My Racal-Dana 1992
Has a 9462 oscillator module in it.
I've been fooling with little C hacks to make freq and phase
measurements via GPIB.
In my desire to super-tweek the timebase I discovered that small
adjustments take
many minutes to settle down. It's like herding cats.
I noticed that physical shocks would randomize the frequency over a 10
Hz range
at 10 MHz. What's more, the new frequency didn't seem to take 10
minutes to
settle down. So I tapped on the oven with the screwdriver until it
landed right on frequency.
The resulting plot is attached.
Come to think of it, I could use the counter to measure the time between
1pps and WWVB
zero crossing as an ultimate frequency reference as well as detecting
changes in the ionosphere.
--
Chuck Forsberg WA7KGX N2469R caf@omen.com www.omen.com
Developer of Industrial ZMODEM(Tm) for Embedded Applications
Omen Technology Inc "The High Reliability Software"
10255 NW Old Cornelius Pass Portland OR 97231 503-614-0430
Forgot to mention, each point is slightly more than 1 second.
On 12/16/2011 03:19 AM, Chuck Forsberg WA7KGX N2469R wrote:
Has a 9462 oscillator module in it.
I've been fooling with little C hacks to make freq and phase
measurements via GPIB.
In my desire to super-tweek the timebase I discovered that small
adjustments take
many minutes to settle down. It's like herding cats.
I noticed that physical shocks would randomize the frequency over a 10
Hz range
at 10 MHz. What's more, the new frequency didn't seem to take 10
minutes to
settle down. So I tapped on the oven with the screwdriver until it
landed right on frequency.
The resulting plot is attached.
Come to think of it, I could use the counter to measure the time
between 1pps and WWVB
zero crossing as an ultimate frequency reference as well as detecting
changes in the ionosphere.
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.
--
Chuck Forsberg WA7KGX N2469R caf@omen.com www.omen.com
Developer of Industrial ZMODEM(Tm) for Embedded Applications
Omen Technology Inc "The High Reliability Software"
10255 NW Old Cornelius Pass Portland OR 97231 503-614-0430
Chuck, I have a Racal 1992 with exactly the same government label on its
side as in your picture, except that my ink-stamped warranty date is a
few days later than the date on your counter/timer. These must have come
from a large government contract order.
However, while counting the 10 MHz from my HP Z3801A and displaying 0.01
Hz resolution on my 1992, I can rap very smartly indeed with my knuckles
on the outside of the counter case (anywhere) and not change the
displayed count. Further, my counter seem insensitive to orientation or
motion. Of course, I can see only to 0.01 Hz resolution at 10 MHz.
I just recently got this 1992 and have not yet tried to adjust its
Option 04E ocxo (the Racal 9462) so I don't know if my unit exhibits the
same cat-herding issue as yours. Judging from the condition of my 1992,
I doubt it was used much. Mine looks new; it had plastic caps on all the
BNC connectors and those connectors are all bright and shiny. My unit
had a calibration-due date of 9/26/2011; it may not have been used since
that calibration just over a year ago. At present, I'm letting it run to
see if there will be short term drift or room temperature sensitivity of
its ocxo. Later, I'll do what you have done to evaluate long-term drift
more critically.
I also have a Racal 1999 10-digit counter that counts to over 3 GHz. I
like these compact counters! I'm using a TADD-1 to distribute the 10 MHz
from the Z3801A to five devices so the ocxo performance is important
only if I use them portable.
How many TimeNuts have the Racal 1992 and what is your experience? Yes,
I've heard of push button switch issues but so far I've not encountered
any myself.
Larry W6FUB
On 12/16/2011 3:19 AM, Chuck Forsberg WA7KGX N2469R wrote:
Has a 9462 oscillator module in it.
I've been fooling with little C hacks to make freq and phase
measurements via GPIB.
In my desire to super-tweek the timebase I discovered that small
adjustments take
many minutes to settle down. It's like herding cats.
I noticed that physical shocks would randomize the frequency over a 10
Hz range
at 10 MHz. What's more, the new frequency didn't seem to take 10 minutes to
settle down. So I tapped on the oven with the screwdriver until it
landed right on frequency.
The resulting plot is attached.
Come to think of it, I could use the counter to measure the time between
1pps and WWVB
zero crossing as an ultimate frequency reference as well as detecting
changes in the ionosphere.
--
Best wishes,
Larry McDavid W6FUB
Anaheim, CA (20 miles southeast of Los Angeles, near Disneyland)
I bought a 1992 a few years ago and I've been quite happy with it. It
does have a few quirks, of course. Some good, some bad.
-
You can get a 10 sec. gate time and therefore .001 Hz resolution at
10 MHz by holding the 'up arrow' key for about 2 sec. It's in the
manual, once. If you blink you miss it. -
Even a completely warmed-up 04E oscillator drifts after the unit is
turned on from standby mode. However, if you use an external reference,
you can turn the unit on and immediately start measuring to .001 Hz with
no drift. I tried removing the oscillator completely, but it also
provides the clock for the microprocessor so it has to be present. The
changeover to the external reference only happens after the processor
detects the presence of the external reference. -
My 1992 doesn't perform properly when the power glitches. Even on a
UPS, the transfer time is long enough that the 1992 resets. The linear
power supply appears to be working correctly and the capacitors are
good, but it just doesn't have enough headroom to handle the transfer. -
I have one unit with bad switches and one with good ones. The body
on the good switches is white, the bad ones are black. I don't know if
this is the same on all units or not. You can see the body if you pull
a keycap off. No disassembly required.
Ed
On 12/16/2011 11:31 AM, Larry McDavid wrote:
Chuck, I have a Racal 1992 with exactly the same government label on
its side as in your picture, except that my ink-stamped warranty date
is a few days later than the date on your counter/timer. These must
have come from a large government contract order.
However, while counting the 10 MHz from my HP Z3801A and displaying
0.01 Hz resolution on my 1992, I can rap very smartly indeed with my
knuckles on the outside of the counter case (anywhere) and not change
the displayed count. Further, my counter seem insensitive to
orientation or motion. Of course, I can see only to 0.01 Hz resolution
at 10 MHz.
I just recently got this 1992 and have not yet tried to adjust its
Option 04E ocxo (the Racal 9462) so I don't know if my unit exhibits
the same cat-herding issue as yours. Judging from the condition of my
1992, I doubt it was used much. Mine looks new; it had plastic caps on
all the BNC connectors and those connectors are all bright and shiny.
My unit had a calibration-due date of 9/26/2011; it may not have been
used since that calibration just over a year ago. At present, I'm
letting it run to see if there will be short term drift or room
temperature sensitivity of its ocxo. Later, I'll do what you have done
to evaluate long-term drift more critically.
I also have a Racal 1999 10-digit counter that counts to over 3 GHz. I
like these compact counters! I'm using a TADD-1 to distribute the 10
MHz from the Z3801A to five devices so the ocxo performance is
important only if I use them portable.
How many TimeNuts have the Racal 1992 and what is your experience?
Yes, I've heard of push button switch issues but so far I've not
encountered any myself.
Larry W6FUB
On 12/16/2011 3:19 AM, Chuck Forsberg WA7KGX N2469R wrote:
Has a 9462 oscillator module in it.
I've been fooling with little C hacks to make freq and phase
measurements via GPIB.
In my desire to super-tweek the timebase I discovered that small
adjustments take
many minutes to settle down. It's like herding cats.
I noticed that physical shocks would randomize the frequency over a 10
Hz range
at 10 MHz. What's more, the new frequency didn't seem to take 10
minutes to
settle down. So I tapped on the oven with the screwdriver until it
landed right on frequency.
The resulting plot is attached.
Come to think of it, I could use the counter to measure the time between
1pps and WWVB
zero crossing as an ultimate frequency reference as well as detecting
changes in the ionosphere.
- You can get a 10 sec. gate time and therefore .001 Hz resolution at 10 MHz by holding the 'up arrow' key for about 2 sec. It's in the manual, once. If you blink you miss it.
What page please - I never spotted that ...
Regards,
David Partridge
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On Behalf Of Ed Palmer
Sent: 16 December 2011 18:09
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] My Racal-Dana 1992
I bought a 1992 a few years ago and I've been quite happy with it. It does have a few quirks, of course. Some good, some bad.
-
You can get a 10 sec. gate time and therefore .001 Hz resolution at 10 MHz by holding the 'up arrow' key for about 2 sec. It's in the manual, once. If you blink you miss it.
-
Even a completely warmed-up 04E oscillator drifts after the unit is turned on from standby mode. However, if you use an external reference, you can turn the unit on and immediately start measuring to .001 Hz with no drift. I tried removing the oscillator completely, but it also provides the clock for the microprocessor so it has to be present. The changeover to the external reference only happens after the processor detects the presence of the external reference.
-
My 1992 doesn't perform properly when the power glitches. Even on a UPS, the transfer time is long enough that the 1992 resets. The linear power supply appears to be working correctly and the capacitors are good, but it just doesn't have enough headroom to handle the transfer.
-
I have one unit with bad switches and one with good ones. The body on the good switches is white, the bad ones are black. I don't know if this is the same on all units or not. You can see the body if you pull a keycap off. No disassembly required.
Ed
On 12/16/2011 11:31 AM, Larry McDavid wrote:
Chuck, I have a Racal 1992 with exactly the same government label on
its side as in your picture, except that my ink-stamped warranty date
is a few days later than the date on your counter/timer. These must
have come from a large government contract order.
However, while counting the 10 MHz from my HP Z3801A and displaying
0.01 Hz resolution on my 1992, I can rap very smartly indeed with my
knuckles on the outside of the counter case (anywhere) and not change
the displayed count. Further, my counter seem insensitive to
orientation or motion. Of course, I can see only to 0.01 Hz resolution
at 10 MHz.
I just recently got this 1992 and have not yet tried to adjust its
Option 04E ocxo (the Racal 9462) so I don't know if my unit exhibits
the same cat-herding issue as yours. Judging from the condition of my
1992, I doubt it was used much. Mine looks new; it had plastic caps on
all the BNC connectors and those connectors are all bright and shiny.
My unit had a calibration-due date of 9/26/2011; it may not have been
used since that calibration just over a year ago. At present, I'm
letting it run to see if there will be short term drift or room
temperature sensitivity of its ocxo. Later, I'll do what you have done
to evaluate long-term drift more critically.
I also have a Racal 1999 10-digit counter that counts to over 3 GHz. I
like these compact counters! I'm using a TADD-1 to distribute the 10
MHz from the Z3801A to five devices so the ocxo performance is
important only if I use them portable.
How many TimeNuts have the Racal 1992 and what is your experience?
Yes, I've heard of push button switch issues but so far I've not
encountered any myself.
Larry W6FUB
On 12/16/2011 3:19 AM, Chuck Forsberg WA7KGX N2469R wrote:
Has a 9462 oscillator module in it.
I've been fooling with little C hacks to make freq and phase
measurements via GPIB.
In my desire to super-tweek the timebase I discovered that small
adjustments take
many minutes to settle down. It's like herding cats.
I noticed that physical shocks would randomize the frequency over a 10
Hz range
at 10 MHz. What's more, the new frequency didn't seem to take 10
minutes to
settle down. So I tapped on the oven with the screwdriver until it
landed right on frequency.
The resulting plot is attached.
Come to think of it, I could use the counter to measure the time between
1pps and WWVB
zero crossing as an ultimate frequency reference as well as detecting
changes in the ionosphere.
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.
Larry wrote:
However, while counting the 10 MHz from my HP Z3801A and displaying
0.01 Hz resolution on my 1992, I can rap very smartly indeed with my
knuckles on the outside of the counter case (anywhere) and not
change the displayed count. Further, my counter seem insensitive to
orientation or motion. Of course, I can see only to 0.01 Hz
resolution at 10 MHz.
That has been my experience, as well, with all of the 1992s I've
seen, even using the 10-second gate. I think Chuck must have one
with a bad oscillator.
Ed wrote:
- Even a completely warmed-up 04E oscillator drifts after the unit
is turned on from standby mode.
Mine do not drift very much when turned "on" from "standby" -- you
have to be in the 10-second gate mode to see it, and even then it is
only a count or two (1 or 2 mHz). The random variation seems to be 1
count or less -- mine often display only 2 or 3 different counts for
months at a time. (Unfortunately, not "000.000000 e3" and
"000.000001 e3", because of the lack of a smooth fine adjustment on
the oscillator -- the closest I can usually get when I adjust them is
"000.000045 e3" or so.) The two I own were selected as the best of a
lot of 12, but none of the others was much worse.
- My 1992 doesn't perform properly when the power glitches. Even
on a UPS, the transfer time is long enough that the 1992
resets. The linear power supply appears to be working correctly and
the capacitors are good, but it just doesn't have enough headroom to
handle the transfer.
All the ones I've played with do this, too. The answer is an "on
line" or "double conversion" UPS. Like ebay item # 190565787556 (no
connection to seller, but I have purchased from him in the past with
good results. Note that he will ship without the old batteries,
which saves significantly on shipping -- you'd want to start with
fresh batteries in any case, which are readily available). With this
kind of UPS, no problem with the 1992s or any other equipment
(including the old Dell that monitors the Thunderbolts and
Symmetricom disciplined oscillators, which always crashed with the
slightest power disturbance). Double conversion UPS's also condition
and regulate the output voltage.
- I have one unit with bad switches and one with good ones. The
body on the good switches is white, the bad ones are black. I don't
know if this is the same on all units or not. You can see the body
if you pull a keycap off. No disassembly required.
IME, white switches are substantially more common -- perhaps 4:1 or
5:1 compared to black switches. I have seen bad switches in units
with both black and white switch bodies with about equal probability,
so I don't think body color is a reliable guide to which are likely
to fail. (I have repaired around 25 1992s, and replaced the guts in
over 200 switches. It is the recurve "rubber" spring/washer that
fails, by cracking.) The one thing I have observed is that if one
switch fails in a counter, many more will, too. I have long wondered
if it may be a difference in soldering or board-cleaning at the time
of manufacture that determines which units are problematic and which are not.
Best regards,
Charles
On 12/16/2011 5:57 PM, Charles P. Steinmetz wrote:
Larry wrote:
However, while counting the 10 MHz from my HP Z3801A and displaying
0.01 Hz resolution on my 1992, I can rap very smartly indeed with my
knuckles on the outside of the counter case (anywhere) and not change
the displayed count. Further, my counter seem insensitive to
orientation or motion. Of course, I can see only to 0.01 Hz
resolution at 10 MHz.
That has been my experience, as well, with all of the 1992s I've seen,
even using the 10-second gate. I think Chuck must have one with a bad
oscillator.
Ed wrote:
- Even a completely warmed-up 04E oscillator drifts after the unit
is turned on from standby mode.
Mine do not drift very much when turned "on" from "standby" -- you
have to be in the 10-second gate mode to see it, and even then it is
only a count or two (1 or 2 mHz). The random variation seems to be 1
count or less -- mine often display only 2 or 3 different counts for
months at a time. (Unfortunately, not "000.000000 e3" and "000.000001
e3", because of the lack of a smooth fine adjustment on the oscillator
-- the closest I can usually get when I adjust them is "000.000045 e3"
or so.) The two I own were selected as the best of a lot of 12, but
none of the others was much worse.
My 04E must be less stable than yours. I was monitoring a fully
warmed-up Efratom FRK rubidium and saw a drift of ~ 0.04 Hz (i.e. 40
counts) over two hours after I turned on my 1992 from standby. Are
more than one type of oscillator used for option 04E? Mine is a model 9462.
- My 1992 doesn't perform properly when the power glitches. Even
on a UPS, the transfer time is long enough that the 1992 resets. The
linear power supply appears to be working correctly and the
capacitors are good, but it just doesn't have enough headroom to
handle the transfer.
All the ones I've played with do this, too. The answer is an "on
line" or "double conversion" UPS. Like ebay item # 190565787556 (no
connection to seller, but I have purchased from him in the past with
good results. Note that he will ship without the old batteries, which
saves significantly on shipping -- you'd want to start with fresh
batteries in any case, which are readily available). With this kind
of UPS, no problem with the 1992s or any other equipment (including
the old Dell that monitors the Thunderbolts and Symmetricom
disciplined oscillators, which always crashed with the slightest power
disturbance). Double conversion UPS's also condition and regulate the
output voltage.
Yes, I'm familiar with the on-line style of UPS. Unfortunately, they're
much more expensive than the regular type. The auction you mentioned
isn't really an option for me. I'm in Canada and the seller has
specified the most ridiculous of UPS's many ridiculous shipping
methods. My shipping costs are shown as $413. Even taking out the
batteries won't reduce that to a reasonable level. Not mentioned is the
UPS brokerage charges and taxes which would add about $50 - $75 above that.
I wondered what the long-term effect would be of changing the line
input setting on the 1992 from the 115V to the 100V position. It would
give better headroom at the expense of more heat. I'd also have to
check for transformer saturation. Another possibility would be to
replace the 04E with any old 10 MHz oscillator and use an external
standard. This would reduce the load on the +5V by a few hundred
milliamps and might make the difference.
- I have one unit with bad switches and one with good ones. The
body on the good switches is white, the bad ones are black. I don't
know if this is the same on all units or not. You can see the body
if you pull a keycap off. No disassembly required.
IME, white switches are substantially more common -- perhaps 4:1 or
5:1 compared to black switches. I have seen bad switches in units
with both black and white switch bodies with about equal probability,
so I don't think body color is a reliable guide to which are likely to
fail. (I have repaired around 25 1992s, and replaced the guts in over
200 switches. It is the recurve "rubber" spring/washer that fails, by
cracking.) The one thing I have observed is that if one switch fails
in a counter, many more will, too. I have long wondered if it may be
a difference in soldering or board-cleaning at the time of manufacture
that determines which units are problematic and which are not.
Thanks for that info. A sample of only two (one good and one bad)
doesn't give much confidence. How did you replace the switch guts?
Where did you get the parts? I looked at mine and had some success
installing a tiny spring to restore the switch's operation. You lose
the tactile feedback, but at least the switch works. I didn't pursue it
because the other unit is okay (so far!).
I hadn't thought about cleaning causing the switch failure. I just
assumed it was old age with different brands or lots of switches being
made with different recipes for the rubber and therefore different
lifetimes.
Ed
Ed wrote:
My 04E must be less stable than yours. I was monitoring a fully
warmed-up Efratom FRK rubidium and saw a drift of ~ 0.04 Hz (i.e. 40
counts) over two hours after I turned on my 1992 from standby. Are
more than one type of oscillator used for option 04E? Mine is a model 9462.
There are actually two Racal part numbers that you sometimes see in
documentation as being used for Option 04E -- 404386 and 454879. The
9462s in the US military contract 1992s that I have seen (pretty much
all of the 1992s one sees in the US are from the mil contract, IME)
are marked "9462 454879." I have not seen an oscillator marked
"404386," so I do not know if these are Model 9462 oscillators or
another model.
As with any crystal oscillator, there is no doubt a range of both
stability and warm-up drift in the 9462s you find, but IME not a very
large variation. (I'm assuming that you clocked the warmup after the
1992 had been in standby -- plugged in with the red power button "on"
-- for a week or more [preferably for a month or more].)
An oscillator with greater warm-up drift will not necessarily be less
stable, after warmup, than one with less warm-up drift.
How did you replace the switch guts? Where did you get the parts?
They pull straight out the front, after removing the key cap. We
bought new switches from TOKO (the only OEM for that part), but they
were retired from production around 20 years ago and I have not seen
any available in the pipeline for years now. Unfortunately, the
parts units you find for sale invariably have bad switches. I have
more than once tried to use switches from a donor unit to repair a
1992, and they have all promptly failed during post-op testing. In
one case, six successive switches failed within five presses after I
installed them (that was way back before I figured out that you can
replace the innards from the front, when I was desoldering and
replacing the switches whole).
I hadn't thought about cleaning causing the switch failure. I just
assumed it was old age with different brands or lots of switches
being made with different recipes for the rubber and therefore
different lifetimes.
That was my original thought, too, but I believe the data indicate
otherwise. Soldering and cleaning appear to be the next most likely culprits.
I haven't seen a capacitor with detents. Could the fine adjustment
be a multi-turn pot?
I assume that is the case -- that is exactly what it feels like --
but I have not disassembled a 9462 to see.
David wrote:
Strange, the fine adjustment on mine doesn't have any detents.
Not surprising -- I assume many of the parts were obtained from
multiple sources, and if detented multiturn pots of the appropriate
value were out of stock, it would be natural to replace them with
their much more common non-detented counterparts. Or perhaps Racal
specified a regular, non-detented pot and when a vendor shipped them
detented pots by mistake, they used them anyway for one lot of
oscillators and I happen to have a couple of those.
The labels on 9462s indicate there were several revisions. I do not
know what is different internally between revisions.
Best regards,
Charles
On 12/18/2011 12:00 PM, Charles P. Steinmetz wrote:
Ed wrote:
My 04E must be less stable than yours. I was monitoring a fully
warmed-up Efratom FRK rubidium and saw a drift of ~ 0.04 Hz (i.e. 40
counts) over two hours after I turned on my 1992 from standby. Are
more than one type of oscillator used for option 04E? Mine is a
model 9462.
There are actually two Racal part numbers that you sometimes see in
documentation as being used for Option 04E -- 404386 and 454879. The
9462s in the US military contract 1992s that I have seen (pretty much
all of the 1992s one sees in the US are from the mil contract, IME)
are marked "9462 454879." I have not seen an oscillator marked
"404386," so I do not know if these are Model 9462 oscillators or
another model.
Mine is marked "9462 454879 Rev A"
As with any crystal oscillator, there is no doubt a range of both
stability and warm-up drift in the 9462s you find, but IME not a very
large variation. (I'm assuming that you clocked the warmup after the
1992 had been in standby -- plugged in with the red power button "on"
-- for a week or more [preferably for a month or more].)
Yes, mine was always in standby mode. It was probably running for many
months like that.
An oscillator with greater warm-up drift will not necessarily be less
stable, after warmup, than one with less warm-up drift.
Agreed. But it was annoying to leave it in standby mode and then still
have some drift. At first, I thought the drift was due to the rest of
the unit warming up. I was quite surprised when my tests showed that
the drift was 100% due to the oscillator. Using an FRK as an external
reference, I can turn on my 1992 (not from standby) and immediately
measure my Z3801A as 10.000 000 000 MHz.
Ed
Ed wrote:
Agreed. But it was annoying to leave it in standby mode and then
still have some drift. At first, I thought the drift was due to the
rest of the unit warming up. I was quite surprised when my tests
showed that the drift was 100% due to the oscillator. Using an FRK
as an external reference, I can turn on my 1992 (not from standby)
and immediately measure my Z3801A as 10.000 000 000 MHz.
I put one of mine in standby earlier this afternoon. I'll turn it
back on tomorrow afternoon and report the results. How long does
yours have to be in standby before it comes on 40 counts off?
Best regards,
Charles
Charles,
I don't know the answer to that question. Since it's a thermal issue,
it's a question of how long does it take for the temperature to
stabilize? 24 hrs. should be lots of time. I'd check it before going
to bed. A quick turn-on and measure won't raise the temperature much
and will quickly show if there's a change from normal. And, as
mentioned, it takes mine about 2 hrs. to stabilize after being turned
back on.
Ed
On 12/18/2011 4:35 PM, Charles P. Steinmetz wrote:
Ed wrote:
Agreed. But it was annoying to leave it in standby mode and then
still have some drift. At first, I thought the drift was due to the
rest of the unit warming up. I was quite surprised when my tests
showed that the drift was 100% due to the oscillator. Using an FRK
as an external reference, I can turn on my 1992 (not from standby)
and immediately measure my Z3801A as 10.000 000 000 MHz.
I put one of mine in standby earlier this afternoon. I'll turn it
back on tomorrow afternoon and report the results. How long does
yours have to be in standby before it comes on 40 counts off?
Best regards,
Charles
My Racal1994 oscillator is labeled "9462 454879" on one face of the can and "404386C" on another.
Aart Olsen
----- Original Message -----
From: "Charles P. Steinmetz" charles_steinmetz@lavabit.com
To: "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Sunday, December 18, 2011 12:00:58 PM
Subject: Re: [time-nuts] My Racal-Dana 1992
There are actually two Racal part numbers that you sometimes see in
documentation as being used for Option 04E -- 404386 and 454879. The
9462s in the US military contract 1992s that I have seen (pretty much
all of the 1992s one sees in the US are from the mil contract, IME)
are marked "9462 454879." I have not seen an oscillator marked
"404386," so I do not know if these are Model 9462 oscillators or
another model.
I wrote:
I put one of mine in standby earlier this afternoon. I'll turn it
back on tomorrow afternoon and report the results.
(All results using 10 second gate): Yesterday, the 1992 had been
reading "000.000000 E-3" stably for several days, and I switched it
to standby. After 16 hours in standby, it came on 4 counts (4 mHz)
high, rose to 9 counts high in 5 minutes, hung there for about 15
minutes, then slowly came down to 0 over the next hour. It undershot
0 by the tiniest bit (several minutes of occasionally reading
"999.999999 E-3". It has been reading 0 stably for over 3 hours now.
So, mine also has an error with a long time constant when it comes
out of standby, but of lesser magnitude than Ed's (peak error 9x10
E-10). I presume the magnitude of the error may change with ambient
temperature (in my case, 74.1 F throughout).
Chuck speculated several days ago that the 9462 fine adjustment may
be an oven parameter (temperature?) adjustment. The long,
oscillatory nature of the settling when the fine adjustment is
changed does suggest the action of a somewhat underdamped control
loop. Both the ringing and the long time constant tend to support
the oven temperature hypothesis (but I am still reluctant to believe
they would design a frequency adjustment this way).
Best regards,
Charles
Charles,
My restart followed the same pattern as yours. Start high, go higher,
drift down, undershoot, recover. The start was 10 MHz +51 counts, rise
to +60 counts after 5 minutes, drift down to a minimum of +8 counts
after 90 minutes and recover to +11 counts after 10 more minutes for a
total shift of -40 counts (peak error of 4e-9). I'm going to redo all
these tests, but I think I'll be sticking to the external standard.
Ed
On 12/19/2011 12:33 PM, Charles P. Steinmetz wrote:
I wrote:
I put one of mine in standby earlier this afternoon. I'll turn it
back on tomorrow afternoon and report the results.
(All results using 10 second gate): Yesterday, the 1992 had been
reading "000.000000 E-3" stably for several days, and I switched it to
standby. After 16 hours in standby, it came on 4 counts (4 mHz) high,
rose to 9 counts high in 5 minutes, hung there for about 15 minutes,
then slowly came down to 0 over the next hour. It undershot 0 by the
tiniest bit (several minutes of occasionally reading "999.999999
E-3". It has been reading 0 stably for over 3 hours now.
So, mine also has an error with a long time constant when it comes out
of standby, but of lesser magnitude than Ed's (peak error 9x10 E-10).
I presume the magnitude of the error may change with ambient
temperature (in my case, 74.1 F throughout).
Chuck speculated several days ago that the 9462 fine adjustment may be
an oven parameter (temperature?) adjustment. The long, oscillatory
nature of the settling when the fine adjustment is changed does
suggest the action of a somewhat underdamped control loop. Both the
ringing and the long time constant tend to support the oven
temperature hypothesis (but I am still reluctant to believe they would
design a frequency adjustment this way).
Best regards,
Charles
This is typical for an underdamped second order servo. The oscillation
is the penalty for a shorter settling time. A critically damped system
would not oscillate, but approach the final value smoothly in a slightly
longer time. It may have been designed for a slight overshoot...
Don
Ed Palmer
Charles,
My restart followed the same pattern as yours. Start high, go higher,
drift down, undershoot, recover. The start was 10 MHz +51 counts, rise
to +60 counts after 5 minutes, drift down to a minimum of +8 counts
after 90 minutes and recover to +11 counts after 10 more minutes for a
total shift of -40 counts (peak error of 4e-9). I'm going to redo all
these tests, but I think I'll be sticking to the external standard.
Ed
On 12/19/2011 12:33 PM, Charles P. Steinmetz wrote:
I wrote:
I put one of mine in standby earlier this afternoon. I'll turn it
back on tomorrow afternoon and report the results.
(All results using 10 second gate): Yesterday, the 1992 had been
reading "000.000000 E-3" stably for several days, and I switched it to
standby. After 16 hours in standby, it came on 4 counts (4 mHz) high,
rose to 9 counts high in 5 minutes, hung there for about 15 minutes,
then slowly came down to 0 over the next hour. It undershot 0 by the
tiniest bit (several minutes of occasionally reading "999.999999
E-3". It has been reading 0 stably for over 3 hours now.
So, mine also has an error with a long time constant when it comes out
of standby, but of lesser magnitude than Ed's (peak error 9x10 E-10).
I presume the magnitude of the error may change with ambient
temperature (in my case, 74.1 F throughout).
Chuck speculated several days ago that the 9462 fine adjustment may be
an oven parameter (temperature?) adjustment. The long, oscillatory
nature of the settling when the fine adjustment is changed does
suggest the action of a somewhat underdamped control loop. Both the
ringing and the long time constant tend to support the oven
temperature hypothesis (but I am still reluctant to believe they would
design a frequency adjustment this way).
Best regards,
Charles
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.
--
"Neither the voice of authority nor the weight of reason and argument
are as significant as experiment, for thence comes quiet to the mind."
R. Bacon
"If you don't know what it is, don't poke it."
Ghost in the Shell
Dr. Don Latham AJ7LL
Six Mile Systems LLP
17850 Six Mile Road
POB 134
Huson, MT, 59846
VOX 406-626-4304
www.lightningforensics.com
www.sixmilesystems.com