Two different kind of Nortel GPSTM

[tkamiya]

GPSDO discussions belong here, right?

I have a Nortel brand GPSTM, which is an Trimble OEM with 10MHz Oven clock inside.  This one is my favorite and is acting as a master clock to my lab.

Just recently, I purchased another set of GPSTM which looks identical to above but made by Symmetricom.  Oddly enough, Oven Crystal inside is 4.91520MHz.  Circuit board is entirely different.

What kind of math does it take to get it to 10MHz??  Are there any implication to all important 10MHz output purity and stability?

[TheSteve]

The 4.9152 MHz is(or was) common with CDMA mobile phone systems. In the GPSDO's I have seen that do also have a 10 MHz output it is designed to lock test gear etc to, but is likely not as clean as a GPSDO with a native 10 MHz reference.

[tkamiya]

That's what I was afraid of.  I'll have to do a comparison study on them.

[ArthurDent]

If you check the date codes on the Trimble and the Symmetricom I think you'll find the Symmetricom unit is a few years newer, maybe made around 2012. If you look at the board you will find a rectangular with a spot for a small multi-pin connector. This area is for the version of this GPSTM that has a SA.22c rubidium oscillator. I have both variations, one OCXO, and one Rb.

[tkamiya]

I was hoping someone here has both.  Thanks for replying.

I have two Trimble kind, and they are working well.
I just bought 4 Symmetricom kind, and only one came with an oscillator.  (OCXO)  Rest did not have Rb or OCXO.  But oddly enough, it does produce 10MHz output.  Looking closely, I found a VCO in output chain.  It's really  hard to see but legend on PCB says it's VCXO.  It also says 10MHz.  Just with this oscillator, I never get the GPSLOCK. 

Do you have any documentation for this version?  I have a full set of docs for Trimble kind but not the Symmetricom kind.

[tkamiya]

At what frequency is your Rb oscillating?  Square or sine?  I don't see heat sink.  Is it just bolted to the PCB?

[mark03]

What kind of math does it take to get it to 10MHz??  Are there any implication to all important 10MHz output purity and stability?

4.9152 MHz = 3 * 2^9 * 3200
10 MHz = 5^5 * 3200

so there is no practical [integer] PLL scheme that would get you 10 MHz out.  You would need to divide the 4.9152 MHz down to 3.2 kHz (!) and then multiply by 3125...

[tkamiya]

I am now very curious to run a comparison study.....

[ArthurDent]

At what frequency is your Rb oscillating?  Square or sine?  I don't see heat sink.  Is it just bolted to the PCB?

These smaller Rb units don't generate too much heat so the circuit board works with large plated areas and many via as a heatsink. The Rb output is a 9.83040 Mhz sine wave which may at first seem odd but if you divide it by 2 you will get 4.91520 MHz, the same as the OCXO in your unit.

[tkamiya]

I see....  I think SA.22.c is user programmable for frequency....

Among 3 different types you own, did you notice any performance difference?

[tkamiya]

I did a comparison study between Trimble type and Symmetricom type.  Test equipment was TICC and base clock was Cesium.  They are basically neck to neck.  I'm pleasantly surprised!  (because I have a stack of them)

[ArthurDent]

I have found if I trigger my scope from my GPS standard with the timebase set to 5ns/div then look at the Trimble on one channel and the Symmetricom on the other channel, the Trimble is rock solid but the Symmetricom is swinging back and forth rapidly within a 5ns window. Check this out.

[tkamiya]

You are absolutely right.  I fed trigger from Cesium, Channel 1 from Trimble, and Channel 2 from Symmetricom.  Symmetricom has a noticeable jitter. 

When I had it apart, I noticed 10MHz is generated from what presumably a monolithic oscillator less than 1" square.  Perhaps that is aggressively pulled by PLL.  I guess it got averaged out on Adev graph.  I got to resolve this.  When I saw it my idea was replace that oscillator with oven xtal.  (which I already have) and see if I can improve on it.

Thank YOU for bring it to my attention.