FS: HP 5328a
No master oscillator. Complete and untested otherwise. has counter to 512MHz. GPIB. 35$ plus shipping
73 de Norm n3ykf
Considering some people were exiled from this list, told to take their
discussions elsewhere (DST, Emperor's New Timescale, etc.), why is this
mini-eBay type of announcement tolerated?
I personally have no problems with the once, twice a week offers, just
intrigued by the apparent double standard. Is there a charter document for
the list, specifying what is allowed, what is encouraged, discouraged,
censored?
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of normn3ykf@stny.rr.com
Sent: Wednesday, August 17, 2011 10:23 AM
To: time-nuts@febo.com
Subject: [time-nuts] FS: HP 5328a
No master oscillator. Complete and untested otherwise. has counter to
512MHz. GPIB. 35$ plus shipping
73 de Norm n3ykf
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.
This is not a mini ebay. It is a straight for sale listing.
Any issues pm the moderator. In other words: Take it off the list.
73 de Norm n3ykf
Hi Jose,
We/I have no written policy on this, but my feeling is that anyone
that wants to mention their precision time related company or T&F
specific gear for direct sale, or on eBay - maybe once or twice a
year - is more than welcome to do so. We all win when modern or
vintage time & frequency instruments are personally handed from
one time-nut to another. Rarely has anyone abused this privilege.
The intro/charter page is www.leapsecond.com/time-nuts.htm
There is no double standard. When I sense threads get way too
far off-topic or when threads belong in the LEAPSECS list (UTC,
timescales, leap seconds) instead of time-nuts, then I send out
polite email off-line.
/tvb
Considering some people were exiled from this list, told to take their
discussions elsewhere (DST, Emperor's New Timescale, etc.), why is this
mini-eBay type of announcement tolerated?
I personally have no problems with the once, twice a week offers, just
intrigued by the apparent double standard. Is there a charter document for
the list, specifying what is allowed, what is encouraged, discouraged,
censored?
Gee , I think under $50 puts it in the "free to a good home catagory"
Complainers should get a life!.
--- On Wed, 8/17/11, normn3ykf@stny.rr.com normn3ykf@stny.rr.com wrote:
From: normn3ykf@stny.rr.com normn3ykf@stny.rr.com
Subject: [time-nuts] FS: HP 5328a
To: "time-nuts@febo.com" time-nuts@febo.com
Date: Wednesday, August 17, 2011, 10:22 AM
No master oscillator. Complete and untested otherwise. has counter to 512MHz. GPIB. 35$ plus shipping
73 de Norm n3ykf
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.
Understood. As I said, I wasn't complaining about the offers here, just
curious about what is welcome and what isn't. Your post clarified it.
BTW, thanks for leapsecond.com - excellent site - technical with a sense of
humor.
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of Tom Van Baak
Sent: Wednesday, August 17, 2011 4:04 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] FS: HP 5328a
Hi Jose,
We/I have no written policy on this, but my feeling is that anyone
that wants to mention their precision time related company or T&F
specific gear for direct sale, or on eBay - maybe once or twice a
year - is more than welcome to do so. We all win when modern or
vintage time & frequency instruments are personally handed from
one time-nut to another. Rarely has anyone abused this privilege.
The intro/charter page is www.leapsecond.com/time-nuts.htm
There is no double standard. When I sense threads get way too
far off-topic or when threads belong in the LEAPSECS list (UTC,
timescales, leap seconds) instead of time-nuts, then I send out
polite email off-line.
/tvb
Considering some people were exiled from this list, told to take their
discussions elsewhere (DST, Emperor's New Timescale, etc.), why is this
mini-eBay type of announcement tolerated?
I personally have no problems with the once, twice a week offers, just
intrigued by the apparent double standard. Is there a charter document for
the list, specifying what is allowed, what is encouraged, discouraged,
censored?
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.
At 01:22 PM 8/17/2011, normn3ykf@stny.rr.com wrote...
No master oscillator. Complete and untested otherwise. has counter to
512MHz. GPIB. 35$ plus shipping
Good deal. I'd think most around here have a 10 MHz source to feed it
with, so the lack of OSC shouldn't be an issue. I'd jump, but already
have two.
I will point out that the HP 5328a counter has a function where it can
act as a (1:10-1:10e7) decade divider. It's in the documentation, but
not obvious. One nice thing is that the input is subject to the level
and slope controls, which are much more capable than what's found on
most simple dividers, so you can work with a wider range of input
signals. It may work even without an internal OSC!
HP: "START A, STOP. Sets the counter to totalize the number of events
at the A input until STOP is selected, for N=l on the RESOLUTION
switch. For N>l, the number of counts divided by N is totalized. The
scaled output (i.e., frequency of A/N) is available at the Time Base
Out rear panel connector."
I don't know that anyone has characterized the jitter when using this
feature, though.
Mike wrote:
I will point out that the HP 5328a counter has a function where it
can act as a (1:10-1:10e7) decade divider. It's in the
documentation, but not obvious.
But note that the military version, which seems to be by far the most
commonly available, does not have this feature.
Best regards,
Charles
Hello Folks,
I am looking for an instrument that is better than Fluke 103a comparator.
The purpose of this piece of equipment is help me learn more about oscillators and characterizing them
The HP5370B is the TIC I keep hearing about but I am not glued to that make or model.
I wou;d like to hear suggestion from the group.
Another way to go might be to build one of those units where there is a common oscillator is split and feeds the LO of 2 mixers.
The RF side comes from the DUT and the REF
Sorry I cannot remember the acronym... MDMM????
I do have that Keithly model 776 Counter time with GPIB I am really fuzzy on this aspect...
Comments welcome
Thank you
PaulC
W1VLF
From: To: Sent:
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 18/08/11 05:10, Paul Cianciolo wrote:
Hello Folks,
I am looking for an instrument that is better than Fluke 103a comparator.
The purpose of this piece of equipment is help me learn more about oscillators and characterizing them
The HP5370B is the TIC I keep hearing about but I am not glued to that make or model.
I wou;d like to hear suggestion from the group.
Could get you started. For good quality oscillators the resolution will
still be a limiting factor for short taus. For longer taus it should
give you good bang for the buck. GPIB interface should be something you
also invest in.
Another way to go might be to build one of those units where there is a common oscillator is split and feeds the LO of 2 mixers.
The RF side comes from the DUT and the REF
Sorry I cannot remember the acronym... MDMM????
Dual Mixer Time Difference (DMTD)
You may have to look for a suitable offset oscillator, since you do want
a low noise one. Also consider that you need to feed it two signals, one
reference and one DUT. You will see the sum of both signals.
With a third mixer you can do three-cornered hat things to resolve the
phase noise of the individual oscillators. Naturally there is some
problems with all that, and isolation of the sources becomes important.
Ah well.
Cheers,
Magnus
Paul Cianciolo wrote:
Hello Folks,
I am looking for an instrument that is better than Fluke 103a comparator.
The purpose of this piece of equipment is help me learn more about oscillators and characterizing them
The HP5370B is the TIC I keep hearing about but I am not glued to that make or model.
I wou;d like to hear suggestion from the group.
Another way to go might be to build one of those units where there is a common oscillator is split and feeds the LO of 2 mixers.
The RF side comes from the DUT and the REF
Sorry I cannot remember the acronym... MDMM????
I do have that Keithly model 776 Counter time with GPIB I am really fuzzy on this aspect...
Comments welcome
Thank you
PaulC
W1VLF
From: To: Sent:
DMTD or dual mixer time difference.
An all digital version (DDMTD like that used in CERN's White Rabbit) is
also possible
3 (and M) input versions are also feasible.
Either an FPGA or a CPD plus a microprocessor are required to ensure low
cost.
Sub 10ps resolution is readily achievable without significant difficulty.
Bruce
Hi
The real answer (as always) is "that depends". Keeping things very simple
(ADEV only):
At one second Tau, a good eBay oscillator will be in the vicinity of
3x10^-12. The 5370 has a single shot resolution of 20 ps RMS. That might get
you to 2x10^-11. The 5370 won't let you do this level of 1 second ADEV
directly.
The same oscillator might get to 1x10^-12 at 100 seconds. The 5370's 20 ps
would give you 2x10^-13 at 100 seconds. Between 1 and 100 seconds, the 5370
becomes useful for measuring ADEV in this case.
In both cases, you still need a reference that's as good as or better than
the oscillator you are testing. The numbers are all a comparison of
reference to device under test. This is true no matter what technique you
use. Since it's always a comparison, there are other approaches.
A quick and simple approach with what you have is a single mixer. Offset
(tune) the device you are testing by 1 to 10 Hz from your reference. The
Rakon would be at 10.000001 MHz and the Z3801 at 10.000000 MHz. Plug the
Rakon into one RF port on a double balanced mixer. Plug your reference into
the other RF port on the mixer. Take the beat note (1 to 10 Hz audio) out of
the IF port. Filter it a bit to get rid of the RF on it. Feed it into your
Keithly 776 counter.
When the device you are testing moves by 0.1 Hz, the beat note does the
same. At 10 MHz 0.1 Hz is 1x10^-8. At 1 Hz 0.1 Hz is 10%. The mix down
"amplifies" the change in frequency by 1x10^7 in this case. If your Keithly
could read 9 good digits off of the 1 Hz, you would have 1x10^-18 data.
Life is never this simple. Converting the audio sine wave to a square wave
is a noisy process. That limits the resolution you can actually get. Your
counter input is likely not ideal for this task. There are a variety of op
amp circuits you can use to significantly improve on it. You still won't get
1x10^-18, but you can get to 1x10^-12. With some care and luck you might
even do better than that.
A chunk of perf board, and the parts you have sitting in a junk box will
probably do well enough to get you started. To buy simple stuff and wire it
up on perf board should be below $50. A full blown approach with a fancy
multi stage filter/limiter and a PC board might get you into the $200 range.
All this assumes that you offset the device you are testing from the
reference. That lets you use a single mixer. If you want a more general
approach where the device under test is not offset, then the DMTD is the way
to go. More money (more mixers), but it removes a constraint.
Lots of ways to go.
Bob
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of Paul Cianciolo
Sent: Wednesday, August 17, 2011 11:11 PM
To: Discussion of precise time and frequency measurement
Subject: [time-nuts] any HP 5370B Available or other TIC
Hello Folks,
I am looking for an instrument that is better than Fluke 103a comparator.
The purpose of this piece of equipment is help me learn more about
oscillators and characterizing them
The HP5370B is the TIC I keep hearing about but I am not glued to that make
or model.
I wou;d like to hear suggestion from the group.
Another way to go might be to build one of those units where there is
a common oscillator is split and feeds the LO of 2 mixers.
The RF side comes from the DUT and the REF
Sorry I cannot remember the acronym... MDMM????
I do have that Keithly model 776 Counter time with GPIB I am really fuzzy
on this aspect...
Comments welcome
Thank you
PaulC
W1VLF
From: To: Sent:
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.
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.
El 18/08/2011 08:37, Bruce Griffiths escribió:
DMTD or dual mixer time difference.
An all digital version (DDMTD like that used in CERN's White Rabbit)
is also possible
3 (and M) input versions are also feasible.
Either an FPGA or a CPD plus a microprocessor are required to ensure
low cost.
I've lying around one of these little things:
http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&CategoryNo=139&No=593
(that really I purchased only for amusement along with a bigger and
quite more expensive FPGA board). It lacks I/Fs except GPIOs (it only
has an USB that really are more intended for JTAG functionalities than
for other things), but seems a nice unexpensive core for a DDMTD. It
could be a fun project :) Anyone would like to join? ;)
Regards,
Javier
Javier Herrero wrote:
El 18/08/2011 08:37, Bruce Griffiths escribió:
DMTD or dual mixer time difference.
An all digital version (DDMTD like that used in CERN's White Rabbit)
is also possible
3 (and M) input versions are also feasible.
Either an FPGA or a CPD plus a microprocessor are required to ensure
low cost.
I've lying around one of these little things:
http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&CategoryNo=139&No=593
(that really I purchased only for amusement along with a bigger and
quite more expensive FPGA board). It lacks I/Fs except GPIOs (it only
has an USB that really are more intended for JTAG functionalities than
for other things), but seems a nice unexpensive core for a DDMTD. It
could be a fun project :) Anyone would like to join? ;)
Regards,
Javier
Javier
I've already done some preliminary test using 74HC164s as mixers to see
what the limits are.
With 5.000055MHz clock and 5MHz data inputs the noise floor was around
1E-11/Tau.
However better estimates of the beat frequency transitions than that
achievable using a 5370A to measure the relative delay between the mixer
output transitions are necessary.
CERN make all the required DDMTD VHDL code freely available.
One potential advantage is that the data input frequency could be any
harmonic of 5MHz with a 5.000055MHz shift register clock.
Thus one could compare 10MHz and 5MHz standards directly using a DDMTD.
Bruce
Hi,
Well, I do have an interest in fpga applications to mess about in the 10 ps rms ballpark.
It's along these lines:
http://www-ppd.fnal.gov/EEDOffice-w/Projects/ckm/comadc/PID765918.pdf
Together with another list member I'm working on a tdc implementation in fpga. We have a 50/50 xilinx/altera split... I'm working on a Spartan 6 (http://www.digilentinc.com/atlys), and he's working on a Cyclone IV. Luckily it turns out that the ideas and architecture carry over well enough between different fpga's since we get similar results.
The interests are roughly in this area:
- frequency counter
- time interval counter
- time stamper
- gpsdo
- dmtd
So while I am not directly looking to divide my tinkering time between this project and yet another project, if what you plan to do in the fpga precision timing realm coincides enough ... it might be useful to combine some things. At the very least we could swap some ideas.
So what did you have in mind?
regards,
Fred
From: Javier Herrero jherrero@hvsistemas.es
To: Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Thursday, August 18, 2011 9:46 PM
Subject: Re: [time-nuts] any HP 5370B Available or other TIC
El 18/08/2011 08:37, Bruce Griffiths escribió:
DMTD or dual mixer time difference.
An all digital version (DDMTD like that used in CERN's White Rabbit) is also possible
3 (and M) input versions are also feasible.
Either an FPGA or a CPD plus a microprocessor are required to ensure low cost.
I've lying around one of these little things: http://www.terasic.com.tw/cgi-bin/page/archive.pl?Language=English&CategoryNo=139&No=593 (that really I purchased only for amusement along with a bigger and quite more expensive FPGA board). It lacks I/Fs except GPIOs (it only has an USB that really are more intended for JTAG functionalities than for other things), but seems a nice unexpensive core for a DDMTD. It could be a fun project :) Anyone would like to join? ;)
Regards,
Javier
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.
El 18/08/2011 22:16, Tijd Dingen escribió:
So while I am not directly looking to divide my tinkering time between this project and yet another project, if what you plan to do in the fpga precision timing realm coincides enough ... it might be useful to combine some things. At the very least we could swap some ideas.
Yes, that would be the idea, to swap ideas :)
So what did you have in mind?
My main interest is a TIC with better resolution than a 5370 and also
with higher data flow, or with the capabilities to do some
post-processing of a bunch of data. Really it is an idea I've from long,
but not yet though too seriously on it (usually not too much time to
dedicate to it :) ). I was at first thinking on an interpolator based
TIC similar to the PICTIC idea (the ADC in the board seems handy for
that), I've also read some papers about FPGA implementation of ~10ps
resolution TICs, and then also the DDMTD idea has come. I really bought
that little board only to play around (I could not resist the $79, that
is only a bit over the digikey price for the FPGA alone, and that all
the I/O pins are 2.54" spaced so it is very easy to play around with the
signals and to build a "carrier" board for interfacing it) - the idea
that could be the core for a nice TIC came after :) And well, if
something good comes from it, we could "industrialize" the result at
time-nuts level if there is some interest in the list :)
Regards,
Javier
--
Javier Herrero
Chief Technology Officer EMAIL: jherrero@hvsistemas.com
HV Sistemas S.L. PHONE: +34 949 336 806
Los Charcones, 17 FAX: +34 949 336 792
19170 El Casar - Guadalajara - Spain WEB: http://www.hvsistemas.com
On Thu, Aug 18, 2011 at 11:26 PM, Javier Herrero jherrero@hvsistemas.eswrote:
My main interest is a TIC with better resolution than a 5370 and also with
higher data flow, or with the capabilities to do some post-processing of a
bunch of data. Really it is an idea I've from long, but not yet though too
seriously on it (usually not too much time to dedicate to it :) ). I was at
first thinking on an interpolator based TIC similar to the PICTIC idea (the
ADC in the board seems handy for that), I've also read some papers about
FPGA implementation of ~10ps resolution TICs, and then also the DDMTD idea
has come. I really bought that little board only to play around (I could not
resist the $79, that is only a bit over the digikey price for the FPGA
alone, and that all the I/O pins are 2.54" spaced so it is very easy to play
around with the signals and to build a "carrier" board for interfacing it) -
the idea that could be the core for a nice TIC came after :) And well, if
something good comes from it, we could "industrialize" the result at
time-nuts level if there is some interest in the list :)
We are also working on a TDC in Spartan 6:
http://www.ohwr.org/projects/tdc-core/wiki
The result will be licensed under LGPL. The reference hardware platform will
be the SPEC board (http://www.ohwr.org/projects/spec/wiki) using a simple
digital I/O mezzanine (http://www.ohwr.org/projects/fmc-dio-5chttla/wiki).
Both cards are licensed under CERN Open Hardware Licence (
http://www.ohwr.org/projects/cernohl/wiki) so no strings attached on either
code or hardware.
Cheers,
Javier
Hi Javier,
My main interest is a TIC with better resolution than a 5370 and also with higher data flow, or with the capabilities to do some post-processing of a bunch of data.
That sounds highly similar to the plan over here.
I was at first thinking on an interpolator based TIC similar to the PICTIC idea (the ADC in the board seems handy for that)
Generally speaking, any ADC approach that uses /affordable/ ADC's (read: no GHz sampling rates) will result in lower number of timestamps per second than with the TDC in fpga approach.
So if you only have to timestamp at a moderate rate, then the adc could very well suit your needs. If you want to do continuous timestamping at a high rate then the adc approach doesn't cut it. Or if it can, I'd be interested in some good reading material about that...
Besides with the fpga approach it is solve problem once, instantiate many times == multi-channel TDC for a nice price.
regards,
Fred
From: Javier Herrero jherrero@hvsistemas.es
To: time-nuts@febo.com
Sent: Thursday, August 18, 2011 11:26 PM
Subject: Re: [time-nuts] any HP 5370B Available or other TIC
El 18/08/2011 22:16, Tijd Dingen escribió:
So while I am not directly looking to divide my tinkering time between this project and yet another project, if what you plan to do in the fpga precision timing realm coincides enough ... it might be useful to combine some things. At the very least we could swap some ideas.
Yes, that would be the idea, to swap ideas :)
So what did you have in mind?
My main interest is a TIC with better resolution than a 5370 and also with higher data flow, or with the capabilities to do some post-processing of a bunch of data. Really it is an idea I've from long, but not yet though too seriously on it (usually not too much time to dedicate to it :) ). I was at first thinking on an interpolator based TIC similar to the PICTIC idea (the ADC in the board seems handy for that), I've also read some papers about FPGA implementation of ~10ps resolution TICs, and then also the DDMTD idea has come. I really bought that little board only to play around (I could not resist the $79, that is only a bit over the digikey price for the FPGA alone, and that all the I/O pins are 2.54" spaced so it is very easy to play around with the signals and to build a "carrier" board for interfacing it) - the idea that could be the core for a nice TIC came after :) And well, if something good comes from it, we could "industrialize" the result
at time-nuts level if there is some interest in the list :)
Regards,
Javier
Javier Herrero
Chief Technology Officer EMAIL: jherrero@hvsistemas.com
HV Sistemas S.L. PHONE: +34 949 336 806
Los Charcones, 17 FAX: +34 949 336 792
19170 El Casar - Guadalajara - Spain WEB: http://www.hvsistemas.com
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,
I just checked http://www.ohwr.org/projects/tdc-core/wiki ... under Status it says:
Basic data path of the TDC (delay line + encoder + LUT) designed. Timing is met at 125MHz with a 400-tap ~12ns delay line. Total latency is 5 cycles (40ns).
Any particular reason for choosing such a relatively low frequency for sampling? Because the number of taps for the TDC will be pretty large that way (400 in this case). Which I'd think would result in a larger accumulated uncertainty at the far end of the tdc, when compared with a shorter tdc. I did a couple of experiments, and those confirm that the measurement uncertainty slowly grows as a function of TDC length. Which is not so strange since you'll have dispersion while the pulse travels down the carry chain. So it seems a shorter TDC length (i.e higher sampling rate) would be preferable.
I'm meeting timing in a 390 MHz design with 192 taps (admittedly after a quite a bit of messing with the tools to get P&R to do what I want). The pipeline obviously is more stages... But maybe you guys are doing something different. Or maybe 125 MHz is enough, and you decided you don't want design headaches. Anyways, just curious as to why the 125 MHz sampling rate.
regards,
Fred
From: Javier Serrano javier.serrano.pareja@gmail.com
To: Discussion of precise time and frequency measurement time-nuts@febo.com
Sent: Thursday, August 18, 2011 11:50 PM
Subject: Re: [time-nuts] any HP 5370B Available or other TIC
On Thu, Aug 18, 2011 at 11:26 PM, Javier Herrero jherrero@hvsistemas.eswrote:
My main interest is a TIC with better resolution than a 5370 and also with
higher data flow, or with the capabilities to do some post-processing of a
bunch of data. Really it is an idea I've from long, but not yet though too
seriously on it (usually not too much time to dedicate to it :) ). I was at
first thinking on an interpolator based TIC similar to the PICTIC idea (the
ADC in the board seems handy for that), I've also read some papers about
FPGA implementation of ~10ps resolution TICs, and then also the DDMTD idea
has come. I really bought that little board only to play around (I could not
resist the $79, that is only a bit over the digikey price for the FPGA
alone, and that all the I/O pins are 2.54" spaced so it is very easy to play
around with the signals and to build a "carrier" board for interfacing it) -
the idea that could be the core for a nice TIC came after :) And well, if
something good comes from it, we could "industrialize" the result at
time-nuts level if there is some interest in the list :)
We are also working on a TDC in Spartan 6:
http://www.ohwr.org/projects/tdc-core/wiki
The result will be licensed under LGPL. The reference hardware platform will
be the SPEC board (http://www.ohwr.org/projects/spec/wiki) using a simple
digital I/O mezzanine (http://www.ohwr.org/projects/fmc-dio-5chttla/wiki).
Both cards are licensed under CERN Open Hardware Licence (
http://www.ohwr.org/projects/cernohl/wiki) so no strings attached on either
code or hardware.
Cheers,
Javier
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 Fri, Aug 19, 2011 at 12:44 AM, Tijd Dingen tijddingen@yahoo.com wrote:
I'm meeting timing in a 390 MHz design with 192 taps (admittedly after a
quite a bit of messing with the tools to get P&R to do what I want). The
pipeline obviously is more stages... But maybe you guys are doing something
different. Or maybe 125 MHz is enough, and you decided you don't want design
headaches. Anyways, just curious as to why the 125 MHz sampling rate.
125 MHz is a natural frequency for us because the FMC carrier will in normal
operation extract 125 MHz from the White Rabbit link (GbE), which will be
phase compensated to account for fiber-induced delay, and will have very
good long-term stability because it is derived from an GPSDO hooked to the
White Rabbit master node. This TDC will then produce proper TAI time stamps,
combining a integer number of 125 MHz ticks and a fractional part. We could
of course double the 125 MHz and use that instead. For the time being, we
are just playing.
Cheers,
Javier