We’ve wired up the Tiwi-R2 chip directly to MMC1 on the am335x (i.e no voltage translation chips), and also wired VDDSHV1,2,3 with 1.8V with the aim of running everything in the same voltage domain thus negating the need for intermediate voltage translators.
However, I cannot get the sdio to run properly - I can detect the chip, and get the modules loaded, but anything more than that (i.e. bringing up the interface, transferring firmware) produces a stream of sdio errors.
I then found, on some fairly old release notes for the AM335x-PSP kernel (http://processors.wiki.ti.com/index.php/AM335x-PSP_04.06.00.07_Release_Notes) a note in the ‘Known Issues’ section that: ‘MMC: hsmmc driver doesnt support 1.8v MMC cards’.
In release notes for later versions (i.e. http://processors.wiki.ti.com/index.php/AM335x-PSP_04.06.00.10_Release_Notes) this is no longer listed as a known issue, although at no point is the issue listed as having been fixed.
Does anyone know if this is actually an issue with the silicon, or if it is indeed a limitation of the driver software - or if it has actually been fixed in later kernels.
Has anyone managed to successfully interface the tiwi-r2 directly to a 1.8V mmc port an AM335x? We’re looking at getting a rev B underway assuming this 1.8v issue is the problem, but I’d appreciate any more knowledgeable input on the issue as it’s not 100% that this shouldn’t work as is.
I have no experience with SDIO at 1.8V on the AM335x. I know on the DM3730 we used 1.8V and there were some issues with early WIFI modules.
Pretty much every forum post I can find with someone successfully integrating an AM335x with the TiWi-R2 mentions the use of some kind of voltage translator (probably due to proliferation of the beaglebone design). The only design I can find with the wifi chip directly wired is the pandaboard with its OMAP4460, so the 1V8 sdio works for that TI part, but that is not much help here.
Do you have any recommendation regarding the best way to ask a question to TI? I’ve posted to the TI-E2E forums here: http://e2e.ti.com/support/low_power_rf/f/307/t/264043.aspx, but have so far had no response - it has only been a few days though. I will also try contacting LSR, maybe they’ll have some idea. We get a 1-hour overlap with US Central Time first thing in the morning, so I might even try calling.
Failing that there will be a serious amount of cut/pasting of the beaglebone/wifi cape schematic on the cards - we thought we were being so smart not using voltage translators!
You need a translation for sure as it is 1.8V. To make the 1.8V work on AM335x is a system issue… You have to make those rails 1.8V for all of the signals that use those power rails. I have no idea how you use those other pins, so only you can answer that question.
You could just look at the BeagleBone cape and see what TI did there for translation. It should apply to pretty much all of the modules.
As to silicon issues, I can’t answer any of those. I don’t deal with those folks. Long story there. You could ask the BeagleBoardToys folks. They may know but I would not count on it.
Thanks again. I’m pretty sure I’ve got all the power supplies wired correctly for the pins I use - as far as I am aware there is nothing needed to be changed within the kernel itself to get the 1V8 to work, it is just a matter of wiring the 1V8 to the correct power inputs on the AM335x - as far as I am aware anyway.
I think I’ll go ahead and get a new PCB done with an exact copy of the Beaglebone/wifi cape setup - there’s plenty of other work to do without reinventing that particular wheel.
Thanks again for your help.