Lars DIY GPSDO with Arduino and 1ns resolution TIC

[ZigmundRat]

Good progress. Does it seem like it wants to lock now?

[Mike99]

Good progress. Does it seem like it wants to lock now?

I'm still waiting for the HC160 to arrive to complete the build. Next week sometime!

Mike

[Mike99]

OK the PCB build is complete. There are two differences between it and the breadboard: the PCB uses a 74HC160 counter instead of the 74HC390 that Lars specified, and it has two voltage regulators whereas my breadboard had four. I can always add more if there is a noise problem.

The resistor network that restricts the tuning range is the same as before and on the PCB the frequency shift between DAC extremes is 1.13Hz, giving a gain of 580. This isn't far from the number that Lars suggested should work with a good OCXO.

I set the DAC to 32768 and adjusted the pot until the frequency error was zero (well 0.002Hz low). filtx10 was around 9800 so I set the TIC offset to 980.

Although it claimed to be locked I still got "Missing 10MHz" and the DAC never went below the 40000s. This was with the default TC=32.

I decided to use PaulV's method of determining the values of TICmin and TICmax which involved setting TC=4, and when I did this the Missing 10MHz went away. Having established and saved the values I set TC back to 32 and it has been locked ever since.

Curious behaviour, but it seems stable at the moment 

My frequency counter is showing an error of -0.005Hz.

Mike

[Mike99]

I ran it overnight with TC=4 and the resulting MDEV plot suggests an optimum TC of 600 so I will try that next.

Mike

[Solder_Junkie]

Hi Mike

I just checked the setting of my Lars/Murray GPSDO (same board as Mike for anyone still following this thread).

F4 returns these settings:
TICmin = 120
TICmax = 1012
x2 = 100
TIC_Offset = 500
filterDiv = 2
warmUpTime = 400
damping = 600
tempRef = 280
tempCoeff = 0
dacValueOut = 34459
gain = 110
timeConst = 60
k = 48

[Mike99]

It's all gone wrong again after I switched it on again ...

The Missing 10MHz came back and even with TC=4 it won't lock.

I decided to reduce the gain by removing the resistor network. The measured tuning voltages were:

h1: 1.616V
h32768: 2.692V
h65535: 3.766V

The voltage range is therefore 2.15V and with my previously measured tuning sensitivity of 3.5Hz/V that equates to 7.525Hz, or a gain of 65535/752.5 = 87.

I entered this with TC=4 and default settings for the other parameters and now it looks like this:



It's very frustrating because I didn't have this problem five years ago when I built my first Lars design.

Any suggestions for what to look at would be gratefully received!!

Mike

[Mike99]

Running out of things to check here so I triggered my scope on the 1PPS signal and looked at the relationship to the 1MHz.

It seems to be drifting on successive triggers. Is that a clue to the problem?

Mike

[Solder_Junkie]

Mine steps back and forth the same, yet it is locked and working.

Dave (SJ)

[Mike99]

Mine steps back and forth the same, yet it is locked and working.

Dave (SJ)

Yes that does appear to be correct. I don't know why it took me so long, but adding an earth braid between the Arduino and TIC counter as previously suggested has stopped the Missing 10MHz ... 

Mike