ZX Spectrum Update

ZX Spectrum mainboard

Some time ago my ZX spectrum had another failing in the video RAM department, but at the time I couldn’t easily diagnose the problem. Some parts of the machine still appeared to work such as the display circuit which could still manage to display the usual pattern when video RAM isn’t working. I put it aside to work on other projects until I could build up some knowledge.

Recently reading around the internet I came across a youtube channel with a nice guide for the basics of repairing ZX Spectrums. His name is JoulesperCoulomb, he has a number of good videos about repairing Spectrums and some other electronics such as amplifiers. He also sells some replacement memory modules for spectrums from his website. I watched the first video and decided I should check the power circuits on my unit.

Following his instructions I found a few problems with the power circuitry. Firstly the resistance through the 7805 was low by his standards at 22K ohms and the inductor seems to have a resistance value between windings that it shouldn’t. So there’s two issues to look at straight away. Further testing revealed that two transistors TR4 and TR5 have failed, which are a part of the power supply circuit. Apparently these are all common problems.

Luckily important components like the Z80 CPU and ULA seems to be working from what I can tell with my scope.

So my Spectrum is one sick puppy. The faulty power circuitry explains the problem completely, as I found out later that the -5V line to the video memory is not getting power whilst the 12V and 5V lines appear to be working. I’ll have to replace the two transistors and do some more testing. I may also replace all the electrolytic capacitors as they are all old with a few having been replaced by a previous owner. The inductor and voltage regulator I will have to investigate as measurements could have been affected by other components.

Motherboard: DataExpert TX531

This week I’m taking a look at the first of many socket 7 boards I have in my collection. The Socket 7 standard lasted a long time compared to most others because of the wide array of hardware support the chip-sets had for CPU and memory technology. The faster Super-Socket 7 even competed with the next generation of chip-sets from Intel, having comparable memory transfer speeds for quite some time. AMD and Cyrix both stuck with the Super Socket 7 for much longer.

Today’s board is actually my older brothers, one he bought in the first PC he bought for himself. It was an AMD K6 rated at 233Mhz equivalent speed, which wasn’t bad for when he bought it. There are a few things that set this board apart, it was made by a company called DataExpert (also known as ColorExpert) and the chip-set was made by Acer.

Judging by the date codes on the chips the board was made late 1997 or early 1998, which was very much a transitional time in terms of memory technology. SDRAM was taking over from the older 72-pin EDO and Fast Page RAM that was the staple of the late 486 and pentium machines. Many boards like this one included sockets for both, but almost all only allowed one type of memory be installed at a time. Unusually this board did allow both types to be installed at the same time, it was configured with 16Mb of each at purchase. Although this was un-documented.

I couldn’t find out much about the Acer chip-set, except that it is a clone of an Intel design. It doesn’t have much in the way of on-board hardware, just the usual storage controllers, serial, parallel and keyboard interfaces. There is some on board cache in the form of two UMC chips that make up the 512Kb pipeline burst cache.

Another interesting thing you don’t often see on digital electronics are the two glass-envelope style diodes shown here. I’m used to having seen these as a signal diode in a crystal set or basic AM radio, I’m sure that’s not the case here.

You’ll notice that the battery clip was broken, Dad accidentally did this late in this boards life when changing the cell. He installed this strip of metal as a kludge to keep the machine going for just a little longer.

In service we probably got at least 10 years (maybe a bit longer) out of this board. Which I think is pretty good considering the manufacturer! I did eventually find a manual for it, but DataExpert have long since disappeared. The manual is a bit hard to read as it suffers from engrish, but it does have the information you need for installing the board.

From the service technicians point of view it’s reasonably easy to work on, but not perfect. Setting the processor and bus speed is done via DIP switches, luckily there is a silk screen reference for these, but some details on some jumpers are absent from the silk-screen. You should be able to install and set this board up without the manual.

With information about this board scarce on the internet it’s hard to pass much more comment about it. Our example was reliable over it’s life-span, and performed reasonably for the tasks given to it at the time. It isn’t really all that remarkable really, there were many different manufacturers making Socket 7 boards of this standard and better. Perhaps that is why DataExpert isn’t around any more, they were crowded out of the market.

Clyde’s Adventure for DOS

Having not played a DOS game in a few weeks and having my other plans fall through, I’ve recently had a quick go at the original Clyde’s Adventure. It was made by Moonlite software back in 1992 and is a puzzle-platform game. It has a sequel that I played quite some time ago.

The game uses EGA graphics and has a simplistic but very effective art style. I was impressed with the animations and colouring, which in some places actually looks like it was done with VGA capabilities. There are also animations for many of the static tiles, which brings a bit of life to the levels. Even the birds in the background move around. I did notice a little flicker under Dosbox, but I suspect that wouldn’t happen on real hardware.

Sound support comes in the form of simple Adlib effects and music. Unfortunately it seemed to not sound very good under dosbox on my windows XP machine, but was ok on my macbook. Perhaps I have a setting wrong on my PC, but on real hardware (or when working under Dosbox) the music during the levels is kinda catchy and the sound effects are quite nice. However it seems there is only one song that repeats over and over, I didn’t find it annoying, but if you do there is an option to turn off the music.

The game was quite unique at the time. Firstly there are no enemies to dodge or shoot and there are only really two ways to die: running out of energy, or touching something deadly such as spikes or lava. Energy is basically your health, and unusually you use energy when you move, jump or fall a long distance. Luckily you can find more in the level, but it is generally quite sparse. This means you really need to plan your path and know the levels in order to successfully complete them. Because of this it took quite a few goes and a few hours to even beat the first level.

Clyde carries a wand to temporarily destroy some bricks, and this combined with switches and magic triggers will reveal new areas, Gems to collect or bombs that can help open areas up. These create some interesting puzzles, especially when you’ve got limited energy.

Clydes Adventure is quite challenging, much more challenging than it’s sequel. I spent quite a lot of time learning each level, dying and having to try again. But strangely it wasn’t really frustrating, perhaps because there is no lives system, and you only really die by making a mistake with the controls or path you take. Being insta-killed on spikes because I jumped a little short did annoy me a little though, especially closer to the end of a level.

Because it is quite challenging and time consuming I didn’t get to play as far into it as I would have liked. So I might have missed seeing some of the features of the game. Still I really enjoyed what I did get to play, only it’ll almost certainly take me several weeks to finish it, if I have time to play.

This slideshow requires JavaScript.

Motherboard: Aopen MX46-533gn

With things post storm back to normal I took a look at one of the more “modern” mother boards I have. I was given this board quite some time ago, and it looks like it hasn’t been used. I think it was probably new old-stock as it was quite out of date by the time I got it. I’ve kept it as a spare as my Dad is still using a P4 based machine, and I’m currently using a P4 based system for my server.

It’s an Aopen mx46-533gn which supports a socket 478 processor with FSB up to 533Mhz. This board is so far un-tested as I have not had any need for it yet and it isn’t really a great performer as we will see.

The highest CPU officially supported is a Northwood P4 @2.8Ghz with 533Mhz FSB, but the board only supports 333Mhz DDR RAM. So it’s fairly likely that the CPU would end up waiting for the system memory quite a bit.

Additionally it has an 651 SIS North bridge and 962L SIS South bridge. SIS were never known for making high performance parts, although boards that used their chipsets were frequently significantly cheaper. I remember replacing a SIS based board in my brothers machine with an N-Force 2 based board and we got a significant improvement in performance, but other value chipset makers such as VIA also usually had better performing parts.

So this is clearly a commodity consumer board, designed to be cheap and cheerful. It has on-board video like most commodity boards, but being SIS graphics it doesn’t compare favourably. There is at least an AGP 4x slot, so you’re not stuck with it. The on-board LAN and audio are the usual Realtek devices of the time. A Winbond Super IO chip provides legacy ports such as serial, parallel and PS/2 ports for mouse and keyboard.

Realtek LAN chip

Some of you may have noticed that some boards such as this will produce low levels of digital like noise with their on board sound. Looking at this board I wouldn’t be surprised if it produces such a noise. The reason is mostly the layout of the board. The nearest chip to the Audio output is the Realtek LAN chip, and not far from that is the ROM and Winbond Super IO. The audio chip is the one at the bottom left of the board. This means the audio output traces have to go past a number of high speed digital lines, which due to cross talk are the likely reason for any noise.

Working on or installing this board isn’t as easy as it could be. Whilst there is no need to set the CPU speed with jumpers, other headers such as the front panel and audio are not labeled or labeled poorly. This makes it pretty much mandatory to refer to the manual or easy installation guide.

For then end user it was probably fine, as long as you weren’t after performance. Boards like this one sold in large numbers to people who just wanted a basic computer for a bit of web surfing, email, and maybe some solitaire, and for that it would have been fine. I can’t comment on reliability as I haven’t used one in service, and we didn’t sell these where I worked.

My Old Power Bank

I was given a power bank quite some time ago that I’ve used sporadically when I’m away from home to charge my devices. Recently when I went to charge it I found that the connector was intermittant and about to fail. So I decided to take it appart and see if it is repairable.

Taking it appart was fairly easy, I just peeled off the upper label, removed 4 screws and it came appart. Here’s a photo of the board.

The PCB

The board is of a fairly simple design, but there are some notable deficiencies. Firstly you’ll notice the lack of any bulk storage capacitors, but just a few small surface mount ones. The main inductor is the only component on the other side of board, and it is pretty small. So I suspected that this power bank might have severe problems with ripple on the output. I decided to investigate, so I got out my oscilloscope.

Ready to measure

Basically I hooked up a load to the power bank, my phone and my tablet in turn, and used my oscilloscope to measure the waveform on the output pins of the power bank. Here’s a screen grab from the scope showing the waveform with my phone on charge.

Yikes! The ripple is really quite bad and appears to contain higher and lower frequency components. The most disturbing thing is the range of the voltage. 5.84V is greater than the usually 10% tollerance accepted, which could damage the charging circuit of a phone. This is likely because the board lacks any bulk storage capacitors.

Charging my tablet which imposes a larger load only seems to make things much worse with the ripple magnitude and frequency increasing. The voltage peaks at 6.4V which is way to high, understandably I’ve taken this power bank out of service because of how bad this is. I was going to repair the charging problem which turned out to be a faulty input socket, but I guess that’s no longer necessary.

I wondered how the unit would go with some bulk capacitance added, I noted that the Arduino that I have has exactly such capacitance on it’s input circuitry, so I tried connecting it and measuring the ripple. I didn’t get a screen capture, but the ripple became almost flat and settled at a peak of 5.13V. If they had added the bulk capacitance needed, I’m sure the ripple would have been reduced to usable levels.

I’m certainly no expert when it comes to electronics, but I can tell when something is not good. If you’re more interested in power banks I’d suggest you look into reviews and tests of them on Gough Lui’s blog/website. He performs much more detailed tests and has tested a number of power banks over time.