Wired network Core interest?

@sej7278, the Core will consume 35ma when in sleep mode (NOT deep sleep). @wgbartley and I did a lot of testing and characterization on another topic. If you need any help with this, let me know. :smile:

1 Like

I am certainly interested. From what I can see, the W5100 is likely to be a better way forward than trying to do a software interface using the ENC28J60 for instance. The W5100 has an SPI interface, so presumably a normal WiFi Spark Core could be connected up using the A2/A3/A4/A5 SPI pins. And there is an Arduino Ethernet Library, so that would presumably make things a bit easier too. Link

I imagine that the serial/usb console would be needed for people who want to configure a static IP rather than use DHCP. It appears that the ease or difficulty of the task is how closely the API for the CC3000 is to that of the W5100.

When using external magnetics, there would be room for pins where the existing patch antenna is located. If you square off the corners you should be able to get the LED pins in too I am guessing. I can see a second version of this device that has the magnetics on a longer board, although I am not sure about the mechanical strength with the existing PCB thickness.

Maybe it is time to order a W5100 board from ebay and to start playing

1 Like

It will blow chunks to do this on the 'F103 with an ENC28J60 - the TCP stack would just chew up too much of the RAM - and if we were taking that route, the 'F107 would be a better choice.

I think the W5100 would be a good proof-of-concept vehicle, since it has the TCP/UDP stack implemented internally, and presents a socket-like interface, like the CC3000; but there are subtle differences: for instance the W5100 doesnā€™t do DHCP, the CC3000 does, so thatā€™s another module that needs to be written/ported and will take up some RAM.

Most of the bad sleep current behaviour is due either directly or indirectly to the CC3000, it doesnā€™t have a true interrupt to wake up the 'F103, so the 'F103 can never really sleep while connected to the cloud. The W5100 might be better in this aspect.

I could see POE being interesting.

I agree with @Swieter, the real value is in the cloud services, so comparisons with USB or RS232 kind of miss the point, IMHO.

3 Likes

Iā€™m looking at this:

http://www.linino.org/wordpress/wp-content/uploads/2014/02/data-sheet-of-Chiwawa-Collar-A4-size.pdf

I have just ordered a couple of W5100 modules, just to see how they work. They ended up being just over US$10 each from eBay including postage.

Definitely agree here. Having code that will work with the same software interface, be it WiFi or Ethernet would be useful for allowing this device to become a real embedded tool. I know that potential customers of mine would prefer some of my designs to be Wireless, and some to be wired.

Time to finally download a build environment for the Spark, and W5100 documentation I think

Darryl

[Edit... I might start a thread about hacking a a wired Ethernet driver to the Spark Core. That would be for people who wanted to talk about the internals, and it could leave this thread for application and 'should we do it' type discussions...]

1 Like

I think I will go with the more expensive, but probably more reliable sparkfun board (not knowing what ebay board you got, but having bought plenty of cheap shit on ebay over the yearsā€¦)

Then the question becomes what development platform to use - I have an unfair advantage being involved with the :spark: team, so I will most likely just depop a CC3000 off a genuine :spark:. However, the Olimex STM32-H103 is probably a good vehicle too, if taking a heat-gun to a :spark: is not in your plans.

2 Likes

@AndyW, that looks a lot like this module I found on eBay!

:slight_smile:

1 Like

Good conversation while I was otherwise occupied.

I agree, we can work on the W5100 platform as a test. I was reading up on the W5300 recently, and though that looks like it would be a great final implementation as there are less external components and a smaller footprint I think. It looks like the breakout board isnā€™t really a breakout board for the W5300 as it has some micro in there for initial config.

Either way, getting W5100 to work initially should prove the concept and not take up too much more to get a W5300 running then. (one feature of the W5300 that was in the press release is using the unused socket memory as general application memory)

Iā€™ll digg out my W5100 modules. Iā€™ll order a couple more Spark Cores. I was thinking of popping the top on the Cores and then adding in wiring to a module.

2 Likes

Donā€™t know if the Flyport boards might suit? I was using this previously for a project that Iā€™ve moved onto a spark core, as one of my main parameters was size. They have WIFI and Ethernet and GPRS versions.
J

Hi Jimbol

Thanks for this. The Flyport looks like it is based on a similar Microchip Ethernet chip, and then from the schematic seems to use an external MicroChip device probably to do the TCP/IP layer. Quickly looking at this, I don't think this has any advantages to us. AndyW has suggested an upgrade to the Spark CPU would be needed to include an entire TCP stack in the Spark, or at very least using a co-processor which has size, power and cost implications.

By going with a WizNet device we can share a mostly common hardware line with the WiFi version

Darryl

1 Like

Lets agree on a W5100 as a proof of concept vehicle.

We can think about optimizations and BOM costs later.

I would encourage anyone thinking about the ENC28J60 to look instead at the STM32F107, which has an ethernet MAC on-chip, but the extra load of the TCP/IP stack is a battle I donā€™t want to fight right now.

3 Likes

Agree. W5100 for IC, Spark Core as development. At least for my setup. Iā€™ll find out which W5100 module I have.

I was thinking of W5100 wired into the same I/O as the CC3000, since that would be gone. This idea is that you either have a wi-fi core or a network core. Not a dual function, but one or the other.

Thanks @AndyW for helping to settle on this as the proof of concept.

2 Likes

Confirmed - the test vehicle should have the W5100 wired in place of (and replacing) the CC3000.

I can start a git branch for this fun and games, and we should share wiring designs. Film at 11.

2 Likes

Hello there

Iā€™m doing an ethernet version,
I therefore created a unique topic about the Ethernet Version
topic proto ethernet

++

1 Like