[time-nuts] FW: Question about the PLL of Trimble Thunderbold

[Tom Van Baak]

> With the idea that, when the adjustment loop is deactivated, an external signal
> can be supplied to the Thunderbolt, and the Time Interval circuit could show the
> difference in between this signal and the feedback of the VCO.

I understand what you're trying to do. Your idea would work for most GPSDO because the PCB contains lots of separate parts like: GPS chip, TIC, MCU, DAC, OCXO/EFC. You've probably read how GPSDO work and know this.

Note that the GPS chip or module is likely made by a 3rd party and contains a cheap little XO which is used to sample the RF, run the ASIC/MCU, and output a 1PPS. The 1PPS will have clock cycle quantization, so-called sawtooth error. The GPSDO deals with this.


But the TBolt is different. Since Trimble is itself a GPS receiver manufacturer they don't need the GPS receiver to be a separate device from the rest of the GPSDO. They are integrated. Since they already have an extremely good OCXO onboard there is no need for the cheap little XO inside the GPS chip. Instead the receiver uses the main OCXO clock. As a result there is no separate receiver 1PPS; there is only the OCXO 1PPS. There is no sawtooth error either. It's a wonderful design and only possible because Trimble is both a GPS receiver company and a GPSDO company. A TBolt can do low-level magic inside that a traditional GPSDO cannot. This is one reason that why we hobbyist like the TBolt.

Since part of your project is to deliver a "fake" or external 10 MHz and 1PPS into a GPSDO, the TBolt is probably not the right choice for you. A more traditional GPSDO design would be better.

What some of us have done is substitute the TBolt 10 MHz OCXO for a better OCXO or even Rb. But that's a separate issue from your request to substitute the 1PPS as well. It is worth reading the detailed ThunderBoltBook2003.pdf document to further understand the unique TBolt architecture.

/tvb