Sunday, March 1, 2015

Solving pressure and condensation build up in your Embedded System.

Have you ever had a problem with pressure build up or condensation in one of your embedded devices? Gore-Tex Vents are a good solution today, but I thought you might find The Rest Of The Story interesting. What follows is slightly edited version of a message exchange of mine from the gEDA-User mailing list.

> Dave McGuire wrote:
> > Bob Paddock wrote:
> > I'd put them in a sealed box, with a Gore-Tex Vent so that the
> > enclosure can 'breath' but not pass water.
> This is an interesting idea. Can Gore-Tex be found in small
> squares for this type of application, or would one be stuck
> destroying an expensive jacket to get some?

Salvation Army or Good Will would be a good place to look if you want to go the clothing route, but there is more to the story.

The whole story goes like this: Gore-Tex was invented ~1978, used in clothing as everyone knows, and didn't really find many other uses then. Jump forward to 1982. I was designing a hand held control to run some 50 Ton Coal Mining Equipment. The control was a sealed box with a membrane switch on the front. We very shortly ran into problems. Taking it down into the Mine would cause a pressure reduction that would suck the switches in, activating the switches. Having a stuck switch on a 50 Ton machine with sharp cutting bits is *bad*. Very **bad**. Also found that taking it up in a plane, as non-pressurized luggage, would deform the switch by causing it to balloon out to about four times what it should be. From flat to nice dome. You then had worthless junk, it did not recover.

I had recently read about the properties of Gore-Tex, tracked down one of the engineers in the factory and asked him if he thought it would make a good vent for such an application. He said he had no idea, but he would send me several different types of the stuff to try out. Which he did.

Putting cloth over a whole in mining equipment would last a few minutes, on on optimistic day. So a colleague of mine, Don F., came up with this labyrinth sandwich to put the Gore-Tex between.

Take two flat disks, we used thick fiberglass, each about 1/4" thick. Mill out a pocket in both disks, place them face to face, then drill a hole through both disks at one of the ends of the milled circle. Now rotate a single disk 180 degrees. Put the Gore-Tex between the disks and epoxy. If you try to stick your Sharp Pointy Coal Mining Implement into the hole you hit the fiberglass and not the Gore-Tex. The assembly was then held in the box by a screw in the four corners. Pressure problem was solved.

Went back to the fellow at Gore-Tex to get more material, which he supplied. Thought our idea was a good one and filed for the patent. So he got it and Don and I did not.

Today you can buy Gore-Tex pressure relief vent off-the-shelf. Most a screw in type plug, but they come in all kinds of sizes.

Q: Why does my FET, IGBT, SCR keep going up in smoke? A: Spot Heating

A common problem I see in message boards is that someone's circuit failed, and went up in smoke. People in the more esoteric realm's blame this on things like "Subtle Energy" overload and other such minutia. Here is the far more realistic explanation:

The very old "GE SCR Manual" goes into all of the Gorey details of what is happening inside the part, when the "Magick smoke comes out", as it is unlikely you have the Manual at hand, in a nut shell:

What lets the Magick Smoke out of IGBTS, FETS and SCRs in most cases is turn them on to slowly, causing 'Spot Heating' of the die.

Think of a FET as hundreds of thousands, possibly millions, of very small resistors all in parallel, where each one can be turned on and off individually. The 'resistors' closest to the gate turn on first, and as the gate potential spreads across the die the rest turn on. The ones farthest from the gate turn on last.

With a slow gate turn on, a few of the small resistors nearest the gate are trying to carry all of the load, which they can't do, so they burn up, but the device does not fail quite yet. The next time the device is turned on, which may be only milliseconds away depending on your switching frequency, or days away depending on the application, some more of the resistors further in burn up. When the point is reached that there is simply not enough of the 'resistors' left to carry the load is when the Magick Smoke escapes, and the part dies a catastrophic death.

This is why the parts generally run "for a while" before failing. If it fails as soon as you fire it up the first time, you either had a catastrophic short in the load, possibly shorted caps that take a bit of time to 'wake up' before they hold a charge, generally fixed with 'Soft Start', or the gate drive really sucked big time.

There needs to a be a few *amps* of current pumped in the gate of the larger parts, for short periods of time, to get the gate potential to spread across the entire die as fast as possible.

You also want to get the thing turned off as fast as possible.

If you are not familiar with the concept of Magick Smoke, this is where all electronic parts run on Magick Smoke, because once the smoke comes out of the part, it no longer runs...

Saturday, February 28, 2015

Is there a rule of thumb for estimating the cost of getting circuit boards assembled?

A reoccurring theme I see on message boards is why does it cost so much to have some electronic widgets manufactured. Here is some background for you that might help explain that.

Is there a rule of thumb for estimating the cost of getting circuit boards assembled?

In a past life I worked for a large Contract Manufacturer, Matric Limited . I don't mean this to a plug for them, but the view of the place is helpful for the discussion.

To a CM it is all about *Time*. When it comes to parts, the actually part cost is really insignificant as far as cost contribution to assembly cost goes. Most of the cost goes to the time it takes to setup and tear down.

For a broad brush overview of cost steps:

One shot fee for getting your project into the system. Someone has to enter your Bill of Materials (BOM), and schedule into the amorphous blob known as "The System". Any change that you do triggers a recalculation, that you either pay for or is amortized across your boards. Every future order you place will have a small "trigger fee" to pay for someone to enter your order.

Included in that is a fee for someone to do a time analysis of the number of operations that your project will require. A unit time value is assigned to each operation, and each operation has a cost, that is, as far as I know, calculated by Magick (All CM's use Magick for this step to my knowledge).

If you supply the parts there will be fees for entering a carrying fee per new part number into The System. They will also charge higher fees if you send them parts that require extra steps to handle such as reels of parts without leaders etc. Some cost analysis guru at GM, long ago, decided to simply have a number in The System carries a charge of $50 or so per year per part number. The accountants just love to beat up the engineering department for "we have to many parts in the warehouse". Company owner wants to keep inventory turnover high. Also cost for physically getting your parts into The System, such has putting them in the warehouse, typing in the data etc.

There will be a scrapping fee to get your stuff out of The System if you take your project someplace else.

Those Non-Recurring Engineering (NRE) fees you either pay up front, or are amortized across the number of boards. This is why the range is highly variable between different CMs. Some hide the fees, some don't.

Also, when you supply the parts the price of each part will be market up by a *minimum* of 33% (More Guru calculations). If you don't mark the price up by this amount, you lose money each time you touch the part. You are charged for the use of the warehouse space, like renting a storage unit.

Now lets say you let the CM supply the parts, in general this will get you a lower per part cost for the commodity parts. As they will be using 100,000 0.1 uF 0603 caps a day, the pick and place machine will have that loaded. So you don't have to pay for loading your reel of much smaller volume part. Also the CM will have negotiated a much better price than you got from DigiKey. The downside here is that you lose some measure of control, which can be a problem if you have to meet UL/MSHA/FDA etc. regulations.

There are extra charges for projects that involve FDA paperwork, such a per lot tracking etc. Other acronyms apply as well, UL, FCC etc.

There is a fee for having the solder paste stencil made.

Now that the NRE's are out of the way, lets build a thousand Widgets.

Someone answers the phone and enters an order into The System to build a thousand Widgets.

The System checks the warehouse to see what parts are in stock. Your order is then routed to Purchasing to get the parts that are out of stock, or routed to planning to get your order into the build Que.

When your build hits the top of the Que:

Someone pulls the parts from the warehouse, at the minimum your PCB; time.

Your bare boards are put in an oven and baked to drive out any moisture, you pay the handling and electricity; time.

Your parts are loaded on the Pick-and-Place Machine; time.

The board go from the oven to SMT Assembly; time.

Someone pulls your stencil out of the rack and puts it in the paste machine; time. Paste is applied to your board; time and paste costs.

Your board is put into the Pick-and-Place and your parts are mounted; paste, electricity and time.

Your board then goes through the IR reflow oven for soldering; electricity and time.

The boards are then cleaned; fluids and time.

Any of your parts left on the P-and-P are removed, and put back in the warehouse, when Widget #1000 comes out the end of the machine; time.

The stencil is cleaned. You pay for whatever the cleaning fluid is and time.

The stencil is put back in the rack; time.

If your boards are in a array, they are then cut apart. It is cheaper to build arrays, but it adds this cutting fee; time.

If there are parts that could not be mounted in the P-and-P Machine they are done by hand, or put through the wave solder machine, then cleaned a second time. There is a big hit in costs for anything done by hand such as connectors, transformers, cable assemblies etc. Time.

The boards then go to Quality Control for the level of inspection that you paid for. Simple visual to full functional test. Time.

Boards leave QC and go to shipping where they are put in Anti-Static bags and cardboard boxes and shipped off to you. Time. You pay the shipping one way or the other.

There could also be Added Value items such as your boards are put in an enclosure. You pay for someone to do it, right down to the number of seconds it takes to tighten down the screws.

By now you probably have gotten the idea that Time is important. When you were looking up stuff in the DigiKey catalog were you billing yourself the time it took to do it? Probably not...

The 500 piece cost for the electronic parts from Digikey is about $4.20

Did you include the Anti-Static Bag, the yellow Anti-Static Sticker that seals the bag, the solder (price of metals is going up every day), and any board cleaning fluid chemicals / deionized-water in that price of $4.20, and the time to do those items? I didn't think so...

A good CM knows the cost of every operation and will be around a long time. A new CM doesn't know his costs. Hence the wide variation in CM quotes.

Matric developed a reputation for being a high price CM, and customers would leave based on cost, rather than value. However many of them would return after a while saying "we got what we paid for", and never left again.

In the end my advice is to analyze the value of the services you are paying for, not the cost of the parts.

Tuesday, July 1, 2014

Real Time Cyber Attack Map. 'War Games' comes to life

If you wonder why your Internet has been getting slower and slower, it is due to the escalating Cyber War going on.

Norse Corporation, to showcase their technology, now lets us watch the war in real time, just like at the end of the movie War Games:

The Map shows the Attacker, the port they are attacking, and the geophysical location being attacked.

The U.S. and China keep trading places for who is the top attacker.

One of the *Attackers* I find troubling is Merck & Co. in Woodbridge New Jersey. They are attacking the SNMP Port 161, which is used to monitor network connected devices, as 19:57 EDT July 1st 2014.

There are three possibilities here:

1) Merck & Co. is actively doing the attacking. Leading to the question 'Why?'.

2) Their system is compromised, and is being used by a third party to do the attacking.

If it is compromised how do you know that their products and vaccines have not been adulterated by someone with malevolent intent, messing with production line equipment settings or worse?

In their defense let us assume they know better than to have production equipment open to the Internet, and this is some office computer than has been taken over.

3) The mapping technology is flawed.

Going back to possiblity #2:

The Sohdan Device Search Engine shows (for a fee) what devices are found on Internet, that probably should not be, or are open to the public when their owner does not know it, such as Web Cams. Are you sure your own devices are secured?.

Sunday, March 23, 2014

U.S. Aims To Give Up Control Over Internet Administration

TIA Announces Intent to Transition Key Internet Domain Name Functions



Domain Name System

Printer-friendly version
March 14, 2014
News Media Contact:
NTIA, Office of Public Affairs, (202) 482-7002,

WASHINGTON - To support and enhance the multistakeholder model of
Internet policymaking and governance, the U.S. Commerce Department's
National Telecommunications and Information Administration (NTIA)
today announces its intent to transition key Internet domain name
functions to the global multistakeholder community.  As the first
step, NTIA is asking the Internet Corporation for Assigned Names and
Numbers (ICANN) to convene global stakeholders to develop a proposal
to transition the current role played by NTIA in the coordination of
the Internet's domain name system (DNS).

NTIA's responsibility includes the procedural role of administering
changes to the authoritative root zone file - the database containing
the lists of names and addresses of all top-level domains - as well as
serving as the historic steward of the DNS.  NTIA currently contracts
with ICANN to carry out the Internet Assigned Numbers Authority (IANA)
functions and has a Cooperative Agreement with Verisign under which it
performs related root zone management functions.  Transitioning NTIA
out of its role marks the final phase of the privatization of the DNS
as outlined by the U.S. Government in 1997.

"The timing is right to start the transition process," said Assistant
Secretary of Commerce for Communications and Information Lawrence E.
Strickling.  "We look forward to ICANN convening stakeholders across
the global Internet community to craft an appropriate transition

ICANN is uniquely positioned, as both the current IANA functions
contractor and the global coordinator for the DNS, as the appropriate
party to convene the multistakeholder process to develop the
transition plan.  NTIA has informed ICANN that it expects that in the
development of the proposal, ICANN will work collaboratively with the
directly affected parties, including the Internet Engineering Task
Force (IETF), the Internet Architecture Board (IAB), the Internet
Society (ISOC), the Regional Internet Registries (RIRs), top level
domain name operators, VeriSign, and other interested global

NTIA has communicated to ICANN that the transition proposal must have
broad community support and address the following four principles:

Support and enhance the multistakeholder model;
Maintain the security, stability, and resiliency of the Internet DNS;
Meet the needs and expectation of the global customers and partners of
the IANA services; and,
Maintain the openness of the Internet.

Consistent with the clear policy expressed in bipartisan resolutions
of the U.S. Senate and House of Representatives (S.Con.Res.50 and
H.Con.Res.127), which affirmed the United States support for the
multistakeholder model of Internet governance, NTIA will not accept a
proposal that replaces the NTIA role with a government-led or an
inter-governmental organization solution.

From the inception of ICANN, the U.S. Government and Internet
stakeholders envisioned that the U.S. role in the IANA functions would
be temporary.  The Commerce Department's June 10, 1998 Statement of
Policy stated that the U.S. Government "is committed to a transition
that will allow the private sector to take leadership for DNS
management."  ICANN as an organization has matured and taken steps in
recent years to improve its accountability and transparency and its
technical competence.  At the same time, international support
continues to grow for the multistakeholder model of Internet
governance as evidenced by the continued success of the Internet
Governance Forum and the resilient stewardship of the various Internet

While stakeholders work through the ICANN-convened process to develop
a transition proposal, NTIA's current role will remain unchanged.  The
current IANA functions contract expires September 30, 2015.

For further information see: IANA Functions and Related Root Zone
Management Transition Questions and Answers

About NTIA

NTIA is the Executive Branch agency that advises the President on
telecommunications and information policy issues. NTIA's programs and
policymaking focus largely on expanding broadband Internet access and
adoption in America, expanding the use of spectrum by all users, and
ensuring that the Internet remains an engine for continued innovation
and economic growth. To find out more about NTIA, visit

While the US Government clearly has issues, do you think other countries will be as supportive of any open system like Internet is today? What will happen to the Internet if control of the root domains end up in the hands of the UN or an organization where each country gets a single vote on governance?

What is truly needed is a decentralized Domain Name System. Namecoine is being considered as one alternative. Are there others?

Tech Companies Say Better to Import More Workers Than Retrain Experienced Ones

"WASHINGTON, March 19,2014 /PRNewswire-USNewswire/ -- Addressing a media conference call today, Scott Corley, executive director of Compete America, asserted that the large high-tech companies he represents would rather bring in more H-1B temporary workers than retrain experienced American employees."


Corley replied: "If it could be done as easily, there would be less value in the worker. ... You're saying it's easy to be trained into these fields, but if that were true, there would be no value; they wouldn't be high-paying jobs. You would be able to find them anywhere."

"Mr. Corley made clear companies would rather use the H-1B visa to hire younger, cheaper temporary workers," IEEE-USA President Gary Blank said.



See also You'd Rather Import Than Retrain. Why? by Carolyn Mathas.

People wonder why the younger generation has no interest in getting in to the STEM fields. They are not stupid, they read garbage like the above just like you and I can. Why pursue such an education when the jobs only go to the cheapest people?

Saturday, February 15, 2014

Does 'Drop in replacement' strike fear into your Soul? It should...

I have been using IrDA (Infrared Optical transceivers) for several years for setting parameters and doing software updates in my designs. Takes far less space and power that WiFi or Bluetooth.

There was once many people making IrDA transceivers, back when they were used in Cell Phones before the era of Bluetooth, now there are only a couple of manufactures left. Once I had three sources for a footprint compatible part with the HP3003 that I originally started with. Over the years the only remaining footprint compatible part was the Vishay TFDU4300.

Sadly the day came when even the 4300 was discontinued, however I was assured that I was not going to have to do a new board spin, "you can drop our new TFDU4301 right in". While true that it was footprint compatible, it was not software compatible. The 4300 would not receive its own output and would drive its Rx line high during transmit. The 4301 echos the transmitted data, meaning all of the bootloaders had to be updated to turn off the receiver when transmitting. Annoying, but at least that was documented.

What was not documented is shown in the scope captures below. The first one shows the discontinued 4300 and the second shows the new 'drop in compatible replacement'. Does it look compatible to you?

Sadly this problem was discovered while I was far away trying to get my wife help with Intracranial Hypotension, more commonly known as a Cerebrospinal Fluid (CSF) Leaks, to stop her excruciating headache resulted from cerebrospinal fluid leakage. I made the mistake of putting my job first and we came home to Pittsburgh, when we should have stayed at Duke in North Carolina. This is one of the small events that accumulated to result in my wife's suicide.

Now does 'drop in replacement' strike fear into your Soul too?

Thursday, January 2, 2014

Internet of Things (IoT) possibly Carcinogenic to Humans

An obscure part of the World Health Organization, the International Agency for Research of Cancer (IARC), published an even more obscure paper: IARC Classifies Radiofrequency [RF] Electromagnetic Fields as Possibly Carcinogenic to humans in 2011.

I bring this up as Epidemiology just published (January 2014 - Volume 25 - Issue 1 - p 23-27;doi: 10.1097/EDE.0000000000000028) Commentary: Mobile Phones and Cancer: Next Steps After the 2011 IARC Review. I've covered this area before so I'll refer you there rather than repeat it here. See also Something Is Rotten in Denmark: Danish Cancer Society Plays Games with Brain Cancer Rates and IARC Monographs on the Evaluation of Carcinogenic Risks to Humans.

Something else bothers me about the push for the (unneeded?) Internet of Things (IoT). Lots of places talk of or show how to power the low power sensors and such from 'ambient' RF power taken from the environment, such as a local TV or Radio Station. At what point does the source transmitter get overloaded with the million(s) of IoT devices drawing power from it? What do the owners of these transmitters think of this? Isn't this technically illegal under the same laws that prevent farmers from drawing power from the power company via inductive coupling to their mile(s) long electric fence wires?

Ponder this, as related to many receivers drawing from a single transmitter: If I strike a tuning fork in a room of a 1000 identical tuning forks what happens? Leave your answer in the comments.