Electronics > Projects, Designs, and Technical Stuff
3.3V AND 5V supply to Digital Isolator / Voltage Translator
DBecker:
You might want to step back and describe the actual problem.
Why are you using one of the ADUM chips that transfers power? It's primary advantage is that it transfers a tiny bit of power to the isolated side without additional components. If you aren't using that capability, there are faster, cheaper, more reliable solutions.
I've designed several boards that use a commodity DC-DC converter, a Si8421 two channel isolater (or one of the larger parts with more channels) and a standard CAN transceiver. It's an approach that has proven to be simple and uneventful. I've tried to use the sexy ADUM parts, there is always something that don't quite fit: interface levels, not enough power out, too much heat generated, speed, etc.
max_torque:
er, i'm not using the Digital isolators that have the built in DC/DCs (That's the "5" series parts)
https://www.analog.com/media/en/technical-documentation/data-sheets/ADuM1200_1201.pdf
^^ two channel, without DC/DC
https://www.analog.com/media/en/technical-documentation/data-sheets/ADuM5200_5201_5202.pdf
^^ two channel with DC/DC
I'm using a separate DC/DC because under some circumstances the interface actually powers the slave (for things like desktop programming etc) so i need much more power than the small DC/DC built into the isolator can provide. I'm therefore using the basic two channel device, that needs to take a small amount of external power to operate, and making use of the fact that the isolated side can be powered from two different potentials so as to achieve the ability to interface with both 5v and 3.3v systems!
Someone:
--- Quote from: max_torque on June 29, 2019, 06:33:37 pm ---It's looking like using some sort of analogue switch to select either the raw 5v, or 3.3v from an LDO running off that 5v is the way to go!
--- End quote ---
If you wish to simplify further most 3.3V regulators can be bypassed with just a jumper/short/switch across their input-output pins. Recall all those application notes about reverse protection diodes? Single quadrant devices (most regulators) will happily sit there and output no current into a voltage above their set point and below the input voltage. The generic LD/LT/REG/LM/TLV/ZLDO 1117 regulators are ok with this, as are many other simple/cheap designs.
Ian.M:
4.8V - 4.9V is probably good enough for interfacing to 5V logic.
Take an adjustable LDO with as low a dropout voltage as possible. *DON'T* use a 3 pin one - it needs to have a Gnd pin Supply it from Iso+5V and select the feedback potential divider for 3.3V out. Grround its Adj (feedback) pin using a switch or optocoupler to force its pass transistor hard on to get nearly 5V out (less its dropout voltage).
If you want to do better, as 'Someone' has just pointed out, you could use a LDO that will tolerate up to Vin on its output, and put a P-MOSFET in parallel with it.
max_torque:
OK, possible change of plan.... I now have a micro on the isolated side (to deal with some other functionality) so it makes some sense to just use that ti run a basic buck converter. Most of the time the pass device will just be hard on (ie 5Vin -> 5Vout) but i can use a "spare" Pwm pin on the micro to adjust the duty cycle to get around 3.3V out (as mentioned, in this application, the exact voltage isn't that important it just needs to be low enough to enable the 3.3V TX line from the slave device to trip the "High" logic threshold). Means no messing around with switches or multiple regulators etc!
Navigation
[0] Message Index
[*] Previous page
Go to full version