Racal Dana 1996 Frequency counter

[saturation]

I've been waiting for a good version of either a 1992, 1996, 1998 or 1999 to pop up, and one finally did after some years of waiting.  There are no tear downs of this model series on the Internet so here is an offering for those interested in acquiring any within the series and to discern internal differences [ the 1995 is without GHz capability, the 1996 is an upgrade of the 1994; they are designed to be rack mountable, while the 1992 design was transportable.]

The 1994-96 were made in the late 1980s and offered slightly more calculations or control over input signals than the 1992.  They were less plagued with switch defects that appear in used models 1991, 1992.  Because they were not made to be transported, the rack mount series are likely to be in better physical shape.

Large photos:

http://www.flickr.com/photos/igsaturation/8241558668/#





All boards and components are easily accessed on this design compared to the compact 1992.  The large green PCB is the dedicated 68000 computer board that does all the calculations.  The gold board is the data input and preconditioning board. The "copper wire" seen crossing from the front to rear in the middle is a type SMB connector cable to the divide-by-64 prescaler for 1.5 GHz rated Input C.  Item marked "B" in yellow is the ovenized oscillator.

[slburris]

I have a 1992 and a 1998 counter, and I just love these Racal/Dana counters.

My 1998 is currently out of service because it has developed the front panel
switch problem that plagues these units, and I haven't yet figured out a nice
way to machine replacement switches.

I also have an EIP 545A microwave counter I'm just starting to play with.

Scott

[saturation]

The 199x series are all great units, a model for their entire class and still in demand.  I found recently that BKPrecision makes a series of counters that at a high of ~ $500, will match the capabilities of a 1999 but without the risk of age related failures; so getting an old Racal is really tied to what price/performance you can muster over eBay.

Thanks slburris for your prior posts on them and photos.  Worrying about having to replace those switches over time made me wait for the right one to come along, although appearance wise, the whole 199x series seems to have the same design but it maybe tied to when they were made, yet some have them, some not.

Do you know the build date of your 1998 and 1992?  There are stickers or stamped dates all over the internals.

I have a 1992 and a 1998 counter, and I just love these Racal/Dana counters.

My 1998 is currently out of service because it has developed the front panel
switch problem that plagues these units, and I haven't yet figured out a nice
way to machine replacement switches.

I also have an EIP 545A microwave counter I'm just starting to play with.

Scott

[muvideo]

Nice machine saturation.
The heart (counter and interpolator) seem similar
to the 1992, the single shot resolution is the same, but
this one has many more functions: rise and fall times for
pulses from 5nS, gate window programmable, higher voltage
capability and statistical calculation, so more useful than the
1992, good catch!
edit: also one more digit on the display 10 vs 9, easier to
read for high resolutions.
I found some specs here:
http://www.teknetelectronics.com/DataSheet/RACALDANA/RACAL_199590030.pdf

What timebase do you have?  If it's the 9462 (04E) should
be good, my 1992 holds pretty well against Rb clocks.

The 199x switch problems are annoying, but thanking to
them I was able to pay few dollars for very good counter, in
mint conditions, the fix was few hours of work and few
cheap switches

Fabio.

[saturation]

Thanks muvideo, and also thanks for your photos on the archives.  That helped me a lot too.  I think the prescaler board is identical too, I have blow ups on the flikr site and I compared it to your photos.  The additional functions are great as it give me the capability to do some measurements that normally you'd need a 1 GHz grade scope to do.

Yes, it has the ovenized reference, 04e option.  Its good to know it hold well against a Rb clock, so I can hold off on getting one for now.

Regardless, I agree with your comments on any effort repair these!  These counters are well worth the effort, my only fear is a firmware or memory failure, but in reviewing prior posts on any model in this series, no spontaneous failures have been reported unlike some early HP3456 DVM. 

My unit had a small pouch will the old firmware ROMs still in them inserted on the mobo, so I have a backup in case of diasaster 

FWIW, BK's model:



Is actually fairly close to the 1999.  Listed at $470,  you can deal it down to the high $300s, which will match the highest end price some eBay sellers want for many 199x series.

Nice machine saturation.
The heart (counter and interpolator) seem similar
to the 1992, the single shot resolution is the same, but
this one has many more functions: rise and fall times for
pulses from 5nS, gate window programmable, higher voltage
capability and statistical calculation, so more useful than the
1992, good catch!
edit: also one more digit on the display 10 vs 9, easier to
read for high resolutions.
I found some specs here:
http://www.teknetelectronics.com/DataSheet/RACALDANA/RACAL_199590030.pdf

What timebase do you have?  If it's the 9462 (04E) should
be good, my 1992 holds pretty well against Rb clocks.

The 199x switch problems are annoying, but thanking to
them I was able to pay few dollars for very good counter, in
mint conditions, the fix was few hours of work and few
cheap switches

Fabio.

[slburris]

My 1998 has circuit boards with '88 date stamps and a QC sticker with an Apr '89 date handwritten
on it.  I haven't cracked open my 1992 yet since it's working like a charm.

I saw another 1992 pop up on Ebay (it's still there) for $100, but it's from Surplus Test
Mart in Israel, and for some reason a lot of their stuff looks like it got run over by a tank.
This one is no exception -- it looks rather squished.  Maybe you could harvest front
panel switches from it, but since they all seem to go bad eventually, that's probably
not a good investment.

And now darn it, I'm going to have to look at that BK model too. 

The 199x series are all great units, a model for their entire class and still in demand.  I found recently that BKPrecision makes a series of counters that at a high of ~ $500, will match the capabilities of a 1999 but without the risk of age related failures; so getting an old Racal is really tied to what price/performance you can muster over eBay.

Thanks slburris for your prior posts on them and photos.  Worrying about having to replace those switches over time made me wait for the right one to come along, although appearance wise, the whole 199x series seems to have the same design but it maybe tied to when they were made, yet some have them, some not.

Do you know the build date of your 1998 and 1992?  There are stickers or stamped dates all over the internals.

I have a 1992 and a 1998 counter, and I just love these Racal/Dana counters.

My 1998 is currently out of service because it has developed the front panel
switch problem that plagues these units, and I haven't yet figured out a nice
way to machine replacement switches.

I also have an EIP 545A microwave counter I'm just starting to play with.

Scott

[nctnico]

Surplus Test is pricy and they are not very eager to accept a reasonable offer for something that obviously was picked up at a scrap yard and needs a lot of TLC to get it working.

Anyway, if you are going to spend a lot of cash on a frequency counter it might be interesting to know that many spectrum analysers have this feature as well. Another problem I've found with some frequency counters is that their sensitivity at higher frequencies is quite poor.

[saturation]

Yes, its impressive they want so much money for what I think are consistently the most battered equipment I've seen on eBay.

Yes, if you do get a modern SA, the counters in them are exceptional, like the new Rigol SA815.   A good counter is a bit redundant now, also good SAs accept external references for improved accuracy.  But until Rigols 815, most SA of some decency are in the $5k range, and in eBay, many sub $1000 used SA don't have high accuracy counters.

The Racal's are fairly sensitive throughout their range, you can read it on their spec sheet.  The input C board for 1.5 GHz is 10x more sensitive than the mainboard A,B inputs.

Yes, I was under the impression that the 199x series was the year it was made, but its just a model number.  The series begins in late 1980s and any future revisions can be found on the mobo and in the manuals, I think the most recent in mid 1995.

My 1998 has circuit boards with '88 date stamps and a QC sticker with an Apr '89 date handwritten
on it.  I haven't cracked open my 1992 yet since it's working like a charm.

Surplus Test is pricy ...
Anyway, if you are going to spend a lot of cash on a frequency counter it might be interesting to know that many spectrum analysers have this feature as well. Another problem I've found with some frequency counters is that their sensitivity at higher frequencies is quite poor.

[poodyp]

Well that's interesting. I too have a 1996 off ebay and it had spare EEPROMs inside it. Are they included from the factory?

[saturation]

Not sure, but its the prior firmware revision in mine, so I have a backup should the old one die, FWIW!

Well that's interesting. I too have a 1996 off ebay and it had spare EEPROMs inside it. Are they included from the factory?

[slburris]

I've convinced myself I need multiple counters for my setup.

Normally, I run a Trimble Thunderbolt as an external 10Mhz reference to
all of my equipment.  So I have one counter running off the Thunderbolt.
I use a second counter as a sanity check on the 10Mhz output of the
Thunderbolt.

Currently my 1992 is that sanity check.  I'm quite impressed with the OXCO
in it.  At maximum resolution, I'm seeing 10Mhz on the dot +- 1 digit even over
temperature variations in my lab room.

I do have a couple of rubidiums as backup, but I rarely power them up.

The talk of firmware reminds me I need to image my firmware in case of problems.

There isn't a list of firmware revisions floating around somewhere, is there?

Scott

[poodyp]

Here's my 1996. It's reading a second 9462 I got along with the 1996. They're "off" .05ppm relative to each other. Unfortunately I can't figure out what the long term drift and/or aging is. The scope is showing 10Mhz from the second 9462, and 5Mhz from before the frequency doubler on the first one.

[saturation]

@poodyp: nice.  You'll need time to calculate the long term drift rates; as these Racals are old units the good news is very likely the reference have stabilized, and at this point the drift rate would be at its lowest, unless is about to break!

@slburris: I think its always good to have a minimum of 2 of each type of equipment, if its mission critical to you.  How can one know its off if we have but one device?  I personally try to acquire 3 types of each so there is a tie breaker.

Can you compare the Racal  XCO to the Thunderbolt?  Being the latter is more accurate, I'd like to know how the Racal crystal compares.  I've always read it compares very favorably, but never heard its quantified, say its off ppm or ppb.

[slburris]

Oooh, there's a 1992 on eBay right now for $80 with no bids.  Looks to be in pretty decent shape.

I will not buy it.  Say it with me.... :-)

Scott

[saturation]

I see it 

I think final sale price of a 1996 is often close to a 1992; they have similar specs as counters but the 1996 has a stat function built in and voltage readouts too.  I find the stat function very useful, as it simplifies making stability calculations and the large size of the chassis makes it easier to work with.

1992 is nevertheless, a good machine, looking for a good home!

[poodyp]

You'll need time to calculate the long term drift rates

Oops, I meant to say I couldn't figure out what the rated long term drift was other than a single line in the manual stating "Aging: < 5 * 10^-10 per day". I know it's highly unlikely to be too far off, I just need to get a thunderbolt or something to check against. The oscillator that came in my 1996 has the serial number of 300 something, so it could be very old, but I doubt it's drifted by ~5E-10 * 24 * 365.

[alm]

5e-10 * sqrt(24 * 365) might be a better estimate, although you should be able to find at least 1y specs.

[saturation]

... yes in the end, a best test would be to compare it empirically against a Rb clock or a GPS disciplined one.  The Racal's  ovenized XO is not that far off.  I've read 2 -3 reports of comparisons against a Rb clock stating its fairly close, alas, but not quantified with a number, one is by Fabio on this thread.

In his brief post he summarized all the key differences between the 1992 and 1996.

I meant to say I couldn't figure out what the rated long term drift was other than a single line in the manual stating "Aging: < 5 * 10^-10 per day"...

Nice machine saturation.
The heart (counter and interpolator) seem similar
to the 1992, the single shot resolution is the same, but
this one has many more functions: rise and fall times for
pulses from 5nS, gate window programmable, higher voltage
capability and statistical calculation, so more useful than the
1992, good catch!
edit: also one more digit on the display 10 vs 9, easier to
read for high resolutions.


What timebase do you have?  If it's the 9462 (04E) should
be good, my 1992 holds pretty well against Rb clocks.

The 199x switch problems are annoying,
Fabio.

[muvideo]

... yes in the end, a best test would be to compare it empirically against a Rb clock or a GPS disciplined one.  The Racal's  ovenized XO is not that far off.  I've read 2 -3 reports of comparisons against a Rb clock stating its fairly close, alas, but not quantified with a number, one is by Fabio on this thread.

Hello, it happens that now I have on the bench
a GPS, some Rb clocks, the counter and some
logging scripts on the pc. So I could try to quantify
the statement I made before, I need only to find
some free time for some time tests...

Fabio.

[saturation]

  Awesome!   As your time allows, thank you, Fabio!

... yes in the end, a best test would be to compare it empirically against a Rb clock or a GPS disciplined one.  The Racal's  ovenized XO is not that far off.  I've read 2 -3 reports of comparisons against a Rb clock stating its fairly close, alas, but not quantified with a number, one is by Fabio on this thread.

Hello, it happens that now I have on the bench
a GPS, some Rb clocks, the counter and some
logging scripts on the pc. So I could try to quantify
the statement I made before, I need only to find
some free time for some time tests...

Fabio.

[muvideo]

Well, I couldnt resist the tentation to make a quick check
of the absolute error of my unit.
Please consider that I'm still learning to deal with precise
time measurements, so feel free to check and correct my
procedures. Also the room temperature is far from ideal.

Measuring the output of a pair of FE5680A that I trimmed
against GPS (manually, no GPSDO still in my hands), the
counter was aroud 70-75 mHz high, so I was expecting
the internal oscillator a little on the low side, and it was low
by about 7.3x10^-9 :



The interesting part is top right, where I have data and linear
interpolation together. The line is the phase between
the GPS pps and internal 10MHz reference, so a positive
slope means that the 10MHz is lagging the pps -> slow.
It's not bad, the counter was never touched by me, and
probably not trimmed from a lot of time. According to the
manual it was still in spec, since the manual calls for recal
only past +-3x10^-8 deviation.

Ok time to try to trim the counter, I lightly touched the
fine adustment, after turning the pot the counter moves
back and forth slowly a pair of times, it takes several minutes, and
then stabilizes very slowly. Not an easy process, better to take
note of the screwdriver movements, and wait many minutes
between the adjustments. This is how the phase changed
from lagging to tracking the pps while I corrected the frequency:



This means that the two signals track each other, and
a quick check with the FE5680A confirms that I'm near,
somewhere around 2x10^-10 slow:



Now time to go to bed. As I have more data I will post them,
I think it will be difficult to show aging effects (unless I log the
unit for many days or the quartz is gone), since the unit
is aged and the aging should very slow now. Probably
it will be easier to try to see temperature effects.
Fabio.

[saturation]

Most excellent Fabio, this is the best characterized data for the 199x series with OXCO against a better standard.  Your performance improvement is fantastic but I'd risk it only myself if I had a better standard to compare it against, as you have. 

[lewis]

Here's the schematic if anyone needs it: http://www.schematicsunlimited.com/?z=racal_dana

[muvideo]

Most excellent Fabio, this is the best characterized data for the 199x series with OXCO against a better standard.  Your performance improvement is fantastic but I'd risk it only myself if I had a better standard to compare it against, as you have. 

Thank you saturation, but that was only early data,
against a gps.
I kept retouching the trim in last two days, last adjustments
are very very touchy. Now I'm unexspectly seeing a second
order change in drift, so I'm seeing the oscillator change
frequency, and I still dont know if it is it's natural aging
or it's  stabilizing after the trimming. The fact that it's slowing
is compatible with the fact that it was low, I was not expecting
such a visible drift, but I've still not wotked the numbers, I will
wait few days and log the behaviour.
I'm now regretting to have changed the trim so early,
I should have logged the counter for a week before,
just to see the stability level it had before trimming.

I will post more details net week, and will describe
the setup I'm using, it's easy to setup.
Fabio.

[slburris]

I keep meaning to ask, where are you finding the BK counter under $470?

I was also wondering if you had an opinion on the 1856D, their 3.5 GHz counter for about $525?

Scott

Thanks muvideo, and also thanks for your photos on the archives.  That helped me a lot too.  I think the prescaler board is identical too, I have blow ups on the flikr site and I compared it to your photos.  The additional functions are great as it give me the capability to do some measurements that normally you'd need a 1 GHz grade scope to do.

Yes, it has the ovenized reference, 04e option.  Its good to know it hold well against a Rb clock, so I can hold off on getting one for now.

Regardless, I agree with your comments on any effort repair these!  These counters are well worth the effort, my only fear is a firmware or memory failure, but in reviewing prior posts on any model in this series, no spontaneous failures have been reported unlike some early HP3456 DVM. 

My unit had a small pouch will the old firmware ROMs still in them inserted on the mobo, so I have a backup in case of diasaster 

FWIW, BK's model:



Is actually fairly close to the 1999.  Listed at $470,  you can deal it down to the high $300s, which will match the highest end price some eBay sellers want for many 199x series.

Nice machine saturation.
The heart (counter and interpolator) seem similar
to the 1992, the single shot resolution is the same, but
this one has many more functions: rise and fall times for
pulses from 5nS, gate window programmable, higher voltage
capability and statistical calculation, so more useful than the
1992, good catch!
edit: also one more digit on the display 10 vs 9, easier to
read for high resolutions.
I found some specs here:
http://www.teknetelectronics.com/DataSheet/RACALDANA/RACAL_199590030.pdf

What timebase do you have?  If it's the 9462 (04E) should
be good, my 1992 holds pretty well against Rb clocks.

The 199x switch problems are annoying, but thanking to
them I was able to pay few dollars for very good counter, in
mint conditions, the fix was few hours of work and few
cheap switches

Fabio.