XK-1 and XScope

[russf]

Hey Al,

The XK-1 simply expected the opposite interpretation of UP and DOWN on both JTAG connectors, so a chain works as you'd expect, it's just inconsistent with the new direction chosen by the XScope and the newer boards. That means we need to change both the IN and the OUT connectors at the same time, and also port allocations in XK-1 examples will now need to be reversed (?).

Look at this page for my proposal for the mods

I'd welcome some checking, before I get my Dremel out.

Best,

--r

[Folknology]

Hmm and how do you know it was the XK1 that was wrong and not the Xtag2 in the first place, all the other boards came after those two?

regards
Al
;-)

[russf]

Excellent question. 1st of April has everybody's scepticism on high.

I'm going by the newer schematics, as pointed out by XMatt in this post in another thread.

Edit: The draft plan is complete. I won't be making this change over the weekend. I'll wait until the XMOS crew have a chance to comment. Someone there must have done this already (?).

[Folknology]

I think it might be easier to just have an Xtag2 adaptor board that swaps the link pins for the XK1 and any other boards may have followed suit. Clearly in the XK1's case the upstream XSYS connectors would still be different but as long as you were only connecting other faulty XK1's it should work right?

Of course if Xmos go and fix the faulty XK1s they have in stock and action the fix in any new builds, those boards won't be compatible with the older faulty XK1s.

Or Xmos could ship adapters with the faulty XK1s.

That would also mean shipping adaptors with the newer builds so they could be used with the older XK1 when daisy chaing of course.

Similar must apply to the distribution stock also.

XMatt what will Xmos do?

*Update I have just added a project for such an XSYS adaptor

regards
Al

[russf]

I think it might be easier to just have an Xtag2 adaptor board that swaps the link pins for the XK1 and any other boards may have followed suit. Clearly in the XK1's case the upstream XSYS connectors would still be different but as long as you were only connecting other faulty XK1's it should work right?

Yes, that would work, but there is the question about what link one is defining for communication in the downstream modules.

I will probably go ahead with this patch, unless someone spots an issue. But as I said, I'll leave it until Monday or later. Interested to see XMatt's comments.

--r.

[Folknology]

Also for Xmos they could use this to fix the current and distribution stock just by adding some adaptors. having to rework the boards would be much more difficult and expensive, not to mention the stock withdraw logistics and supply chain probs.

They could also offer to ship an adaptor to their existing customers.

regards
Al

[russf]

They could also offer to ship an adaptor to their existing customers.

Oooh! Yes! With a 20% rebate-by-post coupon on the next XK-1. Get those channels turning over.

--r.

[XMatt]

The XK-1 was initially designed and produced before the requirements for XScope were complete and so the board supports JTAG and UART only communication with the XTAG-2. The interface for this part of the tools has now been standardized and all reference designs and new boards which support this feature are now wired the same way. The XK-1 was designed to work as a chain of XK-1's as discussed earlier and this operates correctly in this mode.

[Folknology]

Hi Matt and thanks

Just to clarify a little further:

You seem to be suggesting that the XK-1 was designed to be incompatible with the Xtag2 XSYS specification, in particularly with the pinouts of the XSYS xlink pins which were obviously known at the time and have always been part of the Xtag2 XSYS spec.

The XTAG2 has shipped with the XK1 as standard since the beginning, the XTAG2 manual has included the link pinouts as part of XSYS since V1.0, everyone I know here on Xcore was under the impression that the linking between the 2 devices were correct and not designed in opposition. I have seen nothing until this thread that would indicate anything different.

Would you agree that the problem arises with the opposing wiring of the links within the XSYS specification, rather than anything specific to Xscope to which you refer?

regards
Al

[russf]

I like the idea of Al's connector-oriented fix, but I want to go ahead with my patch.

I'd be tickled to death if someone could find any holes in my plan, before I start cutting traces.

--r.