Electronics > Metrology

DIY GPSDO - Divide or Multiply?

(1/3) > >>

t1d:
Re: Noob Metrology Knowledge Level – Novice/Hobbyist EE Knowledge – A nice lab for testing.

I am investigating designing a DIY GPSDOs. As with achieving any electronic design goal, there are very many solutions that might be employed. Accordingly, I will attempt to pose my final question in such a manner as to solicit a very narrow answer…

DIY GPSDOs typically source their reference frequency from a Ublox Clone. The clone captures the atomically accurate NMEA 1PPS. Thereafter, the clone, depending on its model, can output various synthesized frequencies, 10MHz being commonly desired. However, for the 10MHz frequency, the method of synthesis multiplication, or rather its signal result, is generally problematic, due to inherent mathematical constraints.

Consider the common NEO-M8N. It may be programmed to output 10M waves per second, however, due to said mathematical constraints, the waves are not of equal duration. At least two wave duration times are interleaved to place 10M waves into the one second with the greatest accuracy.

This means that nowhere within the one second period can a true 1/10M sample be found. Accordingly, the signal is complicated, if not wholly useless, to use as a reference to discipline a 10MHz OCXO.

We are left with two possible paths…
1) Divide the clone’s true 1PPS signal by 10M and compare that to the OCXO’s output, or
2) Multiply the OXCO’s output by 10M and compare that to the clone’s 1PPS signal.

And, here, we reach the overlying question. Given...
- That the goal shall be to maintain the greatest magnitude of frequency accuracy of the output reference waveform with which to adjust the OCXO,
- That DIYers work on limited budgets, and
- Factoring in the limits of technology in commonly available electronic components of today...
What do you think would be the better circuit to attempt to design? One that divides (#1, above,) or one that multiples (#2, above.)

I understand that this is a very broad consideration, deserving of a full thesis. What I am hoping for is that one of the two options is clearly the better, or worse, for some reason that I have not yet learned. If I can discover the better approach, it will save a great deal of time and effort in development.

Thank you for your interest and your help.

iMo:
NEO's output gives a "clean" frequency when the divider is an integer N= 4,5,6,7,8...
Like 48MHz/N, thus targeted 10MHz does not provide a "clean" signal.
Go with 12 or 8 or 6MHz instead.
That is so with NEO 6/7, not sure how it is with 8.

t1d:
Thanks, imo. Yes, that is the math issue, of which I spoke. Yes, its true of the NEO 8, as well. My gear uses 10MHz external references, so that is the frequency that I need.

I have thought about a combination method, as well... Take a clean 5MHz from the NEO 8 and multiple it by two. But, with my lack of knowledge with frequency circuits, I still need help to even considering that... So, I surely do appreciate your input.

Wallace Gasiewicz:
How are you going to compare the two signals?
Maybe it would be easier to bring down the 10 MHz to 10 KHZ and the GPS up to 10 KHZ. then you might have an easier time getting components to deal with the comparison.

DimitriP:
Your impression of what makes a GPSO "tick" (pun intended !)  seems to be little skewed or incomplete.
There are a lot of GPSDO designs out there,  just a google search away so starting off to "design another one" is another wheel re-invention without having a specific existing design deficiency to overcome.

Navigation

[0] Message Index

[#] Next page