[time-nuts] low noise multiplication to 100 MHz

EWKehren at aol.com EWKehren at aol.com
Tue Jan 26 02:56:42 UTC 2016


We have looked at the LMK devices but with my 74 years would not try to  
solder it. There are other neat parts out there but again who is able to 
solder  them.
Bert Kehren
 
 
In a message dated 1/25/2016 8:11:14 P.M. Eastern Standard Time,  
bill at hsmicrowave.com writes:

Aaah -  but then you need a microprocessor (and its noise if you're not 
careful)  to control it. IMHO - too complicated an approach. Hard to beat 
a  "careful" straight multiplier approach for simple or a phased locked 
100  MHz VCXO for the best phase noise.

Bill - N6GHz

On 1/25/2016  9:20 AM, Graham / KE9H wrote:
> There are clock distribution parts  designed to do this low noise 
frequency
> conversion and  distribution.
>
> Consider TI  LMK04100
>
>  Ignore PLL1
> Put your 10 MHz as the reference input to PLL2.
>  Set Internal VCO to ~1200 MHz
> Set the internal dividers to get 100 MHz  out, and 10 MHz back to the PLL2
> phase detector.
>
> Get  reasonable noise and 100 MHz output with your choice of 2VPECL, LVDS,
>  LVCMOS output levels.
>
> If you have a dirty input  clock/reference, or multiple sources, you can 
use
> PLL1 and an external  crystal in a VCO to clean it up before you multiply 
it
> to 1200  MHz.
>
> And you can get up to four other frequencies out of the  part at the same
> time.
>
> 150 fs class  jitter.
>
> $13 cost, quantity one.
>
> ---  Graham
>
> ==
>
>
>
>
>
> On  Mon, Jan 25, 2016 at 9:22 AM, Bert Kehren via time-nuts <
>  time-nuts at febo.com> wrote:
>
>> If not good enough an XOR  with filter and one of the Crystek VCXO's
>> previously mentioned may  do it.
>> Bert Kehren
>>
>>
>> In a  message dated 1/25/2016 10:01:33 A.M. Eastern Standard Time,
>>  magnus at rubidium.dyndns.org writes:
>>
>> Also, it  will  be systematic, with idle tones. Because of the delay
>> elements  used,  they will not be long-term static but move  around.
>>
>> I agree, this is  quite noisy. If the  noise is tolerable, it is indeed a
>> small solution. 100  ps  1-sigma for 5 MHz in 100 MHz out isn't what I
>> would consider   low.
>>
>>  https://www.idt.com/document/dst/570-datasheet
>>
>>  Cheers,
>> Magnus
>>
>> On  01/24/2016 11:12  PM, Bruce Griffiths wrote:
>>> Unfortunately the ICS570   (like all zero delay buffers) has an output
>> jitter approaching  about 1000  times the likely RF ADC internal sampling
>> jitter.  The resultant SNR  degradation may be a little excessive for  
this
>> application..
>>>    Bruce
>>>
>>>
>>>     On Monday, 25 January 2016  11:00 AM, Bert Kehren via  time-nuts
>> <time-nuts at febo.com>   wrote:
>>>
>>>    With all the discussions  in a small  100 MHz source I asked my 
project
>>> partner  Juerg in Switzerland to run  some data on the ICS 570 that we  
use
>> on the
>>> majority of our projects  with  excellent results. Using the HP53132A we
>>  see
>>
>>> + - 1  count at E10-11 ignore the large  jumps those come from the Tbolt
>>>   frequency   change to correct the 1 pps. Depending on the application
>>  this  is an
>>> excellent  device.
>>>  Bert  Kehren
>>>
>>>
>>> In a message  dated 1/23/2016 6:02:23 P.M.  Eastern Standard Time,
>>>  dk4xp at arcor.de writes:
>>>
>>> Am    22.01.2016 um 22:40 schrieb jimlux:
>>>> the oscillator is a  HCMOS  output,  so figure swinging about 3.5V
>>>>  Output.. I'm feeding  differential clock  inputs on ADCs.  I'll  bet a
>>>> +/-  300mV swing would   work.
>>>>
>>>>> 4)Title said "Low   Noise"  needs better  definition as to what kind  
of
>>>>>   noise and how far down. Are we to   be  concerned about harmonic and
>> spur
>>>>>  content as compared to  real random white   noise?
>>>> This is time-nuts.. it has to be    perfect..
>>>>
>>>> But realistically, my source  is probably  going to be  about -90dBc/Hz
>>>> at 1  Hz, -125 at 10Hz, -145 at  100 Hz.  I'm  going up by a factor  of
>>>> 10, so I'd expect  20 dB worse plus a   little..(nothing is perfect,
>> eh?)
>>>> Call it  maybe -100 to -95 at  10 Hz, -125 to  -120 at 100 Hz and  so
>> forth.
>>>> harmonics are    interesting: it's the sample clock into an ADC. So
>>>>  harmonics of  the  100 aren't a big deal.  harmonics of the 10  or 20
>>>>   are.  If  you have significant  90 or 110 contaminating the 100,  
then
>>>> you get   weird spurs..  (I had this problem on a  software radio  
where
>>>> the  50 MHz sample clock was  contaminated  with some 66 MHz from the
>> CPU)
>>>>    Spurs cause the same issues.
>>>>
>>>> ON the  other   hand... spurs that are pretty low don't make  much
>>>> difference  if  you're digitizing a signal  that is close to the noise
>>>>   floor: the  spur  multiplied by the desired signal is usually lower  
 and
>>>> down in the  noise.  Strong CW in band  signals,  though, are a real
>>  pain.
>>>>
>>> <
>>>
>>  
https://picasaweb.google.com/103357048842463945642/Tronix#607927018804883377
>>>    8
>>> I think that top left board would not be far   away:
>>>
>>> in :  10 MHz LVDS or  CMOS
>>> in:   3V3
>>> out: 100 MHz CMOS  3V3
>>>
>>> just a  few hours wall  clock  time from layout to working as a
>>> ham radio weekender,   so  please excuse my diy home board
>>> production  process.
>>>
>>>   Ok, the use  of a  4046 descendant may not be the last word
>>> from a  timenut  perspective,  but I'll redo it with an osc of
>>> my  own  anyway. Divider 100/10 is a LVC163  (161?) +   lvc04.
>>>
>>>
>>> <    http://www.crystek.com/crystal/spec-sheets/vcxo/CVHD-950.pdf    >
>>>
>>> Digi-Key has 153 of them on a tape  and  441 of a  similar one  , even
>>> cheaper that  seems to point to the same data  sheet.
>>>
>>>  <
>>>
>>  
http://www.digikey.de/product-detail/de/CVHD-950-100.000/744-1213-ND/1644128
>>>    You can get the few dB missing close-in by transfer from your
>>  reference.
>>> In the picture:
>>> The bottom row of  boards is  a doubler  100->200 MHz using 2*BF862,
>>  slight
>>> gain,
>>>   and diode doubler 200  ->  400 MHz, SAW filter to get rid of
>>>    100/200/300/500/600 +/-10  etc,
>>> post amp to get a usable  level  again.
>>>
>>> Still missing   400-> 800, 800->1600 to  feed  _my_ ADC clock  input..
>>>
>>> regards,    Gerhard
>>>
>>>    _______________________________________________
>>>  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.
>>>
>>>
>>>    _______________________________________________
>>> 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.
>>>
>>>
>>>    _______________________________________________
>>> 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.
>>>
>>  _______________________________________________
>> 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.
>>
>>  _______________________________________________
>> 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.
>>
>  _______________________________________________
> 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.

_______________________________________________
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_lists.febo.com mailing list