<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;">Pete and all,<div><br></div><div>Awesome note on QFN soldering. Given how popular this packaging is today, this is a very valuable information. I look forward to trying your techniques out with the power and accelerometer designs I am working on. </div><div><br></div><div>I also appreciate all the help you Allan and Jeff have provided on my antenna problems. </div><div><br></div><div><geek humor> I read<a href="http://nyti.ms/1kugCeQ"> this article </a>about scientists using quantum entangled electrons to communicate securely between two points. After all this education on RF, quantum teleportation is looking like it may be a less complicated approach to solving my transmission issues. </ geek humor></div><div><br></div><div>I am going to take a stab at building a proper dipole antenna and adding some of the components like a balun and isolation to survive a storm. I may even look to move to a slightly highs ISM frequency to avoid an potential regulatory issues and to make some of the antenna components smaller. If I ever do manage to turn this into a commercial product, I may try an industrial strength platform such as the one <a href="http://www.bb-elec.com/Tech-Support/Blogs/B-B-Insider-by-Mike-Fahrion/May-2014/A-Second-Warm-Welcome-to-Wireless-Mesh-Networking.aspx">suggested </a> y a vendor I recently met - B&B Electronics. </div><div><br></div><div><br></div><div>Thanks,</div><div><br></div><div>Chip</div><div><br></div><div><br><div><div>On May 31, 2014, at 10:49 AM, <a href="mailto:triembed-request@triembed.org">triembed-request@triembed.org</a> wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite">Send TriEmbed mailing list submissions to<br><span class="Apple-tab-span" style="white-space:pre"> </span><a href="mailto:triembed@triembed.org">triembed@triembed.org</a><br><br>To subscribe or unsubscribe via the World Wide Web, visit<br><span class="Apple-tab-span" style="white-space:pre"> </span>http://mail.triembed.org/mailman/listinfo/triembed_triembed.org<br>or, via email, send a message with subject or body 'help' to<br><span class="Apple-tab-span" style="white-space:pre"> </span>triembed-request@triembed.org<br><br>You can reach the person managing the list at<br><span class="Apple-tab-span" style="white-space:pre"> </span>triembed-owner@triembed.org<br><br>When replying, please edit your Subject line so it is more specific<br>than "Re: Contents of TriEmbed digest..."<br><br><br>Today's Topics:<br><br> 1. QFN/LGA soldering hazards (Pete Soper)<br> 2. Re: Di-Pole Dilemma (Pete Soper)<br><br><br>----------------------------------------------------------------------<br><br>Message: 1<br>Date: Fri, 30 May 2014 21:24:05 -0400<br>From: Pete Soper <pete@soper.us><br>To: Triangle Embedded Computing Discussion <TriEmbed@triembed.org><br>Subject: [TriEmbed] QFN/LGA soldering hazards<br>Message-ID: <53892F35.5000600@soper.us><br>Content-Type: text/plain; charset=ISO-8859-1; format=flowed<br><br>At last month's meeting I passed around an Adafruit QFN32 breakout board <br>with an Atmega328 soldered on it. I did this by hand with hot air and it <br>was "just like in the movies", with the solder holding the chip in <br>perfect alignment and the chip bouncing back into place when nudged to <br>one side before letting the solder cool. Perfect the first time, and <br>hugely easier than I expected. The rest of this is a comparison of the <br>Adafruit breakout board and two that I had made by OSH Park.<br><br>Here's why I expected it to be harder. I'd tried many times to solder a <br>QFN-16 accelerometer/magnetometer chip onto a board with a reflow oven, <br>hot air, dogs barking at the moon, etc. Out of several attempts I've got <br>two functional boards and several objects too small to be effective <br>paperweights or door stops. Actually, I will pull those chips back off <br>and try to reuse them.<br><br>And here's a photo of a board after I removed the chip following a <br>failed soldering attempt. The picture is edge-on showing the corner of <br>the chip pads and an outline of where the package sits:<br><br>http://triembed.org/images/badqfn16+chipoutline+text.jpg<br><br>The OSH Park soldermask is so thick there are "ledges" of it holding the <br>chip off the pads.<br><br>But that's nothing. Later I made some boards for a different <br>magnetometer that uses a "leadless chip carrier" package. This is <br>similar to a QFN but with this chip the pads on the underside of the <br>package are very narrow. After several attempts I only managed to get <br>one or two sides of the board soldered and the other sides sat above the <br>PCB pads by a tiny distance.<br><br>This picture of the board should make it obvious why the package pads <br>had trouble "kissing" the PCB:<br><br>http://triembed.org/images/badbadbad.jpg<br><br>As you can see, not only are there ledges on the corners, the PCB pads <br>are in "holes" of soldermask because the pads are so narrow there's room <br>for mask in between. Weirdest of all, but probably hard to see clearly, <br>is that the solder mask in the corners is hugely thicker than the mask <br>around the pads. It's as if there are two layers of mask. That's because <br>there is a ground plane around the perimeter: the "taller" mask layer is <br>sitting on copper. Had I not added that "pour" and just used a single <br>ground plane on the bottom layer I might have had an easier time.<br><br>And here is the Adafruit board. Notice how the pads are sitting well <br>above the surrounding board all the way around:<br><br>http://triembed.org/images/goodgoodgood.jpg<br><br>The two solutions to my QFN misery are:<br> 0) Don't use a library footprint that allows solder mask within the <br>package outline area of a QFN (or LCC or land grid array) package! Add a <br>tstop rectangle large enough to keep the chip from sitting on solder <br>mask, period.<br> 1) No copper pours into the corners of the chip footprint. A tkeepout <br>rectangle will prevent encroachment.<br><br>A few other precautions are called for. The router may put vias and/or <br>traces very close to the pad area, tempting the solder bridge monster. <br>Also watch out for misaligned wires that encroach on the space between <br>two pads. That is, a wire may go to a pad and sit diagonally or offset a <br>bit, so the edge of the wire is narrowing the gap between two pads. You <br>can suppress routing of vias and traces under the chip with trestrict <br>and brestrict rectangles (important for magnetometers!)<br><br>-Pete<br><br><br><br><br>------------------------------<br><br>Message: 2<br>Date: Sat, 31 May 2014 10:49:04 -0400<br>From: Pete Soper <pete@soper.us><br>To: triembed@triembed.org<br>Subject: Re: [TriEmbed] Di-Pole Dilemma<br>Message-ID: <5389EBE0.4030403@soper.us><br>Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"<br><br>I'd like to add a few details to Jeff's excellent info in this and his <br>previous post.<br>The penalty for cross-polarization (i.e. transmit and receive antennas <br>being "at right angles" wrt polarization) may be greater than 3db <br>because of the relative advantage (or absence of advantage) of ground <br>return. That is, with horizontal polarization the ground acts as a <br>reflector and often reinforces the field taking a direct path to/from <br>the antenna.<br>The lack of a decent RF ground is surely at the root of some <br>disappointment with vertical antenna arrangements, just as Jeff said. <br>Because the RF ground is effectively the "other leg" of a vertical (i.e. <br>vertically polarized) antenna system, if it's wimpy the efficiency of <br>the antenna system can be pitiful. Arranging a decent ground plane for <br>the "piece of wire" (ostensibly 1/4wl) vertical antenna should make a <br>difference. Running a feedline to an "outboard" monopole/ground plane <br>setup would also be totally valid and possibly very effective (except <br>for the lack of a "DC ground". See next bit).<br>A well designed dipole on the end of a decent piece of coax cable that <br>is properly connected to the radio could make a huge difference for a <br>number of reasons. This is going to sound weird, but in my opinion the <br>really huge difference is that a DC ground connection to the antenna <br>could be arranged such that the coax shield can act as a "lightning <br>rod", with a decent connection to earth ground. This might enable to <br>antenna system to up several feet and still have a rat's chance of <br>surviving a season of southern thunderstorms.<br>But you can't just stick coax on one end and solder the other two pieces <br>of wire and get a dipole antenna that performs properly. The RC plane <br>example and example where the guy uses the mouth-watering good Agilent <br>test instrument are surely improvements, but without a balun or <br>equivalent matching network on the antenna side what you really have <br>with those two examples is a radiating feedline that's part of the <br>antenna system. Worse, the length of the coax feedline gets involved in <br>the tuning of the antenna. This can drive a person to distraction in no <br>time flat. So forget about cutting those wires to the super-precise <br>lengths and expecting to be able to duplicate the match with resonant <br>frequency shown on those instruments: it would only happen by luck. <br>Also, *the gauge of the antenna wires is important*. So it's fine to say <br>"I cut these wires to 77.4mm", but if it's 22 gauge wire vs 14 gauge <br>wire that will make a big difference with respect to tuning the antenna <br>to the desired resonant frequency. Finally, he said "The length of each <br>side is 77.4mm." BZZZZZT! WRONG! That defines a "full wave dipole", not <br>a half wave dipole (the universal meaning if somebody just says "dipole" <br>is "half wave dipole"). A full wave dipole has a very high feedpoint <br>impedance. This tells me that by fantastic luck (or maybe design that <br>the forum poster didn't share for some reason), the feedline constitutes <br>a transmission line transformer that translates the 50ohm feed of the <br>RFM69 to the multi-thousand ohm impedance of the full wave dipole. Or <br>else he just wasn't thinking and quoted the length of both wires and so <br>each is really 77.4/2mm.<br>But back to the DC ground. It's possible to have the dipole be a solid <br>wire/rod that's connected directly to the coax braid and the coax center <br>conductor feeding a gamma match or something similar. This is dead <br>common with amateur radio yagi antennas and has the huge advantage that <br>the entire antenna is very well connected to the feedline coax braid and <br>the transceiver and of that feedline can have its braid well connected <br>to a copper ground rod. (The ground rod does not have to be pounded <br>vertically into the ground. Burying it horizontally a few inches under <br>the soil surface is effective.) Finally, a gas discharge tube/capacitor <br>arrangement could be used to isolate the radio from the feedline in the <br>case of a storm-induced high potential, shunting the energy back into <br>the ground path before it can get in an destroy the radio chip.<br>Yep, this sounds totally off the wall to a casual hobby person. But have <br>you ever noticed that folks are not climbing all over cell towers after <br>every area thunderstorms? Likewise, have you noticed TV stations rarely <br>go off the air during storms despite the fact that their towers are <br>taking direct lightning strikes frequently during the storm? That's <br>'cause the equipment is designed to handle storm situations. So, as I <br>imagine Chip having 100 or 500 trail counters spread all over this part <br>of the state I want to imagine them all humming along vs 5-10 of them <br>being blown to bits every week or two.<br>The catch-22 is that if the trail counter antenna can be kept on the <br>ground, preferably right next to a low lying creek bed, the equipment <br>might be eternal, but communication may be a real problem.<br>But another approach might be to focus the time and money into the <br>relatively small number of data collection systems. With those you could <br>use a (DC-ground), high gain yagi (cocked 90 degrees to maintain <br>vertical polarization) that could be aimed at the various trail <br>counters. Or if you want to go nuts, use a parabolic dish. This would be <br>big at 915mhz, but could provide 20-35db gain that might compensate for <br>the terrible performance of the logger radios. On the other hand, aiming <br>a dish would probably require some specific equipment 'cause the <br>beamwidth at those gains is razor-thin.<br>As for induction of RF back into the sensor wires, one could use <br>relatively cheap shielded cable, possibly in combination with ferrite <br>filters, and I'm confident that would eliminate that issue. But it's <br>very useful for Jeff to point that out, as this is the kind of problem <br>can be very subtle and initially mystifying. The noise margin for I2C <br>has got to be close to pitiful as it is.<br>Gotta go do family stuff, but I thought I'd share some notes.<br><br><br>On 05/28/2014 09:51 PM, Jeff Highsmith wrote:<br><blockquote type="cite">Chip,<br><br>Four to six feet could be enough, but the forum post where the user <br>mentions 1.5 mile range at similar antenna height states that that was <br>line-of-sight, and he only got 900 feet through "thick woods" (though <br>at 900MHz). Given the choice, I'd accept a little transmission line <br>loss in exchange for increased height, especially in rough terrain. <br>You don't want a big chunk of ground (dirt-wise and electrically) or <br>wet foliage between your antennas. Luckily, height is an easy variable <br>to experiment with.<br><br>Another popular antenna option is the twin-lead J-Pole. It does not <br>require a reflecting ground plane, so you can hang it high without <br>worry. For your application, it'd be about 17" long:<br>http://www.qsl.net/wa3yxk/jpole.html<br><br>If your making your own antennas, you might want to seek out the use <br>of an SWR analyzer (ask at local ham club) to help you quickly dual in <br>the length.<br><br>Devil's Advocate bit here: I see what you mean about directional <br>antennas being more work to setup, as you'll be aiming them with a map <br>and compass. The gains from yagis could be significant, though (around <br>12dB, depending on the number of elements). Are you trying to hide the <br>sensor nodes from view? Is it a mesh network topology or is there one <br>home station receiving multiple sensor nodes. Directionality would be <br>ruled out if you need to tx/rx different directions.<br><br>One more thing to keep in mind will be RF interference. This is <br>getting beyond my experience, but my understanding is that if the wire <br>between the sensor and the node is an effective antenna at your <br>frequency, you might get strange issues while transmitting.<br><br>Jeff :)<br><br><br>On May 28, 2014, at 11:22, Charles McClelland <chip@mcclellands.org <br><mailto:chip@mcclellands.org>> wrote:<br><br><blockquote type="cite">Jeff,<br><br>Thanks for the response, very helpful. I had not considered your <br>point about the rubber ducky - I doubt my small sensor node will <br>provide a very good or very large ground plane.<br><br>As for height, I guess that getting the antenna off the ground would <br>be helpful. My sensor is i2C based so I don't want to have too long <br>a wire between the Moteino and the sensor. At the same time, my <br>understanding is that the longer the length of the cable that <br>connects the board to the antenna the more losses there. My first <br>attempt will put the antenna close to the board (3-6") and the sensor <br>6' from the Moteino, If I place this well, that would place the <br>antenna 4'-6' off the ground. Does that sound reasonable?<br><br>Based on your answer, it seems like the first option below may be <br>preferable as it will include a full quarter wavelength ground and <br>emitter arm and it could be "tuned" by trimming the wires. Agree?<br><br>Thanks,<br><br>Chip<br><br><br>On May 28, 2014, at 10:17 AM, Jeff Highsmith <jeff@jeffhighsmith.com <br><mailto:jeff@jeffhighsmith.com>> wrote:<br><br><blockquote type="cite">Chip,<br><br>Forgive me for my rusty RF theory, but my understanding is:<br>. Dipoles are still somewhat directional, just not as much as a <br>"Directional" antenna like a Yagi. Your best radiation and reception <br>is at 90 degrees to the wire. In other words, the performance drops <br>in the directions that the tips of the wires point. If you are <br>vertically polarized, this isn't as much of an issue, since you are <br>unlikely to be above the tip of the antenna.<br>. Polarization matters. You lose 3dB or so when switching <br>polarizations, so make all your antennas horizontal or make all your <br>antennas vertical.<br>. So-called "Rubber Ducky" dipoles rely on the radio's case or, more <br>frequently, the human body holding the case to be the other side of <br>the dipole. If you don't have a good ground like that under your <br>rubber ducky, it won't do as well.<br><br>At what height relative to the ground are you placing your antennas?<br><br>I hope that helps. Others, feel free to correct me-it's been awhile <br>since my ham days.<br><br>Jeff :)<br>WJ3FF<br><br><br>On May 28, 2014, at 9:42, Charles McClelland <chip@mcclellands.org <br><mailto:chip@mcclellands.org>> wrote:<br><br><blockquote type="cite">To all,<br><br>I am playing with the Moteino and recently tested the range at a <br>local park. I was a bit disappointed and discussed this at the <br>last meeting. As with all things embedded, there seems to be a <br>number of things that can be done to improve performance both <br>hardware and software. For this note, I wanted to ask for any <br>advice on the hardware front. Once I have that set, I plan to <br>tweak the software settings (primarily data rate but some others as <br>well) to see how much range I can get - ideally 1.5 miles outdoors <br>through wooded terrain.<br><br>The Environment:<br>- Moteino <http://lowpowerlab.com/moteino/> base and sensor nodes<br>- I went with the lowest frequency transmitter <br><http://lowpowerlab.com/blog/2013/06/20/rfm69-library/> - 433MHz - <br>and the highest power<br>- The Moteino site states that there is not much improvement from <br>monopole antennas <http://lowpowerlab.com/moteino/#antennas><br>- In the forums however, there is a posting promising greater range <br>from dipole antennas <br><https://lowpowerlab.com/forum/index.php/topic,112.msg288.html><br><br>I understand that a directional antenna would be better but my <br>application would not easily support this as the sensors nodes may <br>need to be moved frequently.<br><br>My limited understanding of a dipole antenna theory is that there <br>are two elements - one with signal and one with ground with each <br>sized to match the specific frequency. I plan to add an SMA <br>connector to the board so I can install a commercial antenna and I <br>found two options that both claim to be sized to the 433MHz frequency:<br>- The first <br><http://fpvlab.com/forums/showthread.php?3156-433MHz-Half-wave-dipole-for-LRS> looks <br>like what I expected a dipole antenna to look like and could be <br>easily made<br>- The second <br><http://www.taoglas.com/images/product_images/original_images/TI.15.3113%20433MHz%20Connector%20Mount%20Antenna%20290909.pdf> would <br>be much better for packaging and use on a sensor that may be moved <br>often - but it does not look like a dipole antenna to me.<br><br>Any advice on which might be the better choice or whether I should <br>look at other options?<br><br>Thanks,<br><br>Chip<br><br>_______________________________________________<br>Triangle, NC Embedded Computing mailing list<br>TriEmbed@triembed.org <mailto:TriEmbed@triembed.org><br>http://mail.triembed.org/mailman/listinfo/triembed_triembed.org<br>TriEmbed web site: http://TriEmbed.org <http://triembed.org/><br></blockquote></blockquote><br></blockquote><br><br>_______________________________________________<br>Triangle, NC Embedded Computing mailing list<br>TriEmbed@triembed.org<br>http://mail.triembed.org/mailman/listinfo/triembed_triembed.org<br>TriEmbed web site: http://TriEmbed.org<br></blockquote><br>-------------- next part --------------<br>An HTML attachment was scrubbed...<br>URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20140531/c62dd36a/attachment.html><br><br>------------------------------<br><br>Subject: Digest Footer<br><br>_______________________________________________<br>TriEmbed mailing list<br>TriEmbed@triembed.org<br>http://mail.triembed.org/mailman/listinfo/triembed_triembed.org<br><br><br>------------------------------<br><br>End of TriEmbed Digest, Vol 12, Issue 25<br>****************************************<br></blockquote></div><br></div></body></html>