I have a Symmetricom ExacTime ET6000 that had a Motorola Oncore M12+ mounted to an ACE-III interposer made by Synergy systems that hadn't suffered from a rollover yet, but instead had a 3 month leap second induced offset. I went the (mildly easier) route with a Furuno GT-8736D, then modifying the interposer with a level shifter for the serial lines as the ET6000 uses 5v TTL logic, and the Furuno chipset is only rated for 3.6v maximum (and so was the M12+, for that matter!).
Several years ago I fixed an MTK GPS module in a smartphone with the same problem by RE'ing the firmware and patching the date boundary in it. I wonder if that's possible with this module too.
Seems this guy[0] could do just that if only he had access to a SyncServer S200? The very author of this blog post replied to the person and told them about his interposer solution, he’s not unaware.
> It may also be possible to build similar patches for other units, such
as the S200/S250. But currently I only have access to S350 and XLi.
Alex’s solution doesn’t patch the module itself, but it patches the FW of the MCU in the SyncServer motherboard that drives the module.
People have looked into patching the module itself, but I haven’t seen any progress on that. It’s possible that the M430 MCU doesn’t allow dumping of the FW that is stored in on-chip flash.
They’re less than $1 on AliExpress and $4 on Amazon. Do you really want to spend time optimizing for something cheaper just because you don’t want to waste some MIPS?
The cost of buying an off the shelf GPS module to replace the original is so high that any price difference in microcontrollers or even an SoC is negligible. You could drop an entire raspberry pi in there and still be saving money.
There’s also the fact that I have RP2040-Zeros ready to use in a drawer. They’re so convenient, tiny, cheap, still lots of pins, always enough performance, easy to program with just a USB cable. Any other MCU, and I’d have add power regulator, caps, coils and a USB connector. Add the cost of shipping from Mouser/Digikey (at least $5) or LCSC (long wait time) and it only makes sense for high volumes. IOW: never.
A nice Symmetricom or ptf unit is going to offer much better time accuracy and holdover accuracy than those cheap units from China. The really cheap ones have no OCXO so time will drift fast during GPS outage. You get even better holdover accuracy with Rubidium but that's really only worth it if you need a real ref clock for test equipment. Expect to pay at least $500 for a cheap unit with an OCXO.
Obviously most people need no holdover accuracy so a cheap unit is fine. Most people aren't running data centers in places where GPS jamming is expected. Although it's not uncommon for truckers to be running GPS jammers that can inadvertently knock out a whole sea port.
I have a very lovely Symmetricom in my rack that has a busted GPS receiver but still provides NTP and IRIG-B using a Rubidium source as it's 10MHz reference.
I’ve seen them for $80 on eBay, and you could build something like that yourself with a $25 thin client[0] and a USB GPS dongle. But I doubt you’d have the local oscillator for hold over, you wouldn’t have the 10MHz and 1PPS inputs and outputs, and let’s not forget the lack of a cool looking case with VFD. :-)
I have no need for an NTP server, this was just a fun deep dive into something I found at the electronics flea market. But there must be reasons why these things still go for much more than $80 on eBay after they’ve been upgraded with a new GPS module.
Even the most basic Symmetricom (microchip now?) and ptf units have all the options you need to survive an outage and configure the thing anyway you want. I've got a busted Symmetricom that won't see anything in the sky but still provides NTP and IRIG-B using a calibrated Rubidium source for its reference clock. I don't need the time to be accurate but I need it to be precise.
