Adrian’s electronics blog

On one of my very old computers I had one of those LCDmodkit PC status screens that consisted of a 20×4 LCD housed in a 5.25″ bay and was interfaced to an internal USB header on the motherboard via a USB2LCD controller. As there are no Microsoft certified drivers available for this device it was no longer used in my newer Windows 10 or later machines. I removed the USB204B board off the back of the LCD and used the LCD for another project. That left me with a controller which was pretty much useless. On inspection I found it had an ATmega8 microcontroller and a bit of digging I found that this is based off the LCD2USB project by Harbaum (link to original project here) and it appears that the LCDmodkit controller is basically a redesigned original using SMT instead of surface mounted parts.

So I thought it is highly likely that the board could be reprogrammed and wasn’t locked. So I added a couple of headers to the former LCD connector and found, as in the schematic found on Harbaum’s project several of the I/O ports are broken out onto this connector. The rest are used for V-USB and serial and there is also a group of six solder pads labelled CN5 to the upper right of the microcontroller. I suspected this was for programming and poking around revealed this to be the case. The MOSI, SS and MISO pins were on these solder pads so I connected it to an arduino working as an ISP and uploaded a couple of basic sketches. Worked a treat but one thing to watch out for is the onboard crystal is 12Mhz not 16Mhz so anything that uses timers won’t work properly. However it’s great for very basic applications that involve sensors and activating outputs.

I’ve mapped out the six solder pads with pin one being to the right hand side of the photo. In other words pin one is nearest the edge of the board and pin six is nearest the microcontroller. I connected these via mod wire to a standard 6 pin ISP header on the perfboard.

Pins are 1= ADC6, 2= Reset, 3= MOSI, 4= SCK / SS, 5 = AVCC, 6 = MISO. 5V and power were taken from the 16 pin LCD header. The rest are easily found out by measuring continuity between each of the 16 pins and the pins on the microcontroller using a pinout as a reference. I used a 7805 regulator to power it but even if reprogrammed the original USB connector could be used to power the board.

Continue reading →

Here I have a couple of defective enterprise grade access points that were decommissioned as they were suspected as faulty and were out of warranty. They are manufactured by Xirrus and are meant for outdoor high capacity Wi-Fi for public locations e.g. Stadiums, festivals etc. I thought I’d pull one apart and see what was inside before they were sent for scrap so I will share some photos here and give brief descriptions on what each major chip is and what it does.

I did a video below of the teardown but there is no narration. It’s good enough to get the point across but after all this is just an access point but a very expensive, enterprise grade one costing several hundred dollars.

I bought two lots of faulty laptops and tablets from eBay with the intention of repairing and reselling them as mentioned in my previous blog posts. Without repeating myself this was largely a success more so with the laptops and not so much with the tablets. The problem with the tablets not being profitable is because they are old and go out of date more quickly than laptops and the newer ones are more difficult to repair as they are usually glued together and cannot be taken apart easily. So, with this in mind I bought older models and avoided ones with obvious cracked screens. The majority of the faults were broken USB sockets which were replaceable but given the cost of the parts and eBay fees I only just broke even.

The other thing I found was some of these tablets had obviously come from computer repair shops and had missing parts and / or multiple faults as they had been used to make a good tablet out of several faulty ones and almost all parts in the ones I bought were faulty. Some were made to look like they had never been opened. Take this one for example; a Samsung  nexus 10 tablet which had a broken power / USB socket and a broken power button. I replaced the USB connector and power button but it still would not power up although the battery was charging. I removed the metal shield covering the CPU and flash chips and found that the flash chip had holes drilled into it. Now the seller had made a really good job of reassembling this tablet and making it look like it had never been opened. I was clearly ripped off here. Some of the others were in a similar state.

I thought I’d mention this for anyone who is thinking of buying faulty tablets to try and repair them. Watch out as most are junk unless you are buying from genuine private sellers selling three of them at most. Also they are often account locked and even if you fix them they are useless without being de-registered by the previous owner. This is why I avoid apple products but the latest Android devices are doing the same thing too.

So, take my advice. Don’t bother buying bulk faulty tablets and phones unless you like wasting your money.

I bought several laptops and netbooks off eBay with a view to refurbish and sell them; I got around 7 laptops and 9 tablets and netbooks for just £100. One of the netbooks was an Asus X205T that had it’s eMMC drive encrypted with bitlocker amongst other issues such as a dead battery and faulty keyboard. The netbook would only boot into the bitlocker recovery screen and would not allow booting from USB as it just returned to the bitlocker screen. To get around this problem and seem as I wanted to remove the existing data anyway I selected skip this drive on the bitlocker recovery which took me to the standard windows 10 repair recovery menu. Resetting the to factory settings was out as the storage drive was locked. So to get around this problem I went into the command prompt and removed all partitions with the DISKPART utility.

However upon rebooting the netbook just went straight into the BIOS and would not allow any boot options to be changed. It was set to USB for Windows to Go but would not boot anything. To cut a very long story short in order to get Windows 10 to install you must use a old USB2 stick (not USB3) and use the 32 bit version of windows. Only installation media created with the official Microsoft tool would work. Plugging this in and rebooting made it boot to the windows installer.

The reason for all this palaver is because the UEFI firmware is only 32 bit and will not see 64 bit code. Neither will it see USB3 pendrives, USB sticks etc. So if anyone ever has this problem now you know. This could apply to any older netbook and tablet running Windows 8.1 or later with a UEFI BIOS. See my previous post for a similar issue with a Dell Venue 8 Pro tablet.

I have some largish 7 segment displays which are 0.8″ in height but I found that they were dim when driven from a MAX7219 even at full brightness. I found that this was because the LED displays I were using need around 15 mA for a decent amount of brightness, 25mA ideal. However the MAX7219 only has a maximum peak current of 40mA per segment; now that is peak current so that means with a multiplexed display that will be divided by the number of digits to give an average current of 5mA, or if you prefer the equivalent DC current. Now as my displays need 15mA that simply isn’t enough hence the dim display.

So to get around that problem I used a UDN2982 as a source driver between the segment outputs and the LED segment anodes. 47 ohm current limiting resistors were used (more on that later) and a ULN2803A was used as the sink driver on the digit outputs. Now as the digit outputs will be at a logic low level I had to use an 74HCT04 inverter between the MAX7219 and the ULN2803A as the ‘2803A needs a logic 1 to ‘turn on’ it’s outputs.

I used an LED supply voltage of 7V as 5V wasn’t enough to get a decent amount of current through the displays as the UDN2982 drops about 1.6V across it and the ULN2803A about 1 volt. Together with the voltage drop across the LED display I was only left with 0.4V to play with at a supply voltage of 5V. Both the ‘2803A and the ‘2982 have a absolute maximum current of 500mA for the entire chip so I aimed for 60mA peak current per segment which gives an maximum current through the UDN2982 of 480mA when the worst case scenario of the displays showing 8 and the decimal point is lit. 60mA divided by 8 digits still only gives 7.5mA average segment current – still not enough but as I was using only 4 digits I set the MAX7219 scan limit to only 4 displays which doubled it to 15mA equivalent DC current. This resulted in an acceptable display without overloading the driver IC’s. I did try it on 5V with 6.8 ohm resistors giving around 7mA current but it didn’t work.

Continue reading →