A few weeks ago, we caught wind of a very tiny, very inexpensive WiFi chip TI is producing. Everything required of an Internet connection – TCP/IP stack, configuration utilities, and your WEP, WPA, and WPA2 security tools is included in a single tiny chip, making this a very cool device for an Internet-connected microcontroller project. There’s only one problem: TI put this chip in a really, really weird package, and there aren’t any breakout boards for it.
That is, until now. [Vince] was convinced to spend some time in Altium to design a breakout board for this tiny WiFi chip. Now, if you can get your hands on a sample of the CC3000 from TI, you can breadboard out a circuit with the help of [Vince]’s design.
Included in [Vince]’s git are the board files for this breakout board, schematics, and the necessary parts if anyone has the inclination to make an Eagle library. If anyone wants to spin a few of these boards and put them up on a Tindie Fundraiser, that’d be fine by us, and [Vince] would probably appreciate that as well.
I’m really excited for this chip. Even if you need to roll your own SSL layer via PolarSSL or something, it’s still so inexpensive that it’s worth it. My sample’s on order.
I received a sample roughly one week ago. I don’t have a hot air station to solder this thing onto a board yet, but I’m glad someone took the time to design this. I was planning on doing the same thing myself at some point. Thanks Vince, looking forward to putting this chip to use. Should be interesting.
Thank you [Vince]!
I would suggest getting rid of those 90° paths. 45° is the best practice even for power lines.
i prefer to do power lines as smooth splines
I don’t see too many professional products done that way…
And is there a reason why not? Do electrons only move in 45 degree multiples?
The reason professional boards don’t do it is curved lines make it difficult to pack traces close together. There’s no electrical reason to do anything other than pick an appropriate size trace and watch the impedance / lengths in high speed circuits.
Actually at high frequencies square edges become emitters of EMI because they are pointy. Given that that chip is modulating @2.4Ghz the power rails will also pull current at that frequency unless sufficient decoupling is given, this can be a source of interference.
Also in mass production 90 degree angles are frowned upon because they can become acid pockets and cause board failure later. It slowly eats the internal traces away, provoked by moisture and voltage, think electrolysis.
Interesting that appears to be something that the utilities in these area don’t practice The private contractors who build the oilfield secondary
lol still if size and power usage aren’t much of an issue and you need wifi and sh#t it’s cheaper to buy a used android phone… BT, GPS, GPRS, 3G, a bunch of sensors and more than enough computing power for most uses… kind of surprising they (android phones) are that cheap… damn I can get some for more less the price of a wifi shield for arduino
Problem is often finding the spec, or a spec that’s useful among all the alibaba posts and alldatasheet spam you find googling.
why would you need a datasheet ? just write an app that gets the phone hardware to do what you want and you’re done… this is exactly the method I used here > https://www.youtube.com/watch?v=V3Akt3SKhAs nothing special I know just controlling one relay but this method is fine for vast majority of projects
Spud gun?
“Android phone used as a remote ignitor ” ..wow ..interesting..
..but, why a Attiny is needed ? can u tell us more about how u generate the high voltage? u use the Attiny to generate the high freq AC?
any way..interesting !
High voltage is generated with a bbc spark ignitor module and the relay simply turns it on and off… and I might have mentioned attiny somewhere in the comments because I don’t feel like inluding an arduino board into every little project I create- so yeah it now works with attiny instead of arduino. That was a cool little project – I might have made it even more interesting by inluding a bbc spark ignitor module there rather than using it to switch a lamp on/off or openning a garage as the module generates quite a lot of EMF and I had to do some research into decoupling and stuff
As some one who does RF layout for a living I thought I might give you a few pointers. Thermals are bad, bad, bad, bad, bad, bad, bad! At 2.4GHz the small webbing behaves as inductors. Also you will want to thicken the traces to the decoupling caps (this will reduce the inductance of the traces).
Well, improve it and share. TIA.
Heaven forbid a professional give some pointers for free without spending his own time making the necessary changes to an amateur’s work…
I think DerAxeMan has already been quite generous.
I’m not an RF engineer, but Vince’s design doesn’t follow any of the advice given in TI’s PCB guidelines. (PDF warning!) http://www.ti.com/lit/ug/swru331/swru331.pdf page 16 onwards
Where does one GET 52.2mil (1.3mm) PCBs made, anyway?
The boosterpack board does have thermals on the gnd pads, though…
@Roel,
I could just do that but then no one would know what I did or why I did things. Its some times better teach someone and let them do it.
could u redraw the PCB for open hardware world??
thanks ..
No, no. Free market opportunity is knocking. Build a better wifi enabled mousetrap and everyone here would know what you did and why you did it.
I see what you are saying about the thermals, but I think you could get away with just thickening them as a solid chunk of copper will be much harder to solder to.
Doing a lot of pcb design I see a few other pitfalls here. Vince you need to modify your trace and via rules so minimum thickness is about 6 to 8 mills, but recommended is 12 or so. Also those vias look impossibly small. There are rules for those as well in altium. The default rules in that program make an un-producable board. (Run it through sparkfuns batchPCB checker and you will quickly understand)
I think un-producible is not the right term to use. They are producible, but not at a reasonable cost for hobbyists. For example, our supplier can make 0.1mm diameter vias using a laser drill. But I get your point, the designer should set his design rules in Altium to meet the requirements of the service he requires at his PCB fab.
True, the layout is terrible. But the part seems to be somewhat forgiving. I know someone who dead-bug soldered it and it works. Looking at the output with a spectrum analyser showed some degradation of the wifi signal (probably no longer meeting spectral mask) but no out of band emissions. He did solder the decouple caps directly to the bottom of the device instead of going via wires (as that would likely add too much inductance)
Design Engineer with no RF Design experience here, thanks for your pointers! Do you have any recommended reading about this topic?
Interesting advice. I started working on a board for this chip myself, trying to follow TI advice as best I could. Didn’t know about thermals at high freq. I’m concerned I’d never be able to solder the darn thing on without them tho; I’ve tried doing without thermals on a ground pour region and soldering is a nightmare. Would a hot air station actually be able to solder the thing on without thermals? Maybe reflow skillet. Hm. Anyway, glad to see this article. I’m re-motivated to progress on this and maybe put the thing up on Tindie as indicated.
How do you explain TI’s PCB layout guidelines (see post below) have thermals however? Do the multiple ground pads alleviate the inductive effect?
I wish the manufacturers would just publish a reference design in a open format that everyone could use, instead of these low resolution jpegs.
The thermals are part of the reference design though, there is enough of them to help aliviate the inductance created while still keeping the board easily solder-able.
Though they could be widened slightly I’m seeing a bigger production issue with the thickness of most of the signal traces and especially the via sizes. Altiums default rules do not a real board make.
THIS. 2.4ghz is no joke. The shape of (nearly) every trace matters at that frequency.
That’s the beauty of SoC, though. There’s essentially only ONE 2.4GHz trace involved here, going from the chip to the antenna.
Well, consider the power draw of this chip. Wouldn’t a component of that also be @ 2.4ghz? And what about interface traces? Wouldnt certain lengths act as weak 2.4ghz antennae? Like I said before, everything matters.
If only the Raspberry Pi could drop their current networking setup and us this instead it would be awesome. I believe that the networking setup they use now takes a quarter of the power that the model B boards draws.
There is no way the Pi Foundation would spend $10 a part for anything other than the SoC. Also, some people actually like having their Pi on a wired connection (faster than wifi, more reliable than wifi.) If you’re really that worried about power consumption though, you could aways wait for the Model A, or follow the B to A conversion that was on here a couple days ago.
It’s $10 for 1k pieces, so 1 cent a piece. Doesn’t sound too expensive to me…
LOLOL
No, its $10 per piece when buying in units of 1000. So $10000.
Owww… :-S
Do i get free shipping with this 10$ for 1k deal ?
TI already makes a boosterpack
http://www.ti.com/litv/pdf/swru331
So about $39.70 gets you a launch pad and this wifi booster pack. The CC3000BOOST is $35.
Any idea how to mount it?
It should be re-iterated what has already been mentioned in a reply above. This PCB violates TI’s design guidelines. Unfortunately not only will this reduce RF performance but http://www.ti.com/lit/ug/swru331/swru331.pdf on page 18 there’s specific mention of the requirement for a “strong ground with ground vias under the module for system stability and thermal dissipation.”
I have used other chips from TI with these requirements and all I can say is that if you don’t follow these guidelines expect your chip to quickly burn up. You’re effectively using it without a heatsink. TI has several packages where the heat generating components are grounded and the pads below the package are direct connections purely for thermal reasons.
The last time I was really clever and thought a simple heatsink would work I ended up sorely disappointed.
Hm, the wireless part looks weird to me. I don’t see any proper RF filtering. The WiFi RF trace should be 50 Ohm, as short as possible, but has to go with a few more caps/inductors to filter things out. (Or has Ti built this stuff in the chip itself?)
Second, I haven’t found out how to do the wifi TX power calibration for this chip. This has to be done per-board and it needs some costy hardware like IQflex. PCIe wifi cards have eeprom for that, routers have a dedicated mtd part like ‘art’ on atheros, nfjrom on realtek, etc. Without proper calibration performance will be crappy.
One thing I noticed, in one of TI’s 802.15.4 chips is that the input impedance is not 50Ohm single ended, but more like 69+j29 differential. So it needs matching and/or a balun. If this is the same then the RF performance of this board is not going to be great…
Just had a look at the CC3000 eval module user guide and they don’t stick anything except a 10 pF cap between the U.fl and the chip….
Oo why are the traces so small? There is enugh space for bigger ones and this would make it possible to etch the board at home.
I think it’s possible to redesign PCB in one side. so it would be cheaper to fabricate.
IPv4 stack? Great..
I just made an Eagle part for this the other day, along with a Sketchup model for anyone who uses Eagleup. I also modeled the A6150/A6250, which are both inexpensive SMT antennas available on Mouser that would be great for this application.
Links:
https://github.com/Carrigan/EagleLibrary
https://github.com/Carrigan/elecmod/tree/master/electronics
(Just noticed the duplicate folder, will fix this tonight)
That’s awesome. I will check ’em out. I made a footprint as well (not yet posted) — making the eagle part was a nightmare in arithmetic. :)
Brian that’s awesome, I think I might just go ahead and take up the Hack-a-Day call to port this board to Eagle this evening.
I have one of those chips. Please post when someone starts making the boards for retail sale.
@Nova’s point to the 90 degree turns. I suspected it was something to do with RF, nut I don’t believe enough info has been given to help anyone who decides to improve this layout or create an entirely different layout. Per nova’s comment it read as the sharp points that are are the problem not so much the 90 degree transition. Even lesser angles in the trace could have sharp point. In my mind what amateur board designers need to know is; are 90 traces with rounded corners permissible higher frequencies. I ask that being ignorant if the design software allows for that.
We don’t really need to guess at these things. TI has design guidelines, and a sample design (the launchpad daughterboard.) It looks like:
1) The only RF that goes off the module is the antenna connector.
2) They want a really solid ground plane underneath the module. Connections on the top, ground plane under that, and your THIRD layer can maybe start to think about traces under the module. The boosterpack has substantial groundplanes on all four layers. It does use a 4-layer board.
3) Except for the Antenna pin and trace restriction, thermal issues seem to outweigh HF issues. This is a ~1W module.
4) the SPI interface can run at 16MHz, which is still a frequency not to sneeze at.
5) Power supply issues are remarkably light. The boosterpack has one 1uF cap on each of V_IO and V_BAT. Presumably stronger bypass issues are handled within the module. The recommendations do call for some heavy power traces (40mil) on V_BAT.
Huh. I am also having trouble reproducing TI’s waveguide calculations, using calculators that are on the net…
You could try taking this up with them in their support forums and see how that goes before trying to get a hold of a support engineer.
I would buy such a board with all parts soldered and ready to be used, connecting to a MCU via SPI or so. Is anybody willing to give me a hint?
THX
Get one of the (eval) modules that already exist for the part. You can either go to ti.com and search for the part number which gets you to a complete cross-reference page to everything having to do with the part, or go directly to estore.ti.com and search for the part number. Some have been available at distributors for months.
Maybe more like this?
http://www.flickr.com/photos/58843278@N00/8511030401/
Darn, but this sort of thing comes with a heavyweight testing burden. You stick an antenna on there based on a TI app-note, but actually measuring how well it works is next to impossible at a hobbyist level. And then “minor” changes in board thickness or composition can change behavior. RF sucks… (No wonder people are willing to pay so much more for a “known working” module!)
+1 bigtime
Oh yah, I took weeks figuring out the F antenna on a USB wifi stick I had, then got really freaked when I unconnected it, and the damn thing still seemed to pull in signal as much as ever. :-D … it’s an intractable little POS… at least for trying to hook to a stub in a cantenna, seems to pick signal off it’s ground shield when you’ve armor plated the little bugger. Anyhoo, end result, unable to make it directional and add gain, and conversely, unable to REDUCE gain by hacking bits off it and sealing it in a leaden tomb… well… as long as you had to have a USB cable connecting it to the smarts, it would seem to use that as an antenna.
Can I buy such a board?
Would you want to, given that we don’t know how well it works (or whether it works at all)? “just a little more fiddling”, and I plan to send it to oshpark…
Do you sell one?
Hmm. Also, apparently the cc3000 module can bypass FCC certification if used with the same “PCB Guidelines” and Antenna as one of the certified reference designs. Alas, that doesn’t include either of these breakout boards :-(
Actually, I think I like this one better. Aside from the 0402 SMT components…
http://www.centerblack.com/
Hey guys, I need a recommendation for a wifi/rf chip. It has to be able to picked up for 1/2 a square mile and be able to read temperatures. RFID doesn’t seem to be possible for that range, would a wifi chip allow that range , also how small could you make something like that? I want to be able to record position and temperature back to a base station…and track things for half a square mile.
Is there anything like that that currently exists? I know maybe a GPS collar but Im thinking something the size of a matchbook that lasts months on a battery. or some kind of solar powered deal..
Any suggestions on a solution that might work?
Thanks
Is there a chip that allows me to track location and temperature of an object, that is the size of a matchbook or playing card and has a range of .5 mile square with a long battery life or a solar powered option solution? Using wifi or rfid or some other technology?
I need to track items in 1/2 square mile, with a chip that can do that.
Does anything like that exist?
Thx
Your footprint is wrong… the center pads are shifted…
There is a error in the Schematic file the eighth and seventh pin of the module are incorrectly connected.
It doesn’t need to have a 50 ohm trace, does it?
Right?