1
\$\begingroup\$

For quite some time I am thinking about making 'passive' usb 'amplifier' - but the main question is how it is possible to detect direction of the transfer by some clever way, so that I can disable one of amplifiers?

Classical approach of using USB-HUB chips to extend cable for another 5 meters is too boring.

\$\endgroup\$
7
\$\begingroup\$

Signal strength is not the only reason for the USB cable length restriction. Two other reasons are reply timing and voltage drop on the power line.

It is possible to watch the bus and know who is sending, but this requires the kind of complexity only available in a IC for any reasonable size. However, that alone isn't good enough either. There are passive pullups/downs at the device end to indicate the device is plugged in and what speed it supports.

In short, this would take considerable trickiness to get right, if you can make it work at all. A USB hub is the way to extend USB another cable length. If you need much longer, then USB isn't the appropriate communication link. Ethernet is more appropriate for longer distances.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ What is the difference between signal strength and power voltage drop? \$\endgroup\$ – Kellenjb Jul 27 '11 at 15:52
  • 2
    \$\begingroup\$ @Kellen: USB has separate signal and power wires. Since devices can draw up to 500 mA, the resistance in extra cable can make the power voltage at the device be below the guaranteed minimum level. This signal is differential and carried on a twisted pair. It will also deteriorate with distance, but different mechanism are at work since the twisted pair ideally acts like a transmission line. It's more of a signal to noise ratio issue with the signal, although extending a cable a few meters probably won't matter in that respect much. \$\endgroup\$ – Olin Lathrop Jul 27 '11 at 16:03
  • \$\begingroup\$ Oh, yes I understand. I was thinking you were talking about power voltage drop on the signal lines. \$\endgroup\$ – Kellenjb Jul 27 '11 at 16:08
  • \$\begingroup\$ @Kellen: Edited to make it more clear. \$\endgroup\$ – Olin Lathrop Jul 27 '11 at 21:40
3
\$\begingroup\$

I'm not sure this is possible without intelligence. The signal looks the same any which way it comes from; you'll at least have to decode the communication to tell them apart.

Kellen mentions the advantage that retransmitting means that you start with a clean signal again, and that's true, but I'm more worried about latency introduced by the decoding. I don't know how much time is allowed between transmitting and receiving, and if the decoding and retransmitting won't take too long.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ Plus the advantage of decoding and retxing is that you basically reset the SNR. \$\endgroup\$ – Kellenjb Jul 27 '11 at 15:08
  • \$\begingroup\$ If the delay is acceptable with a hub I would bet it would be acceptable with decoding and re-txing if done in a similar method as the hub. \$\endgroup\$ – Kellenjb Jul 27 '11 at 15:40
  • \$\begingroup\$ @Kellenjb - Unless the hub also sends a message back, or that during negotiation the topology is determined. \$\endgroup\$ – stevenvh Jul 27 '11 at 15:46
  • \$\begingroup\$ Hence the "if done in a similar method as the hub" \$\endgroup\$ – Kellenjb Jul 27 '11 at 15:48
1
\$\begingroup\$

This is not possible passively. You will end up damaging eye or violate chirp sequence or something else. Since USB is kept building on top of old stuff, there are tons of overloads of intelligence over existing stuff.

Hub is one option. But, it is a complex piece of silicon and believe it or not, there are only a handful of silicon vendors that can support hub properly. The TT (transaction translator) inside the hubs is complex and requires tons of tests to make them compatible with all the FS and HS devices in the market.

The other very smart alternative in the market is from a company. (Name escapes me now). They basically NAK to the sender even though they get the data properly, in the mean time forward the data, get the response and when the host retries, they send the proper response since they have received it from the device that they have extended. (Hope this 10K feet explanation gives you the general idea) They had a solid patent on this and were able to build a business out of it.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.