020
The HMC store in memory the CONFIG reg value it will write at address 0x2F42024
The HMC store in memory the TEMP reg values read at addresses 0x2F42200..0x2F22FF

And this is the memory content:

ic>dd 2f42000 2
 0x0000000002F42000  0x01010003 0x010F0004  0x0000035B 0x00000326  ........[...&...
 0x0000000002F42010  0x000005C2 0x000003A0  0x00000000 0x00000104  ................
 0x0000000002F42020  0x000060A0 0x000060B0  0x00000000 0x00000000  .`...`..........
 0x0000000002F42030  0x01BB8A4B 0x00000082  0x00000000 0x00000100  K...............
 0x0000000002F42040  0x00000004 0x00000000  0x00000000 0x00000006  ................
 0x0000000002F42050  0x00000000 0x00000000  0x00000000 0x00000000  ................
 0x0000000002F42060  0x00000000 0x00000000  0x00000000 0x00000000  ................
 0x0000000002F42070  0x00000000 0x00000000  0x00000000 0x00000000  ................
ic>dw 2f42200
 0x0000000002F42200  0x2911 0x2911 0x2911 0x2911  0x2911 0x2911 0x2911 0x2911  .).).).).).).).)
 0x0000000002F42210  0x2911 0x1489 0x1489 0x1489  0x1491 0x1499 0x14A1 0x14A9  .)..............
 0x0000000002F42220  0x14B1 0x14B9 0x14C9 0x14D1  0x14D9 0x14E1 0x14E9 0x14F1  ................
 0x0000000002F42230  0x14F1 0x1501 0x1501 0x1509  0x1511 0x1511 0x1519 0x1521  ..............!.
 0x0000000002F42240  0x1521 0x1529 0x1531 0x1531  0x1539 0x1539 0x1541 0x1541  !.).1.1.9.9.A.A.
 0x0000000002F42250  0x1549 0x1549 0x1551 0x1551  0x1551 0x1559 0x1559 0x1561  I.I.Q.Q.Q.Y.Y.a.
 0x0000000002F42260  0x1561 0x1561 0x1569 0x1569  0x1569 0x1571 0x1571 0x1571  a.a.i.i.i.q.q.q.
 0x0000000002F42270  0x1571 0x1579 0x1579 0x1581  0x1581 0x1581 0x1581 0x1581  q.y.y...........
ic>dw
 0x0000000002F42280  0x1589 0x1589 0x1589 0x1589  0x1589 0x1591 0x1591 0x1591  ................
 0x0000000002F42290  0x1591 0x1591 0x1599 0x1599  0x1599 0x1599 0x1599 0x15A1  ................
 0x0000000002F422A0  0x15A1 0x15A1 0x15A1 0x15A1  0x15A1 0x15A9 0x15A9 0x15A9  ................
 0x0000000002F422B0  0x15A9 0x15A9 0x15B1 0x15B1  0x15B1 0x15B1 0x15B1 0x15B1  ................
 0x0000000002F422C0  0x15B9 0x15B9 0x15B9 0x15B9  0x15B9 0x15C1 0x15C1 0x15C1  ................
 0x0000000002F422D0  0x15C1 0x15C1 0x15C1 0x15C1  0x15C1 0x15C1 0x15C9 0x15C9  ................
 0x0000000002F422E0  0x15C9 0x15C9 0x15C9 0x15C9  0x15D1 0x15D1 0x15D1 0x15D1  ................
 0x0000000002F422F0  0x15D1 0x15D1 0x15D1 0x15D1  0x15D9 0x15D9 0x15D9 0x15D9  ................

So the update delay explains the first invalid reading.

But It doesn't explain why it takes 8-9 seconds before I can get a temperature update after CGF reg is changed.

PS:

if I wait 1 second also before the first TEMP read, then the value is valid, but it still need the same 8-9 seconds before being updated.

In all cases, the refresh rate should be 4Hz (default value).