Re: Kerry or Bush

From: John Larkin (jjlarkin_at_highSNIPlandTHIStechPLEASEnology.com)
Date: 10/20/04 

Date: Wed, 20 Oct 2004 10:09:05 -0700

On 20 Oct 2004 03:29:47 GMT, arthurj@aol.comet.net (James Arthur)
wrote:

>On 10/15/04 John Larkin jjlarkin@highlandSNIPtechTHISnologyPLEASE.com wrote:
>
>>Speaking of electronics (is that allowed?) I have a nasty problem. I
>>use an LC oscillator in my delay generators; when we get a trigger, we
>>start the osc running at 50 MHz and use that to time delays in 20 ns
>>steps, with a little analog ramp afterwards for the fine (picosecond)
>>bits. The oscillator is coarse tuned with a ceramic piston cap, and
>>fine-tuned with a varicap, so we can phase-lock it to a crystal for
>>long-delay accuracy. The vco pull range is +-2500 PPM, which seemed
>>good at the time.
>>
>>But just lately, some, too many, units are losing lock. It appears
>>that they are humidity sensitive on a very slow (weeks/months) time
>>scale, with high humidity pulling down the frequency. I did find some
>>defluxing gunk in some of the piston caps, from the boad cleaning
>>operation, and that added a nasty positive capacitance-humidity effect
>>to tha cap. But with clean caps, there's still a problem.
>
> John,
> Can you be sure it's the capacitors rather than the inductors
>that are changing?
>

Seems so. We've humidified boards then swapped Ls immediately before
datalogging, and the horrible f-creep was still there. We *did* once
have a problem with Coilcraft inductors creeping, but we don't use
them any more.

> Building a separate oscillator for torture-testing might be instructive.
>
> Alternatively, tacking an extra capacitance onto the tank, measuring
>oscillator frequency before & after, then solving the simultaneous
>equations would reveal which component is changing, & by how much.

Maybe, but the change is small enough that the math might be tricky.
The LC osc is phase-locked to a crystal, so what we measure/log is VCO
DAC code, not frequency.

It's looking like the Voltronics ceramic piston caps change C with
humidity, and the PCB does a little too. Guess I'm going to have to
fix the cap problem or find a way to increase lock range without
having to get into the beastly Actel FPGA to change the algorithm.
Analog kluge time!

>
> --James Arthur

Are you "my" James Arthur, friend-of-Dan, Altoids-box-oscillator guy?

John