Bob Bownes wrote:
Comments inline.
On Mon, Dec 20, 2010 at 11:17 PM, Chris Albertson<albertson.ch...@gmail.com
wrote:
I looked. I think we should keep the design goals modest for a first
revision. Shoot for a spec that can be hand built on perf board. So
I'd relax those numbers by a factor of 1000. The top frequency is in
Mhz, not Ghz and the time resolution closer to ns than ps. It's good
to have a cheap option. Many people are happy with an FCC1 Try for
the next step after that with a goal of actually matching the state of
the art in steps.
My initial thinking was to be better than a pictic ii, preferably on par
with, or better than a 5370. I'm not sure you can do that on perfboard. I
suppose if the speed is kept low it can be done that way.
Around 25ps jitter is the likely lower limit with such construction
techniques.
Others have pointed out (offline) 20Ghz isn't reasonabe with a decent noise
figure or prescalers. What do people think a reasonable number is? What
about resolution? I'd like to get better than a ns, preferably a lot better.
1ps resolution is trivial just use a sufficiently high resolution ADC
(16bit) with a short TAC interpolator range (10ns?)
Achieving a commensurate jitter is somewhat trickier.
Around 3ps or so rms short term jitter should be easy enough, the
Wavercrest 2075 does this without using anything particularly esoteric.
10ps jitter should be very easy.
The latest Agilent time interval counters counters achieve around 9ps or so.
1ps rms jitter shouldn't be beyond reach (at least with the ringing tank
method).
However a good layout together with at least a 4 layer board will likely
be necessary.
Adequate grounding and shielding will also be necessary.
Why no through holes? I don't see the point of banishing them. I to
agree with the rest. SMT that is hand solderable by a skilled tech
but now reflow ovens or solder past masks should be required . You
might place a limit on component size too like 0.5mm lead pitch or
whatever is reasonable.
I've been prototyping a lot of late and restricting the number of through
holes makes the job much much easier and quicker. No other real reason.
Precluding the use of RF dams (these use arrays of plated through
holes) is probably counter productive.
I didn't place a limit on the lead pitch because a) I felt that limited the
component selection and b) pretty much even the finest pitch can be hand
assembled with care, solder wick, and 20x magnification. But if folks are
very against it, it can go in the 'desired qualities' list. My only fear is
the limit it might put on critical parts like a FPGA.
Mechanical assy is going to be a killer. Let's start with overall form
factor.
Rack mount or bench format?
If rack mount, 1U or more?
Commercial project enclosure (ala the VNA Hammond box) or do we take
an existing form factor like a disk drive as you suggest.
Heck, if we go with a disk drive size, it could be built/slid into anyone's
lab PC case and use ribbon cable as a back plane... 1/2 :)
Inadequate shielding?
I like 1U because it matches up with the rest of the test equipment on the
bench and it gives it a professional feel. And there are many many surplus
1U cases out there with decent +/-12vdc,+5vdc (even some with 3.3vdc) power
supplies.
Linear supplies or perhaps extremely low noise/well filtered switchmode
supplies will likely be necessary.
Bob
Bruce
On Mon, Dec 20, 2010 at 6:20 PM, Bob Bownes<bow...@gmail.com> wrote:
Well said Chris. Take a look at the initial spec in the OpenCounter Gogle
group and tell me what you think with respect to your item #1. I think
the
core counter is going to be the really difficult part of the module list.
Item #2 is going to be a tough one methinks. I love Eurocard, but, as you
say, it is very expensive, if only for the connectors. In cases like this
I'm a fan of either repurposing commercially available connectors (PCI
and
memory DIMMS are two I have used in the past) because they can be a)
purchased off the shelf, b) are manufactured in enough volume to make
cheap,
and c) are common enough that the really cheap amongst us can get them
off
of scrap boards someplace for little or nothing. The N2PK VNA is built to
fit into a particular HAmmond enclosure that I like but again, there are
many options. My feeling is that the enclosure should not dictate any
functional design decisions.
#3 - I've created a group and appointed myself benevolent dictator. We
can
discuss things, propose designs or design criteria, call for a concensus,
accept, and draft volunteers to design that section to the defined spec.
If
there are multiple competing designs, so much the better, as long as we
all
agree on the interfaces. Sound like a process? Can you tell I've done
this
once or twice? :)
Step one will be to agree on the overall functional spec. If we get
enough
participants, I'd like to nail that down by mid January. The next step is
to
agree on the interfaces between the modules. Same process, discuss,
propose,
draft, get concensus, close and move on.
Then we get folks working on the individual modules.
On Mon, Dec 20, 2010 at 8:43 PM, Chris Albertson
<albertson.ch...@gmail.com>wrote:
On Mon, Dec 20, 2010 at 2:51 PM, Dave M<dgmin...@mediacombb.net>
wrote:
.. never got off the launchpad because of their inability to come to
consensus on a set of features. I had to conclude that too many cooks
spoiled the broth. Everyone that had input to the project was
unwilling
to
yield on anyone else's ideas. Hopefully, our project won't degrade
into
another such fiasco.
THAT is the number one problem to solve. Technical issues are easy
I think the solution is to
(1) chop the project up into small enough parts, each on it's own PCB
so that each part is "easy" and has some wider user outside the
project.
(2) Find a mechanical standard so all the PCBs can be mounted in some
kind of chassis. I'm thinking now that maybe a 160-3U Eurocard would
be about right size. But the parts are expensive.
(3) Need some sort of design process that allows everyone to
contribute. And everyone can. Projects always are lacking technical
writers and quality control people
Of those a "process" and "mechanical standard", I think are the
hardest. We always give managers a hard time but that is what is
needed. The person who will make this happen will be a manager and
organizer maybe not a designer.
--
=====
Chris Albertson
Redondo Beach, California
_______________________________________________
time-nuts mailing list -- time-nuts@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@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
--
=====
Chris Albertson
Redondo Beach, California
_______________________________________________
time-nuts mailing list -- time-nuts@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@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@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
