Changing FE-5650A frequency?
Hi Alec,
I am going to agree with Robert (G8RPI) on his assessment, particularly as the dip switches are clearly visible in one of the pictures. Although I wonder why Robert feels unfortunate about the [ two chip DDS arrangement ] ?
It matters not the number of chips that make up the DDS. Why ? Because the actual physics package and its electronics has nothing to do with the DDS in the older designs. That is, the signal output of the physics package is a fixed frequency (around 50.255 MHz) that you could capture and feed to a more modern DDS if necessary as replacement of the old one is probably not going to happen. However, if it is working then once you get to a frequency you need you probably not touch it again. So no loss there.
As for calibration, there is a hole on one side of the can that has a multi-turn pot that adjusts the "C-field" current. The C-field is a very fine frequency adjustment. You first adjust the C-field pot to its lowest setting (as in frequency), then set the DDS to the closest frequency on downside of where you want to go and then adjust the C-field pot to come up on to the frequency. The C-field pot is a really fine adjustment, so we are talking very small movement that will require patience, a better reference and at least a
oscilloscope to watch the drift rate over a very long time. A truly time consuming experience.
However, if you leave that C-field pot alone, you will most probably be inside 1 part in 10 to the minus 9th.
The most recent designs have the DDS in the control loop for the physics package. In the new internal design in order to change frequencies you need to actually change the BASE crystal frequency, the DDS and the firmware to get a new output frequency. So basically a real pain in the ass.
As for the foam that is sandwiched in between the boards, I agree with Robert that it is for temperature stability in varying environments. If the unit is kept in a normal room all the time then the foam is not a concern and could be carefully cleaned off. There is some other components that have white stuff that looks more like an RTV type substance used to hold down a component like what is seen in photograph IMG_20130407_102937.jpg at the two toroids, That I would leave alone.
Your finding the 12+ MHz just confirms that the design is very similar to that described in the FTP file I provided. Make sure you record the switch settings before you change them so you can return to a reference point.
Although you did not indicate, I assume you downloaded the FTP file ? Please let me know so i can reduce the storage level as it is not my site although I have use of it on a short term basis.
Thank you,
Bill....WB6BNQ
Alexander Wright wrote:
On 07/04/13 17:56, Robert Atkinson wrote:
Hi Alex,
These units were used as timing references for secure communication systems (Havequick). see ebay item 130832014630. Unfortunatly they are an older design with a two chip DDS. The other problem is that the "M" designation is for military and means they have been partially encapsulated with polyurethene foam. You can cut it away but it is very easy to damage the PCB. Frequency setting is by DIP switches (under the foam) allowing full range of frequency selection. However the output filter is narrow so you can't go far from 800kHz.
Did you buy all seven?
Robert G8RPI.
Robert,
Thanks for the info! I bought one of these wondering what they were,
took it apart and found the rubidium clock and figured it must be worth
something (more than the 15 quid he was asking) so I bought the remaining 6!
I've taken the foam off the top PCB and i've found that there's a
12.8MHz signal being produced which is then being divided down with an
800kHz logic chip. I should imagine the filter must be on the circuit
board below? I might try changing the frequency
of the DDS chip, removing the frequency divider and tinkering with the
output filter. Do you have any info on what sort of filter I should
expect to find?
Thanks,
Alec M0TEI
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.
wow, a little more picking will get you to the dds pinout wrt the 24-bit
programming word...
Alexander Wright
On 07/04/13 15:10, Herbert Poetzl wrote:
The purpose of the polymer foam is isolation to keep a
stable temperature throughout the boards.
http://www.qsl.net/z/zl1bpu/PROJ/FE5650-2.jpg
The top layer in this image is the DDS board, and it is
basically identical to the FE5680A, which means it can
be adjusted in a wide range, but only does digital synthesis
based on the reference frequency.
But let's see what Bill says to the pictures ... :)
best,
Herbert
Herbert, the board doesn't seem to be the same, the chips which are
visible don't line up with the ones in that picture. Anyway my curiosity
got the better of me and I dug through the foam some more to find a long
row of DIP switches. I bet i can change the divider ratio with some
combination of those:
http://m0tei.co.uk/fe5650pics/IMG_20130407_170929.jpg
Also, you can see that the chip is an AD9955 DDS chip:
http://novatech-instr.com/PDF_files/AD9955.pdf
Alec
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."
De Erroribus Medicorum, R. Bacon, 13th century.
"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
On 07/04/13 21:55, WB6BNQ wrote:
Hi Alec,
I am going to agree with Robert (G8RPI) on his assessment, particularly as the dip switches are clearly visible in one of the pictures. Although I wonder why Robert feels unfortunate about the [ two chip DDS arrangement ] ?
It matters not the number of chips that make up the DDS. Why ? Because the actual physics package and its electronics has nothing to do with the DDS in the older designs. That is, the signal output of the physics package is a fixed frequency (around 50.255 MHz) that you could capture and feed to a more modern DDS if necessary as replacement of the old one is probably not going to happen. However, if it is working then once you get to a frequency you need you probably not touch it again. So no loss there.
As for calibration, there is a hole on one side of the can that has a multi-turn pot that adjusts the "C-field" current. The C-field is a very fine frequency adjustment. You first adjust the C-field pot to its lowest setting (as in frequency), then set the DDS to the closest frequency on downside of where you want to go and then adjust the C-field pot to come up on to the frequency. The C-field pot is a really fine adjustment, so we are talking very small movement that will require patience, a better reference and at least a
oscilloscope to watch the drift rate over a very long time. A truly time consuming experience.
However, if you leave that C-field pot alone, you will most probably be inside 1 part in 10 to the minus 9th.
The most recent designs have the DDS in the control loop for the physics package. In the new internal design in order to change frequencies you need to actually change the BASE crystal frequency, the DDS and the firmware to get a new output frequency. So basically a real pain in the ass.
As for the foam that is sandwiched in between the boards, I agree with Robert that it is for temperature stability in varying environments. If the unit is kept in a normal room all the time then the foam is not a concern and could be carefully cleaned off. There is some other components that have white stuff that looks more like an RTV type substance used to hold down a component like what is seen in photograph IMG_20130407_102937.jpg at the two toroids, That I would leave alone.
Your finding the 12+ MHz just confirms that the design is very similar to that described in the FTP file I provided. Make sure you record the switch settings before you change them so you can return to a reference point.
Although you did not indicate, I assume you downloaded the FTP file ? Please let me know so i can reduce the storage level as it is not my site although I have use of it on a short term basis.
Thank you,
Bill....WB6BNQ
Bill,
Thanks for your input and sorry for the slow reply!
So, I've had a look at the datasheet and it looks like the output
frequency should be Fin*(tuning word)/2^32... I thought I'd check if
this matches up.
The tuning word was set to x4134111F (towards the inside of the board
seems to represent 1 and towards the outside, zero.
Assuming 50.255MHz going in that gives.... 12.799985 MHz output!
Perfect. Pretty much exactly what I'd measured. Alternatively I could
assume that the output is 12.8MHz exactly which sets the input frequency
to 50.2550586MHz - i wonder which assumption I should use when changing
the frequency? I'd like to adjust it to put 10MHz out, skip the 4000
series logic divider and remove/change the output filter.
Regards,
Alec
Success! I've managed to set mine to 10MHz. I've documented this on my
blog:
http://blog.m0tei.co.uk/post/2013/04/20/Mystery-Aircraft-Parts-and-Atomic-Clocks
Thanks to those who offered suggestions.
Regards
Alec
Alexander Wright wrote:
On 07/04/13 21:55, WB6BNQ wrote:
Hi Alec,
I am going to agree with Robert (G8RPI) on his assessment, particularly as the dip switches are clearly visible in one of the pictures. Although I wonder why Robert feels unfortunate about the [ two chip DDS arrangement ] ?
It matters not the number of chips that make up the DDS. Why ? Because the actual physics package and its electronics has nothing to do with the DDS in the older designs. That is, the signal output of the physics package is a fixed frequency (around 50.255 MHz) that you could capture and feed to a more modern DDS if necessary as replacement of the old one is probably not going to happen. However, if it is working then once you get to a frequency you need you probably not touch it again. So no loss there.
As for calibration, there is a hole on one side of the can that has a multi-turn pot that adjusts the "C-field" current. The C-field is a very fine frequency adjustment. You first adjust the C-field pot to its lowest setting (as in frequency), then set the DDS to the closest frequency on downside of where you want to go and then adjust the C-field pot to come up on to the frequency. The C-field pot is a really fine adjustment, so we are talking very small movement that will require patience, a better reference and at least a
oscilloscope to watch the drift rate over a very long time. A truly time consuming experience.
However, if you leave that C-field pot alone, you will most probably be inside 1 part in 10 to the minus 9th.
The most recent designs have the DDS in the control loop for the physics package. In the new internal design in order to change frequencies you need to actually change the BASE crystal frequency, the DDS and the firmware to get a new output frequency. So basically a real pain in the ass.
As for the foam that is sandwiched in between the boards, I agree with Robert that it is for temperature stability in varying environments. If the unit is kept in a normal room all the time then the foam is not a concern and could be carefully cleaned off. There is some other components that have white stuff that looks more like an RTV type substance used to hold down a component like what is seen in photograph IMG_20130407_102937.jpg at the two toroids, That I would leave alone.
Your finding the 12+ MHz just confirms that the design is very similar to that described in the FTP file I provided. Make sure you record the switch settings before you change them so you can return to a reference point.
Although you did not indicate, I assume you downloaded the FTP file ? Please let me know so i can reduce the storage level as it is not my site although I have use of it on a short term basis.
Thank you,
Bill....WB6BNQ
Bill,
Thanks for your input and sorry for the slow reply!
So, I've had a look at the datasheet and it looks like the output
frequency should be Fin*(tuning word)/2^32... I thought I'd check if
this matches up.
The tuning word was set to x4134111F (towards the inside of the board
seems to represent 1 and towards the outside, zero.
Assuming 50.255MHz going in that gives.... 12.799985 MHz output!
Perfect. Pretty much exactly what I'd measured. Alternatively I could
assume that the output is 12.8MHz exactly which sets the input frequency
to 50.2550586MHz - i wonder which assumption I should use when changing
the frequency? I'd like to adjust it to put 10MHz out, skip the 4000
series logic divider and remove/change the output filter.
Regards,
Alec
Hi Alex,
Your alternate assumption is the correct one. Working the formula backwards with the assumptions that the 800 KHz is spot on would then dictate that the DDS out is exactly 12.8 MHz with the final value of the physics package being
50,255,058.6495
based upon my HP-35s calculator.
Many construction and environmental factors affect the actual final frequency of the physics package. Plus there is some small variance about the center frequency with which signal levels are still strong enough to allow functionality. The C-field adjustment allow for tweaking within that small variance. Fortunately, this ability allows for putting the Rb right on the assumed true frequency.
Provided you have a higher reference source (i.e., Cesium or a well tamed GPS arrangement), the way to properly adjust the Rubidium is to adjust the physics package C-field pot to it minimum frequency point, then set the DDS to the closest point just below the desired frequency and then re-adjust the C-field pot to come up onto the proper frequency. This is a slow repetitive process requiring time and patience. The degree of patience will directly correlate with the precision obtained.
However, if you are not dying to have absolute accuracy, but more interested in the stability provided by the Rb, then do not mess with the C-field pot. If you do not have the means for the calibration, it would, most likely, be safe to assume that the Rb is within 1x10-9 if other means are used to make sure some gross error is not evident.
I read your blog and, besides the success, it looked like you were having fun. Good luck,
Bill....WB6BNQ
Hi Alec,
Well done and thanks for the credit. I had not looked under the board for the crystal or spotted the divider.
One comment. in your blog and ebay description you say that the unit runs on 15V. The "M" suffix 5650A is actually designed for a normal aircraft 28V DC supply (typical specification range 22-32V). It may run on 15V but it may be stressing the power supply circuit and / or under-running the oven. The external (28V) supply to the Marconi unit is just filtered, switched and connected to the FE-5650A without regulation. It appears that these units have options 08, 09, 22 and 25 rolled up into the "M" military suffix. By the way, these units can cost $10,000 or more each for specials see http://www.dlis.dla.mil/webflis/pub/pub_search.aspx
Robert G8RPI.
From: Alexander Wright alecjw@gmail.com
To: Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Saturday, 20 April 2013, 2:33
Subject: Re: [time-nuts] Changing FE-5650A frequency?
Success! I've managed to set mine to 10MHz. I've documented this on my blog: http://blog.m0tei.co.uk/post/2013/04/20/Mystery-Aircraft-Parts-and-Atomic-Clocks
Thanks to those who offered suggestions.
Regards
Alec
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.
Hi Alec,
Well done and thanks for the credit. I had not looked under the board for the crystal or spotted the divider.
One comment. in your blog and ebay description you say that the unit runs on 15V. The "M" suffix 5650A is actually designed for a normal aircraft 28V DC supply (typical specification range 22-32V). It may run on 15V but it may be stressing the power supply circuit and / or under-running the oven. The external (28V) supply to the Marconi unit is just filtered, switched and connected to the FE-5650A without regulation. It appears that these units have options 08, 09, 22 and 25 rolled up into the "M" military suffix. By the way, these units can cost $10,000 or more each for specials see http://www.dlis.dla.mil/webflis/pub/pub_search.aspx
Robert G8RPI.
From: Alexander Wright alecjw@gmail.com
To: Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Saturday, 20 April 2013, 2:33
Subject: Re: [time-nuts] Changing FE-5650A frequency?
Success! I've managed to set mine to 10MHz. I've documented this on my blog: http://blog.m0tei.co.uk/post/2013/04/20/Mystery-Aircraft-Parts-and-Atomic-Clocks
Thanks to those who offered suggestions.
Regards
Alec
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.
On 20/04/13 10:03, Robert Atkinson wrote:
Hi Alec,
Well done and thanks for the credit. I had not looked under the board for the crystal or spotted the divider.
One comment. in your blog and ebay description you say that the unit runs on 15V. The "M" suffix 5650A is actually designed for a normal aircraft 28V DC supply (typical specification range 22-32V). It may run on 15V but it may be stressing the power supply circuit and / or under-running the oven. The external (28V) supply to the Marconi unit is just filtered, switched and connected to the FE-5650A without regulation. It appears that these units have options 08, 09, 22 and 25 rolled up into the "M" military suffix. By the way, these units can cost $10,000 or more each for specials see http://www.dlis.dla.mil/webflis/pub/pub_search.aspx
Robert G8RPI.
Robert,
Good point, well observed!
The power-trends thing inside the package is an adjustable 3A switching
regulator which seems to be set to nominally ~15.3V (91k setting
resistor to ground). When i crank up the input voltage, the output
voltage of the regulator settles to around 15.5v once the input voltage
reaches around 19v. According the the datasheet the dropout voltage of
those regulators is 4V, which means these frequency standards should be
happy with any input from 19.5-38v.
Where did you find the info about pricing for these? I searched in that
form for FE-5650A and couldn't find it. I tried searching the whole
"stable oscillator" thing's NSN to no avail. I did expect these must
have been very expensive when they were new though!
On 20/04/13 04:10, WB6BNQ wrote:> > Hi Alex,
Your alternate assumption is the correct one. Working the formula
backwards with the assumptions that the 800 KHz is spot on would then
dictate that the DDS out is exactly 12.8 MHz with the final value of the
physics package being
50,255,058.6495
based upon my HP-35s calculator.
Many construction and environmental factors affect the actual final
frequency of the physics package. Plus there is some small variance
about the center frequency with which signal levels are still strong
enough to allow functionality. The C-field adjustment allow for
tweaking within that small variance. Fortunately, this ability allows
for putting the Rb right on the assumed true frequency.
Provided you have a higher reference source (i.e., Cesium or a well
tamed GPS arrangement), the way to properly adjust the Rubidium is to
adjust the physics package C-field pot to it minimum frequency point,
then set the DDS to the closest point just below the desired frequency
and then re-adjust the C-field pot to come up onto the proper frequency.
This is a slow repetitive process requiring time and patience. The
degree of patience will directly correlate with the precision obtained.
However, if you are not dying to have absolute accuracy, but more
interested in the stability provided by the Rb, then do not mess with
the C-field pot. If you do not have the means for the calibration, it
would, most likely, be safe to assume that the Rb is within 1x10-9 if
other means are used to make sure some gross error is not evident.
I read your blog and, besides the success, it looked like you were
having fun. Good luck,
Bill....WB6BNQ
Bill,
I thought as much, thanks for confirming! I'll keep an eye out for any
callibrated frequency reference I can borrow for 10 minutes to set this
with, but otherwise I'll leave it as is.
On 20/04/13 09:37, GandalfG8@aol.com wrote:> Hi Alec
I see your web page and Ebay auction, are you able to supply one still
complete and unmodified in its original enclosure?
Regards
Nigel
GM8PZR
Nigel,
You mean just the FE-5650A enclosure or the whole "stable oscillator"
thing? Do you want to ping me off list?
Regards,
Alec
Hi Alec,
I'm an expert searcher ;-). I actually searched for FE-5650 and NSN (NATO Stock Number or National Stock Number to Americans) and then plugged the NSN into Webflis. There are lots of websites out there that trawl for part numbers and NSNs and then try to charge you for information that is freely available elsewere. Another good source of data is military specifications. Assist Quick search is good but the search engine is a bit fussy http://quicksearch.dla.mil/ try plugging crystal into the title keyword search.
Robert G8RPI.
From: Alexander Wright alecjw@gmail.com
To: Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Saturday, 20 April 2013, 12:11
Subject: Re: [time-nuts] Changing FE-5650A frequency?
On 20/04/13 10:03, Robert Atkinson wrote:
Hi Alec,
Well done and thanks for the credit. I had not looked under the board for the crystal or spotted the divider.
One comment. in your blog and ebay description you say that the unit runs on 15V. The "M" suffix 5650A is actually designed for a normal aircraft 28V DC supply (typical specification range 22-32V). It may run on 15V but it may be stressing the power supply circuit and / or under-running the oven. The external (28V) supply to the Marconi unit is just filtered, switched and connected to the FE-5650A without regulation. It appears that these units have options 08, 09, 22 and 25 rolled up into the "M" military suffix. By the way, these units can cost $10,000 or more each for specials see http://www.dlis.dla.mil/webflis/pub/pub_search.aspx
Robert G8RPI.
Robert,
Good point, well observed!
The power-trends thing inside the package is an adjustable 3A switching regulator which seems to be set to nominally ~15.3V (91k setting resistor to ground). When i crank up the input voltage, the output voltage of the regulator settles to around 15.5v once the input voltage reaches around 19v. According the the datasheet the dropout voltage of those regulators is 4V, which means these frequency standards should be happy with any input from 19.5-38v.
Where did you find the info about pricing for these? I searched in that form for FE-5650A and couldn't find it. I tried searching the whole "stable oscillator" thing's NSN to no avail. I did expect these must have been very expensive when they were new though!
On 20/04/13 04:10, WB6BNQ wrote:> > Hi Alex,
Your alternate assumption is the correct one. Working the formula backwards with the assumptions that the 800 KHz is spot on would then dictate that the DDS out is exactly 12.8 MHz with the final value of the physics package being
50,255,058.6495
based upon my HP-35s calculator.
Many construction and environmental factors affect the actual final frequency of the physics package. Plus there is some small variance about the center frequency with which signal levels are still strong enough to allow functionality. The C-field adjustment allow for tweaking within that small variance. Fortunately, this ability allows for putting the Rb right on the assumed true frequency.
Provided you have a higher reference source (i.e., Cesium or a well tamed GPS arrangement), the way to properly adjust the Rubidium is to adjust the physics package C-field pot to it minimum frequency point, then set the DDS to the closest point just below the desired frequency and then re-adjust the C-field pot to come up onto the proper frequency. This is a slow repetitive process requiring time and patience. The degree of patience will directly correlate with the precision obtained.
However, if you are not dying to have absolute accuracy, but more interested in the stability provided by the Rb, then do not mess with the C-field pot. If you do not have the means for the calibration, it would, most likely, be safe to assume that the Rb is within 1x10-9 if other means are used to make sure some gross error is not evident.
I read your blog and, besides the success, it looked like you were having fun. Good luck,
Bill....WB6BNQ
Bill,
I thought as much, thanks for confirming! I'll keep an eye out for any callibrated frequency reference I can borrow for 10 minutes to set this with, but otherwise I'll leave it as is.
On 20/04/13 09:37, GandalfG8@aol.com wrote:> Hi Alec
I see your web page and Ebay auction, are you able to supply one still
complete and unmodified in its original enclosure?
Regards
Nigel
GM8PZR
Nigel,
You mean just the FE-5650A enclosure or the whole "stable oscillator" thing? Do you want to ping me off list?
Regards,
Alec
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.