Level Convertion with complicated requirements

[oziphantom]

So I'm looking to connect something to the PRUs of the Beaglebone Black/Pocket. From the data I can find on the datasheets and faqs etc about the Sitara AM3358 is it is LVCMOS, it doesn't say which LVCMOS, but mentions elsewhere that it takes up to 3v3 on the pins. Although I can't find a 3v3LVCMOS spec, so maybe its 2v5 with 3v3 tolerant pins?
The other end is 5V TTL which I assume ( there is no spec sheet ) can handle 1TTL load ( 8ma )although there is a 74LS244 connected to the pins I'm reading, which is buffering and gating the signal internally.
I have a 6bit parallel signal at approximately 16Mhz ( it will be slight faster in some cases and slight slower in others )
The Sitara cautions note that if you put power onto a one of its pins, whilst it is not powered up, it will damage it. So I need the circuit to also only put power on the pins in the Beaglebone is powered.
I only want this bus to be 1 way, 5V -> LVCMOS.
My first though was optoisolator, but getting them to work at 16Mhz seems tricky and the ones that can need a 48ma forward current across the diode.

Ti have the almost perfect chip in that its 6bits converter but it is 5V CMOS to LVCMOS, I've not found something that handles 5V TTL to LVCMOS.

If the source is putting out a signal and it runs "into a wall" so to speak will it also need some resistors to ground or something to stop ringing and damaging the chips putting out the signal?

[capt bullshot]

Check the 74LCX series for 5V input to beaglebone I/O. They have 5V tolerant inputs (even while their supply is off) and supply them from your beaglebone I/O voltage. 74HCT (or faster ACT) series will take care of the other way (beaglebone 3.3V output to 5V).

[T3sl4co1l]

With very few exceptions, a "CMOS" output pin always swings between VSS and VDD, with a Thevenin equivalent output resistance.  The resistance is different in the high and low states (e.g., for 74HC and most microcontrollers, typically 70 ohms high, 40 ohms low), or you can use the average figure.

Resistance is relevant when you need to calculate how much load a pin can supply, for a given voltage drop.

TTL does not pull all the way to the rails, and its high and low states are more asymmetrical.  Low is like 0.2V and 30 ohms, high is like 3.5V and 100 ohms (or more, I forget; I'm hand-waving here -- RTFDS for the actual curves).

TTL also has an off-center threshold, which is perfectly suitable for its output levels, but means 3.3V CMOS can drive it just fine.

5V TTL driving 3.3V CMOS may not be ideal, because it can potentially pull up higher than the 3.3V supply.  A schottky clamp diode might be used to ensure the CMOS gates' input ESD diodes are not turned on.  In turn, some series resistance (maybe 100 ohms) before the diode, to limit current, would be good.  That's the most you'll ever need.

I don't really know what "LVCMOS" means, unless it's merely to distinguish 5V (and below) versions from CD4000/74C (metal gate, up to 18V) CMOS, which has much higher output resistance to go with its higher voltage rating (which makes it quite pitiful at 5V, by the way ).  3.3V most often seems to be "LVCMOS" but unless it's defined somewhere as otherwise, it seems to count for anything down there, usually 1.8 (74ABT, etc.) to 5V.

Tim

[Peabody]

I think the MC74VHCT50A might work for you.  It's a hex non-inverting buffer and level translator.  And when its Vcc is at ground, all the I/O pins are tristated.  The outputs are determined by the Vcc voltage, but the inputs are TTL levels or lower.

I've used this chip, and found that for the I/O pins to be truly shut off, Vcc has to really be at ground, not just not connected.  Considering any capacitors on this chip and your Sitara, you would want to make sure they power up and power down at the same time.

But I'm not sure about its speed, and the smallest package is TSSOP.

[oziphantom]

Looks like this might do the trick https://www.mouser.com/ds/2/408/74LCX541FT_datasheet_en_20140624-1128340.pdf

A schottky clamp diode probably would do it, but won't give me No Power protection right? It will still let the Voltage hit the Sitara when the Board has no power connected ?

MC74VHCT50A. again I need protection when the BeagleBoard has no power connected. I guess I could use a high resistance resistor to join it to the Source's GND, although the source in this case is notorious for its noisy ground.
Personally I would prefer a larger package say DIP would be lovely.

[me.rishabh12]

TXB0108 bi-directional level converter
u can use this it a bidirectional converter work on lower voltages as well
pls check the datasheet to see if it make ur job done

[Peabody]

MC74VHCT50A. again I need protection when the BeagleBoard has no power connected. I guess I could use a high resistance resistor to join it to the Source's GND, although the source in this case is notorious for its noisy ground.
Personally I would prefer a larger package say DIP would be lovely.

I guess I don't understand your requirements.  If you power the 50A from the same supply as the BeagleBoard, then both would automatically shut down together, and the 50A would go tristate and protect the processor from anything connected to the 50A.  You would just need to make sure that when the circuit powers down, it really powers all the way down.

Comes in SOP too.  I use an adapter board for breadboarding it.

Anyway, let us know what you end up doing.

[oziphantom]

Ok so I will state all cases explicitly to be sure sure.

There is Device A - this has its own supply, it is not powered or controlled by the beagleboard in anyway. It is 5V and once it is powered on, and the 5V rail hits 5V it will output data on the 6bit bus. The data is constant, can't be disabled and is in no way controlled by the beagleboard.
Device B is the beagle board, this does not get power from device A. It has it own power supply, which may or may not be connected.

So You have the case where
A is on, B has a power plug in it, B is powered on, B is connected to A via the BUS - fine
A is on, B has a power plug in it, B is powered off, B is connected to A via the BUS - sounds like it will be fine from above
A is on, B doesn't have a power plug in it, B has no power, B is connected to A via the BUS - the case I'm trying to assert is 100% safe still.

So the conversion chip won't have VCC it will be for all intents "floating" to which the it has to a.) not let power into the circuit. So no power on the level translated bus, no power out of its VCC/VDD. It also needs to not try and pull power from the 6bit bus as the bus only supports one TTL load, so if it tries to eat more than 8ma its going to damage device A.
From your text about "at ground" its seems to me the 50A fails the 3rd case.