TMP421 on the preamp board

[Gerard Visser <gvisser@indiana.edu>]

Hall D Electronics:

Hi Fernando,
	I encountered a mysterious effect this week when at CMU... There were 
always spikes visible in the output of the preamp board, at least I can 
say "always I saw them on most channels". I saw this on both the boards 
(mine and CDC's). I thought it was external interference from something 
over there. But we never could track down what, and shielding never had 
much effect on this, though it did help with the 88.3 MHz interference 
(dramatically now as I see from Yves!).
	I left Yves with "my" preamp board (with the 59 Ohm resistors 
installed) and returned here with "his". I set it up yesterday and was 
amazed to find exactly the same spikes occurring, at roughly 100 kHz, 
same as I saw at CMU.
	So it had to be an internal problem; I tried powering from a battery, 
still the same. We tried a Tek 465 scope at CMU, so I knew it was not 
the digital scope generating interference.
	Turns out, it was from the TMP421 circuit - this works by switching 
(apparently rapidly) between two bias current levels in the temperature 
sensing diode, no doubt at a frequency of about 100 kHz. The actual 
problem on this board was a poor solder connection at the bypass cap by 
the TMP421 chip. I resoldered and there is now _no_ detectable noise 
spike. [Triggering the scope on the output noise and averaging the trace 
I get a nice symmetric function, due to the correlation of the noise - 
previously if you did this you got a highly asymmetric function that 
looked just like the response of the ASIC to a charge pulse input.]
	I'm not writing to complain about soldering though - what I really need 
to say is that the middle ASIC, the one where the temperature diode 
sits, may still suffer some effects from the TMP421 circuit. I can't 
test this because I don't have a middle ASIC on either of these boards. 
But you do, right, so you can look into this...
	I think that either
	- It will work fine (generates no noise spikes on any channels), or
	- It generates some noise spikes, but this can be remedied by putting a 
few hundred pF to ground on the diode side of R144,R145 on your 
schematic, or
	- It's hopeless and we should reconsider how to use the internal 
temperature diode, or should just use an external temperature sensor.
	I think we should not have a situation where the board generates noise 
spikes even at a low rate, unless it only happens a short while in 
response to controls command to measure the temperature. It seems that 
the TMP421 is continuously measuring the temperature regardless of when 
it is read out. That is probably acceptable only if it can generate no 
interference.
	Sincerely,

	Gerard