Speaking of Small Packages...

T'was a a dark and stormy night when the news came through. Joe Layout had been both dreading and preparing for years. But it had always been little more than rhumors from a far off land. It was a looming threat, always dancing in the distance, but never quite real.

Until now. 1.27mm, 1.0mm, 0.8mm, 0.5mm, 0.4mm... and now... drum roll please 0.3mm pitch. I just got Shrinking BGA pitchan email announcing an Amkor 8 x 8mm 368 ball BGA at 0.3mm pitch. Yikes.

There's still some controversy over the best way to make a 0.4mm pitch BGA land pattern. Some say says you need to use solder mask defined pads. Some say you still need to use the non-solder mask defined pads. Now we throw something 25% smaller into the mix.

The image isn't to exact actual scale - because I don't know how big your monitor is - but the parts are in relative scale from 1.27 pitch to 0.3 pitch.

Duane Benson
If you can't see it, you shouldn't eat it

Reference Designators

A while back, I wrote about reference designators relative to family panels. Family panels can cause problems because often times, each individual circuit layout will have reference designators that start at the same place.

For example, circuit A, down in the lower right corner of the panel, will have resistors R1, R2, R3... Looking at the other three circuits on this hypothetical family panel, all of them also start their reference designators with R1, R2, R3... That's bad. It can lead to confusion and wrong parts on the board in the wrong spots. If we see this here at Screaming Circuits, we may spend some extra time and sort through it manually or we may ask you to fix it first. Fixing it here is a labor intensive and risky process. It's bad news.

Anyway, to the point of this post: In the original post, I listed one wrong way and three right ways. There are two other wrong ways not in the original post, which I'll list here.

Wrong way number one: R1-1, R1-2, R1-3. Bad. Most assembly software will interpret a dash as meaning a range. It will see "R1-3" as equalling "R1, R2, R3". That can be bad.

Wrong way number two: Leading zeros. Don't do "R1, R01, R001". The leading zeros are stripped and that can cause all of those the be seen as "R1". Just don't put leading zeros in your reference designators.

Duane Benson
Corrigan says Long Beach is actually in Ireland

IS IT VALUE? IS IT PROTO?

SMT Geiger PCB

Looks like it to me. I got a couple of nice PCBs from Sunstone here. The parts, except for the Geiger tubes are here from Digi-Key too. The tubes are someplace between the Ukraine and Canby. All I have left to do is kit it up and place the order here at Screaming Circuits. It may seem silly that I have to go through the effort of placing an order on our web site when I work here. But it's not.

Doing so does two things. It reminds me of what it's like to be a customer. Always a good thing. And, it doesn't disrupt the shop floor with something that is outside of our process.

In case you haven't been following, this is my SMT re-layout of the open source Geiger counter designed by Jeff Keyzer of MightyOhm.com.

Duane Benson
Tick. Tick. Tick. Tick. Tick. (actually, I hope not)

More Geigering

SMT Geiger 2nd I didn't get back to this at the same bat channel and same bat time. Sorry if anyone tuned in and found the Penguin instead. I think I'm ready now though. I did a bit more layout tweaking, moved the MCU bypass cap closer to the supply pins and added in a MAX3232 so that I won't need an external driver / transceiver board if I want data on my PC.

I'm calling it done. I have my gerbers in a .ZIP file, my centroid and the completed BOM . Next step is to get some PCBs fabbed up. I'm trying out the ValueProto service from our partner, Sunstone Circuits. As I said before, if, after assembly, this design actually works and counts Geigers, I'll post all of the design files as needs to be done with Open Source hardware.

Here we are at ValueProto. The PCB is 4" x 2.7". I'm not going to do a quick build and my zip is 97013. $57.40 for one. $34.90 each in quantity ten. That's all the questions I need to answer. I upload my .ZIP file, check a few boxes and the order is now placed. Next, I'll get some parts, kit it up and send it through Screaming Circuits. The NOS Soviet Geiger Mueller tube is one the way from the Ukraine.

I'm not sure what I'll use as a radiation source to test it out with once it's built. I hadn't really thought that far ahead. I'll have to come up with something.

If you're going to be at the Embedded Systems Conference in Boston next month (September 27, 28), we'll be there in booth 615, across the isle from Element14. Stop in and take a look at it. While you're there, ask for fabulous prizes and gifts. We have a smattering of Screaming Circuits shirts and flashlights for the asking.

Duane Benson
No baked beans, please

Open Source The Eskimo

Not long ago, I found myself in the vicinity of the former Trojan Nuclear Power Plant. I say former, because it's not there anymore. The reactor's long gone and the cooling tower was exploded a few years back. I was just under a mile away, and downwind, from the tower when it went down. The cooling tower didn't have exposure to radioactive material when it was in operation, but the dust cloud was still pretty annoying. Interestingly, prior to the nuclear power plant being there, the land was owned by the Trojan Powder Company, maker of explosives. I wonder if the blasting material was made by the Trojan Powder Company. Today, there's just a park and a few abandoned buildings left on the spot.

While I was there recently I pondered the possibility of small amounts of residual radiation in the area. I quickly forgot the thought until I ran across a post by Jeff Keyzer (mightyohm.com) about an open source geiger counter he designed.

SMT Geiger Well, who could resist the chance to build a geiger counter and run it around looking for bad stuff at the former sight of a nuclear power plant. Not me. Here at Screaming Circuits, we specialize in surface mount, so I thought it would make sense to re-layout the PCB for SMT components. Which I did. That's the cool thing about open source. (one of the cool things) The design files are accessible.

The only real challenge I had was in finding a few parts substitutions. The exact match wasn't available in SMT for some of the components. Hopefully, I picked good subs. Doing that is frequently not as easy as it might seem. Especially when the original design is not mine so I don't necessarily know which parameters are in the criticle path.

Now I have to get some parts from DigiKey or Element14, PCBs from Sunstone Circuits and I'll have to run over to Eastern Europe to pick up an old Soviet era geiger muller tube. If my layout actually works, I'll pack up the SMT files and BOM and make them available as all open source is supposed to be. If it doesn't work, I'll quietly try to disavow any knowledge of every having tried to mess with it.

Duane Benson
Everybody's building ships and boats
Except us. We're building prototypes.

 

Via Shifting

Here's an example of what via in pad can do for a small passive component. Other things can happen too, like tomstoning or twisting. But take a close look at this photo. In doing so, you'll note that both sides of  Small fillet passive via in pad the part are soldered down. Sure, it's shifted, but who really cares? It's electrically connected. Right?

In this case, much of the solder on the lower pad flowed into the via. This led to an imbalance in surface tension between the two pads which shifted the part. Some logic might say that since both ends of the part are soldered in and there aren't any shorts, it's all cool.

It is all cool because it's been out of the reflow oven for quite a while, but it's not cool because it's not good workmanship. The IPC created standard IPC-A-610 for just such an issue. Class I is the loosest. This might pass that. I'm not sure though because we don't do anything with Class I here at Screaming Circuits except reject it. Class II is the typical commercial type standard and this shall not pass that standard. Nor would this pass Class III, an even tighter workmanship standard for higher-reliability requirements.

That's the real issue: reliability. With a good, symmetrical solder joint, you not only have a good electrical connection, but you also have a reliable mechanical connection. It will resist flexing and thermal expansion stress. This one may not. Give it some good thermal cycles or bounce it around in a race car engine computer and you may find yourself sidelined.

The moral of the story is to keep those vias out of your pads; even with passive components. Or, put the vias there but fill and copper plate them at the board house.

Duane Benson
Balrogs in pad are bad too

Fun With Electrolytics

I was fiddling with one of my robot boards the other day - popping some passives on and off and checking out subs and alternate values. I was doing this on a couple of boards at the same time. Everything was going along fine until I started to do a power-on test. The first board was fine. The second one would briefly light the power indicator LED. It would start a full brightness and then fairly quickly fade out.

My first thought was that I had been too agressive with my soldering iron and had burnt something out. (who has already guessed what really happened?). Turns out, that wasn't the case. I put it aside and came back to it a few days later. This time, I gave it the finger test and discovered that my regulator was hot. Darn. Next, I found a hot tantalum cap. Nothing looked out of the ordinary/ I stared at it for a while. The + side was on the left in both parts and... The plus side was on the left in both parts. One was supposed to be on the right. Oops. The cap had a high enough voltage rating that it didn't blow up. It just pulled down the supply until the over-current protection in the regulator shut it down.

I've heard a number of folks recommend that you try and keep all of your polarized parts facing the same way. It's not always possible, but it can certainly reduce opportunities for errors like I made here.

Duane Benson
Left, right. Left, right. Left, right. Left, left. Left, right...

Family Reference

I've written a bit about reference designators here and there. There are a few more factors that we run into now and then. Take the family panel. In case you aren't familiar with the term, it means that you have several different designs laid out into in one panel, as opposed to multiple copies of the same design in one panel.

Using a family panel can be a convenient way to deal with a multi-board design and can sometimes save a bit of money. Just a caution, though. Make sure to check with your fab house first. Some don't like family panels and some won't separate them for you. If you do have them separated prior to assembly, either at the fab house or by you, then you don't have any reference designator worries.

S 065 My-9 600 If you leave them in the panel and wish to have them machine assembled, it can get a bit more complex though. "Why?", you say. I'll tell you why. Generally, most people start at "1" for each new design. i.e. "D1, D2, D3... R1, R2, R3..." If the boards go into the machine independently, that's no problem. However, if you send the panel into a smt assembly robot, it may very well see that as your board having multiple D1's, R1's, etc. That would be rejected as an error in most cases.

If you are using the family panel approach, don't restart your numbering when you move to another one of the designs that will be in the panel. Either continue on from the last number in the prior design, add in a hundred's, with each design getting a different hundred's number or add a unique suffix on each board.

  1. Wrong way: PCB1: "R1, R2, R3, R4, C1, C2". PCB2: "R1, R2, R3, R4, C1, C2".
  2. Right way: PCB1: "R1, R2, R3, R4, C1, C2". PCB2: "R5, R6, R7, R8, C3, C4".
  3. Right way: PCB1: "R101, R102, R103, R104, C101, C102". PCB2: "R201, R202, R203, R204, C201, C202".
  4. Right way: PCB1: "R1A, R2A, R3A, R4A, C1A, C2A". PCB2: "R1B, R2B, R3B, R4B, C1B, C2B".

There are a lot of ways to do this. Just make sure that no reference designators are repeated from one board design to the next. I prefer method #3 myself.

Duane Benson
Is it immediate or extended? Does it matter?

Via Current Capacity

Over on the Circuits Assembly blog, Michael asked a question about my Via in Pad Myth #5. He asked:

"I have a question about vias. I have seen charts on the current carrying capacity of traces, but what about vias?"

That's a good question. I've heard that you first need to know the thickness of the via wall. Then, once you know that, you can calculate the trace-width equivalent for the via by using the formula for the circumference (diameter X pi ). For whatever number that gives you, compare the closest smaller trace width.

Via cross section My related questions to all of you PCB fabrication gurus out there are:

Since vias are not created in the same way as the trace plating is, can that simple formula be used? While the trace copper is laminated onto a nice smooth PCB surface, the vias are typically created by deposition of copper dust in the via and then electroplating more copper. Then the surface finish is applied to all of the exposed metal. The via walls would generally be rougher than the flat substrate surface. Does that have an impact on the current capacity of a via?

Further, since airflow will be somewhat restricted in a via relative to a surface, should the via effective width be compared to an internal trace instead of an exposed surface trace? Should it be a compromise between the two?

If you look closely at this via cross I pulled from Wikipedia, you can see that the via wall looks to be thinner that the traces. You'll have to make sure that your board fab house can give you an accurate thickness of the via wall.

Duane Benson
If you know the via current capacity, can you calculate the past and future capacity?

Random Via-In-Pad Myth #5

Myth #5: When you need thermal vias, more is better, bigger is better

Hmmm. Logically, this would seem to be the case. There are limits though; especially if you want a reliably assembled product. Older parts with heat slugs easily accessible for bolting on heat sinks didn't have this issue. Just bolt on a piece of metal and maybe blow a fan across it. It's different with a lot of the new, Padinvia smaller surface mount packages. Many have a heat slug on the bottom which requires carefully placed thermal vias to a copper pad on the underside of the board.

An extreme case of flooding the land with vias can be seen in this illustration here on Padinvia_alt the left. In terms of assembly, you can hack this together for a prototype, but it'll never fly in a production environment.It would be much better to use fewer smaller vias and have the center land covered with solder mask except where the metal on the chip is exposed, as in the illustration on the right.

Duane Benson
Place one carrot seed in each via and cover it with planting soil