Thursday, 11 December 2014

Spectrum Pi Review in Micro Mart

This week's edition of Micro Mart magazine carries a review of my Spectrum Pi, an original ZX Spectrum converted to work as a USB keyboard with a Raspberry Pi installed inside.
'A wonderful fusion of old and new'. 'Superbly built, and an excellent addition for any retro enthusiast’s collection'. Overall 9/10.
I don't think I could have written a better review of it myself!
The Spectrum Pi is available from my Etsy store.
Order your's now to avoid the rush!
Also available in Commodore 64 flavour, and many others.

Tuesday, 9 December 2014

Commodore Pet 8032 Repair and Restoration

Oh dear, here we go again. Hopefully this won't take as long as my previous Commodore Pet 4032 restoration.
This is a Commodore CBM Model 8032, from around 1980/1981. The model is given as 8032-32 B. That is 80 column, 32K, business keyboard.
It is the earlier version of the 8032 with the larger section between the monitor and keyboard. This ebay score appears to have been used for spares, many of the parts are not fixed down, and at least the main board is not the original.
The board is not bolted down, and is the later universal 40 or 80 column, 16K or 32K board, dating from 1983. It is also not working. I will fix this board later.
For the moment, I have another ebay score, an 8032 board form late 1980.
This is the style of board which would have originally been fitted. It was also faulty, but I repaired it in preparation for this.
The transformer is the right type, but not bolted down, and it looks like the monitor lead has been reconnected at some point, and the attached capacitor wasn't clipped down. So this may well not be the original.
The capacitor is marked as being made in West Germany, where as the serial number sticker on the back of the case is marked made in the USA. Pets were built in both locations, so I suspect the power supply came from a German produced unit. The capacitor was rattling around loose and is a bit dented. It should be 22000uF, with had a stated tolerance of 0/+50%. It is now reading 29830uF, still well in spec and low ESR. Not bad for a 35 year old part (far better than modern capacitors which tend to be one of the main causes of failure of consumer electronics these days).
The power connector is a bit mangled, it looks like someone has tried to fix it by squashing the wires down. Given the damage to that, and I've replaced the plug.
A useful tip in this situation is to do this one wire at a time, to minimise the chance of making a mistake. Pin three is left blank for polarisation.
I'm assuming the monitor and keyboard are original, as all the screws are all present and tight. The monitor is an earlier version to the one in the 4032 and 8032SK, but required less clean up that the 4032 monitor.
There were a few issues with the monitor. it was out of adjustment, over bright and the vertical hold had slipped.
The text on the screen was intermittent, down to a dry joint on the video connector. With that fixed, and the monitor adjusted, it was working fine.
The keyboard wasn't too bad, about 90% of the keys were working, but I still went for the full strip down and clean up as with the 4032 keyboard.
As usual, the keycaps were removed and washed, and the PCB cleaned and reassembled.
All keys were working again, a bit yellowed, but fine.
The base cleaned up quite nicely, no need for a respray this time.
The power supply and board were installed and tested. All working well.
The upper case was a bit dirty, but cleaned up. From inside, you can see the extra space in this model hides two slots that look like they could have been for dual floppy drives, but they aren't the right size.
The cleaned case was reassembled with the piano hinge and bonet prop.
I'm replacing fixing where they are missing or tarnished, and cleaning up as I go.
The monitor was reinstalled and the whole setup tested together. After a few runs of memory testing and a long soak test, it's working very well.
So there is another computer restored. Quite a few faults in this one, the monitor out of adjustment and with a few dry joints. The keyboard needing cleaning to get a few keys working. The power supply plug mangled, and the mainboard faulty (to be repaired later).
Here you can see the differences between the 8032 on the left and the 4032 I previously restored, on the right. The the higher monitor stand, the business vs graphics keyboard and the 80 vs 40 character screens. I have to say I prefer the white coloured monitor bezel on the 4032, but they are both very nice machines.
Inside, the different transformers and boards can be seen, and the higher back on the older unit.

Now all I need to do is get a Pet 2001 (and a bigger bench)....

Saturday, 6 December 2014

Commodore 8032 Main Board Repair #2

Here is another Commodore CBM 8032 board to be repaired. This one is slightly later than the previous 8032 board I repaired, but earlier than the universal dynamic pet board in the long running 4032 restoration saga.
Like the other early board, it is 80 columns only, but jumpered for 16 or 32K. A few other differences from the later boards. There are three voltage regulators, two 7805 and one 7812. Later they dispensed with the two 7805s (5V @ 1A each) and replaced them with a single LM323K (5V @ 3A).
The earlier board had gold plating on all three rear edge connectors, but not the side datasette 2 connector. This one like the remainder of the PET boards has gold plated IEE488 / GPIB connector and the rest solder tinned. There is also space for a pin header for the GPIB port. It was populated on the earlier board. Here it isn't, and in later boards it was removed altogether. These earlier boards had the cassette transistors riveted down. Later boards they are left floating in the air.
They also tend to have white sockets on the ROMs and the five main chips. Being white it stands out even more that all five 40 pin chips are all missing. The ROMs are present, and there is a ceramic version of the later 901465-23 BASIC 4.0 ROM.
As you will see from the keyboard and video connectors above, it is not in the best of condition, most of the pins are bent.
It does however, have the same symptoms. Nothing on the screen. No beep.
Obviously one reason it doesn't beep is the piezo disc is missing, but it's probably a lot more than that.
The RAM is made up of 16 ceramic MM5290J-4 4116 clones. These had very long legs which were all mangled and shorting on the rear. Forgot to take pictures of those, but I did clip them all short. The lid of one of the chips was rattling around in the bag when it arrived. All of the bits must have all fallen out.
When I desoldered the chip, all the legs just fell off and left the die and the bottom of the case
The first job was to tidy up all the pins and replace the piezo and the damaged DRAM chip. Powering on with all the chips removed passed, so no problems with voltage or shorts on board. .
Next, a working set of 40 pin chips, from left to right, 6520 (GPIB PIA), 6522 (VIA), 6502 (CPU), 6545 (CRTC), 6520 (Keyboard PIA). After testing, the 6520's were replaced with new WDC 65C21N's, and the 6522 with a new 65C22N.
Time for the ROM / RAM replacement board. That booted fine, but failed with either onboard ROM or RAM enaabled. Some of the DRAM chips were getting very hot, particularly the bottom few, so I started to replace those.
It was at that point that I found that more of the DRAM chips were falling apart. The lids just peeled off. Eventually all of the lower 8 chips fell apart. The top 8 were seemingly intact.
Inside, 16384 bits of DRAM. You can just see a few loose bonding wires in the way.
With those replaced, the RAM tests passed.
The ROMs were next. I'd found the monitor ROM 091474-04 was also hot and not working. Replacing that, the ROMS were fine, and the ROM/RAM board could be removed.
That's complete and has been on soak for a few days with no problems. Another one back in service. I have something lined up for this board in the coming weeks.

Update: This board has now found it's new home in an early 8032.

Tuesday, 2 December 2014

New fully mappable ZX Spectrum+ Replacement Membranes

The Sinclair ZX Spectrum Plus has a much improved keyboard to it's predecessor. However, it uses the same computer board inside so is limited to the ZX Spectrum's 40 keys.
The connectors are the same, one 8 way and one 5 way, making a 40 key matrix. It adds extra keys using a special 4 layer membrane which actually presses two keys at a time, so there is no difference at the keyboard connectors between pressing Break or pressing Caps Shift and Space.
So the arrow keys on the bottom row actually press Caps Shift and the 5,6,7 or 8 key as appropriate. The same goes for all the keys other than 0-9, A-Z space and enter. This makes it difficult to remap the keys when making a USB keyboard, as the controller cannot know if you pressed the left arrow key or shift + 5. I get around this with the remapping in the controller, but it limits what the extra keys go do.
I built a prototype replacement using tact switches for each individual key in a 5x13 matrix. This proved the principle, but wouldn't be practical.
I did design a PCB, but it would have been quite a pain to assemble each one with 58 individual switches, it would also work out quite expensive to manufacture, even as a single sided board
So now I have a new solution. I have designed a replacement membrane for the Spectrum+. This features a 5x13 matrix with each key individually mapped and a single 18 way connector
This fits in the case as an exact replacement for the original spectrum plus one.
Interesting to note I placed the connector to line up with a small cut out in the backplate and insulating layer which is exactly the right size for an 18 way connector. Was this part of Sir Clive's plan all along?
This connects to a modified version of one of my USB keyboard controllers. The new versions now use the Spectrum Plus's reset switch to change mode from direct to remapped. The direct mode for emulators was strange to do as I had to reverse the process. Now I can individually read the arrow keys, I need to translate those back to caps shift + 5,6,7 and 8.
There is a piezo buzzer on the controller which beeps when the mode is changed. I've also set it to make a click sound each time a key is pressed. This is what the original Spectrum used to do, and it is a neat idea, so I copied emulated it. I have found it does make typing easier, particularly on the rubber key Spectrum and the ZX81 (which didn't originally do that).
So there it is, the new fully remappable ZX Spectrum+ USB keyboard. To get a batch of membranes made will be quite expensive, so I would need to get a few preorders in to justify the cost. Please let me know if you are interested.