Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
GPSDO ubitx ?
|
On Mon, Nov 16, 2020 at 05:27 AM, Rafael Diniz wrote:
Did anybody managed to add a GPS to external referente the SI 5351?Might be a round about way to do it. Depending on available pins. In the time that the Nano processor is not doing much, let the Nano read the 1 PPS pulses from a GPS to get a counter value that would represent how far off the 16mhz CPU clock is.? Then, once and a while, let the SI5351 generate a couple of mhz signal (extra unused SI5351 output) that is fed into an input of the Nano. Count that frequency, also. Then calculate a correction factor for the SI5351 values needed to program it to the correct frequency. Just a rough idea. Haven't kept track if any extra pins are still available on the Nano recently. Tom, wb6b |
|
If you are sticking with the Si5351 family for this and are ok with laying out a new board,
I'd use the $2 Si5351C in the QFN package.? It has a refclk input that can work with directly with the 10mhz reference clock often available from GPS modules. ? ? The $1 Si5351A in the MSOP10 that we normally use can take a reference clock in on one of the crystal pins, but spec sheet says it's only good for 25 to 27 mhz. Reports are that it does work with a reference clock of 10mhz, but with significantly more jitter. Jerry, KE7ER ? |
|
Thanks Tom, that is a good idea!
toggle quoted message
Show quoted text
Rafael PU2UIT On 11/16/20 5:26 PM, Tom, wb6b wrote:
On Mon, Nov 16, 2020 at 05:27 AM, Rafael Diniz wrote: |
|
Hi Jerry,
toggle quoted message
Show quoted text
That makes total sense. I have some affordable GPSDO with 10MHz output I got from ebay. I feel your suggestion might be the best solution, of course, at the cost of re-doing the Raduino and adding (may be in the same Raduino board?) the GPSDO with 10MHz output already connected to the refclk pin of the Si5351C, plus a new SMA connector for the GPS antenna. Thanks! Rafael PU2UIT On 11/16/20 6:19 PM, Jerry Gaffke via groups.io wrote:
If you are sticking with the Si5351 family for this and are ok with |
|
Here's an interesting note regarding a 10mhz reference output?
toggle quoted message
Show quoted text
from a cheap NEO-7M GPS module:??? /g/HBTE/topic/69599658#699 ##### It is worth noting that the module only produces a "nice" output at divisions of it's internal reference (48MHz). At any other The output is erratic. A frequency counter (which only counts edges) will show the correct frequency (e.g 10MHz) but a spectrum analyser will show the truth, and an oscilloscope will exhibit multiply triggering. ##### That discussion continues in a couple other threads: /g/HBTE/topic/69627276 /g/HBTE/topic/69910563 Programming the NEO-7M for 24mhz and 12mhz is done with an even divide by 4, and so creates a clean clock. Other divide ratios use a dual modulus divider, and thus the clock has lots of jitter. Another issue, that clock apparently comes from an internal crystal oscillator, it is occasionally disturbed by corrections to keep the edge counts agreeing with what is coming in from the GPS satellite. So edge counts over a second or so are very accurate, but as an RF source the NEO-7M is rather noisy. That may or may not be suitable for use as an Si5351 reference, depends on what SiLabs did with regard to the loop filter on their VCO. The Si5351A in the MSOP10 would probably work fine with a 24mhz reference clock into one of the crystal pins.? The datasheet says 25 to 27mhz, but 24mhz is pretty close. And it actually does work with a reference of 10mhz, but the Si5351 outputs have more jitter. The Si5351C does have a dedicated CLKIN pin that is spec'd for a much wider frequency range. There are better GPS modules than the very cheap NEO-7M that do give a clean reference clock. Jerry, KE7ER uple other threads of interest (a continuation of the discussion On Mon, Nov 16, 2020 at 06:01 PM, Rafael Diniz wrote: That makes total sense. I have some affordable GPSDO with 10MHz output I |
|
So it is worth trying to connect the NEO-7M GPS module (48MHz divided by
toggle quoted message
Show quoted text
4) to the Si5351A and see what happens, as it is already there in the Raduino. Hopefully it will be good enough to provide a precise and jitter-free frequency synthesis! Rafael On 11/17/20 12:07 AM, Jerry Gaffke via groups.io wrote:
Here's an interesting note regarding a 10mhz reference output? |
|
Worth trying, but the NEO-7M is very cheap and does not give a clean square wave.
toggle quoted message
Show quoted text
Any corrections in edge count to better agree with the satellite will disturb that square wave. In the discussion I pointed to, it was suggested the NEO-7M was mostly clean at 24mhz. However, since it is then dividing by two from 48mhz, it can only slow down, by adding an extra 48mhz clock in there.? It cannot speed up, if that's what the satellite wants it to do. Best bet is to spend a little bit more money on a better GPS module that has a clean reference. Jerry On Mon, Nov 16, 2020 at 07:14 PM, Rafael Diniz wrote: So it is worth trying to connect the NEO-7M GPS module (48MHz divided by |
|
I just bought one of the VFO/SigGen kits from QRP-Labs and they already have a GPS disciplined option for it if you want to try it out. It is only 1PPS but apparently does the job. I haven't built my GPS kit yet so I can't give any personal details atm. You have to provide the 5v supply on your own so that is some extra work.
Just a thought in case you hadn't seen that project yet. |
|
Very interesting Jason! Tks!
toggle quoted message
Show quoted text
Rafael On 11/17/20 10:47 AM, Jason Pirok wrote:
I just bought one of the VFO/SigGen kits from QRP-Labs and they |
|
I tried something a year ago or so. I was attempting to reduce the phase noise on Si5351 using an external oscillator instead of the internal oscillator. So, this is how it worked.? 1. You take a 10 MHz TCXO, (subsitute this with a GPS source for a GPSDO) and divide it by 2 in a 74HC74. The output is rich in harmonics at 5 MHz. 2. You feed the 5 MHz square wave into a triple tuned bandpass filter at 25 mhz to isolate the 5th harmonic. This will be weak, add an amplifier of 16 db to bring it up. 3. Feed this to the xin input of the Si5351. Now, your Si5351 is GPS disciplined. On Thu 19 Nov, 2020, 7:35 AM Rafael Diniz, <rafael@...> wrote: Very interesting Jason! Tks! |
|
Dear Ashhar,
toggle quoted message
Show quoted text
Thanks a lot for the detailed solution. The two ubitx are arriving soon home, I'll try to replicate here, I have a GPSDO with a 10MHz output. Rafael On 11/19/20 12:48 AM, Ashhar Farhan wrote:
I tried something a year ago or so. I was attempting to reduce the |
|
I also forgot to mention the project from Scullcom.uk for a 10Mhz reference oscillator. Granted it is designed to be more of a bench reference, but I'm sure you could accommodate for that! His project is based on the Neo 7m for reasons he describes in the video. I've not seen many of those around so if you wanted to do this project you may want to look for the Neo 8M/N or other newer variants that have the 10Mhz capability. ? http://www.scullcom.uk/category/projects/frequency-reference/ |
|
Hi Ashhar,
toggle quoted message
Show quoted text
Is there a way to know if the Si5351 correct got the 25 Mhz input? I still did not do the 10 MHz to 25 MHz circuit (I bought a GPSDO with 10Mhz output from ebay). Any way I can buy a 10 MHz to 25 MHz transverter already assembled? Regards, Rafael On 11/19/20 12:48 AM, Ashhar Farhan wrote:
I tried something a year ago or so. I was attempting to reduce the |
|
You will have to brew it. Program the gps module to output 10 mhz, feed it through a 10 mhz filter and divide it by two using a flip flop, now you get a stable 5 mhz signak, square wave. Filter out the fifth harmonic and use that as the input to the si5351.? It sounds complicated but it can be a single evening project On Thu 7 Jan, 2021, 7:38 PM Rafael Diniz, <rafael@...> wrote: Hi Ashhar, |
|
Thanks Farhan, I'll buy the parts and give a try. As a computer
toggle quoted message
Show quoted text
scientist, my soldering and electronic skills are a bit limited. Rafael On 1/7/21 5:37 PM, Ashhar Farhan wrote:
You will have to brew it. Program the gps module to output 10 mhz, |
|
Through experimentation I've determined that the QRP Labs VFO with a 1pps GPS can be used to provide a reasonably stable and accurate crystal substitute for the uBitx raduino (many thanks to everyone's suggestions on this and other threads). The impetus was to see if I could use the two units together to get decent results for the hamsci.org Eclipse Festival happening today.
?
The 25 MHz crystal is unsoldered from the raduino (using good flux and braid to remove the solder on the two pins), then two ground pads scraped on the backside ground plane to reveal the copper, and a SMA connector soldered.
?
Before soldering the SMA connector to the board, I created an assembly of the connector with a 47 ohm resistor load input (parallel), followed by a 150 pF cap in series for DC blocking. The output of the cap wire goes through the former crystal hole to the si5351 pin 2 (XA) and excess wire cut (and almost forgot to solder it).
?
Experiments with an additional PI (C-L-C) LPF right before the 47 ohm resistor load worked (to reduce harmonics), but proved unnecessary, so I left it out.
?
The QRP Labs VFO is configured to be GPS disciplined (in my case an old Fastrax 1pps GPS).
?
It is then configured to generate 25 MHz out which is also set into the first preset so it runs 25 MHz out on power up.
?
I'm using the CEC firmware with Nextion on my v5 uBitx. I set uBitx calibration to 0 because the VFO is generating 25 MHz exactly (within a couple of hertz). This was confirmed by setting the uBitx VFO to 9.999 Mhz and monitoring the carrier of WWV as a 1kHZ audio output tone and the SpectrumLab software FFT measured it to within a couple of hertz of 1000 Hz. One thing I didn't do is change the BFO but I don't think that is necessary (might have to check that at some point anyway).?
?
The documentation for the VFO recommends a 50 Hz delta (correction threshold) before it resyncs to the GPS time to avoid audible glitches, however, I'm finding that the minimum of 1 Hz seems ideal for my application where I'm not actually listening to the audio but I need the highest possible frequency accuracy due to crystal temperature drifts (not unusual for the 27 MHz crystal to change 20 Hz over the course of a day indoors in the spring.
?
So far the graphs and results look good for such a simple time standard (the Fastrax accuracy is only rated 50 ns RMS in the 1pps output). I do see a 1 or so Hz frequency jumps occasionally when the QRP Labs VFO recalibrates its output 25 Mhz. On the VFO 16x2 LCD I display the output frequency, the 1pps heartbeat, the lock (No/2D/3D), the number of satellites, and the reference frequency (somewhere around 27.005 MHz, as the QRP Labs VFO uses a 27 MHz crystal intended for WSPR applications).
?
I've run into two issues:
1. The power up sequence is manual. I first turn on the VFO and then the uBitx to ensure the raduino's si5351 has oscillator present when it powers up.
2. If the GPS signal deteriorates for whatever reason and the unit goes in and out of lock, the heartbeat "skips" a beat or two and under just the right circumstances VFO's pps detector gets confused (possibly due to a bug or implementation limitation) which then causes the VFO to do a severe recalibration throwing the output frequency off hundreds of hertz until it gets a steady lock again. Must get a better GPS.
?
I still need to build a shielded box for the VFO - I find that the GPS loses signal when the VFO is outputting 25 MHz but is fine when it outputs 5 MHz or less (thereabouts). This is a problem because I have the GPS module next to a window so it only sees half the sky. My temporary solution is to connect the GPS with a 1 meter cable to move it away from the VFO and put a 20 cm x 40 cm ground plane about 5 cm under it (these numbers are arbitrary but seems to increase the number of satellites seen by 1 or 2)
?
Hope this is useful info for those looking for a simple hardware solution.
Roman |
|
Fascinating... We must build this as a module for others. This stuff is the holy grail for the weak signal buffs. On Thu 10 Jun, 2021, 10:49 AM Roman, <rmach1000@...> wrote:
|
to navigate to use esc to dismiss