Date: Fri, 25 Nov 2016 10:39:18 -0800 (PST) From: "Rodney W. Grimes" <freebsd-rwg@pdx.rh.CN85.dnsmgr.net> To: Ian Lepore <ian@freebsd.org> Cc: Michael Sperber <sperber@deinprogramm.de>, freebsd-arm@freebsd.org Subject: Re: Can't get 11.0-RELEASE to boot on Banana PI M3 Message-ID: <201611251839.uAPIdIuV022651@pdx.rh.CN85.dnsmgr.net> In-Reply-To: <1480095986.1889.76.camel@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> On Fri, 2016-11-25 at 09:07 -0800, Rodney W. Grimes wrote: > > > On Fri, 2016-11-25 at 08:42 -0800, Rodney W. Grimes wrote: > > > > > sperber@deinprogramm.de said: > > > > > > > > > > > > > > > > > > Ah, thanks ... but that's not standard RS232, right???(BPI > > > > > > homepages says > > > > > > "TTL".)??If it isn't, what kind of hardware connects to > > > > > > that?? > > > > > The normal setup for RS232 is that the transmit and receive > > > > > signals > > > > > come out? > > > > > of a big chip (SOC, or PCI UART, or USB UART, or ...) and then > > > > > go > > > > > through a? > > > > > level converter which is typically a MAX-232 or one of many > > > > > clones > > > > > or? > > > > > variants.??The "TTL" is telling you that it doesn't have that > > > > > level > > > > > converter? > > > > > chip. > > > > > > > > > > You can either add a level converter chip and then plug it into > > > > > a > > > > > real RS-232? > > > > > port, or find some setup that also doesn't have the level > > > > > converter > > > > > and? > > > > > speaks TTL levels.??Adafruit and probably many others sell a > > > > > USB > > > > > UART without? > > > > > the level converter for applications like this. > > > > > ? https://www.adafruit.com/product/954 > > > > > > > > > > Sometimes, TTL means 3V CMOS levels and 5V from real TTL/CMOS > > > > > will > > > > > fry your? > > > > > expensive chip.??Best to check carefully.??The above part says > > > > > 3V.??It also? > > > > > has an extra power wire that you get to ignore. > > > > Be SURE to ignore that extra power wire!??If your USB/Serial > > > > adapter > > > > also has > > > > a power wire DO NOT CONNECT IT.??Many of these embeded boards > > > > provide > > > > a power > > > > pin with the serial interface that can be used to power something > > > > external, > > > > like a level shifter, and many of the USB/Serial adapters also > > > > bring > > > > out the > > > > USB 5V rail on a wire.??DO NOT CONNECT THE TWO!? > > > > > > > Ummm... say what? > > > > > > I power my rpi boards using the 5v power from the USB serial > > > adapter > > > connected to the 5v pin on the rpi header. ?I can't imagine any > > > reason > > > not to. > > You either have a USB port wiling to supply more than the 500mA > > from the standard, or are running your RPI under very light load. > > This is NOT the recomended path to power a RPI, and if the RPI is > > getting power from the micro usb port your connecting 2 power > > supplies in parallel, which is bad. > > > > And why exactly is two sources of +5v a problem (not that I said I was > doing that, but I have at times done that, and it works just fine)? ``Works just fine'' are famous last words when the smoke comes out of a) your power supplies or b) your board cause of power supply failure. What you are doing is called brute force paralleling of power supplies, and any good EE knows that is just a bad idea(tm) becuase power supplies are not designed to do that. Google up parallel power supplies and read for more information, lots of it out there, some bad, some good. > > > If you plub a USB device into your RPI that draws significant > > current it well probably reboot due to power loss. > > > > For information on how much power your PI wants/needs: > > https://www.raspberrypi.org/help/faqs/#powerReqs > > > > > > > > > > > > > > That being said, there are many aftermarket USB/Serial cables > > > > avaliable, > > > > usually a 3.3V version of these well work everywhere as long as > > > > it > > > > has > > > > 5V tolerent inputs, which most of the newer ones do, check the > > > > specs > > > > from the vendor.??3.3V outputs well satisfy the input > > > > requirements of > > > > a 5V TTL/CMOS circuit and not cause it problems, the opposite is > > > > not > > > > always true. > > > > > > > Usb serial adapter based on Prolific chipsets are NOT 5v tolerant. > > > ?Those based on FTDI chips are. ?Those are the two big names in > > > usb- > > > serial chips, but there are others out there too; you have to check > > > the > > > datasheet to be sure. > > From my read of the Prolific data sheet that is not clear, they > > specify > > 3.3v and 3.3v 5v tolerent, but then they do not clearly state which > > pins are 3.3v and which are 3.3v+5v tolerent.???I'll take your word > > that they don't like 5V on the serial rx pin. > > > > They are actually very clear that of the IO pins, only the RESET_N pin > is 5v tolerant. Somehow I missed that in reading the datasheet. > > > > Watch your lead length and wire sizes if you need to do anything > > > > funny > > > > to get this connected, capacitive loading of any kind on this > > > > type of > > > > signal can cause character loss, especially at speeds above 9600 > > > > baud. > > > > > > > Ummm... that sounds pretty bogus too, considering that I've run > > > ftdi > > > chips at 12mbps using breadboards with a rat's nest of wiring to > > > carry > > > the comms signals to other boards. > > Iout of most of these chips is 4mA,??you are welcome to do the RC > > calculations.??You can often get away with amazingly poor setups, > > then suddenly get bitten by what looks to be just fine but drops > > characters. > > > > Full level RS-232 specifies cable length of 50ft at 19200, your > > not going to get away with that using CMOS 4mA drivers unless? > > you have very low loss cable. > > Nice introduction of a red herring. ?Who said anything about 50 foot > cables? ?Oh wait... you did. ?Only you. Again, I would suggest googling up single ended signaling, serial cmos wire length, ttl over ribbon cable, etc, just cause it works for you on a bench does not make it good design practice and that it well work all over. In general CMOS or TTL signals are not really desinged to go great distances without carefull impedence control and signal integrity considerations, neither of which exist in some random wiring. Sure, I have crammed 10Mhz signals down a meter of ribon cable and gotten away with it, but thats not a sound design practice, genreally TTL/CMOS signars are considered to have a sub 1 meter usable signal length. -- Rod Grimes rgrimes@freebsd.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201611251839.uAPIdIuV022651>