Order Tray | Contact Us | Home | SIG Lists

[time-freq] Query: Interest in clock synthesizer module -- useful for stabilizing PC timekeeping (among other things)

John Ackermann N8UR jra at febo.com
Sat Aug 19 13:56:05 UTC 2006


Hi Mark --

Unfortunately, the phase noise of the ICS525 synthesizer is fairly
horrible (you can see plots at
http://www.febo.com/time-freq/hardware/ICS525).  When I talked to the
factory engineer, he made it clear that the chip was designed for timing
and not RF applications.  So, I don't think plugging it into the IC-746
will be a good idea.

It's also possible that the phase noise will be insufficient for good
sound card performance, but since the signal is being divided so much to
get to the sample rate, that may improve the phase noise enough to make
it workable (multiplying or dividing a signal by 10 yields a 20dB
increase or decrease in phase noise.  If the onboard crystal was running
at 10MHz, and the sample rate is 96kHz, that should improve the phase
noise by 40dB.  However, I haven't actually tried this yet...

The production board will include a 74HC4020 ripple divider chip tied to
the output, so by using that as a divider (it can divide by up to
16384), you can generate a low frequency square wave that, again, should
be pretty clean because of the division ratio.

John
----

Mark Miller said the following on 08/19/2006 09:46 AM:
> John,
> 
> I would be interested in this kit as a replacement for my IC-746pro's 32
> MHz reference frequency.  It would be nice to be able to bring in a very
> precise and stable 10 MHz reference and have it converted to a 32 MHz
> clock for this rig.  I have another rig that uses a 30 MHz reference.
> 
> I very much like your idea of using it as the clock for a PC and sound
> card.  I have often wished for very stable and accurate sampling rates
> for sound cards.
> 
> Great idea.  I have no problem with a surface mount kit.
> 
> 73,
> 
> Mark N5RFX
> 
> At 03:49 PM 8/16/2006, you wrote:
> 
>> I'm working on a project for TAPR (http://www.tapr.org) that might be
>> of interest to the timekeeping community.
>>
>> It's a very simple clock synthesizer called the "Clock-Block" that
>> accepts a reference input in the roughly 2-50MHz range and generates
>> an output in the 5-250MHz range, programmable by a series of jumpers
>> or switches.
>>
>> My primary inspiration for designing this circuit was Poul-Henning
>> Kemp's suggestion that it would work nicely to replace the crystal
>> oscillator on a PC to allow much more stable timekeeping.  For
>> example, it can generate 14.318182MHz or 33.333333MHz from a GPS
>> disciplined oscillator or other stable reference.  A PC with a
>> Clock-Block and external reference should be able to keep *very* good
>> NTP time.
>>
>> But there are likely to be other interesting uses as well; one I'm
>> considering is replacing the clock in a sound card to get precise
>> sample rates.
>>
>> There's information about my prototype (and a picture) at
>> http://www.febo.com/time-freq/hardware/ICS525/
>>
>> The prototype works well and we are now ready to move to production.
>> Before we do that, I'd like to get a sense of the interest level (if
>> any), and in particular whether there's enough demand to warrant an
>> semi-kit or assembled version instead of TAPR's usual bag o' parts.
>>
>> As a kit, the Clock-Block would be quite inexpensive; we haven't
>> finalized the BOM yet but I believe TAPR would be able to sell it for
>> well under $50.
>>
>> However, assembly may be problematic for some folks, because the
>> synthesizer chip is a "SSOP" surface mount package with about 0.5mm
>> lead pitch.  I've had no trouble putting down these chips using a
>> microscope and small iron but not everyone may be willing to take on
>> that challenge.
>>
>> I am guessing that given the likely low volume, a semi-kit (all the
>> surface mount parts soldered down, but the remaining parts -- two DIP
>> switches and four 2 pin headers -- left as an exercise for the buyer,
>> would cost in the range of $80-$100.
>>
>> A fully assembled and tested version would probably be $100-120 (most
>> of the increased cost is testing, not soldering down the few remaining
>> parts).
>>
>> I've tried to be conservative in these prices and I hope the actual
>> price will be lower, but don't know that for sure at this point.  A
>> lot depends on quantity so there's a bit of chicken and egg here.
>>
>> I'm hoping to collect no-obligation expressions of interest from this
>> group, assuming prices in the ranges described above.  If you'd be
>> interested in buying one or more Clock-Blocks, please let me know
>> privately how many, and what version.
>>
>> Thanks!
>>
>> John Ackermann
>> jra at febo.com
>>
>> _______________________________________________
>> time-freq mailing list
>> time-freq at lists.tapr.org
>> https://lists.tapr.org/cgi-bin/mailman/listinfo/time-freq
>>
> 
> 
> _______________________________________________
> time-freq mailing list
> time-freq at lists.tapr.org
> https://lists.tapr.org/cgi-bin/mailman/listinfo/time-freq
> 
> 





More information about the time-freq mailing list