have 10MHz need 19.5Mhz
Recent talk about NTP servers. It seems the limit to their accuracy is the
quality of the crystal that drives the CPU clock. Most of them make really
good thermometers. I'd like to try and replace the crystal on a Raspberry
Pi with a signal derived from a time nut quality 10MHz standard.
The Pi uses a crystal (not a TTL can, a real two lead crystal and a pair of
47pf caps) Both leads of the crystal attach to a pair of pins on an IC. I
figure I can unsolder the crystal and inject a balanced 19.5MHz signal
directly to the IC's pins. I know the ARM CPU just might work on a 20MHz
clock or maybe 15MHz but the video likely would not. I'm going to have to
supply 19.5MHz.
The question is the best way to get from 10MHz to 19.5MHz. I care only
about long term (tens of seconds to days) stability.
I thought of using an AD9850 DDS chip. You can buy these on break out
boards very cheap on eBay but they need a 125MHz clock. I could drive
the 9850 with a 120MHz clock that is multiplied up from 10MHz. what is
the simplest 12x multiplier. I assume getting to 125MHz from 10MHz is to
hard.
New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZhttp://www.ebay.com/itm/New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZ-/400422353936?pt=LH_DefaultDomain_0&hash=item5d3b083c10
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
--
Chris Albertson
Redondo Beach, California
On 02.06.2013 20:59, Chris Albertson wrote:
I thought of using an AD9850 DDS chip. You can buy these on break out
boards very cheap on eBay but they need a 125MHz clock. I could drive
the 9850 with a 120MHz clock that is multiplied up from 10MHz. what is
the simplest 12x multiplier. I assume getting to 125MHz from 10MHz is to
hard.
New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZhttp://www.ebay.com/itm/New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZ-/400422353936?pt=LH_DefaultDomain_0&hash=item5d3b083c10
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
Or use the AD9851 that is more or less the same, but includes an
internal x6 multiplier for the reference. You can find also some at the
same place:
http://www.ebay.com/itm/1pcs-DDS-Signal-Generator-Module-AD9851-0-70-MHz-2-Sine-Wave-and-2-Square-Wave-/400352450150?pt=LH_DefaultDomain_0&hash=item5d36dd9666
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
Am 02.06.2013 20:59, schrieb Chris Albertson:
,
The Pi uses a crystal (not a TTL can, a real two lead crystal and a pair of
47pf caps) Both leads of the crystal attach to a pair of pins on an IC. I
figure I can unsolder the crystal and inject a balanced 19.5MHz signal
directly to the IC's pins. I know the ARM CPU just might work on a 20MHz
clock or maybe 15MHz but the video likely would not. I'm going to have to
supply 19.5MHz.
One of the pins is probably the output of the sustaining amplifier,
the other one is the input. There would be no need for a balanced
signal. Just force-feed the input pin with 100 mV AC or so via cap.
Remove crystal and capacitors.
(I have not verified this against the data sheet, but that's the usual
way.),
80 MHz for the DDS should work. Just double 3 times.
regards, Gerhard
On Sun, 2 Jun 2013 11:59:09 -0700
Chris Albertson albertson.chris@gmail.com wrote:
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
Are there any documents on the PLL and clock network of the RPI processor?
If so, i would assume that you should be able to switch the input frequency
to 20MHz and then fix the PLL settings in the chip. I wouldnt be surprised
if the 19.5MHz happend because they were cheap and had nothing to do
with the internal frequencies.
As or a different approach: How about using an FPGA? The internal PLL/DLL
allows you to upconvert the 10MHz to anything you want, and you can build an
internal clock generator quite easily. The jitter requirements of modern uC
aren't as high as they used to be, as the clock frequency is upconverted
using an internal PLL to something different. And the PLL/VCO combo filters
out quite a bit of the jitter from the oscillator outside.
Attila Kinali
--
The people on 4chan are like brilliant psychologists
who also happen to be insane and gross.
-- unknown
On 06/02/2013 08:59 PM, Chris Albertson wrote:
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
Well, you could do a double-frequency re-generative divider.
Running it at 9.5 MHz and 19.5 MHz in synchronous mode. Should not be
too hard to achieve.
Cheers,
Magnus
On 6/2/13 11:59 AM, Chris Albertson wrote:
Recent talk about NTP servers. It seems the limit to their accuracy is the
quality of the crystal that drives the CPU clock. Most of them make really
good thermometers. I'd like to try and replace the crystal on a Raspberry
Pi with a signal derived from a time nut quality 10MHz standard.
The Pi uses a crystal (not a TTL can, a real two lead crystal and a pair of
47pf caps) Both leads of the crystal attach to a pair of pins on an IC. I
figure I can unsolder the crystal and inject a balanced 19.5MHz signal
directly to the IC's pins.
One pin is an input, the other an output most likely. The usual scheme
has an amplifier on chip to make an oscillator with the crystal between
in and out.
Drive just the input.
(if you can't find out from the data sheet, you can just try em both to
see which one works. or, heck, drive both through a suitable resistor.
The amplifier output won't care)
I know the ARM CPU just might work on a 20MHz
clock or maybe 15MHz but the video likely would not. I'm going to have to
supply 19.5MHz.
The question is the best way to get from 10MHz to 19.5MHz. I care only
about long term (tens of seconds to days) stability.
DDS eval board is your friend. Any of the analog devices parts would
probably work. the older 985x ones are cheap, but draw a lot of power.
Hi
The Pi is simply a standard ARM CPU. The clock system in it is well documented. How you get at the clock system may or may not be so well documented. There are bits you flip to switch it from external crystal to external oscillator. The gain of the stage and the function of the output pin are impacted by the bits. By far your best option would be to figure out how to flip the appropriate bit. Baring that, ignore the output pin and just drive the input pin. Trying to put a differential signal in there will just mess things up.
If you don't need video you might be able to change the clock frequency. The other things that would be impacted are the USB (needs 48 MHz from somewhere) and the serial baud rates. Since the ethernet hangs of of a USB port it's clock would not be impacted.
Bob
On Jun 2, 2013, at 2:59 PM, Chris Albertson albertson.chris@gmail.com wrote:
Recent talk about NTP servers. It seems the limit to their accuracy is the
quality of the crystal that drives the CPU clock. Most of them make really
good thermometers. I'd like to try and replace the crystal on a Raspberry
Pi with a signal derived from a time nut quality 10MHz standard.
The Pi uses a crystal (not a TTL can, a real two lead crystal and a pair of
47pf caps) Both leads of the crystal attach to a pair of pins on an IC. I
figure I can unsolder the crystal and inject a balanced 19.5MHz signal
directly to the IC's pins. I know the ARM CPU just might work on a 20MHz
clock or maybe 15MHz but the video likely would not. I'm going to have to
supply 19.5MHz.
The question is the best way to get from 10MHz to 19.5MHz. I care only
about long term (tens of seconds to days) stability.
I thought of using an AD9850 DDS chip. You can buy these on break out
boards very cheap on eBay but they need a 125MHz clock. I could drive
the 9850 with a 120MHz clock that is multiplied up from 10MHz. what is
the simplest 12x multiplier. I assume getting to 125MHz from 10MHz is to
hard.
New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZhttp://www.ebay.com/itm/New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZ-/400422353936?pt=LH_DefaultDomain_0&hash=item5d3b083c10
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
--
Chris Albertson
Redondo Beach, California
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 Sun, 2 Jun 2013 11:59:09 -0700
Chris Albertson albertson.chris@gmail.com wrote:
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
Another stupid idea: You have a 19.5MHz crystal, probably some
odd PLL and a varicap diode lying around: Build a VCXO out of the
crystal and lock it to the 10MHz using the PLL.
Attila Kinali
--
The people on 4chan are like brilliant psychologists
who also happen to be insane and gross.
-- unknown
Hi
A SI5335 would appear to be a more straightforward way to get the 19.5 MHz. There are many other clock multiplier IC's out there. You don't need anything very fancy, just a divide by two on the input, a multiply by 39, and then a divide by 10. If the internal VCO will go to 400 MHz, you don't need the divide by two. Just do a X39 and a divide by 20.
If you are doing some other logic "stuff" this sort of conversion is drop dead easy for the PLL's in most cheap FPGA's. Cyclone III(?), IV, and V will all do it no sweat. I think the switch from 20 MHz pll to 10 MHz was when they went from the II to the III.
Bob
On Jun 2, 2013, at 2:59 PM, Chris Albertson albertson.chris@gmail.com wrote:
Recent talk about NTP servers. It seems the limit to their accuracy is the
quality of the crystal that drives the CPU clock. Most of them make really
good thermometers. I'd like to try and replace the crystal on a Raspberry
Pi with a signal derived from a time nut quality 10MHz standard.
The Pi uses a crystal (not a TTL can, a real two lead crystal and a pair of
47pf caps) Both leads of the crystal attach to a pair of pins on an IC. I
figure I can unsolder the crystal and inject a balanced 19.5MHz signal
directly to the IC's pins. I know the ARM CPU just might work on a 20MHz
clock or maybe 15MHz but the video likely would not. I'm going to have to
supply 19.5MHz.
The question is the best way to get from 10MHz to 19.5MHz. I care only
about long term (tens of seconds to days) stability.
I thought of using an AD9850 DDS chip. You can buy these on break out
boards very cheap on eBay but they need a 125MHz clock. I could drive
the 9850 with a 120MHz clock that is multiplied up from 10MHz. what is
the simplest 12x multiplier. I assume getting to 125MHz from 10MHz is to
hard.
New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZhttp://www.ebay.com/itm/New-AD9850-module-modest-capacity-AD9851-DDS-Function-Generator-up-to-40MHZ-/400422353936?pt=LH_DefaultDomain_0&hash=item5d3b083c10
Is there a smarter and more direct way to get 19.5MHz for 10MHz?
--
Chris Albertson
Redondo Beach, California
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 Chris,
The question is the best way to get from 10MHz to 19.5MHz.
Must it be the RasPi or can it be another cheap Linux device?
There are some out there which have a frequency which is simpler to reach than
19.5 MHz.
Have a look e.g. at the cheap low-power TL-WR703N, they use 25MHz:
https://forum.openwrt.org/viewtopic.php?id=39829
They don't have as much CPU as a RasPi, but should be enough for a local NTP.
On ebay they cost a bit over 20 EUR delivered from China.
Kind regards,
Gerd
On Mon, Jun 3, 2013 at 4:30 AM, Gerd v. Egidy lists@egidy.de wrote:
Hi Chris,
The question is the best way to get from 10MHz to 19.5MHz.
Must it be the RasPi or can it be another cheap Linux device?
There are some out there which have a frequency which is simpler to reach than
19.5 MHz.
Have a look e.g. at the cheap low-power TL-WR703N, they use 25MHz:
https://forum.openwrt.org/viewtopic.php?id=39829
They don't have as much CPU as a RasPi, but should be enough for a local NTP.
On ebay they cost a bit over 20 EUR delivered from China.
Kind regards,
Gerd
Hi,
I've been using WR703N/MR3020 and WR841N routers as NTP servers with
NMEA and PPS. The PPS drivers are a bit of a hack since I don't know how
to enable GPIO interrupts.
Looks like the CDCE913 is the simplest one chip solution. The frequency in
the subject line is in error. What is needed is 19.2MHz.
So inside the CDCE913 we divide by 25 then multiply by 23. Or said
differenty 9.2MHz = (23/25) 10MHz.
The CDCE193 is a 14 pin chip that will do the above. It has non volatile
EPROM so you can program 23/25 ratio into it once and you don't need a uP
in the final circuit. But you do need a 1.8 volt power supply.
This chip appears to be generally useful. It can synthesize just about
anything and phase lock it from a 10MHz reference.
The price is right, TI offer samples.
On Sun, Jun 2, 2013 at 11:59 AM, Chris Albertson
albertson.chris@gmail.comwrote:
Recent talk about NTP servers. It seems the limit to their accuracy is
the quality of the crystal that drives the CPU clock. Most of them make
really good thermometers. I'd like to try and replace the crystal on a
Raspberry Pi with a signal derived from a time nut quality 10MHz standard.
Chris Albertson
Redondo Beach, California
--
Chris Albertson
Redondo Beach, California
On 06/02/2013 10:21 PM, Attila Kinali wrote:
Another stupid idea: You have a 19.5MHz crystal, probably some
odd PLL and a varicap diode lying around: Build a VCXO out of the
crystal and lock it to the 10MHz using the PLL.
It has been done with an xor gate, a couple of passives, and the varicap
diode. There processor performs as the divider:
This bit of code will configure the gpclk0 block to output 10MHz on pin
7 of the connector:
The signal on the pin is the 19.2MHz crystal multiplied to 500MHz (the
GPU PLL), and then divided back down to 10MHz.
Feed that signal, and the external reference signal into an xor gate or
other phase detector of choice. Put an RC filter of around 50us on the
output, and connect that to the varicap.
Picture of the thing as built:
The only thing on the other side of the board is the connector and the
74HC86 in DIP. I don't know what my varicap is, it came out of an old
gps receiver.
/Kasper Pedersen