[time-nuts] time transfer over USB

DaveH info at blackmountainforge.com
Tue May 14 23:47:24 EDT 2013


I am also interested in CNC machining and for a long time, the parallel
printer port was the only way to get reliable multiple axis movement. USB
did not work because of the issues being talked about here.

If you are machining a complex curve, each X, Y and Z axis step and
direction has to be output simultaneously. If not, your curve will not be
smooth.

Do not know the timing numbers but I do know that if there is something else
on the USB port (for machining, a lot of times it is a webcam), the lag can
be highly variable.

A successful solution was done by the people at SmoothStepper (
http://www.warp9td.com/  ). They have a printer driver that installs on the
host machine and I am guessing that they buffer a few seconds of activity
and reconstitute it on the other end of the USB cable in their proprietary
box-o-magic. They also have an Ethernet solution. Each are well under $200
and have a really nice system.

There may be an issue of intellectual property here, there may also be an
issue of Prior Art but I am wondering if there might be some solution like
this for transferring time data. Thinking of several...

Dave

> -----Original Message-----
> From: time-nuts-bounces at febo.com 
> [mailto:time-nuts-bounces at febo.com] On Behalf Of Peter Monta
> Sent: Tuesday, May 14, 2013 20:11
> To: Discussion of precise time and frequency measurement
> Subject: Re: [time-nuts] time transfer over USB
> 
> > > Bulk transfer might work well on a system where nothing 
> else is going
> >  on,
> > > so "best effort" translates to "now".
> >
> > Does it mean "now" or "next polling cycle"?
> >
> 
> It means "now", or pretty close.  No waiting for start-of-USB-frame or
> -microframe.
> 
> Herbert mentions that all USB transfers are host-initiated, 
> which is true
> (even for the so-called "interrupt" transfers), but the 
> latency is usually
> much less than one frame on a lightly loaded bus I believe.
> 
> I agree with Magnus' remark that this is not rocket science.  
> Maybe it's
> time for a small proof of principle, using something like a 
> Teensy board,
> that takes PPS edges from the outside world and connects them to USB.
> 
> GNSS samplers should be timestamping their packets so that 
> the relationship
> between sample number and device-local time (and ultimately 
> also host time)
> can be known.
> 
> Cheers,
> Peter
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to 
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.



More information about the time-nuts mailing list