GPSDO project

Status
Not open for further replies.
I built the logic and pll section on a piece of perfboard this evening with the ocxo on board.
I double checked the ref voltage from the ocxo, it's actually 3v.
So I'm not sure of the best way to proceed. I want the pll to run on 3v to avoid risk of overvoltage on the oscillator's control pin. I don't want to run the rest of it on 3v because I want 5v outputs, so I need to either add a pd between the hc164 and hc4046 and make the lock capacitor a bit smaller, or put a series resistor and 3v zener across the output. Not sure which to actually do. Going for the pd at the moment.
 
I take it you dont have the datasheet for the ocxo then.
If you at least know the adjustment range you could lash up a circuit with a pot and a 10k resistor inline with the control signal to the ocxo, adjust the voltage till you get to the max adjustment range on the ocxo, if its 5v then thats what it'll take.
If it was 3.3v logic then you'd have 3.3v not 3v.
Theres probably just a bit of a low pass filter and a varicap diode on the control pin, if they bothered with an op amp I'd expect that to be powered from the main rail not 3v.
 
Unbe-effing-lievable.
I powered it on, OCXO now dead. Still heating up, just no output. Same when out of circuit.
So I don't know what has caused this. Power is right way round! Nothing now on Vref and the 0.5v that was showing on the control pin has disappeared.
This is getting expensive way to fail!
I'm going to open it up and see what's going on.
 
Last edited:
Hmmmm
Got the top off with a heat gun. Didn't take long to find the culprit - see photo for where pcb is burned by a transistor. Amazingly when I took it out and tested it, it is still ok. I think it had melted it's connections to the point where it could come disconnected when I knocked the oscillator earlier.
Put it back, oscillator working ok!
Tranny got very hot very quickly when re-connected. I turned the supply voltage down, it cut off a bit above 3v.
So I wonder if this is actually a 5v version of the oscillator? It seems to run ok at 5v!
 

Attachments

  • inside.jpg
    119 KB · Views: 141
  • roasted.jpg
    148.6 KB · Views: 149
Looks like I should have paid attention to the other part number on it: tpn 57964-05. I think the -05 means it's 5v. Fnorking eBay seller listed it as 12v. Grrrrr.... Hope no other damage is done...
 
You did well to fix that.
A lot of ebayers dont have a clue what they are selling, you'd be quite right to get back to them and demand another.
RS components do a Tcxo cheap, less than 4 quid, not quite as good though.
 
In retrospect, a TCXO would have been just fine. It just happens the very first article I read on the subject used an OCXO, so I ran with that... Anyway, it's a /nice thing/!
Not sure I can be bothered getting back to the eBay seller...
 
Tcxo's are cheap, not quite as good though, your circuit will outperform a Tcxo.
My confidence in ebay has gone up a little, I sent someone some cash by mistake, they didnt seem keen to give it back, but ebay did!
I got divers Gps frequency standard to work, last report via its serial port was saying its within 5nS, pretty close.

 
Power, GPS lock & holdover Leds are controlled by the gizmo's inside the unit, the enabled, active & alarm led's are just pilot's controlled via the serial port through on/off commands, I think they are provided to help comms engineers, pulling one of these out thats controlling a large mobile tower probably wouldnt do your boss relationship much good.
Holdover is an algorithm the unit has, it learns the control profile for the oven, ie records the control voltage over 24 hours, then if the Gps signal fails it reverts to holdover where basically it plays back the control voltage over 24 hours, maintaining an accurate timebase without a Gps signal, they claim its very accurate, and I think that explains a lot of the complexity of the device, not something I need but I spose you dont want a mobile tower shutting down because its really cloudy or something.
The oven in this thing is huge.
 
Interesting comparing the HP GPS with my home brew standards.
Assuming the Hp is bang on.
My BBC radio 4 standard is 3 milli Hz fast, which is acceptable.
My Msf standard has a wobble, it cycles up & down by 2 cycles at about once in 90 seconds.
So it looks like possibly the mains is getting onto the tuning voltage line and the low pass in the ocxo is converting that to a 7th 3rd harmonic of the mains freq, the tuning voltage signal wire runs close to the mains power.
So I think I'll unscrew the ocxo and move it well away from the power in and redo the test, I think mains pickup is the most likely cause as noise or atmospherics wouldnt cause a frequency change which shifts up & down equally and with the same period each time.
Of course another cause could be too much deltaF/deltaV in the control loop making the control loop oscillate.
 
50 divided by 3 a total of 7 times.
Dunno why I said it like that must have been late.
If I move the ocxo away from the power line it reduces a lot.
I'm not clever enough to twig that myself, I remember a project on the net where someone had a similar thing.
I might have been better using a torroid trans, there wasnt any in my salvage box at the time.
 
Ah. 3^7. I had to use a calculator before I realised it's not just 3*7. 7th 3rd harmonic just looks so much better than 2187th harmonic. Maybe it's the proper notation too (speculates wildly)
I'm surprised you didn't just keep the mains wiring out of the way of signal paths anyhow
I've still got a week's work to go before I can work on mine (then I got 4 days off, I might even get one of those to myself, yay!).
What actually happened when I powered it on (apart from the oscillator not working) was the lock indicator came on. I don't see why this happened since it had a pulse generator connected to the PPS input, in the absence of a working GPS module. Maybe I made a wiring boo-boo.
 
Well you know when you put stuff togther it doesnt always end up 'ideal', some slight moving things & screened cable might work.

If you had the ocxo disconnected maybe the floating input capacitively connected itself to another input making it seem like a lock, that or like you say you messed up.
 
Ocxo was connected but dead. So it had signal on one input, no signal on the other, but not o/c either. I'll have to see if it's different when it has the proper signals.
 
I bought a universal counter timer off flea Bay! It's a Marconi 2438 in indeterminate condition. It's mounted in half of a 19" rack case, I'm just hoping it's got it's original casing on under there. So we'll see when it arrives anyway. Fairly ancient, I found a manual dated 1979!
I'd been hoping for a Racal 199x, in fact just missed one going for peanuts a couple of weeks ago - sickening. But the Marconi is more than adequate to be going on with anyway.
 
Status
Not open for further replies.

Similar threads

J
  • Question
Project guidance: Real time camera stream over local network
Replies
6
Views
748
ZipZapOuch
  • Question
My Christmas Ornament project
Replies
1
Views
1K
ronsimpson
E
  • Question
Help with my project please
Replies
13
Views
5K
danadak
F
  • Question
My Nixie clock project feels really close to working... but not quite
Replies
2
Views
2K
Roff
B
  • Question
Open-Source ESP32 Robotics Kit for Modular, AI-Powered Projects
Replies
1
Views
349
aliarifat794
A
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…