About 5% of my production run is having
problems with start up of the oscillator. I am
using a 4 MHz parallel crystal Digikey part number
X006-ND, with a pair of 22pf NPO ceramic chips
on either side to ground. Local bypass is good. At first,
because the board is not accessible to customer, I had
been leaving the flux on to save time and environment.
For normal resin flux, this didn't seem to matter, but I recently
switched to organic water soluble flux, and I thought
I would pass along my tentative observation that this stuff
has got to be thoroughly washed off.
But even with a clean board, the start-up problem
persists. Has anyone done any comparison between
quartz crystals and ceramic 4 MHz resonators regarding
reliability of start up?
At 10:43 AM 10/6/97 -0400, you wrote:
>About 5% of my production run is having
>problems with start up of the oscillator. I am
>using a 4 MHz parallel crystal Digikey part number
>X006-ND, with a pair of 22pf NPO ceramic chips
>on either side to ground. Local bypass is good. At first,
>because the board is not accessible to customer, I had
>been leaving the flux on to save time and environment.
> For normal resin flux, this didn't seem to matter, but I recently
>switched to organic water soluble flux, and I thought
>I would pass along my tentative observation that this stuff
>has got to be thoroughly washed off.
>
>But even with a clean board, the start-up problem
>persists. Has anyone done any comparison between
>quartz crystals and ceramic 4 MHz resonators regarding
>reliability of start up?
>
I have heard that making one of the caps on the xtal pins (I think it was
the one on the pin OSC2) larger than the other by a few pF sometimes makes
startup more reliable.
I have not tried this, just heard it.
Sean
Sean Breheny,KA3YXM
Electrical Engineering Student
>About 5% of my production run is having
>problems with start up of the oscillator. I am
>using a 4 MHz parallel crystal Digikey part number
>X006-ND, with a pair of 22pf NPO ceramic chips
>on either side to ground. Local bypass is good. At first,
>because the board is not accessible to customer, I had
>been leaving the flux on to save time and environment.
> For normal resin flux, this didn't seem to matter, but I recently
>switched to organic water soluble flux, and I thought
>I would pass along my tentative observation that this stuff
>has got to be thoroughly washed off.
>
>But even with a clean board, the start-up problem
>persists. Has anyone done any comparison between
>quartz crystals and ceramic 4 MHz resonators regarding
>reliability of start up?
Take a careful look at some of the problem boards. Sometimes flux and other
process residue can get trapped between the crystal and the board. It will
stay moist and cause start-up problems. If you remove the crystal, clean it
and the board carefully and re-install it, do those boards still have a
problem? You may need to use spacers underneath the crystal and/or modify
your clean-up process.
Also, make sure you are setting the oscillator fuse appropriately. And you
might want to experiment with the effect of a resistor in parallel with
the crystal. Both these things affect oscillator feedback, which is an
ingredient in determining how rapidly and reliably the oscillator starts.
> About 5% of my production run is having
> problems with start up of the oscillator. I am
> using a 4 MHz parallel crystal Digikey part number
> X006-ND, with a pair of 22pf NPO ceramic chips
> on either side to ground. Local bypass is good.
- Check clock mode is the correct one
- Power via an I/O pin before power on could cause this as
could slow power supply rise time etc. ((A brown out protection
circuit is always a good idea)).
- Having C2 higher than C1 can help.
- Other values than suggested in the data book may be
appropriate for your design.
> About 5% of my production run is having
> problems with start up of the oscillator. I am
> using a 4 MHz parallel crystal Digikey part number
> X006-ND, with a pair of 22pf NPO ceramic chips
> on either side to ground. Local bypass is good.
- Check clock mode is the correct one
- Power via an I/O pin before power on could cause this as
could slow power supply rise time etc. ((A brown out protection
circuit is always a good idea)).
- Having C2 higher than C1 can help.
- Other values than suggested in the data book may be
appropriate for your design.
>> About 5% of my production run is having
>> problems with start up of the oscillator. I am
>> using a 4 MHz parallel crystal Digikey part number
>> X006-ND, with a pair of 22pf NPO ceramic chips
>> on either side to ground. Local bypass is good.
A tip I heard at an MCT seminar is that the amplitude on both sides of
the xtal should be about the same for most reliable operation -
obviously you'd need a low-capacitance probe (10M R in series with
normal probe?)
> A tip I heard at an MCT seminar is that the amplitude on both sides of
> the xtal should be about the same for most reliable operation -
> obviously you'd need a low-capacitance probe (10M R in series with
> normal probe?)
Actually, even if you want a low-capacitance probe, you need to
correct its frequency response by providing a capacitance across your
series resistor and adjusting it so the capacitance ratio equals the
resistive division ratio.
The resistance ratio is that of the series resistor to the input
resistance of the CRO, while the capacitance ratio is that of the
trimmer cap across the series resistor to that of the coaxial cable AND
CRO input. The input attenuator of the CRO is quite complex with all
these caps and things. In general, a "10X" probe divides the input
capacitance by ten and so with any other ratio.
What I have often meant to build is an active probe using one of the
high frequency LM series buffers or UHF MAR devices for this particular
job of monitoring oscillators. Given that only voltages of 0 to +5 or
perhaps +10 are involved, this should be quite easy.
AVENUETECH.COM> writes -
