PS/2 Keyboard Decoder

I have run into a problem with the decoder software.  I have read many posts on how to turn the keyboard LEDs on and off but they stubbornly remain off.  Normally interfaces to a PC are extensively documented by multiple sources and widely available.  I’m surprised by how little accurate information is available.  This includes some waveforms (diagrams, not captures) that have the wrong number of clock pulses in the transmitted data.  I have been busy recently and unable to work on the issue but during the past week I got back to it and rewrote the code to toggle the LED once more and it still doesn’t work.

Since I can’t find reliable information I will have to measure it.  I bought a few cables, connectors and set up a test PC to talk to the keyboard.  I will capture the waveforms and publish them here so that there is a trustworthy source of information.

Daniel

Reverse Engineering IC’s

A message I posted on the forum about the vulnerability of the programs store in a programmable device created a lot interest and e-mail.  Many e-mails asked for more information on vendors who can read programs from devices with the security bits blown.  I don’t want to encourage piracy so I don’t respond to these requests with more information.  Recently a YouTube video of a presentation on Low-Cost Chip Microprobing was brought to my attention.  It has some details on how chips are protected and how the protection is circumvented.  The talk can be found here https://www.youtube.com/watch?v=b_MsQRpwRlw

PS/2 Keyboard Decoder

We have been working on a comprehensive PS/2 keyboard controller.  PS/2 keyboards are available in many shapes, sizes and specialty versions and most are low cost as well.  A problem with these keyboards are that they require fairly complicated code to read the data from the keyboard and translate it into a simple ASCII character.

There are a few PS/2 controllers on the market so why build another one?  The short answer is that none of the ones on the market meet our needs.  Most (All?) PS/2 controllers produce 5V signals and this is a problem for modern CPU’s that run on 3.3V.  The output ports available on the existing controllers are typically serial ports and some offer I2C interfaces.  We will offer serial, I2C and SPI outputs, simultaneously.  The power supply to the microcontroller is switchable between 3.3V and 5V to allow the keyboard decoder to be used with 3.3V and 5V systems without level shifters.  We also offer IO pins that signal when there is a character available, when the internal buffer is full and the state of the Caps Lock of the keyboard.

The first prototype had a few issues such as the PS/2 connector had to be soldered onto the bottom of the board, the reset signal was broken and some other minor issues.  The second prototype is in much better shape and we are coding away furiously.  Once the board and the code is fully tested it will be for sale at a reasonable price.  The schematic will be available to everyone  and the source code will be available to all who purchase the controller.  You will be able to modify the code to suit your own personal requirements.  The decoder is programmable without any hardware and with free software.  The CPU is quite capable and fast and has several IO pins available so you may be able to fit your entire project onto the controller.

 

5110 LCD Back in Stock

I try to never be out of stock but I managed to get the 5110 LCD’s to sell out before my stock arrived.  I didn’t plan well enough for the Chinese New year and its effect on delivery and shipping.  A batch of displays arrived today and another, smaller batch is on its way as well.

I ordered some new items that may be of interest to some of our users and after I’ve tested and photographed them they will appear as regular stock items.

Daniel

Switch Mode Regulator Drop In Replacement for LM7805

I have completed the design, layout, assembly and initial testing of a switch mode regulator that is a drop in replacement for an overheating LM7805.  It includes all the capacitors on the module.  The module is only a few millimeters wider than a TO-220 package.  With 12V in and 0.5A load current a LM7805 needs a fairly substantial heatsink to keep the device from reaching over temperature cutoff.  The new module can supply 1A continuously with 15V input without overheating.  The back of the module gets noticeably hot but does not burn you when you touch it.

What needs to be done is more testing and some documentation before I’ll offer the part for sale from this web site.  If you are interested in an evaluation module, send an email and I send you a module for test.  I will also offer a 3.3V version.

• Drop-in replacement of 3-terminal LM7805 or equivalent linear voltage regulator.
• Guaranteed 1A output current
• Wide input voltage range up to 7V to 17V
• High efficiency, greater than 70% for loads greater than 1mA, peak efficiency achieved of 90% at 300mA load current.
• Thermal shutdown and current limit protection
• Breadboard friendly
• Board dimensions are 0.51″ x 0.75″ or 12.9mm x 19mm excluding pins. With pins 1.09″ or 27.75mm.
• Definitely no heatsink required.
• All components are mounted on one side of the PCB
• Highest component is the plastic of the connector at 0.15″, Next highest is the inductor at 0.121″ including the thickness of the PCB
• Available with straight or right angled pins.
• Weighs only 0.04 Ounce or 1.3g, ideal for RC model aircraft or quadcopters
• Can drive inductive loads such as DC motors.
• 2.2MHz Switching Frequency

The price is $6-95 in singles with discounts starting from 10 pieces.  If all goes well stocking quantities will be available by next week.

 

Size comparison PSU2-5 with LM7805 in TO-220 Package

Size comparison PSU2-5 with LM7805 in TO-220 Package

Shipping Cost

After fighting with the shipping calculator we now have new shipping rates for USPS First Class mail.  The shipping calculator refuses to offer First Class mail as an option so we added a table based shipping cost calculator that should fix this issue.  For people in the UK, the shipping calculator would routinely produce $50 as a shipping charge and this is ridiculous since our most expensive item is only $59-95.  We now offer world wide shipping for $6-95.

I hope the new shipping rates make shopping a pleasurable experience.

 

New Products

We are always looking for new products that solve real problems and are easy to use and master.  We are almost done transferring the manufacturing of our existing products to the new equipment.  This is a big task because all the pick and place programs were re-written on an unfamiliar machine.  At the same time we built stock so we can spend time on the machine programming the production of new designs without having to stop midway and build some board because stock is running low.

The first time you build a board on a new machine there is a lot of setting up to do.  You have to transfer the CAD data to the machine so you know where to place the parts.  Then you have to mount the reel containing a part on the feeder, tell the machine that the feeder exists and where it can find the part.  This goes quickly for one or two new parts but it gets tedious to do it for every part used in all the designs.  You also have to figure out how the tape goes into the feeder, where the cover tape goes etc.  Once you know, the next time is easier and by the filth one, it is easy.  Then you end up learning how the matrix trays work and the optical alignment.  Everything takes a lot of time the first time round.  One of the strange behaviors of this machine is how it picks up parts from the matrix trays.  It starts from the last position and ‘advances’ to the first position.  My first tray was partially used so I carefully moved the parts to the start of the tray just to discover that the machine starts from the side that I had just emptied out.  So I moved all fifty parts back to the other end of the tray.  What a waste of time.  Now, after having built panels of everything we sell we can move on to new designs.

The first new design is a small switch mode voltage regulator to fit the footprint of a standard TO-220 LM7805 type linear regulator.  When the supply voltage is a little high and the load current more than a few milliamps then the linear regulators run very hot with small heatsinks.  There are switching regulators of this type on the market but the suffer from two problems:  Most need external capacitors and are very expensive, around $10 each.  The new design has all the capacitors on board so you can remove and overheating linear regulator and drop this one in its place without having to find a place to hide a few external capacitors.  The price is going to be less than $6 in singles.

Watch this space.

Pick and Place Arrival

The LE40V arrived yesterday (Dec 18). The pick and place machine arrived as promised in a gigantic crate with the feeders, PC and the machine professionally packed. I was surprised when the truck arrived with just a driver, no helpers. The driver maneuvered the truck under a carport and single handedly offloaded the crate with just a hand operated forklift and the lift gate of the truck. It was raining pretty heavily and it was cold so the crate was left unopened and we waited for the rain to stop. It was around five o’clock when the rain finally let up enough so we thought we could unpack it and get it indoors. The container is constructed as a wooden box that can be lifted with a forklift. It has a million screws. I took the lid off and it revealed a bunch of boxes on top of a wooden partition. The boxes contained the PC, monitor, feeders, bank feeders and some tools. We set all the boxes aside and removed another million screws. We removed all four side panels of the box and some cross beams holding the LE40V in place. In the manual it says that the machine is held onto the base by four bolts so we tried to locate the bolts. No luck. We tilted the pallet to look underneath. That sucker is HEAVY! and surprise! no bolts. We then thought that the feet is holding the machine in place and tried to remove them by unscrewing them. When we lifted the machine to take the weight off the feet, one foot came off. It left a rounded rubber piece that sits in the foot and the foot stuck to the pallet. New plan, lift the machine off the feet. That worked and we ended up with three feet attached to the machine and one stuck to the pallet. Next we carried the machine on top of the pallet closer to the work area and rested it on top of a sturdy box. Here two of us lifted the machine off the base of the pallet, two people removed the pallet and we carried it indoor and put it on top of its table. It couldn’t have been more than 10 yards but it is down a narrow passage and through a door that is just barely wide enough to get through. It felt like half a mile. That machine is heavy. The manual says 250lbs but I’m sure its more, it certainly feels like a lot more. The pallet and content weighed in at 600lbs (according to the waybill).

After maneuvering the machine to its final resting place in the corner the cover would not open, it was too close to the wall. We had to move it (and the table) again.

Assembly was easy as the machine is fully assembled in the factory and then tied down for shipping. The only things to do were hook up the PC, the vacuum pump and remove some straps. This was quick and easy. The hard part was that the software wouldn’t run, it complained about some hardware being missing. Opening the PC and pushing all the cards back into the slots cured that problem.

The LE40 in its shipping container

The LE40 in its shipping container

The LE40V still wrapped in plastic

The LE40V still wrapped in plastic

LE40 with 2 Bank Feeders, four conventional feeders and vibratory feeder

LE40 with 2 Bank Feeders, four conventional feeders and vibratory feeder

 

 

New Pick and Place machine

The delivery of the new Pick and Place machine was confirmed by the shipping company for 17 Dec 2012.  Finally we will have an accurate, fast machine.

Pick and Place machines are car sized purchases and a lot of thought and research went into the choice of the machine.  After sifting through the options for desktop machine only two machines were seriously considered.  The Manncorp 7722FV and the DDM Novastar LE40V.  Both machines are small enough to fit through the doors to our work area.  This is a big deal, kind of a showstopper.  Pick and Place machines can be large and heavy and are made to be rigid.  This often means that the frame is welded or cast and can not be disassembled.  Also, disassembling a machine means re-assembly and alignment.  Alignment is also critical in the machine meeting its accuracy spec.

The runner up was the 7722FV.  Several smaller manufacturers such as Adafruit and Eurobuilder went with the 7722FV and since they must have done their homework I had a very serious look at the machine.  It is a bench-top machine and does not have a welded or cast frame.  It is reasonably light, around 150 lbs.  It has real feeders and optical alignment.  It is reasonably fast at around 2200 components per hour and can handle all the components that I see in immediate future.  Here are some photos http://www.adafruit.com/blog/2009/10/13/our-manncorp-7722fv-pick-and-place-is-here/

In my evaluation of the machine I was slightly uncomfortable with the stated accuracy of the machine and to a lesser extent the way in which the machine is constructed.  The machine’s accuracy is adequate for current parts but it has no room for the feature shrink in packages occurring now.  A while ago, everything was dip packaged and pin pitch was 2.5mm.  Then everything was leaded surface mount.  Now lots of new parts are only available in leadless surface mount and ball grid (or chip scale) packages.  Currently the ball pitch is already down to 0.4mm and t is likely to shrink in the future.  0.4mm is the limit of the 7722.  Also, the 7722 is an open loop machine.  Positioning takes place by counting pulses after a reference was established, the gantry position is not measured.  The reference is found by bringing the part over an up-looking camera and doing a pattern match to a known good position of a part on the nozzle.  The nozzle (and gantry) is then moved to the desired position and the part is placed on the PCB and held in place by the solder paste.  In any well designed machine the movement does not miss steps after the alignment and if the part does not move on the nozzle the only error is the positional error inside the step.  This may appear to be negligible but in practice it depends on exactly where the motor is commanded to stop and the speed at which the motor slows down.  Friction, bending and stretching and temperature effects all add to the complexities.  That is why the accuracy is not equal to the resolution.  The stated resolution of the machine is much better than the accuracy.

My second reservation about the machine is the construction using Aluminium profiles.  When stiffness matters this is not the best way to build the frame.  No doubt the machine meets its accuracy when it is new but I don’t know how it will age.  Secondary things that I took into consideration was the design of the feeders.  The feeder design is ‘classic’.  It has an indexing mechanism and can advance the tape quite precisely and quickly.  The feeder is a mechanical design.  I can’t remember if the machine had an option for ‘smart’ feeders but the default feeder for the LE40 is smart.  A ‘smart’ feeder has an electrical interface to the machine and it can be identified by the machine to verify that it is in place.  Smart feeders also maintain a count of the parts used making inventory tracking easier.

The LE40V was the winner.  The machine base is welded steel and the motion is closed loop in the sense that the actual gantry position is measured using linear encoders.  This gives me confidence that the long term accuracy will be the same as the ‘factory’ accuracy.  The other major factor influencing my choice was that DDM Novastar gave me access to the manuals before purchase and the manuals are complete, legible and in good English.  They even contain circuit diagrams of the machine.  From the manuals I recognized the motion controller used in the machine as a controller made by Galil.  I have first hand experience using their controllers in motion systems and they are used wherever high quality motion control is required.  My current machine uses trapezoidal profiles for the motion and the jerk is not welcome.  S-curve motion is definitely not a luxury.

The price of feeders are also a consideration.  Very often you spend as much in feeders as on the machine.  The LE40 has a very clever bank feeder for 8mm tapes.  8mm tapes are by far the most common size and you can never have enough of them.  The bank feeder is a ‘single’ feeder for 12 8mm tapes and it costs less than a 5 individual 8mm feeders.  Since I have existing products I knew from the beginning that I need more than 12 8mm feeders and this is an economical solution.  The prices for the LE40 feeders are comparable to the feeders for the 7722 but they are smart feeders.  Even though the initial price for the LE40 was higher than the 7722 by the time I factored in the price for the feeders, and external PC and a compressor, the LE40 ended up at roughly the same price.

As they say in the movies, watch this space.  I will provide pictures and details of my experience with the machine.

Happy hacking.

Pick and Place Machine Trials and Tribulations

Our old and clunky pick and place machine can’t keep up with demand for some of the modules we sell.  The current machine was bought second hand from someone who had stopped using it.  When the machine arrived here there were some difficulties with the software and I had the software and machine upgraded by the manufacturer.  This solved most of the problems.

Nothing lasts forever and the pick and place machine is no exception.  There is nothing physically wrong with the machine and it still works.   The vision system is ok but the component picking is not completely reliable and this causes the tape to jam.  Sometimes a component is dropped near the tape or on top of a neighboring tape.  The component then get trapped in the small spaces in the tape and this causes the machine to stop and wait for a manual intervention.  The LS1 level shifter sells well and it has small margins.  To manufacture it economically it has to be made in panels.  There are enough components on a panel so that the machine jams on nearly every panel.

The vision system is optimized for small parts and this causes issues with bigger parts.  When a part is picked up and moved to the up looking camera the vision alignment must find the part somewhere in the picture in order to align it.  When using a parts larger than 144 pin QFP the part nearly fills the frame.  The vision system fails to align the part if any piece of it is outside the frame.  This is limiting the size of parts that we can use in our designs.  Watch this space for more news on our adventure in surface mount assembly.

December 5, 2012

Old2_Smaller