resolved - hardware spec atmega168 maybe related to 2.0
Posted: Fri Apr 16, 2010 1:38 am
I have a number of systems that work, enumerate, and function. And I have another number that does not even attempt to pass data to the host. Using a protocol analyzer it immediately tells me "device problem". I'm using 1/2 watt diodes and 68 ohms for my - + leads. I also have a 1.5k pull up on +. This is a atmega168 so I need the 1/2 watt diodes. Most of the troubles are on newer systems and I think mostly 2.0 ports ( more so in laptops ). Although most computers have 6 ports or more and if it fails on one port on a system it fails on all of its ports, likewise if it works on one port on a system it works on all of them.
Also in the cases where it fails, if I plug in a hub ( in between ) it works. I matched the device descriptors of the ports and I see no differences.
----
Ok I ruled out the diodes.. I found an old board and was able to make it work. I tried switching out the diodes on the trouble board and that didnt help. I would not think the 12mhz caps would matter. The 12mhz is identical. and all 3 resistor have the same value. I'm so confused on this one.
-------
Ok turns out it was the diodes. Note sure where I failed to test that the first time... The surface mounts produce 4.0 and my regular through hole shows 2.7 .. something weird there..
Also in the cases where it fails, if I plug in a hub ( in between ) it works. I matched the device descriptors of the ports and I see no differences.
----
Ok I ruled out the diodes.. I found an old board and was able to make it work. I tried switching out the diodes on the trouble board and that didnt help. I would not think the 12mhz caps would matter. The 12mhz is identical. and all 3 resistor have the same value. I'm so confused on this one.
-------
Ok turns out it was the diodes. Note sure where I failed to test that the first time... The surface mounts produce 4.0 and my regular through hole shows 2.7 .. something weird there..