Comments for Project Gus

Comment on Anatomy of a cheap USB to Ethernet adapter by Glen Turner

Glen Turner — Sat, 18 May 2013 05:21:38 +0000

I bought something similar at the ANU in Canberra, Australia. It even knocked off a Linksys USB ID.

New USB device found, idVendor=066b, idProduct=20f9
New USB device strings: Mfr=1, Product=2, SerialNumber=3
Product: HG20F9 USB2.0
Manufacturer: USBLINK
SerialNumber: 000003

A look for complaints on the web about it not working on Linux showed people from the USA and UK. The kernel people were happy to take a patch, despite the hijaaked but non-conflicting ID.

Comment on Notes on FTDI latency with Arduino by Chris

Chris — Thu, 25 Apr 2013 22:48:54 +0000

Neat!

Comment on Notes using scratchbox2 with debian multiarch by Danny

Danny — Mon, 22 Apr 2013 13:37:01 +0000

Thank you, your multi-arch fix worked for me trying to setup a scratchbox environment to compile code for the raspberry pi.

I have been scratching my head on this one for several hours already!

Comment on Servo pulse width range with Arduino by Mike

Mike — Fri, 29 Mar 2013 22:42:39 +0000

Thank you very much Angus!

Comment on Anatomy of a cheap USB to Ethernet adapter by Peter

Peter — Thu, 28 Mar 2013 12:37:30 +0000

One other thing you could run into is the following : future updates of the driver software won’t work anymore.The guys who made the knockoff chip, actually do not know of all quirks around in the real chip,so they implement only the nescessary functionality.But the guys at ASIX, who know their own software and hardware inside and out can easily build-in some functionality to test if there is a genuine ASIX chip fitted.
I’ve seen this happening with USB-RS232 bridges from prolific.The knockoff hardware only works with a very old driver which they supply on a CD.

Comment on Anatomy of a cheap USB to Ethernet adapter by Leif

Leif — Mon, 25 Mar 2013 14:15:38 +0000

@Galane mentions a story about how some straight through cheap 12V to USB (adapters) came about.

I don’t know anything about that particular story and it’s validity but perhaps part of the reason this happens is that the USB spec says that it should be ok… sort of.

My understanding anyway is that the spec for USB hosts does say 5V and only 5V. However for clients it is written that they are supposed to accept a wide range of voltages (I forget how high) and not blow up.

I guess the people making the spec knew that knockoffs would cut corners and tried to make the device manufacturers build their devices to survive that kind of ‘abuse’.

However, that doesn’t mean it is ok to use an adapter like that. First off, just like you can’t trust the knockoff cigarette plug to usb adapters to do the right thing you can’t trust device makers to fully follow the spec. They might not actually take the higer voltages well. Also.. car power is very very dirty. There are huge spikes in there, especially every time you start your engine. Without a good regulator and filtering in the adapter you can still kill or shorten the life of your device even if it does handle 12 volts.

Comment on Anatomy of a cheap USB to Ethernet adapter by angus

angus — Sat, 23 Mar 2013 23:14:24 +0000

Hardcoreforensics - I've heard that some of the cheap devices only implement USB 1.1 "Full Speed" devices, It seems to me like these ones are getting less common now than they were a year or two ago, though. Which is good. :) I think you might be misunderstood about hub-related slowdowns, though. As long as the hub is a 2.0 hub, devices on downstream ports can be either USB 1.1 or 2.0 without any slowdown. It's only if you have all the traffic going through a USB 1.1 hub somewhere in the line that everything gets slowed down. USB 2.0 hubs actually contain what's called a "transaction translator" that means they speak USB 1.1 to slow downstream ports, but combine those downstream connections into the 2.0 upstream link (see the previous link for a description.) For example, you mention the Raspberry Pi. It's SMSC LAN9512 definitely contains a USB 2.0 hub, USB 2.0 devices should run at high speed (480Mbps) no matter what is plugged into the other hub port. - Angus

Comment on Anatomy of a cheap USB to Ethernet adapter by angus

angus — Sat, 23 Mar 2013 23:02:23 +0000

Russell – thanks for the tips about measuring RFI/EMI. I don’t know how much use I’d get out of such a testbed, even such a readily available one, but it’d definitely be interesting to play around with.

Comment on Anatomy of a cheap USB to Ethernet adapter by Hardcorefs

Hardcorefs — Sat, 23 Mar 2013 22:51:04 +0000

I have a load of research on this crap, because I needed to build a high speed interface to a project I was working on.
One of the most interesting facts, is that these devices are sold as 10/100
When in fact the chips internally are not capable of anything more than USB 1.1 standard.
Basically many of these chips are built around a reltek macro library they sell the products as being USB 2.0 compatible, which some are, but the actual internal Ethernet core is interfaced with USB 1.1…
Which means :
1. That anything you plug it into gets downgraded to 1.1 speed….
2. You can never come close to saturating the link.

you will see this time and time again , if you take a look into actual chipsets used on some products (RPie for example)
The SOC will have two distinct USB infrastructure ports, but then some idiot designs a hub chip between one of them to give say 2-4 ports, but does not route the other port off the PCB.
The net effect is that you plug in one of these devices, the hub sees 1.1 standard and BANG… all of your USB traffic hits the 1.1 wall.
Ideally you have to ensure that all your 1.1 standard devices are off one of the SOC USB ports and that all the 2.0 are off the other.
The sad thing is to get USB working properly, you need to be a systems engineer as regards to which devices you plug in externally to the equipment……
and in some cases it means pulling the chipset data sheets and looking at the INTERNAL spec of the chips..

Comment on Anatomy of a cheap USB to Ethernet adapter by Russell McMahon

Russell McMahon — Sat, 23 Mar 2013 03:52:21 +0000

A simple RFI / EMI test that is often useful is to place an AM band “transistor radio’” immediately adjacent to the device and tune across the while band. while listening for “spuries”. When placed hard against the device almost anything will produce some signal somewhere. Now move the receiver away progressively while checking for the level of major signals found initially. A device that will meed Class B certification will usually “quiet up” reasonably rapidly. If you can detect it from 10 feet away “you may have problems”. Obviously such a test has an immense degree of variability depending on equipment used, variability of environment and how you hold your tongue. It’[s still quite useful as a subjective indication.
- Tuning the FM band is usually not such a good indicator as AM.
- Interesting & useful results can be had from the low cost small screen (7″ or so?) monochrome CRT tube “cube shaped” televisions still available from Asia. These usually have continuously tunable frequency input. Tuning these across their several bands gives both visual and audible indications of spurious outputs and can allow actual harmonics of on board components to be detected. (Continuous tuning makes them much more useful for this purpose than fixed channel units). Possible example only
http://upload.ecvv.com/upload/Product/20118/China_5_5_B_W_TV_TV_50920118142129553.JPG