Why not use a set of chips like the PCA9534 and PCA9534A (two identical ICs that have different default addresses) instead of the current chip to allow for 16 Trellis addresses on a single I2C bus?
I see potential for a Frickin' Oontz with a 16x16 grid, or even an 12x20 monstrosity.
The natural disadvantage is that you would need to produce either two varieties of board or a board with two chips. The former is inconvenient and difficult for the consumer, and the latter would be cramped on such a small board. However, if you used the 16bit versions (TCA6416 and TCA6416A) and made the boards 8x8, then you could probably fit both and the means for selecting between them. The 8x8 boards could be marketed to people wanting to build larger grids.
Datasheets:
8bit
http://www.ti.com/lit/ds/scps124g/scps124g.pdf
16bit
http://www.ti.com/lit/ds/scps194a/scps194a.pdf
Potential for greater # of Trellis addresses
Moderators: adafruit_support_bill, adafruit
Please be positive and constructive with your questions and comments.
- adafruit_support_bill
- Posts: 88086
- Joined: Sat Feb 07, 2009 10:11 am
Re: Potential for greater # of Trellis addresses
Thanks for the suggestion. I'll forward it to the design team.
- pburgess
- Posts: 4161
- Joined: Sun Oct 26, 2008 2:29 am
Re: Potential for greater # of Trellis addresses
Clever! Simple! We were thinking about using a software I2C lib to run a second bus in parallel, but this is fraught with peril.
Gonna require an external power brick though. Too many LEDs for one USB port to drive. :)
Gonna require an external power brick though. Too many LEDs for one USB port to drive. :)
- adafruit2
- Posts: 22144
- Joined: Fri Mar 11, 2005 7:36 pm
Re: Potential for greater # of Trellis addresses
you would need a chip with 32 I/O per trellis - 16 LEDs and 16 buttons. you could put two down per trellis but then in the end you'd still be stuck with 8 max per I2C bus.
Its probably way easier to just go with one hardware and one software I2C bus! or use a switch to select between two sets of trellis' SCL pin like philb did with the back-to-the-future clock
Its probably way easier to just go with one hardware and one software I2C bus! or use a switch to select between two sets of trellis' SCL pin like philb did with the back-to-the-future clock
Na0s wrote:Why not use a set of chips like the PCA9534 and PCA9534A (two identical ICs that have different default addresses) instead of the current chip to allow for 16 Trellis addresses on a single I2C bus?
I see potential for a Frickin' Oontz with a 16x16 grid, or even an 12x20 monstrosity.
The natural disadvantage is that you would need to produce either two varieties of board or a board with two chips. The former is inconvenient and difficult for the consumer, and the latter would be cramped on such a small board. However, if you used the 16bit versions (TCA6416 and TCA6416A) and made the boards 8x8, then you could probably fit both and the means for selecting between them. The 8x8 boards could be marketed to people wanting to build larger grids.
Datasheets:
8bit
http://www.ti.com/lit/ds/scps124g/scps124g.pdf
16bit
http://www.ti.com/lit/ds/scps194a/scps194a.pdf
- na0s
- Posts: 3
- Joined: Wed Jun 19, 2013 12:11 am
Re: Potential for greater # of Trellis addresses
Hmm, that's a good point regarding the 32 I/O. I'm not familiar with the financial implications of increased complexity, but you could go with a 16 I/O button controller for an 8x8 grid of buttons and a separate LED controller (the original Monome has a similar design). There's definitely an incentive to keeping any new designs similar to the current trellis and changing up too many of the ICs might make the two products too dissimilar. I'm trying to design a system that allows for eight 64 button grids (up to a Monome 512) on as few i2c buses as possible. I'd be happy to post my findings if there's an interest.
Please be positive and constructive with your questions and comments.