[time-nuts] HP 5730 data polling technique?

Sun Sep 6 15:41:24 EDT 2015

Thank you everyone for responding to my plea for help doing data collection
with an HP5370B. Let me post my progress so far. These techniques may be
obvious to the experts here, but I think the guide will help novices like
me.

This is for the HP5370B with John Seamon's Beaglebone modification,
http://www.jks.com/5370/5370.html where the original M6800 is replaced with
a modern processor board running the original code in emulation. Most
helpfully, the connection can be via tcp/ip rather than HPIB.

There are several connection points to the Beaglebone (besides the
preserved HPIB):
  ssh on port 20
  telnet on port 5370
  telnet on port 5371
  http on port 5372 (which echoes the 5371 telnet port)

My initial attempt was on 5371, which is a menu-driven interface that
includes a convenient menu:
  ?
  commands:
  d show instrument display including unit and key LEDs
  h <HPIB cmd> emulate HPIB command input, e.g. "h md2"
  h? prints reminder list of HPIB commands
  k <fn1 .. fn4> emulate function key 1-4 press, e.g. "k fn1" is TI key
  k <gt1 .. gt4> emulate gate time key 1-4 press
  k <st1 .. st8> emulate statistics key 1-8 press
  k <ss1 .. ss5> emulate sample size key 1-5 press
  k <m1 .. m6> emulate "misc" key 1-6 press
1 TI only, 2 +/- TI, 3 ext h.off, 4 per compl, 5 ext arm, 6 man rate
  m run measurement extension example code
  s show measurement statistics
  rc show values of count-chain registers (one sample)
  rcl|recall [name]   load key settings from current or named profile
  sto|store name      save key settings to named profile
  r reset instrument
  q quit

Unfortunately, every thing I tried gave only a single data point, rather
than the time series I wanted.

In desperation I tried the 5370 port. John gives a short example with the
command "md2"
(see https://github.com/jks-prv/5370_proc/blob/master/READ_MORE.md)

Again that only gave a single data point. "md0" "md1" and "md3" didn't
help, but then "md4" was magical!
I get a new line for each time the display changes.

So using Poul-Henning Kamp's suggestion for external arming, and the
command line
>  telnet 10.183.180.232 5371 | tee datafile.txt
I get a time series.

Here's my results for:
1pps (Trimble Thunderbolt)
200Hz oscillator (HP200CD)

..
TI = 4.99315145000E-03
TI = 4.96596342000E-03
TI = 4.97499937000E-03
TI = 4.98463980000E-03
TI = 4.97498115000E-03
TI = 4.97546959000E-03
TI = 4.98466773000E-03
TI = 4.97510635000E-03
TI = 4.97518645000E-03
TI = 4.97555330000E-03
TI = 4.97523096000E-03
..

Finally analyzing the data with Tom Van Baak's adev1 (
http://www.leapsecond.com/tools/adev1.htm)

[paul at localhost Documents] $ ./adev1 1 < ../datafile.txt

** Sampling period: 1 s
** Phase data scale factor: 1.000e+00
** Total phase samples: 59110
** Normal and Overlapping Allan deviation:

       1 tau, 5.4296e-04 adev(n=59108),     5.4296e-04 oadev(n=59108)
       2 tau, 2.7584e-04 adev(n=29553),     2.7213e-04 oadev(n=59106)
       5 tau, 7.3463e-05 adev(n=11820),     4.7803e-05 oadev(n=59100)
      10 tau, 4.0074e-05 adev(n=5909),      2.5945e-05 oadev(n=59090)
      20 tau, 1.8440e-05 adev(n=2954),      1.1119e-05 oadev(n=59070)
      50 tau, 6.2975e-06 adev(n=1181),      4.6753e-06 oadev(n=59010)
     100 tau, 4.2792e-06 adev(n=590),       2.4867e-06 oadev(n=58910)
     200 tau, 3.0057e-06 adev(n=294),       1.3930e-06 oadev(n=58710)
     500 tau, 1.6899e-06 adev(n=117),       6.7780e-07 oadev(n=58110)
    1000 tau, 9.2053e-07 adev(n=58),        3.9410e-07 oadev(n=57110)
    2000 tau, 5.0472e-07 adev(n=28),        2.8388e-07 oadev(n=55110)
    5000 tau, 2.2221e-07 adev(n=10),        1.2028e-07 oadev(n=49110)
   10000 tau, 1.7590e-07 adev(n=4),         6.1471e-08 oadev(n=39110)
   20000 tau, 1.7688e-07 adev(n=1),         1.9906e-08 oadev(n=19110)

I suspect I can sample at a far faster rate than 1pps. Time to experiment!

On Thu, Sep 3, 2015 at 6:52 PM, Azelio Boriani <azelio.boriani at gmail.com>
wrote:

> The manual reports that the 5370 is SRQ capable but no example is
> given. Anyway, the operation is the serial_poll and read_status_byte
> sequence, when the status byte reports the error 0, then a measurement
> is complete and the result can be read. How to implement the SRQ
> management depends on your GPIB interface and its software suite
> (ionsrq to install the SRQ handler for the 82357B and its C library,
> for example).
>
> On Thu, Sep 3, 2015 at 12:14 AM, Poul-Henning Kamp <phk at phk.freebsd.dk>
> wrote:
> > --------
> > In message <
> CAP_SGzeoWcQdLA77EE8-10AigkwzzCTh-w5Eim-2uV4mpwiQtA at mail.gmail.com>
> > , Paul Alfille writes:
> >
> >>I am stymied by what should be a simple task: I'd like to gather
> sequential
> >>measurements from my HP5370B.
> >
> > You can do it two ways.
> >
> > Either use the EXT ARM to pace your measurements, and the computer
> > just reads them as they happen.
> >
> > Or you can use the MD2 mode, where measurements only start when the
> > computer sends "MRM" to the counter.
> >
> > I usually use the former method because I get more precise pacing
> > of the measurements (I feed EXT ARM from a HP33120).
> >
> > --
> > Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> > phk at FreeBSD.ORG         | TCP/IP since RFC 956
> > FreeBSD committer       | BSD since 4.3-tahoe
> > Never attribute to malice what can adequately be explained by
> incompetence.
> > _______________________________________________
> > 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.
> _______________________________________________
> 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.
>