[Libre-silicon-devel] Clock generation

[Hagen SANKOWSKI]

Hello David.

On 08/20/2018 06:19 AM, David Lanzendörfer wrote:
> Hi Hagen
> We could theoretically just use a Pierce oscillator, because you've already 
> drawn the inverter circuit and could test this way, which one performs best to 
> generate a square clock signal?
> What do you think? An idea for a test structure?
> https://en.wikipedia.org/wiki/Pierce_oscillator


Well, I am not sure, what you want test with that oscillator.
Usually, this kind of quartz crystals are used since the eighties for
microcontroller, often with both input pins. The frequency is determined
by the crystal while the inverter is used to "shape" (having a
significant magnification, the sinus waveform gets over-amplified -
coined "endless") into a square waveform with cut-off areas (limited by
the power supply).

We can not measure many helpful stuff out of this oscillator for our
test waver. Or I do not see the point.

On Pearlriver we already have a space problem. The test pads consumes a
lot of area for itself and for the distance between each other. We
should concentrate on the values we like to get - what are resistance
and capacitance for Poly, N+, P+, Metal etc and transistor parameters.
If all layer processed fine it would be our bonus to see the ring
oscillators running and measure the delay times over the minimal-sized
gates.

Regards,
Hagen.

P.S.: Usually the inverter is a Schmitt trigger
(https://en.wikipedia.org/wiki/Schmitt_trigger) which has internally an
extraordinary circuitry. Designing the StdCellLib further, a Schmitt
trigger is on the list for input pads. Than I like to measure the
hysteresis - which I can't with an oscillator.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://list.libresilicon.com/pipermail/libre-silicon-devel/attachments/20180820/11de4f67/attachment.sig>