Microstepper
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
Something like this? (see image)
On Wed, Dec 25, 2019 at 1:02 AM Magnus Danielson magnus@rubidium.se wrote:
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Magnus,
Why not just clock a good DDS (AD9854) with the reference frequency, and
run its
I & Q outputs into the motor via suitable LP filters and some power gain?
You might
need to periodically alternate between two different output frequencies to
get the desired
rotation speed (as with a fractional N divider), but with the 48-bit DDS
I'd think that this
would work well. After all, the magnetics in a motor are not all that
accurate anyway-
one cannot get perfection in avoiding salient pole effects etc.
Dana
On Tue, Dec 24, 2019 at 6:02 PM Magnus Danielson magnus@rubidium.se wrote:
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Hi,
That was not exactly what I was aiming for, but you got one aspect
potentially better than I proposed, that the offset generator is running
off the input. That is for sure a variant I did not think about as I
wrote it, but that was in my mind in an earlier variant of thoughts.
Now, the output I&Q DACs then mix with the I&Q of the offset, thus doing
the same as on the input side, but with opposite direction, and then
summed for an output.
The reference to the offset generator may be the input (as you draw it),
a free-running oscillator or a locked oscillator. When locked, the EFC
is from a separate DAC.
I could probably pull this off directly on a Red Pitaya.
Cheers,
Magnus
On 2019-12-25 01:53, Azelio Boriani wrote:
Something like this? (see image)
On Wed, Dec 25, 2019 at 1:02 AM Magnus Danielson magnus@rubidium.se wrote:
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Dana,
What motor? I have no motor involved here.
Alternating between frequencies causes modulation I try to avoid, a DDS
is a good way to avoid that, but I was hoping to keep phase noise low.
Cheers,
Magnus
On 2019-12-25 03:03, Dana Whitlow wrote:
Magnus,
Why not just clock a good DDS (AD9854) with the reference frequency, and
run its
I & Q outputs into the motor via suitable LP filters and some power gain?
You might
need to periodically alternate between two different output frequencies to
get the desired
rotation speed (as with a fractional N divider), but with the 48-bit DDS
I'd think that this
would work well. After all, the magnetics in a motor are not all that
accurate anyway-
one cannot get perfection in avoiding salient pole effects etc.
Dana
On Tue, Dec 24, 2019 at 6:02 PM Magnus Danielson magnus@rubidium.se wrote:
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Hi
Different sort of micro stepper. The 3D printing version runs a stepping motor
to improve the angular resolution. As you mention, the poles in the motor limit
just how well you can do. There are dedicated chips out there these days that
get down below the limits on any motor I’ve ever seen. They are available
on pre-assembled boards for not a lot of money per board.
The micro stepper Magnus is after is one to adjust the phase of the output of
his shiny new (to him) maser. Same name, different beast.
Bob
On Dec 24, 2019, at 9:03 PM, Dana Whitlow k8yumdoober@gmail.com wrote:
Magnus,
Why not just clock a good DDS (AD9854) with the reference frequency, and
run its
I & Q outputs into the motor via suitable LP filters and some power gain?
You might
need to periodically alternate between two different output frequencies to
get the desired
rotation speed (as with a fractional N divider), but with the 48-bit DDS
I'd think that this
would work well. After all, the magnetics in a motor are not all that
accurate anyway-
one cannot get perfection in avoiding salient pole effects etc.
Dana
On Tue, Dec 24, 2019 at 6:02 PM Magnus Danielson magnus@rubidium.se wrote:
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Thanks, Bob, for clarifying what Magnus means by "microstepper". To me,
the term has always referred to operation between the discrete steps defined
by the polepieces and windings. One can do it (albeit poorly) with many
stepper
motors. There is a class of servo motor in which the designers of the
magnetics
went to great lengths to avoid salient poles (think magnetic detent-like
effects),
which are meant to operate with I & Q controlled currents.
BTW, the "phase noise" associated with a DDS is not entirely random. In
fact
most is usually from systematic effects and can be eliminated or at least
dodged
by restricting one's choice of frequencies.
Dana
On Wed, Dec 25, 2019 at 8:00 AM Bob kb8tq kb8tq@n1k.org wrote:
Hi
Different sort of micro stepper. The 3D printing version runs a stepping
motor
to improve the angular resolution. As you mention, the poles in the motor
limit
just how well you can do. There are dedicated chips out there these days
that
get down below the limits on any motor I’ve ever seen. They are available
on pre-assembled boards for not a lot of money per board.
The micro stepper Magnus is after is one to adjust the phase of the output
of
his shiny new (to him) maser. Same name, different beast.
Bob
On Dec 24, 2019, at 9:03 PM, Dana Whitlow k8yumdoober@gmail.com wrote:
Magnus,
Why not just clock a good DDS (AD9854) with the reference frequency, and
run its
I & Q outputs into the motor via suitable LP filters and some power gain?
You might
need to periodically alternate between two different output frequencies
to
get the desired
rotation speed (as with a fractional N divider), but with the 48-bit DDS
I'd think that this
would work well. After all, the magnetics in a motor are not all that
accurate anyway-
one cannot get perfection in avoiding salient pole effects etc.
Dana
On Tue, Dec 24, 2019 at 6:02 PM Magnus Danielson magnus@rubidium.se
wrote:
Hi,
I realize that I lack a microstepper. Consider that I have a stable and
low-noise 5 or 10 MHz but I want to resynthesize to correct frequency
and do phase-steps, and doing so without too much loss of noise.
This has traditionally been done using a variation of techniques, but if
we would use some of the things that happened lately, pretty OK
performance should be possible to achieve without too much hardware.
OK, so after a discussion with Bob, here is one sketch for a
possibility, just to toss one proposal to crush into pieces and propose
improvements or better versions.
So, consider using a modern Silabs chip clocked from an oscillator,
producing a offset generator with I/Q and then do an I/Q mixdown to a
beat frequency, which is digitized by a pair of ADCs. A pair of DACs
produces I/Q which is used to mixed to produce the output signal using
the signals from the Silab. The ADCs/DACs can be either be fed to some
CPU platform or FPGA. With this platform one can choose to servo the
reference oscillator, or just modify the beat received. Maybe use a
Raspberry Pi as platform. Rotating the vector and steering the rate of
rotation should not be extremely hard to do. The input vector can be
added internally or used to steer the oscillator. Either way, the in
loop noise from the Silabs will be fairly well suppressed since it only
acts as a transfer oscillator.
So, suggestions, thoughts and improvements?
God Jul and Merry Christmas!
Cheers,
Magnus
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
and follow the instructions there.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Hi Dana,
On 2019-12-25 17:14, Dana Whitlow wrote:
Thanks, Bob, for clarifying what Magnus means by "microstepper". To me,
the term has always referred to operation between the discrete steps defined
by the polepieces and windings. One can do it (albeit poorly) with many
stepper
motors. There is a class of servo motor in which the designers of the
magnetics
went to great lengths to avoid salient poles (think magnetic detent-like
effects),
which are meant to operate with I & Q controlled currents.
Well, I was completely in T&F thoughts when doing my posting, and
assumed the term to be understood here for what I intended.
BTW, the "phase noise" associated with a DDS is not entirely random. In
fact
most is usually from systematic effects and can be eliminated or at least
dodged
by restricting one's choice of frequencies.
It is completely systematic. However, for a phase/frequency microstepper
you want the full range of narrow frequency to be directly available,
with high resolution. So, therefore my question to collect input from
others.
Yes, so when phase-noise is not complete crap but somewhat useful, I
would not like to completely ruin it and hence trying to ask for input.
I have an few ideas now that I want to try out.
If I where not so busy cleaning up the lab to make it accessable and
thereby useful, I would throw myself at the task of testing it. Also,
being able to locate where I have the Red Pitaya helps.
Cheers,
Magnus
fwiw, looks like the spectradynamics patent expired today:
https://patents.google.com/patent/US6278330B1/en
afaik that design (more or less) is also described in
https://doi.org/10.1109/FREQ.1998.717932
a variation is also https://doi.org/10.1109/EFTF.2012.6502322 (which might
be sold by http://www.skk.it/index.htm - but that website is a bit
spartan..)
(if anyone knows more micro-stepper papers, please post!)
my plan is to present our DIY version of this, an open hardware design, at
EFTF2020.
we have two AD9912:s clocked at 1GHz (derived from the maser) producing two
LO's that drive a mixer-board that locks an OCXO that feeds a PICDIV.
Anders
Hi,
Many thanks for those references. Downloaded and stored for future
reference and reading.
Check out US4358741 and US4417352.
Cheers,
Magnus
On 2020-01-02 18:10, Anders Wallin wrote:
fwiw, looks like the spectradynamics patent expired today:
https://patents.google.com/patent/US6278330B1/en
afaik that design (more or less) is also described in
https://doi.org/10.1109/FREQ.1998.717932
a variation is also https://doi.org/10.1109/EFTF.2012.6502322 (which might
be sold by http://www.skk.it/index.htm - but that website is a bit
spartan..)
(if anyone knows more micro-stepper papers, please post!)
my plan is to present our DIY version of this, an open hardware design, at
EFTF2020.
we have two AD9912:s clocked at 1GHz (derived from the maser) producing two
LO's that drive a mixer-board that locks an OCXO that feeds a PICDIV.
Anders
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.