[McHUG] Less success

[Peter Morton]

Last night and today I did some more checking out of the wb6dhw SI570 board. 
I tried to zero beat the SI570 output with WWV at 10 MHz.  The output 
frequency was about 5.5 KHz low.  This is far outside the 20 ppm specified 
for the LVDS SI570.  At first I thought that maybe the XTAL frequency in the 
program didn't match the crystal frequency in the SI570.  The error at 5 MHz 
was exactly half the error at 10 MHz.

I checked the data and clock signals from the parallel port of my laptop to 
Header no. 2 on the SI570 board and found that a few of the high-level data 
bits drop to half amplitude just after the clock transitions high-to-low. 
It appears that the SI570 is trying to pull the SDA line low for a short 
period of time while the line is still high after the clock transition. 
This occurs on several high-level data bits through out the 6 ms. burst of 
data and clock.   I don't see this contention on the cable when the board is 
disconnected.

The parallel port output high level is 3.3V on my laptop.  All data and 
clock transitions look good with very little overshoot and ringing at the 
board header and at the SI570 pins.  I used RG-174 cable for each of the 
data and clock (SDA and SCL) with backmatching 33 ohm resistors at the 
computer end of the cable (in the 25-pin D-sub connector shell).

One concern I have is that I left pin 1 of Header 2 (Microprocessor VDD) 
open.  The GTL2002 level-translator data sheet states that the GREF pin 
(connected through a 200K resistor to Microprocessor VDD on the header) 
should be at least 1.5 volts greater than the SREF pin (3.3V board supply) 
for best performance.  For thoroughness I'll connect an external 5V supply 
to pin 1 of Header 2 to see if it makes a difference, but I don't think it 
will.

So, has anyone checked out their SI570 board yet and, if so, what did you do 
with the "Microprocessor VDD" pin of Header 2?  Do you see the contention 
that I see on the SDA line?

I'm using pe1nnz's SI570 LPT1 software I found in the files section of 
Yahoo's Softrock-40 Interest Group.

At this time, I really don't know if the SDA contention is the cause of the 
frequency inaccuracy problem.  I've read messages on the Yahoo site where 
some are calibrating out a couple hundred Hertz frequency error in the 
software.  PowerSDR allows you to specify the actual SI570 crystal frequency 
in the program setup to correct for xtal frequency error in the SI570.  But, 
no one mentions an error as large as what I'm experiencing.

I may have a bad SI570.  I should probably get a static mat and wrist strap.

-Pete
(it's past 12-dark-40)