My wife had the CBR600 F4i, very nice bike to ride on.
I had a CBR600F from '82 when I started riding.
Cool bikes, but today it’s VFR for us both.
I have mine some 14 years now and my wife owns her red one for about 6.
And the GPS can follow.
Got to love the German Autobahn 
Bas.
I’ve done some tests with this, and wrote my own firmware for a custom serial/USB converter. It can still talk USB Full Speed / High Speed, but provide timestamps between when the PPS happened and when the USB host picked up the carrier detect message. This was good enough for a time sync error of 7 microseconds.
Full Speed polls at a minimum interval of 1ms, which explains your results.
I have more detail and the code here: ch32v307 dev board, part 4
I’ve been meaning to design USB GPS hardware board with this firmware, but I haven’t gotten around to it yet.
I do not understand this, if you have an RS-232 device, why not simply use a motherboard that has an RS-232 header or use an PCIe RS-232 card.
See here, tiny board, very fast, silent…and has COM1 header:
That is a good question, and if RS232 is an option, that would be easier and well supported.
Not every device can use RS232 or has a free PCIe slot, so having USB as an alternate option is useful.
No it is not…then better get an USB-device instead of a converter.
It will never be as good as native RS-232.
Sorry it’s not.
While I agree that hardware RS-232 should be able to do better, I think there are applications that +/-10us would be acceptable.
At ~10 microseconds I think the limitation is stability and accuracy of software timestamping. Maybe serial port on a PCIe card can be better, but with a typical on-board serial port (on an emulated ISA bus behind multiple bridges?) it’s difficult to get below 10 microseconds.
Here is a test where I measured accuracy of system clock synchronized to PPS on an on-board serial port by comparing it to hardware timestamped PPS on I210:
@mlichvar how do you connect that?
I do have Intel Corporation 82576 controller cards installed and they can do HWstamping.
Chrony also recognizes them for HWstamping.
Bas.
You need a NIC for which phc_ctl reports a non-zero number of external time stamp channels and find the pins where the PPS signal can be connected. The 82576 doesn’t have that.
The I210 card has a nice pin header:
On other cards and on-board NICs you might need to solder wires. See this great post from Dan:
Ok, understand, ordered me a i210 that has this header.
Found more info…the PPS needs to be connected to SPD0 on the 210 NIC.
That is easy enough, just cut a wire on the RS-232 header-wire and solder it to the nic. Should be easy enough to do.
Thanks.
SDP Port / Pin are on 3.3V level.
Take care that you don’t connect 5V or even more…
300 Ohm resistor to make sure it doesn’t kill the pin?
Will that work? It just needs to rise or lower the voltage.
I would try a voltage divider 330 & 680 Ohms to be sure
470 Ohm & 1k should also be fine - would be around 3.4V
You will have to lower it anyway. Either from 5V, 12V or even 24V
The values above are for 5V
Do you have a oscilloscope ? To make sure the pps don’t get “slewed”
Dave, thank you for the hint
Just order one too and it’s already been shiped.
Pressing both thumbs that they are delivering faster as promised 
Got that and it won’t PPS, I used 3x 1N4148 to lower the voltage to 3.2V, I know, it only pulses 0-3.2V not negative. I measured with my voltmeter and I get a 3.2V pulse every second.
But it’s not registerd.
So I’m going to order some PCIe RS232 adapters to connect the Garmin’s with.
That should work a lot faster then via the south-bridge mess and it by far easier to configurate.
As you said the SB and ‘emulations’ hinder precission, so why not directly RS-232 to the PCIe bus that that is directly connected to the Intel CPU.
New experiment
Doesn’t the voltage drop of a diode depend on the current? With very little current sinked by the SDP the voltage could be still too high?
To test the external timestamping you can try ts2phc from the linuxptp package. In chrony configuration the PHC refclock directive needs to have the extpps option to use the external timestamping and also the width option to filter out falling edges as the I210 always provides both edges.
I measured a drop of about 1.8V what is normal for a Silicium diode, they drop 0.6V per diode.
And it measure +3.2V with peak-hold on the pulses.
Without Peak-hold it pulses every second, at 3.2V.
Resistors change drop with current, but I have tried that too, no register.
Anyway, I gave up on this project. Next will be an PCIe RS-232 card, to see if it matters.
See my other topic, I got them pretty good for PPS.
It’s not Chrony that doesn’t register, it’s LINUXPTP etc that reads nothing, it tells me there is nothing to read.
Dave,
got my Module last Saturday 
just need to figure out the 2mm Pin-Header configuration
This motorcycle is really great, Kawasaki motorcycle?
