4 ms·

Nifty idea. If you want to make it even better, have it simulate the iPhone/iPad charger ID circuitry so that I can charge my iPad off any old USB charger (prov

Nifty idea. If you want to make it even better, have it simulate the iPhone/iPad charger ID circuitry so that I can charge my iPad off any old USB charger (provided it's rated high enough). Edit: Actually, scratch that. Leave the data lines connected, but "short" them to the V- line (or shroud, should hopefully be the same thing) with a small capacitor to act as a low-pass filter. I don't have the specs in front of me, but it should be easy enough to filter > 1Mhz down by 3dB and still keep the DC "slew rate" enough to properly ID a charger.

MertsA 13y ago
This is how this USB condom should have been made while making sure to use two separate capacitors instead of just shorting the data lines together. The only problem with this approach is that it's physically possible to make a special USB host that pumps enough current down the data lines to fill up that tiny capacitor every cycle. Might want to throw in an inductor for the extra paranoid.
- benjamincburns 13y ago
  It depends on the size of the capacitor. And even with smaller capacitors, any device which is capable of producing a signal strong enough to overcome even a simple passive 3dB filter well enough to get through the USB handshake, let alone actual device operation, would almost certainly be quite large as compared to a regular charger. Adding an inductor would screw with things, but if we're being really paranoid here, active circuits can measure the resonant frequency of the line and overcome it. Or, even if we're not being paranoid, you've now given someone a nice RCL trampoline to bounce a nice high current into your phone's USB data lines. There are a few ways to protect against all of that if you want to be really paranoid. To start, you could go with a higher order active filter built from a cheap op-amp circuit. More complex varieties could include the use of a tiny 8-bit uC programmed to control a digital pot on the protected side, and an optoisolator somewhere in there just in case there's some weird failure mode which causes a signal path to short from protected to unprotected. The benefit of something like this would be that the controller could also control a light or buzzer to alert the user when a signal is detected on the "unprotected" side. Cheapest and most reliable might be to pump the output of a simple RC high-pass filter into a simple RC low-pass filter (translates to DC bias) then feed that to a comparator which latches the signal lines open (and sounds an alarm) if signal is detected. Or better yet, make it normally open and close only when signal isn't detected.
dnautics 13y ago
minimum viable product first. Then iterate.

MertsA 13y ago

This is how this USB condom should have been made while making sure to use two separate capacitors instead of just shorting the data lines together. The only problem with this approach is that it's physically possible to make a special USB host that pumps enough current down the data lines to fill up that tiny capacitor every cycle. Might want to throw in an inductor for the extra paranoid.

benjamincburns 13y ago
It depends on the size of the capacitor. And even with smaller capacitors, any device which is capable of producing a signal strong enough to overcome even a simple passive 3dB filter well enough to get through the USB handshake, let alone actual device operation, would almost certainly be quite large as compared to a regular charger. Adding an inductor would screw with things, but if we're being really paranoid here, active circuits can measure the resonant frequency of the line and overcome it. Or, even if we're not being paranoid, you've now given someone a nice RCL trampoline to bounce a nice high current into your phone's USB data lines. There are a few ways to protect against all of that if you want to be really paranoid. To start, you could go with a higher order active filter built from a cheap op-amp circuit. More complex varieties could include the use of a tiny 8-bit uC programmed to control a digital pot on the protected side, and an optoisolator somewhere in there just in case there's some weird failure mode which causes a signal path to short from protected to unprotected. The benefit of something like this would be that the controller could also control a light or buzzer to alert the user when a signal is detected on the "unprotected" side. Cheapest and most reliable might be to pump the output of a simple RC high-pass filter into a simple RC low-pass filter (translates to DC bias) then feed that to a comparator which latches the signal lines open (and sounds an alarm) if signal is detected. Or better yet, make it normally open and close only when signal isn't detected.

dnautics 13y ago

minimum viable product first. Then iterate.