μblog: engineering from the trenches

IBM has recently made available the datasheet for the new PowerPC 750CL processor, the muscle behind the Wii entertainment system. At the same time, Walmart has issued a buyer’s guide for the same system. Let the document battle begin!

( 750cl.pdf )

del.icio.us | digg

I am entering LabView purgatory. When clocking data in or out, there is a requirement for a trigger source. To implement acquisition across multiple cards, we can use the RSTI (real-time standard integration) inputs via a ribbon cable that connects those boards. To synch with external devices, we can use PFI (programmable function input). We can also route internal signals to some of these PFI ports, or other clock sources. The M-series boards that I am using feature a 80MHz, 10MHz and 100kHz time bases (/Dev#/80MHzTimebase, /Dev#/100kHzTimebase, etc) which can be routed to a counter or /Dev#/freqout. Freqout is subjected to a 4-bit divisor and can be routed to any PFI, so in this way, we can synchronize digital output with digital input or something else on the same board. Included is an example VI (not my code) to do this routing.

( 4645.zip )

del.icio.us | digg

For fifty dollars you get the EKK-LM3S811 ARM development board with built in OLED display and a JTAG port, so you can debug other targets providing you make the right software. The board supports virtual UART over the USB interface, so the PC interface is easy. The board comes with a stripped down version of the KEIL suite for development. Finally, there is a design contest with a deadline of 7th Feb 2007.  Seems like a lot of features for the price.

del.icio.us | digg

My new found love for Cypress Semiconductor has led me to look at their other offerings besides the PSoC platform. The development kit was around 100USD with the software, USB programmer, LCD, and a few other toys. There are also a few that are as cheap as 30USD. Everything worked out of the box and I was able to test the first code on the PSoC only minutes after installing the software. Furthermore, Cypress is pretty good about sending samples, albeit in oversized boxes, so testing their design is pretty painless. For these reasons, I am going to try to implement a Cypress, low-speed USB device.

The basis for this design will be the CY7C638xx line of chips, which can be directly connected to the USB port and will even provide regulated 3.3V power to peripheral hardware. These chips also come in PDIP form factor which is good for prototyping. Once the samples arrive, I will try to make something based on the CY4623 USB optical mouse reference, something that will seamlessly fit in with the HID driver and will provide an input device to the computer. Maybe one day, interface it with an msp430 and track mu rhythm to move a pointer, all with a tiny USB device. For now, here is some documentation.

( cy7c638xx_8.pdf ) ( cy4623_mouse_reference_design_19.zip )

del.icio.us | digg

June’s 4G iPod finally kicked the bucket, so she got one of the pink nanos. It seems like the older iPods like to give the broken icon with the support url after an impact. This condition has been fixed by three people that I know of so far by popping open the case and reseating the hard-drive connector. The reason June’s iPod died was because I tried to unplug the connector for the headphones and pause button and it tore the connector off the motherboard instead of disengaging it.

After connecting the new nano to a iTunes, I was prompted for an update which kept failing with a unknown (1417) error. Much searching has lead me to this functional solution: start the update, when the flashing “do not disconnect” message goes away, unplug the cable from the iPod for a few seconds and then plug it back in. For some reason, this allows the update to finish.

del.icio.us | digg

To implement the artifact subtraction system which I presented at SFN, a tunable high-pass filter is needed. After looking around, I settled on a PSoC from Cypress. The CY8C29466 is such an IC that has a 14bit ADCs and 9bit DACs as well as various analog blocks. The chip also has an internal 24MHz oscillator so minimal external parts are required. Finally, the MiniProg is really easy to use, just plug it in and download a binary. The first thing I did was to implement the ADC to LCD example where the chip reads the voltage off a potentiometer and displays it. The filter design stuff seems to be straightforward, so I will try that next week.

( bpf4-design.pdf ) ( 38-12013.pdf )

del.icio.us | digg

A mere ten days after my last post about my distaste for Apple, there is more good news: some video iPods got shipped with a pre-installed windows virus. Although Apple apologizes, they are “upset at Windows for not being more hardy against such viruses.”

del.icio.us | digg

Just got back from Atlanta where the Society for Neuroscience 2006 conference took place. I suppose it is time to get back to work then. Here are some pictures, but only the good ones.

del.icio.us | digg

It is well known that I do not hold Apple Inc. in very high regards. Ever since the switch to Intel, I have not withheld my view that the switch was not for performance reasons, it was because of economics. The claim that IBM was unable to provide powerful enough chips was just a front. Steve Jobs realized that people who bought Apple were no longer buying the computers for their performance; they were buying them for the aesthetics. Apple decided to hop on the Intel bandwagon to save money. Macs were marketed as easy enough to use by novices, same novices who wouldn’t care about PowerPC v.s. x86. Apple has even branched out into the “enterprise” sector lately with sub-par hardware. Our 3.5TB raid system crashed this week, two months after installation, due to a faulty DeskStar drive. One drive crashed and trashed the rest of the Raid5 system corrupting it beyond the repair-ability of the operating system tools. The Apple premium support gurus told me to “try DiskWarrior, a 3rd party product” to rescue the sinking ship. As you can tell, I am not a fan.

Meanwhile, IBM is undeterred by its breakup with Apple and continues to forge ahead with progress. They designed the XBOX360 Xenon CPU (3-way PowerPC), the PS3 Cell Processor (PowerPC with eight symmetric processing engines) and the Nintendo Broadway (PowerPC based, single core). The PS3 and the XBOX360, which has already sold ten million units, give the latest Core 2 Duo chips a run for their money in certain fields. The Broadway, the lowest end chip of the bunch, was offered to Apple as a high performance laptop chip. This is to put things in perspective. Today EETimes posted an article that IBM is in the debug phase of their new Power6 architecture: a dual core 64 bit system with a hard-wired control unit running at a native 4-5GHz clock speed. Not to mention that IBM has been using dual core technology since the Power4 series, compare that to the latest microcode based Intel chips topping out at 3GHz for the dual core units. I guess this is enough for this whiny rant, but then again, everyone has an opinion.

del.icio.us | digg

Every six months or so, I encounter the Miller effect and can never remember what it is. Every six months or so, I open my copy of Sedra and Smith to look up the Miller effect only to forget it over the upcoming six months, hopefully this short write-up will help store it in long term memory. The idea is this, the differential voltage (Vo-Vi) across the impedance Z is dependant on the input voltage and the gain, so for the same change in voltage (per change in time) on the input, different gains can create almost any voltage at the output. This dependence on the gain can lower the effective impedance as the gain goes up for the same input signal. Most amplifier stages have some sort of capacitance and this Miller effect tends to increase the effective capacitance by a factor of 1-G. How much of this effect is seen in real amplifiers really depends on the configuration, especially on the feedback network and the resulting output impedance. In resonant circuits, the Miller effect can be understood, as it linearly depends on the circuit gain, and can sometimes be exploited to get away with using smaller capacitors. On the other hand, filter topologies, such as Sallen-Key, can be used as their pass-band gain is unity and therefore they do not exhibit the Miller effect.

On a side note, there is also the Miller theorem, which allows one to break Z apart into Zi and Zo, which are connected from Vi and Vo respectively to ground. This is the method that is often employed when calculating the input impedance of an amplifier, usually at open loop gain.

del.icio.us | digg