μblog: engineering from the trenches

While looking up some papers on statistics I came across this freely-available copy of William Gosset’s March 1908 paper, almost a year before the its United States copyright expires and it becomes public domain. The paper is a very nice transcription of the original paper (available through JSTOR) which makes it much easier to read.

The point of the paper relates a statistical confidence limit and a set of bounds within which the mean of a set of measurements must lie. To be more specific, the paper addresses measurements whose distribution is Gaussian and goes on to specify how small a error bound you can place around the mean given a finite number of experiments with a finite certainty, say 95%. Furthermore, the usefulness of the paper comes from the fact that the limit is examined with a decreasing number of observations. This is useful since certain experiments cannot be performed many times and we need to be able to say how certain we are of the mean of the observed Gaussian random variable given the finite number of data points.

In a small side note, the p < 0.05 (or >95% “certainty”) is often considered to be a “good” value, but it may seem somewhat arbitrary. There are some people who attribute this to Karl Fisher, more specifically, his publication titled Statistical Tables for Agricultural, Biological and Medical Research (3.3MB). For those looking for Fisher’s historical papers, they can be found at the University of Adelaide.

( 1908student-the-probable-error-of-a-mean.pdf )

del.icio.us | digg

This IC Friday focuses on a BlueCore4-EXT bluetooth chip from CSR. This is the first IC Friday chip that has some extraneous designs on the die featuring some sort of dinosaur and a man in a hat. If anyone has specific knowledge of what they are, please let me know.

del.icio.us | digg

This is a continuation of the learning for free on the internet series, this time with reading. I have had the misfortune of having to study quantum mechanics on a graduate level, after getting an undergraduate in computer engineering, from Introductory Quantum Mechanics by Liboff (Amazon). This book was tedious to say the least and having to study from it to pass a qualifier was even worse. Fortunately, there are people, such as Doron Cohen, who can write coherently and choose to publish their lectures on quantum mechanics on line. Although it may seem that quantum mechanics is not important to the engineer, one can remember that a bipolar-junction transistor cannot be properly described without a little bit of quantum mechanics. Many designs that push performance limits will also find themselves involved with the laws of quantum mechanics. Lastly, when the laws that govern quantum mechanics are extrapolated to the classical world, the result of an argument can become a superposition of states and otherwise reasonable people can be driven to madness.

( lecture-notes-in-quantum-mechanics.pdf )

del.icio.us | digg

In a follow up to a previous post regarding using operational amplifiers as comparators, here is an application note from National Semi on one of their comparators.  Although some specific comparators are addressed, the document is 24 pages long with some in depth discussion of design practices.

( an-74.pdf )

del.icio.us | digg

I decided to upgrade my Razr V3C to a silver Motorola Q by way of Craigslist.  After activating the phone initially, I had some problems with making phone calls. I was able to browse the web using EVDO any time, but I was unable to make any outgoing or receive incoming phone calls. On the outgoing calls, I wouln’t even hear ringing. After a some times, I dialled 911 to see what would happen assuming that the phone would be designed make every possible attempt to make that call go through. Surprisingly, I heard a few rings and quickly hung up before connecting with 911. The icon at the top of the phone indicated CDMA (1X) mode and every subsequent outbound call went through without problems. I figured that the issue was with the CDMA/EVDO switching so I did some searches and used the following steps to correct the problem:

• dial ‘##073887*’ and hit the send button
• enter ‘000000′ for the security code (don’t forget to hold the shift key to enter ‘0′ instead of ‘?’)
• you are now inside the programming menu!
• go to ‘G Test Mode’
• change the status to ‘Enabled’
• hit ‘Back’ then ‘Exit’ and then you can end the call
• dial ‘##*’ and hit the send button
• you are now in the field test mode!
• select ‘B Field Test Menu’
• select ‘E HDR Preference’ and make sure that ‘Automatic’ is selected and hit back
• select ‘F HDR Hybrid’ and make sure that ‘On’ is selected

This will make sure that the phone will switch between EVDO and CDMA depending on the task at hand. Unlike the RAZR, the USB driver for the phone is Microsoft ActiveSync, so that is what needs to be installed on a Windows machine to recharge the device. I will get a few more things posted about this phone after I play with it some more.

del.icio.us | digg

This is the Wifi chip aimed at smartphones from Marvell. Not too much to see with so much metal on the top, however, there are some open areas. A small note of interest is that there are pads on the perimeter of this chip as well as on the bottom leading me to think that the layout was destined for flip-chip bonding, such as in the iPhone, or to be put in a standard package. There is also an orientation mark at the center of the chip. There is some speculation that this chip features an ARM core and given my past experience looking at CPUs, it is reasonable that it can be in there below all of that metal. Although I don’t do RF IC design, I would speculate that the open windows are the last stages of RF amplifier designs and are open to minimize parasitic capacitance. I am not sure if the coils on the surface are used for tuning purposes or in a DC/DC converter application to power the output stage of the RF generator. As always, if you have more insight on the chip, please share!

Thanks to a tip by “marius”, I have made a composite image of the 4x scans (see below) using hugin.

w8686b13-4x-stitched.jpg

del.icio.us | digg

I apologize to those who are sick of seeing iPhone related news clogging the internet. As per request, below are the scans of the two PCBs with and without the chips. It is likely that there will be a few more IC Fridays displaying iPhone chips with the hope of finding some easter eggs and then the grand finale will be my attempt at reading the 4GB flash chip. The 48 TSSOP adapter has been ordered already and with a lot of luck, I may be able to use a method that involves interfacing the flash with a SD card reader. With even more luck, the chip will not be destroyed. Finally, if the planets align, it may be possible to read the contents of the chip in a meaningful way. After that, I will look for other gadgets to dissect. Files are about 3MB each.

iphone-nometal-front256.jpg

iphone-nometal-back256.jpg

iphone-bare-w-cpu-front256.jpg

iphone-bare-w-cpu-back256.jpg

del.icio.us | digg

Someone handed me a a Sega Saturn without a video cable this evening (thanks Craigslist).  Wanting to add it to my 32bit gaming system collection, this problem would have to be remedied. The one beneficial fact is that generally, the audio/video cables for game systems are purely passive. That is, the system will have pins on the A/V connector for all of the supported outputs and it is just a matter of routing the right video connector and you are set. Some designers get clever and add some simple components to let the system identify what kind of cable is attached, such as the original Microsoft Xbox, which can then be used to add additional functionality to the device like a reset button. Here is the pin diagram for the Saturn’s A/V connector. While looking for it, I accidentally found a PIC16F630 based hack to allow you to toggle between 50/60Hz video output (PAL/NTSC).

del.icio.us | digg

While looking at a Zener diode datasheet, I scrolled past a temperature coefficient plot and became puzzled for a few moments. Why was the temperature coefficient negative for low breakdown values and high for others? I then remembered that, unfortunately, the term Zener diode is used describe diodes designed to operate in the reverse breakdown region, both due to Zener and avalanche multiplication effects. For small (<5V) breakdown voltages, the Zener effect usually dominates and has a negative temperature coefficient. For larger breakdown voltages, the breakdown mechanism is typically avalanche multiplication.

(alpha is the temperature coefficient, image is from Wikipedia)

In a true Zener diode, the p-n junction is heavily doped so that the depletion region is very thin. This can cause appreciable tunneling through the depletion region barrier. The probability is inverse exponentially proportional to the square root of the effective carrier mass and the (3/2)nd power of the bandgap energy (reference). It can be shown that the effective carrier mass increases and the bandgap energy decrease with temperature (reference) is the dominant effect. The net result is a negative temperature coefficient. That is, the probability of tunneling across the depletion region energy barrier increases with an increased temperature and therefore the effective Zener breakdown voltage is made less negative.

The story is quite the oposite with avalanche multiplication breakdown. The idea behind this process is that a electron in the depletion region experiences significant acceleration due to the applied electric field accumulating enough momentum to create an electron-hole pair upon impact with an atom in the crystal lattice. The newly liberated electron then also accelerates to required velocity to liberate another electron. Now there are two electrons with the required velocity, and so the current multiplies. What started as just one electron traveling quickly has suddenly turned into an appreciable current given a strong enough applied field. This is where the name “avalanche multiplication” comes from. Typically, in order to have high electron velocity, a thick depletion region is desired so the p-n junction is relatively lightly doped. The reason that this process has a positive temperature coefficient, or the “Zener” breakdown voltage becomes more negative with increased temperature, is that the probability of an electron colliding with an atom is decreased due to the increased thermal jitter. In this sense, the atom becomes more of a moving target so a direct collision is more difficult.

Although this may seem like a trivial matter, it is important in various designs. Beyond ESD protection circuits, “Zener” diodes are frequently used as band-gap voltage references. A good voltage reference should have near-zero temperature dependance over the operating range so a single “Zener” diode may not be good enough. Aditionally, one type of breakdown may be more compatible than the other in a certain manufacturing process.

( czrf52series.pdf ) ( ds18007.pdf )

del.icio.us | digg

This is the second Apple-branded chip on the larger iPhone logic board, the other being the processor/ram combination. Various sources around the industry are speculating that Philips/NXP is responsible for the power management chip for the phone. Judging by the Philips copyright symbol, I would have to agree that it is something like an NXP PCF50626 or PCF60633. Some of the chip’s structure can be seen under the metal pattern as well as some exposed circuits. I can see what looks to be three ADCs/comparators on the right side of the chip and at least four charge pumps in the middle and left side (noted by the square capacitors and large drive transistors).

Again, great thanks to Think Secret for sending me the chips!

Update: Here is a picture of the bare PCB. The place where this chip was attached is boxed in red.

del.icio.us | digg