Yes, that was part of the original troubleshooting. I have not tried it with the USB2 version yet since the firmware flash, but since it worked before and since it works for you, I assume that will not change. The ST4 guiding through the USB3 version is the problem. wvreeven (see a few comments above) found exactly the same problem, during calibration the mount never reverses, but drifts away into one direction, so it can't be my camera or isolated to the settings on my Pi3, as the issue reproduces on his setup.

It’s probably worth noting that my setup doesn’t use an RPi but simply a (2010) Dell laptop with Ubuntu Mate 17.10.1 (artful).

If USB3 version produces same problem, it's probably time to call ZWO about this.

And tell or ask them what exactly? Remember, guidng via ST4 on the USB3 versions works in PHD2. Just not in Ekos...

In PHD2 you're using ZWO driver or INDI driver?

I use the ZWO driver and on-cam guiding. I didn't even know it was possible to use the INDI driver in PHD2.

I was using the ZWO driver that shows up in the PhD2 drop down menu. I don't remember whether that was there before I installed Indi and Ekos and whether the PhD2 ZWO driver is dependent upon the installation of Indi or whether the driver was included in one of the last PhD2 updates.
I agree with wvreeven: This problem's got to be rooted somewhere in Ekos, since the ST4 guiding worked until the January 29 update. Besides, nobody working at ZWO, they are out of the office until February 23 to celebrate Chinese New Year.

Agreed, this is not Pi3 dependent. I also have Zotac PI332 pico with 4 GB of RAM running Ubuntu Mate 16.04. Behaves exactly the same in all respects as the Pi3 does.

I tried this again now with the USB2 version of the ASI120MM and now I replicate the same problem I am having with the USB3 version! Sending guide pulses through the ASI120MM and ST4 connection from inside Indi Control panel and ZWO Guider Control results in movement in one direction only, with the mount rapidly drifting away. Guide pulses sent to the mount directly work fine in all 4 directions. I have not tried guiding in the field yet, will have to wait for at least one week, since I will be out of town again for the next few days.

*face palm*

I found the problem. It was indeed a commit I merged a while back. So while checking the code again, I noted the WE guide function was using NS direction. It's fixed, compile now or wait until tomorrow PPA. Sorry about that, should have checked the code first.

That’s great Jasem! Can’t wait for tomorrow!

I see that I introduced this bug. I tested the code using the ST4 port on an ASI174mini, but I was doing so on my Losmandy Titan, which still needed its worm block adjusted, and I put down the bad RMS to that, rather than a copy and paste error in the code!

Thanks for finding and fixing this.