Test 04: fails (Si570 identification)
- INCAA reported the following problem with Test 04 of the production
test.
- Test 04 needs to be modified to allow the correct identification of the assembly of the right Si570 IC.
From: Bart Sijbrandij [mailto:Bart.Sijbrandij@incaacomputers.com]
Sent: 20 March 2012 14:38
At the moment we are doing the production test of the FMC ADC modules.
Until now we have tested two boards. Both give the same result. All test are OK expect for test 04.
The software reports:
Running test 04
Test 04 error, continuing:[Si570 bad reference frequency configured or
wrong part is mounted]
The log-file reports:
RFREQ : 43.73628905043005……….75
N1 : 9
HS_DIV: 1
The module that we received from you and passes all tests gives here:
RFREQ : 42.01……………………………..500
N1 : 7
HS_DIV: 2
Please note that the part numbers of the used Si570 parts are identical.
If the total part number of the Si570 is the same (570BBC000121DG) then there is no problem. In that case you can ignore the message you got.
The problem is in the test program and a misinterpretation of the datasheet. The technical explanation is that the program checked for the 42, 7 and 2 values. The Si570 device uses those numbers to generate the 100.00000 MHz output out of the used crystal frequency. But on each device the crystal frequency is slightly different and that is compensated for by the factory by programming other values in it.
As there is no register where one can read the 100.00000 MHz (it is only in the human readable part number on the package), one cannot easily check some registers to find out if indeed a 100.00000 MHz device is installed. I think this is actually an omission from Silabs in their design.
An idea to test if the correct device is mounted is to program the connected function generator at a known frequency, say 10 kHz and then sample the data with the SiLabs supposed 100 MHz and verify if you received 1000 full waves (+/-1) in a second (using the 100 MHz as timebase too), ie. taking 100 Million samples.