Back
[00:01:33] <osteri> hmm. Is it good practice to use git submodules for e.g. LUFA, FreeRTOS or something that you don't touch yourself?
[00:06:55] <hypermagic> ofc
[00:07:06] <hypermagic> though i never use anything i didn't read
[00:07:16] <hypermagic> and i usually only use what i wrote
[00:07:33] <hypermagic> for led flashing you might be good with anything
[00:09:10] <osteri> usually you can point these directories within a makefile, so you don't have to modify them
[00:16:29] <osteri> adding these repository files to each project seems wasting space/processing power... but sometimes you have to customize FreeRTOS for one specific project, so centralized FreeRTOS dir is not possible
[00:17:01] <osteri> because if you accidentally delete this centralized folder, all your modifications are gone
[10:21:09] <hypermagic> hi
[12:14:24] <nics> Hi all, I have got a question regarding the USB unit on an Atmega32u4, namely on how, if at all, the RWAL bit in UEINTX relates to the FIFOCON bit if an IN endpoint is operated in single bank mode. More precisely: does FIFOCON=0 imply RWAL=0 in single bank mode?
[12:15:40] <LeoNerd> Mm.. that's a fun-looking question. Sadly, while I do have a 32U4 I've so far only done USB via Arduino with it, which mostly just wraps LUFA I believe.. so I haven't poked it that low-level directly myself
[12:16:16] <nics> The Atmega32u4 datasheet only explicitly states that RWVAL is cleared by hardware when the buffer is full.
[12:17:51] <nics> The reason why I'm asking is that I believe that the Arduino-1.6.4 USBCore.cpp:USB_Send() depends on the behaviour FIFOCON=0 => RWVAL=0 for single bank mode IN endpoints.
[12:36:49] <[w_w]> When the datasheet (for the attiny84) points to a pin and says NOTE: Bottom pad should be soldered to ground. DNC: Do Not Connect... what should I do with the pin?
[12:37:45] <LeoNerd> There are no DNC pins on a 'tiny84
[12:38:12] <LeoNerd> Ohwait, yes there are on the QFN20 package
[12:38:17] <[w_w]> Attiny84A sorry.
[12:38:32] <LeoNerd> Yeah - that's a separate notice. It's just placed al ittle close to the other one ;)
[12:38:38] <hypermagic> :)
[12:38:39] <[w_w]> Oh. the DNC is the legend for the diagram
[12:38:52] <LeoNerd> It's pointing out that the DNC pins mean "do not connect". That's separate from the thermal/grounding pad
[12:39:04] <hypermagic> heatsink pads are either No internal Connection, or Ground
[12:40:07] <hypermagic> even if it is isolated and not connected to anything you could inject noise into the mcu by routing a signal layer to it i think
[12:40:28] <hypermagic> capacitive coupling
[12:40:54] <osteri> once my RTC IC wasn't working, then connected NC pin to GND -> started working
[12:41:06] <hypermagic> :)
[12:41:29] <osteri> i'm a rebel, not following orders
[12:42:03] <hypermagic> you unleashed hidden powers of radiation proof RTC chip :)
[12:42:46] <osteri> yeah, it took tens of hours to solve, not very nice
[12:43:25] <hypermagic> and it is confidential information to connect NC to gnd to make it more stable
[12:44:09] <osteri> haha, right
[12:44:50] <RikusW> NC pin = secret backdoor interface :-P
[12:45:20] <hypermagic> i used an old ide cable to connect my programmer to the avr, i wired it every second pin gnd - so now it works in UDMA mode :)
[12:45:35] <hypermagic> 0 capacitive crosstalk
[12:46:03] <RikusW> 40 or 80 conductor ?
[12:46:14] <hypermagic> 40 was fine.. i only used a part of it
[12:46:48] <RikusW> even 10 should do, like the old style ISP cable
[12:46:54] <hypermagic> you can make 3-4 programming cables from a 40 pin cable
[12:47:39] <hypermagic> (with every second wire grounded)
[12:48:03] <RikusW> does it indeed work better ?
[12:48:13] <RikusW> like > 1MHz ISP clock ?
[12:48:22] <hypermagic> yes
[12:48:40] <hypermagic> i emasured a piece of cable and 2 adjacent wires had about 15pF capacitance
[12:49:06] <hypermagic> that will spike your lines if you don't use lowpass filters...
[12:50:06] <hypermagic> this is why your 40 pin cable will only be good for about upto 5MB/s
[12:50:18] <hypermagic> your cable had crosstalk
[12:50:40] <hypermagic> 80 pin cable has every second wire gnd.
[12:50:50] <RikusW> I rarely go beyond 1MHz anyways...
[12:51:25] <RikusW> it seems on 80 pin the connector has a internal ground for every other wire..
[12:52:52] <hypermagic> yes, this is the only difference
[12:53:23] <hypermagic> so pins will not crosstalk, only be a little more loaded to gnd
[14:29:57] * LeoNerd ooohs at http://www.adafruit.com/products/457
[14:30:10] <LeoNerd> Essentially a 74xx595 with high-current drivers; can sink 150mA
[14:30:16] <LeoNerd> For big LED matrices
[14:34:56] <twnqx> tons of low side drivers out there
[14:35:06] <twnqx> but search for high side drivers.. good luck
[14:36:17] <LeoNerd> Eh.. for that presumably just a string of PNPs
[14:36:28] <LeoNerd> Though to be honest that would do lowside as well
[14:37:17] <twnqx> i'm just using n+p channel posfets
[14:37:53] <twnqx> and a bunch if resistors
[14:38:55] <LeoNerd> Yah