EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: palindrom71 on February 10, 2024, 11:49:13 pm
-
Hi,
as I finally bought rubidium frequency reference, I started to look at my beloved scope. And it looked right back at me, nervously. And rightfully so.
I dug out that clip:
https://youtu.be/kb9P1Am9aFU?t=2080 (https://youtu.be/kb9P1Am9aFU?t=2080)
to find the clock source to be ADF3460 + 25MHz quartz
https://www.analog.com/media/en/technical-documentation/data-sheets/ADF4360-1.pdf (https://www.analog.com/media/en/technical-documentation/data-sheets/ADF4360-1.pdf)
The scope was never intended to have a reference input but it doesn't necessarily stop us.
The schematic got me thinking, what if I added a small board with some PLL/MCU that would take external 10MHz, pick up 25MHz or RF out from ADF3460 and do one of the following:
1. Pull or trim 20MHz crystal (voltage controlled capacitor?) or
2. Offset VTUNE with respect to filtered CP (see schematic / scribble)
to match multiplied reference to either of those signals?
I wonder if anyone tried this or if it's a no-go for some obvious reason I fail to see here (Adding capacitance/delay? Spurious emissions? Jitter?).
Please don't shout at me, there are many ways to tell someone that he's an idiot ;)
-
For this particular model there were some bugs in the past, in which the PLL misbehaved, and that turned into horrible jitter.
https://www.eevblog.com/forum/projects/project-yaigol-fixing-rigol-scope-design-problems/ (https://www.eevblog.com/forum/projects/project-yaigol-fixing-rigol-scope-design-problems/)
That was fixed by a firmware update, Dave made a video about the PLL bug, and a second one after the fix: https://youtu.be/K1IJH9aJvgE (https://youtu.be/K1IJH9aJvgE). There are many, many other long topics about DS1054Z, here on EEVblog, including some posts showing the measured phase noise of the PLL (e.g. https://www.eevblog.com/forum/blog/eevblog-683-rigol-ds1000z-ds2000-oscilloscope-jitter-problems/msg586046/#msg586046 (https://www.eevblog.com/forum/blog/eevblog-683-rigol-ds1000z-ds2000-oscilloscope-jitter-problems/msg586046/#msg586046)).
My point is, the PLL stability was finicky from the start, chances are any modification might make the PLL to misbehave again. However, if you succeed with the modification, that will be very useful. :-+
-
Rb units are not known for their phase noise performance. You should consider whether this will impact the performance of your scope. They would not be recommended for a spectrum analyzer that's essentially a swept radio. You can clean them up with a phase detector and amplifier that drives the EFC pin of an oscillator with low phase noise. You can make it switchable so that it is accurate and phase locked when counting and clean otherwise. Bill Riley published a circuit years ago.
-
The scope was never intended to have a reference input but it doesn't necessarily stop us.
The schematic got me thinking, what if I added a small board with some PLL/MCU that would take external 10MHz, pick up 25MHz or RF out from ADF3460 and do one of the following:
1. Pull or trim 20MHz crystal (voltage controlled capacitor?) or
2. Offset VTUNE with respect to filtered CP (see schematic / scribble)
to match multiplied reference to either of those signals?
Those will not work.
The unit has an oscillator, not a crystal.
You cannot pull the VCO DC signal, as the loop will fight against that.
The simplest is probably to change the 25MHz Oscillator, to a VCTCXO model.
As a first pass, you can simply trim that to a reference
It looks like the PLL can accept a clipped sine reference, which expands your choices.
Digikey show stocks of
O 25,0-JTP32CSV-F-K-3,3-1510-1015-LF = VCTCXO 25 MHz Clipped Sine Wave 3.3V ±280ppb -40°C ~ 85°C 3mA 3225
and also TXEAADSANF-25.000000 ±500ppb (cheaper)
Well suited to a 'trim and forget' setup.
-
What is the frequency at the ADF4360 output? It seems act like a PLL. So you can just generate new precise clock from your reference, instead of using adjusted clock from ADF4360
Oh, its output is close to 2GHz, so maybe the best way will be just to substitute frequency from oscillator with your reference 25 MHz clock.
-
I just leave the insides of a frequency measuring instrument alone and instead ascertain its inaccuracy by feeding in an accurately known signal and thereby obtain a correction factor. On the rare occasions when I need a more accurate reading than what the instrument displays I reach for a calculator and apply the correction factor.
-
Thanks for the replies. I think you successfully talked me out of it. And I'll replace quartz with OCXO for sure. Working on faulty signal generator really annoyed me, as I couldn't tell if frequency shifts, or does my scope fly away.