[time-nuts] Modified Extron DA [WAS: Rb video]

[Bob Camp]

Hi

For a reasonable standard distribution, you probably want one input and many outputs. One in / eight out or one in / 12 out are fairly common. At least the video gizmo we've been dissecting has trouble past one in / 4 out. If you cascade them you are at one in to 3 useful outputs. to get to eight you do a lot of jumping from here to there. Each op amp adds it's noise in a cascade. 

Metal work is an issue, and the video boxes do have a case that comes with them. They also have the connectors already mounted to a PCB. A more custom solution would indeed require a bit of fab for a front plate and / or back plate. You would also have to buy connectors to solder to the pcb. 

With a more custom solution, you could do it "right". If 1 in 12 out makes sense, then you do exactly that. If fab is an issue, move all the stuff past one surface. Power wise, a digital DA will run off  < 1/4 A. A wall wart into a cheap linear regulator on the PCB is the much more rational solution than a built in supply at that level. If you want X out's at 10, Y at 5 and Z outputs at 1 MHz, that's trivially easy once you are running logic. They still make flip flops and divide by 10's. To go nuts, use one of Bert's favorite CPLD's and you can have 10, 5, 1, 0.1 and a couple of others, all off one chip. 

Design wise, there's not a lot to a DA. A simple double sided PCB will do the job. For the ultra cool approach, go with a 4 layer board and a power plane if you want. It will work just fine either way. The cheap board shops will sell you the board for less than the price of the DA's at auction. That's certainly true of a 2 layer, there are places that will do 4 layer (in modest quantity) for those sort of prices as well. 

If you are going to run -185 dbc/Hz phase noise signals, none of these solutions will work. For that stuff you want a totally different approach. The same is true if you are after ADEV at 1x10^-15 at 1 second. For any normal piece of test gear, you don't need that sort of thing. Commercial DA's don't do that stuff. The standards normally used  (like that Rb !!!!) are nowhere near that level. 

All I'm really trying to say here is that the alternative isn't all that tough. You can do it cheap with common parts and not a lot of effort. The time to hack up an existing video box (and do it right) may not be much less than the time to do something much simpler from scratch.

----------------

…. and before it gets asked… yes I've done it, measured it, tested it, use them every day. There have been many millions of those amps (single channel version) built and fielded over the last 20 years. 

Bob


On Aug 9, 2013, at 6:51 PM, Charles Steinmetz <csteinmetz at yandex.com> wrote:

> Bob wrote:
> 
>> I still think that a distribution amp based on logic ic's is cheaper / simpler / lower power / higher performance. A pair of NC7SZ125's will dump 20 dbm into 50 ohms all day long running at 5.5 volts. Good isolation as well. Do the lowpass filter right and the harmonics are not an issue. Two coils / one cap plus dc blocking does it quite nicely.
> 
> The filters add a potential source of close-in phase noise due to the temperature coefficients of the parts.  Granted, this may be academic in the case of a simple reference distribution, and you could always do it with harmonic traps instead of BP or LP filters to minimize the problem.  Also, I'd rather not generate fast edges in the first place if I don't need them -- they can leak out and be pesky.  But these are not religious positions for me.  In some cases the digital solution may be preferable.
> 
> However, I think the large interest in video DAs has more to do with the fact that they are cheap and often need very little modification.  When I first wanted to distribute 10 MHz, before I built my own DA, I grabbed an Extron out of the junk pile and modified it in a couple of hours.  Indeed, if you can tolerate the mismatch and the overall 2 dB loss, video DAs can often be pressed into service with no modification (not my preferred solution, but it's been done).  Digital DAs, OTOH, are more for the person who is rolling his or her own from the ground up.
> 
> Best regards,
> 
> Charles
> 
> 
> 
> _______________________________________________
> 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.