[time-nuts] hm H Maser
kb8tq at n1k.org
Tue Jan 10 20:09:44 EST 2017
The gotcha is that 5065’s never were a popular item in HP’s lineup. As a
result, they are fairly sparse in the surplus market. Those who need them
for this or that application gobble them up on a regular basis. Trying to do
up a couple hundred “improved” 5065’s just isn’t going to happen (at least
without driving the current price up by > 10X or 100X).
Since about the only thing you keep from the 5065 once you are done is the
physics package, that’s a big payout for very few usable parts. You then
modify (and possibly repair) the physics package. If we ever get into this, you
also replace a few parts in there to improve it’s performance. Now you have
even fewer “keeper” parts.
Decide you want a state of the art Rb (what other goal would there be?)
Organize the team
Work out a first pass design
Find a source for *large* Rb cell sets.
Work with them to get the cells right
Design up a physics package in parallel with this effort
Get it all prototyped multiple times and debugged with lash up electronics
Test for about a year once you have the prototype debugged
Order up the tooling on the long lead stuff (cells and some machined parts)
Get the real electronics working in some form
Debug the electronics against the real cells and parts
Test for about a year once you think it’s working
Do the real layouts and packaging, including shielding and all the other nasty stuff
Fit up the first unit
Test for about a year to be sure you have caught all the issues
Redo what is needed
Start building the hundred or so units on order with the cash on hand from those orders.
Lots of fun !!
I’m sure somebody will chime in at this point and claim they can do that all
for about $100 a unit. If so feel free to try. It’s simply liars poker at that point
since nobody ever has to actually do it. Based on having done it and on having
seen others do it … it is not at all cheap to do. Rb *is* cheaper, but it’s still not free.
You might also question the “test for a year” stuff. If you want ADEV style data that has
any meaning, you need sample sizes that are in the 10 to 100X tau range. For a one
week tau, each run will be > 3 months. Testing takes time…..You also need to be
testing multiple units to get any confidence. That takes money.
Even more fun.
> On Jan 10, 2017, at 7:40 PM, Gregory Maxwell <gmaxwell at gmail.com> wrote:
> On Wed, Jan 11, 2017 at 12:25 AM, Bob kb8tq <kb8tq at n1k.org> wrote:
>> I have a pile of stuff. You have a pile of stuff. Others each have their pile of stuff. Doing
>> a design that works only with my pile is possible. Doing a design that works with my pile
>> You have to do it with a fairly standardized
>> design. That means buying (at the very least) kits of parts. Like it or not, the parts kit for a
>> Rb will be cheaper than the parts kit for any of the other devices…..
> I read the occasional posts by PHK on his efforts to upgrade the
> electronics in his 5065a and Corby's SUPER physics package upgrade
> with great interest. I have wondered if the end result may be that
> incremental upgrades to someone elses classic design, adding on modern
> synthesizers and digital control, etc. Might eventually result in a
> 'Ship of Theseus' oscillator, which in its final form is buildable
> from relatively easily sourced parts (plus perhaps a rubidium cell
> that could be group bought at non-absurd prices).
> Presumably taking an already established design and improving it
> incrementally has lower risk and costs than a new design. In
> particular, it can start off with 5065a as "my pile" inputs, but by
> the end it doesn't have them anymore... and not just lest risky but
> also a more natural way to divide the effort up into less
> professionally-sized chunks.
> 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