I understand that the Nano is currently limited to 1500 byte frames.
A product we are designing would benefit from use of the Nano instead of the GigaStax Rugged but we require jumbo frame capability. Is there a way to enable that?
The latest hardware revision of GigaBlox Nano (revision B, BB-GGN-B-1) added a footprint for an EEPROM to be added to the board. This EEPROM can then be loaded with firmware to try to enable jumbo frames on the main ethernet switch.
We’ll test this in the lab and, if we can get it working, will figure out how best to release this.
Thank you, Josh, for that reply. While we won’t be able to use the Nano in the new design yet, the capabilities the Rev B EEPROM may provide could well be vital such as we have now with the UbiSwitch management tool. We’ll be most interested in learning of your progress on that front.
At the moment we only added an EEPROM footprint to the GigaBlox Nano, not to GigaBlox, sadly. But now we know this works, we will update GigaBlox to a new revision with added footprint for the EEPROM. Will expect to roll that out towards the end of this year.
We still have lots of the early revisions with unsoldered expansion header that exposes SDA and SCK. Would it be possible to configure the jumbo frames via this bus instead of the EEPROM? Or even connect the EEPROM to these pins?
So to get it right: we need to desolder or change a few resistors and we need to directly solder the EEPROM to the unpopulated header (or create a small adapter PCB?) ?
Do I get it correctly that those resistors select whether the EEPROM is used at all or not? So if we put them back, the switch will again work as originally? (even with the soldered EEPROM?)
If that’s the case then yes, we’re interested!
I guess you’ll provide the EEPROM contents? What other things can be done this way? I think you also talked about disabling EEE, right?
EDIT: We also have a few newer revisions that have only testpoints and not the unpopulated header. Apart from the ugliness of soldering to testpoints, would it work the same?
Yes you’re pretty much right. Some resistor straps need to be flipped around to tell the chip to load the contents of the EEPROM into the internal chip registers at power on.
And yes you can flip those resistors back to put it back to the non-jumbo mode.
It is, in theory, possible to setup more configurations, but we’d need to engage with the manufacturer of the chip to get that information, and that’s not easy. This isn’t something where we can generate our own firmware files as the registers are not public.
For the new revisions with the testpoints, yes you’d just solder to them, ugly, but it would work.
@nickstar can you help prepare a guide for @peci1 on how to do this on GigaBlox?
@peci1, please see the following steps to enable jumbo frame on BB-GGN-B-1. The next revision (C) will likely have the capacity for both standard and jumbo without customer intervention.
Solder a M24C08-FMC6TG EEPROM to footprint U3. Pin 1 is in the lower right corner.
Power and decouple the EEPROM by soldering a 0 Ohm to R13 and a 0.1uF to C47 (0402 package size).
Remove R41 and place the 4.7k ohm resistor on R11 to enable EEPROM autoload
Remove R42 and place the 4.7k ohm resistor on R12 to correct LED bootstrapping
See the image below for the locations of these components. Please note that the LED indicator for port 2 will flash red but is still communicating properly with jumbo frame.
Thanks @nickstar . However, I was asking for the solution for GigaBlox (not Nano). We have some older revision which has the EEPROM pads on it (it actually looks pretty large, so mabye it’s rather footprint for an unpopulated header. Hopefully it’s BB-GGB-C-1.