[time-nuts] Thunderbolt? (re simple gpsdo.)

David davidwhess at gmail.com
Fri Dec 30 21:48:11 EST 2011

Did you test the LTC1655 INL?  The data sheet says plus or minus 20
counts maximum.

I suspect Linear Technology designed those low DNL high INL parts for
just this sort of application where only monotonic behavior really
matters.  Their equivalent current output DAC costs about twice as
much not including a precision transimpedance amplifier but has an INL
specification of plus or minus 1 count.

Every couple years I consider the design of a digitally adjusted
oscillator and do a search for likely parts.  I wonder if it would be
more cost effective to use an instrumentation ADC to correct a less
expensive DAC design like one based on a PWM.

On Fri, 30 Dec 2011 17:48:24 -0500 (EST), EWKehren at aol.com wrote:

>Over the last two years along with two list members that may want to pipe  
>in, I have spend a large amount of time on D/A's and we went as far as  
>developing a test board using the LTC 2440 and testing numerous D/A's taking in  
>to consideration performance, solderability, cost, availability and the 
>winner  is LTC 1655 by a long shot, is even available in a DIP with 16 bits 
>more than  you need for any Rb and if you want 20 bits, dithering is an option. 
>My testing  consistently shows with OCXO's aging that will in most cases 
>allow operation of  an OCXO for 3 years with out intervention. To top it off 
>the LTC1655 cost less  than $ 10. Testing the old AD 1861 was an eye opener 
>but considering what its  purpose was and its time the best choice.
>In a message dated 12/30/2011 4:24:37 P.M. Eastern Standard Time,  
>timenuts at n4iqt.com writes:
>The DAC  and it's voltage reference looks to be the weak link in the 
>control and the "simple" goal. The CPU I mentioned before on closer look  
>doesn't have a good DAC. The 20 bit TI DAC1220 looks better but not sure  
>can find it in the same package as the CPU. The cheap Rb standards  with 
>digital control would not need a DAC and maybe this points to a  simpler 
>GPSDO that doesn't control the XO with analog but corrects it with  a DDS 
>again finding them both in one chip is the problem. I have seen  OCXO and 
>in the same package and even the DDS and OCXO combined but  they didn't fit 
>the simple goal. Not even sure how good they were. I know  they are hard to 
>time-nuts  mailing list -- time-nuts at febo.com
>To unsubscribe, go to  
>and follow the  instructions there.
>time-nuts mailing list -- time-nuts at febo.com
>To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>and follow the instructions there.

More information about the time-nuts mailing list