Thursday 21 July 2016

Mega:bit


During July last year Andrew Gale (@PocketMoneyTron) tweeted details of his mega:bit which is a lasercut scaled-up model of the BBC Micro:bit. This looked interesting and I thought it would be great to build my own.   


Fast forward to May this year. 



Excellent news, I duly contacted Andrew who kindly sent me the CAD files and a list of components required. He also very kindly sent me the blank PCB’s required for this project, in the meantime I acquired all the other parts needed. 



Step one:  Assembling the PCB's.


I used standard components which I sourced from various suppliers. The only changes I made to Andrew's prototype was to use nice turned pin DIL sockets and turned pin SIL sockets for the resistor networks. 


Blank PCB's
Populated  PCB, LED maxrix side


Populated PCB component side

Step two:  Laser cutting


I am lucky enough to have access to a laser cutter at work. I opened the CAD files  Andrew sent me and cut out all the required parts in 3mm Perspex. I used blue for the back layer and black for the front layer then glued it together using special Acrylic glue. When using the laser cutter, remember Rule Zero. Do not be on fire.



Laser cut parts

































Step three: Putting it all together


Now it was time to put it all together, Andrew had sent some photos of the completed mega:bit and uploaded a video to YouTube showing how it was assembled. There is a ribbon cable connecting the buffer PCB to the phototransistor PCB, this is not shown in the diagram. 

I used 4mm terminal posts for the general purpose input and output pins, plus 3V and GND, these are connected to the Kitronik breakout board. There are also two large push buttons which allows you to trigger or detect a button "A" or "B" click externally, these are also connected to the Kitronik breakout board.


Assembly diagram


The back of the mega:bit
Completed mega:bit




















































Overall I'm very pleased with how this project turned out. On a footnote I recently took my mega:bit to the CAS East Regional Conference, I'm pleased to say there was a lot of interest in it.



Finally if you are interested in having a go at making your own then do get in touch with Andrew Gale.

Tuesday 19 July 2016

Marconi TF2438 Repair

I recently got an old 80’s vintage Marconi TF2438 Universal Counter Timer from a fellow forum member.  The unit looked very good and powered up but was sometimes intermittent.  The unit has an oven-stabilized 10 MHz crystal oscillator fitted which is a bit erratic when warming up but fine after 5 minutes or so.  The following faults still exist even though the previous owner had already done some fault finding.
  1.   It either says 0Hz    
  2.  Or reads high (approximately double the frequency, but not stable). The former, when it occasionally happens, is fixed by gentle "percussive maintenance", so that's clearly an intermittent connection.

After a bit of googling, I found out that a typical failure mode of these instruments made by Marconi Instruments at the time used double-sided PCBs with the two sides connected with what could only be called rivets. Over time, with thermal expansion and contraction, the solder on these rivets would crack and they became very unreliable. Some of the rivets had already been re-flowed.

Anyway I digress. I got the unit home, powered it up and connected the 10MHz standard output to channel B, it did measure the frequency but unfortunately it was up to its old tricks of reading high approx. double the frequency and despite some “percussive maintenance” I couldn’t get it to behave. I found a legible copy of the schematics, courtesy of UK Vintage Radio Repair and Restoration forum.  I guess it was time to check some voltages, there are four main d.c. supply rails,   +5.1V,  -5.1V, +12V and -12V.

The joy of fault finding



















Function Board



















Control Board




















All of the supply rails were reading low approx.  +/- 4.0v.  I checked the rails pre-regulator and these seemed to be correct.  Inspired by a post on the UK Vintage Radio Repair and Restoration forum,  I started checking for shorts and faulty components on the bottom function PCB but I couldn't find anything wrong. I also checked components which share both +5.1V and +12V rails and removed the 10MHz oscillator but this made no difference.   Following some helpful suggestions, I ran some tests with and external power supply connected across D21 (a temperature-stabilised 6.2V Zener) and the + -/ 5.0v rails were ok.  Several suspect tantalum capacitors were changed, two pass transistors in the power supply were changed and an IN4148 diode.  Finally I replaced the IN4148 diode again and repaired a broken track/pad which had lifted when I had removed the original.  The supply rails then sprang back into life. 

These instruments have a nice feature which allows you to plug the top control PCB with display and switch board into the bottom function PCB via a second set of connections which improves access for servicing, therefore you don't need an extender card or any special cables.  

Having sorted the power supply issue my attention moved to the other annoying intermittent faults, which I suspected was due to a bad through-hole rivet somewhere on the control board.   I re-flowed every through-hole rivet I could find and cleaned the push-button switches with some switch cleaner.   Success, I tested the unit again and it now worked correctly.


Overall I’m very happy I managed to repair the annoying intermittent faults.  It is a beautiful instrument and well-made though I suspect that those through-hole rivets will come back to haunt me one day.

Sucess