On 12/6/12 5:01 PM, Bob Camp wrote:

I think that's the interesting aspect of this discussion (because it
mirrors many similar such discussions I've had over the years)..

The challenge is not in building a GPSDO.. there's tons of ways to do
that at minimal hardware cost..

The challenge is "how do you do it without using any additional glue
logic or hardware"  That is, given things you can buy off the shelf, and
no hardware work other than fabricating cables or soldering a wire or
two, what can you do inexpensively.

On 12/6/12 11:20 PM, Don Latham wrote:

This is a general problem with eval boards these days.. They provide a
lot of functionality on the board to make it easy to evaluate the chip
(USB interfaces, buffer memories, etc.)  but that makes it hard to use
the eval board as a sort of glorified chip carrier.

For instance, all those nice serial interface ADC and DAC parts.. it
would be nice to have a little board with the converter, power supply
filtering and maybe an opamp buffer, and bring the serial interface pins
to the edge where you could just wire it to something like a PIC eval
board or Arduino or parallel printer port.  But no.. they have a weird
connector that goes to a mother board with a fancy preprogrammed micro
and dual port memory and stuff.. all so you can just hook up a signal
generator and capture samples to run FFTs to duplicate the databook graphs.

Jim right on target for my 2 cents,
Simple is often hard.
Regards
Paul
WB8TSL

On Fri, Dec 7, 2012 at 10:12 AM, Jim Lux jimlux@earthlink.net wrote:

Hi

Without some qualifier on "how good", we all will be talking about
hourglasses and cesium standards at the same time. That makes sorting things
out a bit difficult. It may get you into building a "controlled hourglass"
that's less accurate than a free running $10 wrist watch.

So far the only suggestion of a "how good" has been TBolt performance. At
least that's how this started out.

Building a "cpu only" GPSDO can be done with a big enough chip. If you are
limited to clip leads to wire it up, the cost advantage over a TBolt may not
be there. That makes the issue of significant performance loss a very real
issue.

Just as surplus deals on TBolts come and go, so do deals on OCXO's and
GPS's. Do I price the OCXO at the $13 that I once saw it for, or the $70
that they sell for today? Is the GPS in the junk box (that I paid $200 for)
free because it's sitting there? Let's not even wonder if any of the stuff
works...

For the sake of moving this forward, I'll suggest that half the fair cost of
a (working) TBolt is the OCXO and the GPS. Put another way, your $200 (or
what ever) is split $100 for the OCXO plus GPS and $100 for:

1. The nice shielded case it's all in
2. The connectors
3. The onboard power regulation and filtering
4. The communications in and out of it
5. The 10 MHz and PPS buffering and generation
6. The PC board it's all built on
7. Putting it all together
8. Programming all the parts in there
9. Testing it and troubleshooting it
10. The smarts to get it to do what it does

I'm sure we could spend at least a year splitting hairs on that division of
costs :)

To "win" the hourglass vs cesium fair trade proposition you need to come up
with a gizmo that:

1. Does 1-10 above
2. Works with just a GPS and OCXO added.
3. Costs less than $100
4. Works as well as a TBolt
5. Has a Lady Heather like monitor program to go with it

For the sake of eliminating crazy GPS cards. Let's say you also need a
source where we can buy both the OCXO and GPS used for less than $100.

If you want to factor in lower performance, I'll grant a waver for the fact
that you can average multiple gizmos and the result will get better by the
square root of the number of gizmos. If it's half as good as a $200 TBolt it
needs to be 1/4 the price.

Bob

-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of Jim Lux
Sent: Friday, December 07, 2012 10:13 AM
To: time-nuts@febo.com
Subject: Re: [time-nuts] GPSDO Alternatives (GLUE)

On 12/6/12 5:01 PM, Bob Camp wrote:

$5 for pretty much all of them. No need to re-invent the wheel. The gotcha
is that you can't do it 100% with internal CPU peripherals. You will need
some glue.

I think that's the interesting aspect of this discussion (because it
mirrors many similar such discussions I've had over the years)..

The challenge is not in building a GPSDO.. there's tons of ways to do
that at minimal hardware cost..

The challenge is "how do you do it without using any additional glue
logic or hardware"  That is, given things you can buy off the shelf, and
no hardware work other than fabricating cables or soldering a wire or
two, what can you do inexpensively.

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.

Bob
Yes would agree these are the attributes of a solution thats interesting.
Numbers of folks have created solutions over the years. But the
reference architecture should be as good as or better then the tried and
true tool we have like the HP 38XX or Tbolt. Otherwise is just another
GPSDO.
Regards
Paul
WB8TSL

On Fri, Dec 7, 2012 at 12:15 PM, Bob Camp lists@rtty.us wrote:

Hi

At least one should compare similar to better than approaches. This is a
multi year design. Pick a goal first, then try to fit a specific approach to
the goal. If the approach makes sense, move on with the design.

Bob

-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of paul swed
Sent: Friday, December 07, 2012 1:38 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] GPSDO Alternatives

Bob
Yes would agree these are the attributes of a solution thats interesting.
Numbers of folks have created solutions over the years. But the
reference architecture should be as good as or better then the tried and
true tool we have like the HP 38XX or Tbolt. Otherwise is just another
GPSDO.
Regards
Paul
WB8TSL

On Fri, Dec 7, 2012 at 12:15 PM, Bob Camp lists@rtty.us wrote:

for)

of

up

fact

below

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.