We are looking to add some PWM signals to the expansion connector. This will be valuable for robotics and other functions. We are considering two different ways to accomplish this. Our goal is to make this as non-disruptive as possible and to have a zero cost impact.
The current plan is to replace the following pins on the OMAP3530 with different pins that will not change the available functions on those expansion pins, but it will add the PWM signals to them. One side affect is that the pin muxing will change for these pins on the Rev C. There are ID pins on the REV C board that will allow the SW to identify if the board is a REV C or REV B. Based on that reading the drivers could set the pin muxing accordingly to make the pins be in the correct mode. No functionality is lost because of this change. Below are the affected signals:
The other option would be to add pins to the expansion connector and add only two PWM signals. Those added signals in affect will only have use as a PWM signal as the other signals on those pins already exist on the expansion connector. This could also result in the connector having to move from its current location on the board.
I would like to get everyones input on this as soon as possible. We are planning to release the board in a week and I need to get this change in.
I'd love to see more I/O pins on the expansion connector, but it seems
to me that moving the connector location would potentially obsolete
any existing / planned expansion boards. Tweaking software to check
the board ID prior to talking to GPIO seems like a lower impact way to
add PWM functions.
Adding PWM pins without moving the connector would be the best of
both, especially if you could add pins whose other muxed functions
aren't already available on the connector. If that's not possible
though then the assignments you've given above seem like a reasonable
compromise.
I would also like more I/O, but I would also like to see the existing
expansion header kept in its current location. This is for a selfish
reason, but I am probably not the only one doing this. I am building
an expansion card that mates with that expansion header. I actually
received the first prototypes of the board I am building from the PC
fabrication company this morning.
The proposed pin changes look good to me and, checking the board rev
in software and doing the pin configuration based on the results works
for me.
Actually, you are getting more I/O options via the LCD expansion headers. If you look at the pin muxing, I think you will find that some of these signals have alternate pin functions, like GPIO and UART1.
Indeed! Table 7-4 of spruf98b indicates that GPIO 66-90, UARTs 1 & 3
and McSPI3 are all available as alternate modes of the DSS pins. If
you power down the DVI framer (via GPIO 170) then these would be
available for other purposes. Nice.
If you tried to install one of these connectors, you will know they are very difficult to solder and the pins are very soft, so they bend really easy. I would have used a ZIF connector, which is far easier to install and if you used the pin layout of the PSP1000 display, you could connect a low cost LCD without the need for an adapter.
Actually, it isn’t that difficult. Use liquid flux and fine solder wick. Apply liquid flux, solder all the pins. Don’t worry about shorts, Use fine solder wick to remove the shorts. Simple.
Good point. Still, the ZIF connector is more robust and the adapter can be
placed anywhere, whereas the 1.27 header only allows the adapter board to be
placed behind the BeagleBoard.
Unless you are trying to fit this into a specific enclosure. The ZIF solution gives you the flexibility to put the adaptor/LCD/protoboard anywhere you want. Anyway, it sounds to me you are too far along to make a change at this time, so this isn’t really a problem. I will just make a 1.27mm header to ZIP adapter. Simple solution and everyone is happy.
I already have an adapter board for the zif connector that is on the
way. I also have an adapter board that does level translation to
whatever voltage you need, most lcds don't want 1.8v, so most likely
the zif connector wouldn't go on the beagle without additional level
translators.
The zif connectors are probably harder to solder than the connectors
proposed, they are spaced .5mm
