[time-nuts] Improved Symmetricom ET-6000

[Erik Thomassen]

Good evening/morning -depending, group :-)
 For the perfectionists of you, pass this post ;-)  For the rest, here are my results in improving an Symmetricom Exacttime ET-6000.  My problem was that the supposedly stable 10MHz, off the programmable outputs just wandered "all over the place" with the supplied TCXO. Refered to any free running OCXO or rubidium the output was useless even in the locked state.

There was an upgrade-set avaliable that also involved a software change but that was not an option for me. 

The original TCXO is relatively easy to remove. The solder holes are left alone, better connection points for experiments exists on the PCB. Check for connectivity the 10MHz output, and Vcorr.  The 5V connection i left alone.Before this project arose i had a 5V/10MHz Bliley OCXO unit previously bought off Ebay from China for some USD20.-  I did not think about the output being square wave.... The main thing however is that this OCXO matches the Vref range of the TCXO.

Yesterday was the day:  Took out the original TCXO.  The new Bliley unit, for the sake of the experiment, was set outside the main PCB with long wires soldered between the units.  Due to the higher current drain on the OCXO i decided to supply it via an external simple 7805 regulator. This was mounted directly to the bottom of the chassis for good heat dispersal during warmup. 12V was taken from the main PCB.

After power-up things looked usable - but did not work....  No lock, the system searching up to+/- 200Hz trying to find a lock point. Setting the DAC-value to "zero beat" with a similar standard did not help.The combination of quite higher output voltage from the OCXO combined with a very distorted waveform (due to the square-wave output) made the 6000-unit to discard the OCXO.  

To improve the waveform I used an old trick: A 10MHz crystal in series with the signal.  This improved things a bit, I now can see a distorted sinus wave on the output....  A 50k 10-turn potmeter in series with the crystal solved the level problem -.

A new connector mounted on the rear for 24VDC supply and an surplus Rifa 24V to 5/12/-12V converter module from a Instrument Landing System transmitter solved the mains problem.... No more power surges :-)

Inserting a jumper (J27) told at least the display that there is an OCXO in place.
After some 6 hours of use now, the internal monitoring system claims that output accuracy for the time beeing is +/- 40 e-12, almost two decades better than the original TCXO, for all what that statement is worth....
The next question coming up: Why not use a simple 12V supply, sinus output OCXO? Yes, why not try that?  With the level problems sorted out - a simple potentiometer, the only problem withstanding is the Vcorr/ref voltage. It may have altered with the setting og J27 - I did not measure that, there, the software change probably kics in - or not :-)
Erik/