[time-nuts] Using commercial video amplifier for 10MHz clock distribution.

[Charles Steinmetz]

Luciano wrote:

>     The CLC409 datasheet show all the test response up to +12dBm (around 1Vpp on 75/50Ohm load)
>     see: pdf.datasheetcatalog.com/datasheet/nationalsemiconductor/DS012748.PDF

I'm reporting what I have observed with the CLC409 (and LMH6702) when 
operated on the "absolute max" supply rails of +/- 6v.  Of course, I 
would never design a commercial product using abs max specs without 
having a serious conversation and understanding with the manufacturer 
(if then).

Clipping commences at just over +/- 2v (a touch over +16dBm).  This is 
with a 50 ohm build-out or "back termination" resistor -- you can get an 
additional 6dBm if you do without the build-out resistor and terminate 
only the load end of the cable in 50 ohms, which actually works fine as 
long as the output loads are matched at 50 ohms.  I terminate both ends 
"just in case" they aren't.  Harmonic distortion does not rise until 
there is visible clipping.

>     I agree that harmonic distortion is an important factor to keep in mind however -45dBc is a very interesting value considering the market alternatives. In my document I have enclosed on page 10 a little research I did online on similar products and as you can see the -45 are positioned at a decidedly higher level than other commercial solutions
>     see [table]:

IMO, pretty much all commercial distribution/isolation amplifiers fail 
miserably to hold distortion down to tolerable levels -- one reason why 
I have always designed my own.  Interestingly, some older, vacuum-tube 
models did better.

>     An extremely important factor not to be overlooked is the maximum output power / level of the amplifier. Measurement systems such as the HP3048A for phase noise measurement use mixers that require minimum signal levels of + 15dBm at higher PO1 and the CLC409 will never meet these requirements.

As noted above, it will if it is operated on abs max supplies, even with 
a 50 ohm build-out resistor.  Without the build-out resistor, it will 
supply +22dBm (+20dBm on +/- 5v supplies).

>     In your modification document: Residual phase noise of <-165dBc at 10 MHz
>     I suppose you mean @100kHz, The buffer I propose is -170dBc @100KHz

I normally cite 10kHz measurements, but I don't recall if I cited 100kHz 
or 10kHz measurements (my bad for the incomplete spec).  One or the other.

>     In the end I don't think this should be a competition among Distribution Amplifiers.

Nor I.  I built a few dozen of the modified Extron DAs, which I've 
passed on to other time nuts over the years.  I still have one around, 
but it is not in daily use.  I currently use two different designs, both 
scratch-built, one IC-based and one discrete.  As I've said before, I 
distinguish between distribution amplifiers (lots of outputs) and 
isolation amplifiers (where only 2 or maybe 3 outputs are needed, but 
distortion, residual PN, and transit delay must be truly 
state-of-the-art for sensitive measurements).

On re-reading my original post, I see the tone could be taken as a bit 
confrontational.  That was certainly not my intent -- I just wanted to 
defend IC designs against some "received wisdom" (*NOT* from you) that 
they are inferior.  Thank you for giving me the benefit of the doubt and 
responding as a gentleman.

>     For a "ready" solution at low cost like DA the Extron I think is a great solution.

Me, too, which is why my first efforts as a time-nuts newbie were made 
in that direction.

Best regards,

Charles