Nortel NTBW50AA/NTGS50AA (Trimble Thunderbolt) GPSDO


I tested 5 Nortel/Trimble GPSDO units, each having a different size/shape/model oscillator inside. On the the Trimble label the oscillators are called: "T2", "C1", "O", "T", and "V". No GPS antenna was connected so this is the unlocked or free-run 10 MHz oscillator performance. Two summary plots are given below. The time/frequency analyzer was a TimePod 5330A and the plots were generated with TimeLab. The PN (phase noise) plot was made using a low noise quartz reference (ULN). The ADEV (Allan deviation) plot was made using a high stability atomic reference (H-maser).

An addition comparison follows. Here is the current Trimble Thunderbolt-E (fancy blue box), the original Trimble Thunderbolt (fancy red box), two OEM/eBay/surplus TBolt's (plain gold box), along with three of the Nortel's:

Run 1 - Using ULN5/ULN10 for best phase noise

Reference oscillator ULN5 and ULN10 are commercial Wenzel ULN. These have an exceptionally low phase noise. But the mid-term ADEV is rather poor, so ignore points right of tau 10 s.

Run 2 - Using ULN15 for better ADEV (1 s to 100 s)

Reference ULN15 is a selected FE-405B 15 MHz oscillator. It has good short-term ADEV and a good phase noise floor. But the close in phase noise is poor, so ignore points left of 200 Hz.

Run 3 - Using 5071A for better ADEV (long term)

Using a 5071A as a reference is only a good idea for runs lasting days to weeks. For tau less than 1000 s the ADEV "hump" of a 5071A gets in the way. The same is also often true for using a GPSDO as a reference. The plots below are incomplete. I decided quickly to use a H-maser instead of cesium for a reference.

In summary, just ignore the above 5071A runs. It's the wrong kind of frequency reference to use for this class of OCXO.

Run 4 - Using H-maser for better ADEV (long term)

To avoid the ADEV hump of the 5071A I switched to a H-maser for better mid- to long-term stability. The phase noise and short-term ADEV of this maser was not ideal due to 100 ft inexpensive coax but it was not practical to move the maser closer to the Nortel units or move the Nortel units to the maser.

The frequency plots below go from 0.1 1 10 100 1000 second averaging:

The phase noise plot below is limited by the maser and the long cable, not the DUT. So ignore it:

Note: It looks like the Nortel(4) and Nortel(5) runs had some sort of disturbance. If it was important this could be investigated. Or additional runs performed. Or one week of data instead of 12 hours.

Run 5 - Nortel(3) with 6 different references

Run 6 - Different Thunderbolts (short-term)

Run 7 - Different Thunderbolts (long-term)


I randomly labeled the boards #1 to #5 as I pulled them out of the shipping box. Each of the 5 Nortel/Trimble GPSDO units had a different size/shape/model oscillator. I did not break the factory seals so peeking inside for a look at the oscillators was difficult. The following are a set of poor photos (taken through one of the tiny air vent holes).