[time-nuts] Austron 2201, Tbolt, HP 3801 comparison question

[Tom Van Baak]

Hi Ulrich,

I want to re-iterate how difficult it is to compare one make of GPSDO
with another. A lot depends on antenna, and software configuration,
environmental controls, and the particular OCXO that you happen to
get with the unit. You can see significant difference in N TBolts; you
can see significant differences in N HP Smartclock's.

I don't believe there's anything magic about the hp Smartclock. The
main goal back then was to reduce the effects of S/A. Maybe that
was clever 15 years ago, but S/A hasn't been around for a decade.

In order to investigate the Smartclock algorithms in detail it would
be possible to replace their algorithm with your own. That is, take
a 58503 or Z3801 and keep the Oncore, keep the OCXO, keep
the DAC, keep the p.s., but insert your own TIC and your own
PC-based disciplining algorithm. FYI: here's info on their DAC:
http://leapsecond.com/pages/z3801a-efc/

If without too much effort you match HP's performance, then there
is no magic in Smartclock. In other words, the performance they
get is mostly the Oncore and the 10811 and a decent TIC & DAC
and nothing extraordinary about the software.

On the other hand, if after weeks of work HP still beats your best
effort, then I would agree there's something clever and hidden in
their implementation.

Realize that the HP Smartclock was one of the very first GPSDO.
Since then there have been tens (hundreds?) of different GPSDO
products, both commercial and amateur. It's really hard for me to
believe that any stone has been left unturned.

Still, I welcome anyone who wants to test the Smartclock algorithm
as suggested above. If you have the time but not the Smartclock,
let me know and I'll make a loaner available.

Thanks,
/tvb

----- Original Message ----- 
From: "Ulrich Bangert" <df6jb at ulrich-bangert.de>
To: "'Discussion of precise time and frequency measurement'" <time-nuts at febo.com>
Sent: Saturday, April 07, 2012 3:13 AM
Subject: Re: [time-nuts] Austron 2201, Tbolt, HP 3801 comparison question


> Gents,
> 
> one of the things that MAY be responsible for the differences in performance
> is that the Z3801 uses HP's Smartclock technology while the TBolt does not.
> The TBolt works with a fixed set of parameters (unless we change them)
> which's default values are far from optimal but ensure a fast lock of the
> pll. The Smartclock in difference seems to be able to adapt regulation
> parameters to its "measurements" of ocxo stability and long term drift. 
> 
> Unfortunately there are only a very limited number of publications available
> about Smartclock technology with most of them only scratching the surface.
> That is why I believe that HP & Agilent still make a big secret out of it
> even today. I guess we time nuts could learn a lot if we had an in depth
> description available on how Smartclock works.
> 
> Best regards
> Ulrich Bangert 
> 
>> -----Ursprungliche Nachricht-----
>> Von: time-nuts-bounces at febo.com 
>> [mailto:time-nuts-bounces at febo.com] Im Auftrag von paul swed
>> Gesendet: Samstag, 7. April 2012 02:09
>> An: Discussion of precise time and frequency measurement
>> Betreff: Re: [time-nuts] Austron 2201, Tbolt, HP 3801 
>> comparison question
>> 
>> 
>> A lot of great comments.
>> Hope I do not drop any responses.
>> I am sure its older also it was $135 and picked it up 
>> recently. Have to say for the $ its actually quite fine and I 
>> am happy.
>> 
>> I have had the 3801 at least 10 years now. Picked it up early 
>> on and did some of the original tinkering and reverse 
>> engineering. So perhaps is does have the better oven. I m 
>> looking at what the 38xx software program says its doing at 
>> the numbers match the 2201 very nicely. So beginning to 
>> believe that what the 2201 says may be pretty accurate..
>> 
>> Tom asked if someone was going to make down converters. I 
>> might believe that I was involved in those threads. But it 
>> would have been attempt to make. Not produce. I have produced 
>> 2 main approaches with a number of other sub approaches. They 
>> do not emulate the RF down converter but are dependent on 
>> older less integrated receivers. The key is being able to get 
>> to the signals.
>> 
>> First version
>> Used the odetics antenna and then up converting the 35.42 to 
>> 75.42 Mhz. Easy to say harder to implement then imagined. 
>> plus building a 10 Mhz to 40 Mhz multiplier. though this all 
>> worked for a year I don't think many could reproduce it.
>> 
>> Second most recent approach thats really working very well.
>> A novatel starview 2 receiever provided by another Time-nut. 
>> The G2015 chips quite a jewel. Its made by zarlink now for 
>> $7.50. Any how it produces a 40 Mhz clock and has nice 
>> filtering and such for the 35 Mhz. Mix them and you are in 
>> business. Though I had been using active mixers with mixed 
>> results. (Pun) I went brute force a week or so ago using a 
>> minicircuits SRA1 type mixer. Boy does that work nicely in 
>> fact every things rock solid. No muss no fuss. Thats the 
>> right kind of design. The hardest part of this effor is 
>> attaching the IF wire and 2 wires for the pecl 40 Mhz clock. 
>> Right about at my limits for soldering.
>> 
>> So I do believe the 2201s are really quite a good receiver. 
>> There are numbers of tricks to actually getting them going. 
>> But after all the work I do believe worth it. I have made an 
>> offer for a second antenna-less unit. Considering my first 
>> was $5 I am offering more then that. But not going crazy 
>> either such as the silly prices I see on e... for a faulted unit.
>> 
>> Hope every things covered and thanks.
>> regards
>> Paul
>> WB8TSL
>> 
>> 
>> 
>> 
>> On Fri, Apr 6, 2012 at 5:43 PM, Ed Palmer 
>> <ed_palmer at sasktel.net> wrote:
>> 
>> > Hi Paul,
>> >
>> > I'm sure you've followed the discussions in the past on Tbolt 
>> > performance tuning.  Have you jumped through all the appropriate 
>> > hoops?  Things like precision survey, autotune the oscillator 
>> > parameters, good antenna visibility, mask angle, etc. come 
>> to mind.  
>> > Having said that, I've found that my Z3801A performs 
>> somewhat better 
>> > than my Tbolt.  For example, the 1 PPS out of my Tbolt has 
>> a Standard 
>> > Deviation of ~ 550 ps and a max-min range of ~ 4 ns.  My Z3801A is 
>> > ~200ps and ~2 ns. so call it twice as good.
>> >
>> > FYI, my best GPSDO is a Z3817A with a Standard Deviation of 
>> < 100 ps 
>> > and a max-min range of < 1 ns.  That one has an E1938 oscillator.
>> >
>> > Ed
>> >
>> >
>> >
>> > On 4/6/2012 2:24 PM, paul swed wrote:
>> >
>> >> Recently I put a 2201 back into service with  a home brew down 
>> >> converter. I am a bit surprised that when I use it to measure the 
>> >> Tbolt and then the HP 3801. The 3801 comes out always better by a 
>> >> decade actually. Granted what I am seeing is way down 
>> below a e-12th 
>> >> and in fact what I am reading seems nuts to me.
>> >> But can a 3801 run that much better then the Tbolt?
>> >> I kind of thought they would both be in the same region.
>> >> Thanks
>> >> Regards
>> >> Paul
>> >> WB8TSL