A look at my Symmetricom GPSDO / 10MHz reference (OCXO + Furuno receiver)

[wkb]

hi there,

Last week I received a Symmetricom UCCM module, sans OCXO. Carefully inspected the module, it does not appear to have any damaged/lost components.  Just had to remount C1, close to the PPS connector.

I have tried to marry it with 3 different OCXO types:
- Murata-Erie OC2501DT
- Bliley NV47A1282
- Trimble 34310T

All powered according to their specs, so 12V or 5V. The Trimble is now connected, and powered directly from a 12V DC supply. The 5-12V boost converter does not seem to like the Trimble, too much whining and drawing 2A from the 5V supply. Anyway, this should work. I repurposed a Sarantel helix antenna which I took out of a handheld GPS years back, adding a 5V-3V LDO to it. The antenna sits now on the side of the roof, just outside the window of my shack, which roughly faces South. Position hold mode is what we need, so that all appears OK to me.

OK..  Sofar the stage.

Now the question: looking at the attached Heather screenshot, I wonder out the sawtooth purple PPS trace. On the command line the Status of the UCCM shows me "Assessing Ref".  I have been powered on now for say 18 hours or so.

Is this the UCCM trying to figure out if the GPS derived PPS is stable enough for meaningful reference to the PLL? The green line in the graph is not moving in the slightest.

When I had the Bliley OCXO connected the green line was more like a block pulse (not good I think). See the 2nd screenshot.

Do note this is new territory for me, so please be gentle 

Wilko

[thinkfat]

Neither of the two appear to be working.

[bingo600]

DAC seems to be maxed out , and PPS looks weird , Elevation is a bit low (5)

Heater says Oper Mode initilizing ... Not good if thats the permanent state.

/Bingo

[ZigmundRat]

It looks like in both cases it looks like the DAC is railed (+/- 99 or 100%). Check the EFC voltage on the OCXO to see if this is true. I don’t know if the UCCM expects positive (an increase in EFC voltage results in an increase in frequency) or negative slope, but the OCXO needs to match of course and the range (or sensitivity) needs to be considered as well.

[PA0PBZ]

If you can't get it working I can send you an original UCCM OCXO. Just promise to send it back if you still can't get it working for some reason.
If interested PM me your address.

[wkb]

After a nice walk I performed some measurements. Mind you, this is a homegrown counter, albeit with a OCXO reference that was checked against a Rb.

Trimble:
- Vref = 5.91V
- fmax = 10.000 000 @ EFC= Vref    <---- that I do not like
- fmin =  9.999 993 @ EFC= 0V

Bliley:
- Vref = 3,97V
- fmax = 10.000 009 @ EFC = Vref
- fmin = 9.999 994 @ EFC = 0V

So both are positive slope.

The Bliley numbers align with the datasheet (see attached). I am not overly fond of the Trimble maxing out at 10.000 000 ?

So... maybe the original Symmetricom OCXO is negative slope EFC?

Wilko

[wkb]

That would be nice to verify that my module itself is working properly! 

Wilko
de PA1WBU

[PA0PBZ]

Well, that sucks    I hooked up the spare UCCM OCXO to test the tuning and the output frequency doesn't change whether I connect the EFC to Vref or ground.
Sorry Wilko

[wkb]

So.. appears you have a duff spare OCXO.  That is indeed 

In the meantime I got a bit ambitious: wanted to find out why the Trimble did not go > 10,000 000. 
So I hooked up an adjustable lab PSU to its EFC input. With (say) 6.8V at the EFC I get 10,000 001
(stupid freq counter is 1Hz resolution; that never bothered me but now it is frustrating).

So.. looks like the Trimble has either a duff Vref output (too low), or the oscillator itself needs a higher
EFC voltage (aging??) than the Vref output supplies to make it go > 10,000 000  <sigh> 

Wilko

[wkb]

Would still be nice to confirm that the original Symmetricom OCXO is a negative slope EFC.

Wilko

[PA0PBZ]

Just took the OCXO out of a working UCCM:

EFC @ 0V  = Fmin
EFC @ Vref = Fmax

Vref = 5.132V
Delta F about 8Hz
Current during warmup 0.48A

Tested the broken one again: still broken 

So it looks you're good with the other OCXO's

[wkb]

Thank you!  [time to Dremel-open the broken original  ]

So the Bliley has a wider delta-f than the original OCXO, ~15 Hz versus ~8 Hz. It does that for a delta V of ~4V
The original OCXO does ~8 Hz for delta V of ~5V

So the Bliley responds more "aggresively" to a delta V on EFC.

I guess that EFC for a normal healthy unit is somewhere in the middle of the 0-Vref range?

hmm

Wilko

[ZigmundRat]

@MKatana reported back in post #1090 the same Bliley OCXO worked for him, at least on a ‘Samsung’ unit. Assuming your Symmetricom unit is compatible in the same way, we’re back to making sure the EFC DAC info is actually being reported correctly. This can be checked fairly easily.

Keep in mind the unit does not measure the DAC output directly, it just sees the timing offset and steps the DAC accordingly. So if the output frequency doesn’t actually change it will keep stepping and wind up at the DAC limit. This can also happen if the timing isn’t measured correctly and it thinks it needs to adjust the frequency. I’ve had GPSDO with both failure modes.

Another possibility is that due to the EFC sensitivity difference, the loop parameters for the original OCXO cause an oscillation when applied to the Bliley. You’ll probably be able to see this in the Lady Heather DAC graphs if you watch it from startup (although it might depend on the DAC value at startup).

[wkb]

[ Specified attachment is not available ]So, with the Bliley connected the EFC voltage appears to
be in line with what LH reports.

EFC goes from GND to Vref and viceversa. So the DAC works.

Next step is to see what the output freq does in response to that..

Wilko

[PA0PBZ]

Pay attention to this, it looks like they couple the 10Mhz output via an easy to damage L20:
https://www.eevblog.com/forum/projects/a-look-at-my-symmetricom-gpsdo-(ocxo-furuno-receiver)/msg967071/#msg967071

[wkb]

Well, indeed, my thinking exactly. So, I have been studying L20 under a stereomicroscope. It looks fine to me. And there is 10MHz on "the other side" of L20 as well.

Using the Trimble instead of the Bliley produces different behaviour which increases my belief that the 10MHz signal has a closed feedback loop to the control circuit.

Does that make sense?

Wilko

[ArthurDent]

I’ve used a Trimble Thunderbolt with the internal OXCO removed and the leads brought out to test or adjust an external oscillator for a few years now. Here is an 8 hour test/adjustment on a Rb LPRO-101 with the EFC provided from a stable adjustable power supply. You can see how I got the Rb oscillator to be within 10nS over 2 hours.

I normally set the EFC voltage to a value I want then adjust the oscillator coarse adjustment to give me the 10.00000Mhz output to make sure I have plenty of range on both sides of 10Mhz. A BG7TBL FA-2 counter is a cheap way to check it out.

[uski]

Hi,

(op here)
Cool to see this thread had so many replies !

I left the GPSDO featured in the beginning of this thread in my previous workplace. I miss it... so I am going to build myself a new one !
I already ordered a more recent GPSDO from eBay (with a LEA-6T GPS receiver from uBlox), a STP2878LF.

Can anyone please give me the dimensions of the PCB ? They all look the same and I am trying to order an aluminum enclosure ahead of time.

Thanks

[thinkfat]

70mm by 115mm should be close enough, I believe.

[ve7xen]

Well, indeed, my thinking exactly. So, I have been studying L20 under a stereomicroscope. It looks fine to me. And there is 10MHz on "the other side" of L20 as well.

Using the Trimble instead of the Bliley produces different behaviour which increases my belief that the 10MHz signal has a closed feedback loop to the control circuit.

Does that make sense?

Wilko

Perhaps this is obvious, but I've also fiddled a bit with the Samsung UCCM board and different OCXOs. While the board has multiple OCXO footprints, they're not shorted together. The EFC and RF pins are somehow selected, either by the software or at assembly time. I wasn't able to reverse engineer this, but shorting them with bodge wires to the 'expected' pins did seem to work for me with several OCXO variants. Without doing this the symptoms were very similar to what you're describing when trying to use an OCXO with a different pinout than what shipped with the module.

[PA0PBZ]

Received the UCCM pcb from wkb and placed an original OCXO on it. Now the wait begins but it looks good so far.

[wkb]

Sofar so good..

Wilko

[PA0PBZ]

So I let the thing run until Monday morning and just when I wanted to get a screenshot there was a power failure.  This only happens once in 5 to 10 years here so I guess it was my lucky day. Powered the stuff up again and just took the screenshot. Please note that a) this is an indoor antenna because the real antenna is occupied and b) I have no idea what happened to the date but I decided to ignore it instead of doing a 3rd run.
So the thing seems to work fine with an original OCXO, what's next?

[wkb]

 

So, we now know that he PCB is working like it should with an original OCXO.

So I can now either find myself an original OCXO, or try to figure out how to get one of my collection of as of yet not working OCXO models working.  In the meantime I have found a 3rd model OCXO in my shack. Yet more options to test I suppose.

The original OCXO is a positive slope EFC, so higher EFC voltage -> higher frequency?

Wilko

[wkb]

What Vref and EFC range does the original OCXO use btw?

Wilko