Hi
I've built my own arduino clone. It uses an ft232rl chip to upload sketches onto the atmega. I want to be able to isolate my ft232rl chip ( and the rest of the board) from the usb interface so whatever happens I don't blow up a £500 laptop. Is Is this possible? I 'm thinking I can archive this using an opto-isolator but am unsure since I've never used one.
Most hubs are not isolated, and isolated hubs are expensive. I would isolate the UART side, not the USB side. UART signals are much slower than USB.
+1 for isolating the UART, not the USB. Here's a USB to UART board I made earlier that has isolation:
http://partiallystapled.com/~gxti/circuits/2012/03/13-usbjumper_sch.pngIf you're only worried about blowing up your host PC, you don't necessarily need full isolation. Most of the risk comes from overloading the port which you can solve by using a separate power supply to run the board (which can also power the UART chip).
You can use an USB isolator. E.g. Farnell code: 1843660
http://pt.farnell.com/analog-devices/adum3160brwz/ic-isolator-digital-usb-16soic/dp/1843660?Ntt=ADuM3160
I recommend using an USB hub whenever possible, but some micros don’t work very well when connected to external hubs…
http://uk.rs-online.com/web/p/optocoupler/7098388/Would this one be ok? I took a look at the datasheet it says it's compatible with usb2.0 low/full speed.
This chip is pretty expensive however it's worth it if it does protect my £500 laptop/ PC.
I've never used one is there anything I need to watch out for?
@ GXTI You say most of the risk comes from overloading the port but I specifically configured the ft232rl to be self powered. On my board it has a 5v regulator (with output protection) that powers the atmega as well as the ft232rl. If I managed to creat a short on my power rail could that not potentially blow up my pc as well? Also what do you mean by most of the risk? What are the other risks?
The only way to damage a USB port is to inject voltages above 5V or below ground.
As for overloading, each USB port (should) have overcurrent protection, so no problem there.
It shouldn't be too hard to check that the signals are wired correctly before plugging the board?
The only risk with many USB port implementations is the ferrite beads that is often used in the
ground wire. Excessive ground current and this baby can get fried. But that is only possible
with desktop PCs, which have the system ground connected to mains earth. (Unless there are
devices connected to other ports which complete the loop, then a laptop might suffer too.)
The only way to damage a USB port is to inject voltages above 5V or below ground.
As for overloading, each USB port (should) have overcurrent protection, so no problem there.
It shouldn't be too hard to check that the signals are wired correctly before plugging the board?
The only risk with many USB port implementations is the ferrite beads that is often used in the
ground wire. Excessive ground current and this baby can get fried. But that is only possible
with desktop PCs, which have the system ground connected to mains earth. (Unless there are
devices connected to other ports which complete the loop, then a laptop might suffer too.)
So the only way to damage a USB port is to put slightly too high voltages in, or too much current? Could never happen.