[time-nuts] Clock level conversion 5V -> 3.3V

Mark A. Haun haunma at keteu.org
Thu Oct 2 23:56:32 EDT 2014


Thanks to everyone who replied.  I probably should have explained that
I am familiar with the various options for logic level conversion.
It's how those options affect clock noise that I was fuzzy on.  Bob's
summary definitely helped.  SN74LVC1T45 looks to be the winner.

Regards,

Mark

On Thu, 2 Oct 2014 07:07:41 -0400
Bob Camp <kb8tq at n1k.org> wrote:
> Hi
> 
> It will indeed be better for phase noise to do away with the
> resistive divider and get faster edges. 
> 
> Of course there are indeed resistive dividers that don’t slow things
> down. It’s unlikely that a divider with a 10 ohm output impedance is
> going to tack on to the output of an OCXO.
> 
> ————————
> 
> The real point is that (in this case) you can get the job done for
> less than 10 cents with a single gate chip. They are available in
> many packages from many people. There are indeed chips that have a
> bit less noise than others. With the OCXO that we’re talking about
> here, it’s not worth going crazy to find this or that. This being
> Time Nuts, if you do decide to go crazy - as long as it’s saturated
> silicon, faster is more quiet than slower. 
> 
> Bob
> 
> On Oct 1, 2014, at 9:21 AM, David McGaw <n1hac at dartmouth.edu> wrote:
> 
> > Would it not be better for phase noise to use a logic gate with a
> > fast transition than a resistive divider that would be slower due
> > to the load capacitance?
> > 
> > David
> > 
> > 
> > On 10/1/14 7:09 AM, Bob Camp wrote:
> >> Hi
> >> 
> >> Ok, so it’s not a super duper low phase noise OCXO. It’s also at a
> >> reasonably high frequency.
> >> 
> >> I’d just drive it into a 5V tolerant input and move on. There are
> >> lots of logic gate chips out there that will run from 3.3 and
> >> accept 5V inputs. Use something reasonably fast and it will do a
> >> pretty good job.
> >> 
> >> Bob
> >> 
> >> On Sep 30, 2014, at 10:11 PM, Mark A. Haun <haunma at keteu.org>
> >> wrote:
> >> 
> >>> Hi Bob,
> >>> 
> >>> The OCXO is one of those 26-MHz ebay Pletronics from a couple
> >>> years back.  I would like to not degrade its close-in phase noise
> >>> (quoted as -100 dBc @ 10 Hz, -130 dBc @ 100 Hz).  Thinking about
> >>> Said's suggestion to phase lock a higher-frequency sampling clock
> >>> to this, with a loop BW somewhere in the 10-100 Hz range.
> >>> 
> >>> I have seen a resistive divider used in a similar application, but
> >>> wondered if I could save the couple dozen mA they were spending.
> >>> 
> >>> Mark
> >>> 
> >>> On Tue, 30 Sep 2014 20:18:56 -0400
> >>> Bob Camp <kb8tq at n1k.org> wrote:
> >>>> Hi
> >>>> 
> >>>> How quiet does it need to be?
> >>>> 
> >>>> Put another way - how good is the OCXO?
> >>>> 
> >>>> What frequency are we talking about?
> >>>> 
> >>>> What is the phase noise “need” after you get to 3.3V (is there a
> >>>> system spec)?
> >>>> 
> >>>> Bob
> >>>> 
> >>>> On Sep 30, 2014, at 5:46 PM, Mark Haun <haunma at keteu.org> wrote:
> >>>> 
> >>>>> Is there a "best" way to do this without adding phase noise?
> >>>>> For example, a 5V OCXO into an ADF4002, or a 3.3V or even 1.8V
> >>>>> logic input.  Is a resistive divider the way to go?
> >>>>> 
> >>>>> Mark


More information about the time-nuts mailing list