WWVB teensy BPSK early experiments
Hello to the group. Wanted to update the everyone thats interested in
what I have learned so far on the Teensy and audio codec. No complete
solution yet. Much of my experimentation and knowledge has come from Frank
and Chris, who built the complete wwvb AM time receiver. In addition
and important is Johns KD2DB BPSK receiver. There is a reason this matters.
The teensy combination is powerful and somewhat easy to use. (Has to be for
me). So over the week or so it's been getting used to the audio libraries
and how pieces are connected in software and then seeing the results. All
of the base experiments worked very quickly. Simple things like signal
generators, multipliers and filters. Things already accomplished by Chris
in the wwvb AM receiver.
But the question really is what to accomplish?
If its the wwvb bpsk timecode. Simply buy an ES100 and be done.
The interest that I have is a locked reference. Minimizing soldering and
construction. This is the point things get interesting.
A NCO can be created in Teensy but it tends to be low frequency and a
multiple of 60 KHz. Stability sort of isn't. But if it could be created
then a complete frequency reference in the teensy could be accomplished.
That makes for a heck of a low power receiver 1 watt, inexpensive, and
little soldering.
The above path literally follows the old Spectracoms and Truetime direct
conversion receivers.
Have to look at their schematics because they do lock a useful reference.
But that means something external has to come into the teensy. Get the
soldering iron hot.
The other approach is essentially Johns KD2BD receiver in software with an
external reference chain delivering 50KHz and 10 KHz to the teensy. Well
this is getting ugly now because that external chain is made up of a
classical divider 10 MHz to 50 KHz etc. But does give a very nicely locked
useful wwvb reference. Its really a hybrid because it significantly reduces
the soldering required in a true KD2BD receiver but isn't the pure in
a chip solution.
All of this is just for fun because the fact is the GPDSOs we use are
better.
If a receiver is built a natural by-product is the time message. Its just
not my focus or interest.
Much more to learn.
Next steps
Start to reuse the wwvb teensy AM receiver.
Chop out all of the display software. Its all very nice but for me at this
stage gets in the way of understanding things.
With respect to I&Q generation several suggestions have been made. But the
teensy supports multiple multipliers. Sort of thinking, use the sine
wave oscillator and add a 90 degree delay to a second path to a second
multiplier. An alternative inject the delay in the wwvb signal also. How
fine a delay is a serious question.
Much to learn and potholes to fall into.
Regards
Paul
WB8TSL
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as cheap.
Bob
On Oct 31, 2020, at 1:47 PM, paul swed paulswedb@gmail.com wrote:
Hello to the group. Wanted to update the everyone thats interested in
what I have learned so far on the Teensy and audio codec. No complete
solution yet. Much of my experimentation and knowledge has come from Frank
and Chris, who built the complete wwvb AM time receiver. In addition
and important is Johns KD2DB BPSK receiver. There is a reason this matters.
The teensy combination is powerful and somewhat easy to use. (Has to be for
me). So over the week or so it's been getting used to the audio libraries
and how pieces are connected in software and then seeing the results. All
of the base experiments worked very quickly. Simple things like signal
generators, multipliers and filters. Things already accomplished by Chris
in the wwvb AM receiver.
But the question really is what to accomplish?
If its the wwvb bpsk timecode. Simply buy an ES100 and be done.
The interest that I have is a locked reference. Minimizing soldering and
construction. This is the point things get interesting.
A NCO can be created in Teensy but it tends to be low frequency and a
multiple of 60 KHz. Stability sort of isn't. But if it could be created
then a complete frequency reference in the teensy could be accomplished.
That makes for a heck of a low power receiver 1 watt, inexpensive, and
little soldering.
The above path literally follows the old Spectracoms and Truetime direct
conversion receivers.
Have to look at their schematics because they do lock a useful reference.
But that means something external has to come into the teensy. Get the
soldering iron hot.
The other approach is essentially Johns KD2BD receiver in software with an
external reference chain delivering 50KHz and 10 KHz to the teensy. Well
this is getting ugly now because that external chain is made up of a
classical divider 10 MHz to 50 KHz etc. But does give a very nicely locked
useful wwvb reference. Its really a hybrid because it significantly reduces
the soldering required in a true KD2BD receiver but isn't the pure in
a chip solution.
All of this is just for fun because the fact is the GPDSOs we use are
better.
If a receiver is built a natural by-product is the time message. Its just
not my focus or interest.
Much more to learn.
Next steps
Start to reuse the wwvb teensy AM receiver.
Chop out all of the display software. Its all very nice but for me at this
stage gets in the way of understanding things.
With respect to I&Q generation several suggestions have been made. But the
teensy supports multiple multipliers. Sort of thinking, use the sine
wave oscillator and add a 90 degree delay to a second path to a second
multiplier. An alternative inject the delay in the wwvb signal also. How
fine a delay is a serious question.
Much to learn and potholes to fall into.
Regards
Paul
WB8TSL
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.
Just a thought, Paul --
If the teensy can generate an output that is accurate but jittery, you
could use a simple PLL to lock a crystal to that with a time constant
that smooths out the jitter.
John
On 10/31/20 1:47 PM, paul swed wrote:
Hello to the group. Wanted to update the everyone thats interested in
what I have learned so far on the Teensy and audio codec. No complete
solution yet. Much of my experimentation and knowledge has come from Frank
and Chris, who built the complete wwvb AM time receiver. In addition
and important is Johns KD2DB BPSK receiver. There is a reason this matters.
The teensy combination is powerful and somewhat easy to use. (Has to be for
me). So over the week or so it's been getting used to the audio libraries
and how pieces are connected in software and then seeing the results. All
of the base experiments worked very quickly. Simple things like signal
generators, multipliers and filters. Things already accomplished by Chris
in the wwvb AM receiver.
But the question really is what to accomplish?
If its the wwvb bpsk timecode. Simply buy an ES100 and be done.
The interest that I have is a locked reference. Minimizing soldering and
construction. This is the point things get interesting.
A NCO can be created in Teensy but it tends to be low frequency and a
multiple of 60 KHz. Stability sort of isn't. But if it could be created
then a complete frequency reference in the teensy could be accomplished.
That makes for a heck of a low power receiver 1 watt, inexpensive, and
little soldering.
The above path literally follows the old Spectracoms and Truetime direct
conversion receivers.
Have to look at their schematics because they do lock a useful reference.
But that means something external has to come into the teensy. Get the
soldering iron hot.
The other approach is essentially Johns KD2BD receiver in software with an
external reference chain delivering 50KHz and 10 KHz to the teensy. Well
this is getting ugly now because that external chain is made up of a
classical divider 10 MHz to 50 KHz etc. But does give a very nicely locked
useful wwvb reference. Its really a hybrid because it significantly reduces
the soldering required in a true KD2BD receiver but isn't the pure in
a chip solution.
All of this is just for fun because the fact is the GPDSOs we use are
better.
If a receiver is built a natural by-product is the time message. Its just
not my focus or interest.
Much more to learn.
Next steps
Start to reuse the wwvb teensy AM receiver.
Chop out all of the display software. Its all very nice but for me at this
stage gets in the way of understanding things.
With respect to I&Q generation several suggestions have been made. But the
teensy supports multiple multipliers. Sort of thinking, use the sine
wave oscillator and add a 90 degree delay to a second path to a second
multiplier. An alternative inject the delay in the wwvb signal also. How
fine a delay is a serious question.
Much to learn and potholes to fall into.
Regards
Paul
WB8TSL
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.
Good suggestions. The Teensy runs at some 580 MHz. I thought on the crystal
I had seen something like 166. These things are seriously small. But the
schematic shows its 24 MHz.
Inside the teensy is a PLL that creates the high speed system clock. So
thats potentially a good answer. Use something like either a 20 or 25 Mhz
vcTCXO as John did in the KD2BD receiver. It will be seriously tough
getting the micro 4 pin crystal out without damaging the board.
Secondary effect will occur by changing the xtal. Such as audio sample
rates. Not sure there are real world effects to really consider.
Time to pull the microscope and see.
Regards
Paul
On Sat, Oct 31, 2020 at 5:55 PM John Ackermann N8UR jra@febo.com wrote:
Just a thought, Paul --
If the teensy can generate an output that is accurate but jittery, you
could use a simple PLL to lock a crystal to that with a time constant
that smooths out the jitter.
John
On 10/31/20 1:47 PM, paul swed wrote:
Hello to the group. Wanted to update the everyone thats interested in
what I have learned so far on the Teensy and audio codec. No complete
solution yet. Much of my experimentation and knowledge has come from
Frank
and Chris, who built the complete wwvb AM time receiver. In addition
and important is Johns KD2DB BPSK receiver. There is a reason this
matters.
The teensy combination is powerful and somewhat easy to use. (Has to be
for
me). So over the week or so it's been getting used to the audio libraries
and how pieces are connected in software and then seeing the results. All
of the base experiments worked very quickly. Simple things like signal
generators, multipliers and filters. Things already accomplished by Chris
in the wwvb AM receiver.
But the question really is what to accomplish?
If its the wwvb bpsk timecode. Simply buy an ES100 and be done.
The interest that I have is a locked reference. Minimizing soldering and
construction. This is the point things get interesting.
A NCO can be created in Teensy but it tends to be low frequency and a
multiple of 60 KHz. Stability sort of isn't. But if it could be created
then a complete frequency reference in the teensy could be accomplished.
That makes for a heck of a low power receiver 1 watt, inexpensive, and
little soldering.
The above path literally follows the old Spectracoms and Truetime direct
conversion receivers.
Have to look at their schematics because they do lock a useful reference.
But that means something external has to come into the teensy. Get the
soldering iron hot.
The other approach is essentially Johns KD2BD receiver in software with
an
external reference chain delivering 50KHz and 10 KHz to the teensy. Well
this is getting ugly now because that external chain is made up of a
classical divider 10 MHz to 50 KHz etc. But does give a very nicely
locked
useful wwvb reference. Its really a hybrid because it significantly
reduces
the soldering required in a true KD2BD receiver but isn't the pure in
a chip solution.
All of this is just for fun because the fact is the GPDSOs we use are
better.
If a receiver is built a natural by-product is the time message. Its just
not my focus or interest.
Much more to learn.
Next steps
Start to reuse the wwvb teensy AM receiver.
Chop out all of the display software. Its all very nice but for me at
this
stage gets in the way of understanding things.
With respect to I&Q generation several suggestions have been made. But
the
teensy supports multiple multipliers. Sort of thinking, use the sine
wave oscillator and add a 90 degree delay to a second path to a second
multiplier. An alternative inject the delay in the wwvb signal also. How
fine a delay is a serious question.
Much to learn and potholes to fall into.
Regards
Paul
WB8TSL
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.
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB
speeds change. So, basically, the USB stops working, and you need to
set your serial port to something like 112.8 * 10/28 (and it takes a bit
of fiddling to get it to work right).. I sort of cheated, and switched
back and forth - signal generator to 28MHz, load and debug software,
start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are
the wrong length.
One could probably figure out a relatively few patches to the
Teensyduino code base that would fix all this (clock rate is a variable
- you can run the teensy at multiple clock rates, even with the same
crystal)
Hi
Looking at the data sheet for the MCU, they really do want 24 MHz and that’s about it. I suspect you would
do better to take your 10 MHz OCXO and run it into one of the frequency converter chips to get the 24.
Then feed that into the board. One more chip, but you now don’t have a bunch of stuff to hack up.
Bob
On Oct 31, 2020, at 7:17 PM, jimlux jimlux@earthlink.net wrote:
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB speeds change. So, basically, the USB stops working, and you need to set your serial port to something like 112.8 * 10/28 (and it takes a bit of fiddling to get it to work right).. I sort of cheated, and switched back and forth - signal generator to 28MHz, load and debug software, start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are the wrong length.
One could probably figure out a relatively few patches to the Teensyduino code base that would fix all this (clock rate is a variable - you can run the teensy at multiple clock rates, even with the same crystal)
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.
Jim
Thanks for the details. I took a serious look at popping the xtal out and
am afraid its a bit beyond me since there are 4 pads that need to be
heated. I have worked on very small stuff under the microscope. But this
seems problematic. I sort of thought all the bits would get upset. No free
lunch.
No matter not popping the xtal. Mainly because if anyone else did want to
build the magical solution it would be as bad as soldering lots of chips.
Super fine wires to very small pads.
But at least at the moment perhaps thats not critical to developing
something.
I did tinker with delay and will need to use a scope at this point to see
the effects.
Regards
Paul.
On Sat, Oct 31, 2020 at 7:35 PM jimlux jimlux@earthlink.net wrote:
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the
soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so
solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that
comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as
cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB
speeds change. So, basically, the USB stops working, and you need to
set your serial port to something like 112.8 * 10/28 (and it takes a bit
of fiddling to get it to work right).. I sort of cheated, and switched
back and forth - signal generator to 28MHz, load and debug software,
start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are
the wrong length.
One could probably figure out a relatively few patches to the
Teensyduino code base that would fix all this (clock rate is a variable
- you can run the teensy at multiple clock rates, even with the same
crystal)
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.
On 10/31/20 4:46 PM, Bob kb8tq wrote:
Hi
Looking at the data sheet for the MCU, they really do want 24 MHz and that’s about it. I suspect you would
do better to take your 10 MHz OCXO and run it into one of the frequency converter chips to get the 24.
Then feed that into the board. One more chip, but you now don’t have a bunch of stuff to hack up.
Yeah.. you can spin the dial on the signal generator and move the
frequency up and down, but.... Nothing is guaranteed to work right. Who
knows what sort of little DPLLs are on that chip that have narrow
ranges, etc.
This experiment was with a Teensy 3.1 - I had a lot of them, so I wasn't
afraid to hack it up.
Bob
On Oct 31, 2020, at 7:17 PM, jimlux jimlux@earthlink.net wrote:
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB speeds change. So, basically, the USB stops working, and you need to set your serial port to something like 112.8 * 10/28 (and it takes a bit of fiddling to get it to work right).. I sort of cheated, and switched back and forth - signal generator to 28MHz, load and debug software, start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are the wrong length.
One could probably figure out a relatively few patches to the Teensyduino code base that would fix all this (clock rate is a variable - you can run the teensy at multiple clock rates, even with the same crystal)
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.
On 10/31/20 5:42 PM, paul swed wrote:
Jim
Thanks for the details. I took a serious look at popping the xtal out and
am afraid its a bit beyond me since there are 4 pads that need to be
heated. I have worked on very small stuff under the microscope. But this
seems problematic. I sort of thought all the bits would get upset. No free
lunch.
It's only hard if you want to save the crystal for later use <grin>.
Soldering iron right on the top of the crystal package, all 4 pads get
soft, scrape it off with the xacto knife.
Then, solder on the tiny wires from a coax pigtail. Blob of superglue to
keep it from breaking off.
I had about 50 of them, so if I broke it, I had more to try, but it
worked ok.
But, as noted, not particularly useful.
No matter not popping the xtal. Mainly because if anyone else did want to
build the magical solution it would be as bad as soldering lots of chips.
Super fine wires to very small pads.
But at least at the moment perhaps thats not critical to developing
something.
I did tinker with delay and will need to use a scope at this point to see
the effects.
Regards
Paul.
On Sat, Oct 31, 2020 at 7:35 PM jimlux jimlux@earthlink.net wrote:
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the
soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so
solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that
comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as
cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB
speeds change. So, basically, the USB stops working, and you need to
set your serial port to something like 112.8 * 10/28 (and it takes a bit
of fiddling to get it to work right).. I sort of cheated, and switched
back and forth - signal generator to 28MHz, load and debug software,
start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are
the wrong length.
One could probably figure out a relatively few patches to the
Teensyduino code base that would fix all this (clock rate is a variable
- you can run the teensy at multiple clock rates, even with the same
crystal)
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
The gizmo you want is a hot air rework tool. They are not all that expensive
( < $200 for a fancy one, quite a bit less for a simple one). They normally
come with a bunch of tips. That lets you “focus” the heat on the part you want
to pull.
One of an infinite number, randomly chosen:
https://www.amazon.com/gp/product/B006FA481G/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1 https://www.amazon.com/gp/product/B006FA481G/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1
Bob
On Oct 31, 2020, at 8:42 PM, paul swed paulswedb@gmail.com wrote:
Jim
Thanks for the details. I took a serious look at popping the xtal out and
am afraid its a bit beyond me since there are 4 pads that need to be
heated. I have worked on very small stuff under the microscope. But this
seems problematic. I sort of thought all the bits would get upset. No free
lunch.
No matter not popping the xtal. Mainly because if anyone else did want to
build the magical solution it would be as bad as soldering lots of chips.
Super fine wires to very small pads.
But at least at the moment perhaps thats not critical to developing
something.
I did tinker with delay and will need to use a scope at this point to see
the effects.
Regards
Paul.
On Sat, Oct 31, 2020 at 7:35 PM jimlux jimlux@earthlink.net wrote:
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the
soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so
solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that
comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as
cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB
speeds change. So, basically, the USB stops working, and you need to
set your serial port to something like 112.8 * 10/28 (and it takes a bit
of fiddling to get it to work right).. I sort of cheated, and switched
back and forth - signal generator to 28MHz, load and debug software,
start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are
the wrong length.
One could probably figure out a relatively few patches to the
Teensyduino code base that would fix all this (clock rate is a variable
- you can run the teensy at multiple clock rates, even with the same
crystal)
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
On Oct 31, 2020, at 9:45 PM, jimlux jimlux@earthlink.net wrote:
On 10/31/20 4:46 PM, Bob kb8tq wrote:
Hi
Looking at the data sheet for the MCU, they really do want 24 MHz and that’s about it. I suspect you would
do better to take your 10 MHz OCXO and run it into one of the frequency converter chips to get the 24.
Then feed that into the board. One more chip, but you now don’t have a bunch of stuff to hack up.
Yeah.. you can spin the dial on the signal generator and move the frequency up and down, but.... Nothing is guaranteed to work right. Who knows what sort of little DPLLs are on that chip that have narrow ranges, etc.
This experiment was with a Teensy 3.1 - I had a lot of them, so I wasn't afraid to hack it up.
Your OCXO will have a much narrower tuning range at the extreme’s of it’s EFC than the tolerance on
the typical crystal. The MCU PLL’s will run over the OCXO tune range ….
Bob
Bob
On Oct 31, 2020, at 7:17 PM, jimlux jimlux@earthlink.net wrote:
On 10/31/20 11:42 AM, Bob kb8tq wrote:
Hi
…..errr…..
Can you pull the clock oscillator off the Teensy board? (Yes, the soldering
iron would be involved).
Will the clock input to the MCU accept something like 10 MHz? If so solder
on a cable ….
At that point whatever the Teeny does is locked to the 10 MHz. If that comes
from one of the $3 eBay OCXO’s, steer that with a DAC output … now you
have a WWVB GPSDO.
Indeed, if the Teensy needs 28 MHz, then the OCXO will not be quite as cheap.
Bob
I've tried this - It will run just fine, but all the UART and USB speeds change. So, basically, the USB stops working, and you need to set your serial port to something like 112.8 * 10/28 (and it takes a bit of fiddling to get it to work right).. I sort of cheated, and switched back and forth - signal generator to 28MHz, load and debug software, start it, then switch generator to 10 MHz.
And of course, all the functions that are time based, like delay() are the wrong length.
One could probably figure out a relatively few patches to the Teensyduino code base that would fix all this (clock rate is a variable - you can run the teensy at multiple clock rates, even with the same crystal)
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.
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.
On 10/31/20 7:26 PM, Bob kb8tq wrote:
Hi
On Oct 31, 2020, at 9:45 PM, jimlux jimlux@earthlink.net wrote:
On 10/31/20 4:46 PM, Bob kb8tq wrote:
Hi
Looking at the data sheet for the MCU, they really do want 24 MHz and that’s about it. I suspect you would
do better to take your 10 MHz OCXO and run it into one of the frequency converter chips to get the 24.
Then feed that into the board. One more chip, but you now don’t have a bunch of stuff to hack up.
Yeah.. you can spin the dial on the signal generator and move the frequency up and down, but.... Nothing is guaranteed to work right. Who knows what sort of little DPLLs are on that chip that have narrow ranges, etc.
This experiment was with a Teensy 3.1 - I had a lot of them, so I wasn't afraid to hack it up.
Your OCXO will have a much narrower tuning range at the extreme’s of it’s EFC than the tolerance on
the typical crystal. The MCU PLL’s will run over the OCXO tune range ….
Bob
I was thinking if you tried to run it at 10 MHz.. yeah, if you get a 24
MHz OCXO, no problem.
But who knows what sort of weird timing stuff might happen inside
running at less than half the speed.