Monthly Archives: June 2014

Back to the HP.

A brief bit of troubleshooting today. Upon starting the HP up, I got a click from the speaker. The voltages were out of whack but not horribly so.

Apologies for the terrible photographs. I hadn’t eaten and had the shakes.

Set up on the kitchen table.

HP and Tek

The thing was kinda alive. No bee-doop when it started, but there was life on the pins of the diagnostic connector. REALLY fugly waveforms.

HP heartbeat

At that point I decided to adjust the timing for the power supply. At that point the +5V rail dropped to 3.4V and it all died.

More PSU troubleshooting when I feel like it.

Compaq… “Compaq”.

Yet another old computer.

This one has broken into geek culture recently, via AMC in the form of the tv show Halt and Catch Fire.
Yes, of course, it’s the Compaq “Compaq”, affectionately known as the Compaq Mk1 or Luggable. It’s about the size of a sewing machine, and has about the same processing power as a modern computerized one, too.

It was another rescue machine, like the TRS-80. I decided to bring it home. Connected to a long extension cord, I powered it up to see what happened.
Yes, of course it went bang.

Several tantalum caps exploded. Those things go off like firecrackers. Cue the strip-down of the device.


I’m actually quite impressed with the layout of the machine. It’s fairly easy to work on, considering how compact it is. (No, Compaq is an amalgam of Compatibility and Quality).

Compaq HT cage and PSU

There’s the high voltage cage that contains the monitor, and to the right the power supply that runs the entire shebang. It’s the highest powered, smallest switch-mode supply of the era, apparently.

With the exploded capacitors on the board replaced, I decided to use an external power supply to bring the motherboard to life. Connected up to a TV set’s composite input, I present to you the first signs of life in many years!

First signs of life

After a fair bit of hunting about to find out what the DIP switch settings do on the motherboard (there’s two banks of 8 switches each. One sets things like floppy disk drive numbers, graphics display mode etc., and the other is for memory size, which the BIOS completely ignores), I powered it up in the chassis. The readable version of what it was trying to put on the TV screen. The video card needed a little work 🙂

301 401 601

Multitude of error messages- no keyboard, various other memory problems and so forth. I think those are IBM error codes.

Now.. this machine was upgraded over the years by the looks of it. We’re talking some real high dollar items- a Hayes 28.8 dial-up modem, a 21Mb Plus HardCard hard disk drive and a Diamond Flower multi-function card (includes RAM, real-time-clock, serial ports, parallel port and game adapter). By default the computer came with a single card instaled- the CGA-compatible video card. It has input for a light pen.
The floppy drive controller card also has a parallel port built in. That was an optional extra. Not the parallel port, the floppy drives and controller card.

HardCard 20

The hard drive is plastered with all the usual “removal of this sticker will void warranty” labels. It came apart, because the rubber bump-stop had dissolved and turned into a sticky tar that the arm kept getting stuck to, and the optical location sensor (the pale grey glass piece attached to the arm at the top) was filthy and wasn’t working properly.
With the bump stop repaired with a piece of my aquarium’s bubble-stone pipe and the rest of the optics cleaned up, the seek time was back to factory spec.

It has bad sectors. So what, deal with it. There aren’t any more appearing. The RAM likes to go bad more than the hard drive.

I got it all powered up and discovered the keyboard didn’t work. Took it to bits and found that it’s an odd design- aluminized Mylar discs are brought plastic-side down towards two pads on the PCB. The change in capacitance makes the circuit register the keypress. Unfortunately years of being stored in Louisiana’s humidity had made the Mylar discs disintegrate. I discovered that I could type by touching my fingers to the PCB.

Flat keyboard

Booted up into DOS.. it was useable, despite making my fingers tingle. Strange circuitry, for a keyboard, despite seeming to have a standard AT (with proprietary connector) interface to the motherboard.

Keyboard controller

Someone’s had a go at it before. Decided ultimately that the keyboard needed a case, so I bought a foil thermal blanket from Academy Sports, a pot of wood glue from Home Depot and a foam sheet from Michael’s (beginning to sound like a commercial advertisement), and set about making silvered foil discs:


And began the horrible task of putting them into the keyboard:

New capacitances

It’s not great, but it does work. That allowed me to get it all together and start using it as a computer.

Together at last
(My Nixie clock making an appearance.)

Norton Disk Doctor
(Norton Disk Doctor. The de facto standard for so many years.)

Xenon 2
(Xenon 2. Games! In 320×240 in 3 shades of green!)

Got my niece and nephew interested in it also. They were fighting over the keyboard to play Reader Rabbit.
Reader Rabbit

Got a couple other games for it. I’d read up on some of Sierra’s early games being written for CGA, displayed on an NTSC-timed composite TV. Had a poke about at the composite-out, repaired one of the issues it had:
Back porch, color burst, luma

The vertical stripes found in a lot of their games serve a purpose- on an NTSC TV set the timing quirks cause other colors to appear. Not bad for a 4-color output:

Police Quest

The Apple ][ used the same tricks to cause NTSC color. This is why the PAL versions were monochrome. You can’t screw up PAL and make color.

So.. that wasn’t enough. I had a dig about in my stash of old computer junk and found me a network card. A 3Com Etherlink device, with the old thicknet AUI connection, a co-ax for thinnet and- joy- twisted pair RJ-45 jack. Got it set up, discovered that only the 8-bit portion of the ISA slot is used for data throughput. Utilizing the wonderful mTCP suite, I present to you… an 8088 online! IRC, anybody?


Of course, that also wasn’t enough, so an old router (on top of it in the last picture) was flashed with Tomato and set up as a WiFi bridge. It was then set up to be put inside.

I built a PWM circuit to automatically dim the lights in the panel I made up to fit in the front.
PWM circuit

Bought a panel to fit two half-height drives, a 3.5″ and a 5.25″. The 3.5″ operates up to 720k single density on the original controller.
Floppy drives

Stripped down the 5.25″ drive, fitted red LED’s in both to match the originals and painted the beige plastic satin black:
Black drives

Fitted up the light panel, wired the router in. Looks real neat.
Updated front panel

Powered up, it looks as though it should have been there. Plus, you can’t go wrong with blinkenlichts.

Now fully wireless, apart from the power cord!
Mobile computing
Flight Sim 3

One day, I shall bring it to a Starbucks and confuse the MacBook-turtleneck-beret crowd. I’m a geek.

TRS-80 Model III

This machine, and a couple others are really the starting point and inspiration for this site. I figure, in that case, I ought to give them a mention.

Starting with this, one of three machines left over that were in good enough shape to recover from years of damp storage. Wikipedia has a page about it (TRS-80 Model III).

I say that about this site because I came by this via a work colleague. His brother in law was clearing a house he had for sale and these were in storage inside, no climate control (ever read the operational/storage parameters for electronics? 120F/80% rH non-condensing – Storage) in the wonderful swampland of south-east Louisiana. He asked if anyone was known who’d be interested in taking them, rather than it all going straight in the dumpster. My name was mentioned. The rest, as they say, is history.

Anyhow, this is the beast in question, after a rather hasty scrub-up to remove about a decade of grime from storage and then operational gunge.

TRS-80 Model III first sighting

48kb of RAM.
2 MHz of Zilog’s finest 8-bit processing power.
Incompatible text display.
Floppy drives.

In the true spirit of things, and considering I was at work, the first thing that happened was it got pulled apart.
This one is a fairly advanced model, having a pair of floppy drives in. That means there are two Astec power supplies inside. One runs the screen (12V, it’s basically a portable TV without the tuner), the serial card and motherboard, the other runs the floppy drives and controller card.

I’ve worked with those Astec power supplies before. They were used widely in Acorn machines, notably the BBC Microcomputer. They have a tendency to go bang when powered up, and were notorious even when new.
Therefore, in went a slightly modified Dell desktop power supply, rated at about double both the Astecs put together. Dismantled the brightness and contrast pots because they had seized solid, and presto:

TRS-80 first power up

You can see the PSU there on the left. I ended up mounting it so the fan draws in cool air from underneath the floppy drives and vents out through the grille in the top. As it forms a chimney, the fan never runs at anything other than idle and is nearly silent due to the convective nature of the box.
Works a charm there, booted up into BASIC.

It migrated around for a while, at home being poked and prodded:

TRS-80 troubleshooting

In so doing I sat down and went through the entire motherboard, chip by chip to find out how it worked.
It’s from an era when you can tell there were one, maybe two people involved in the overall architecture, and you can get inside their heads a little when you look at the design:

TRS-80 chip layout -Motherboard

Got it running properly, with the obligatory “Hello, World!”:

Hello, World!

…and as the PSU is no longer switched by the main power switch on and off, the soft-on is controlled instead. Hooked up an LED into the reset button (top right of keyboard) to show power’s applied, using the clock keep-alive rail of the PSU:

Power light

Decided to run it flat-out as a test…

TRS-80 failing at pi.

Gave BASIC some floating-point recursive calculation to do. Calculate pi to as many places as possible. It nearly got there.

Buttoned it all up and powered up the other bonus haul for it- the daisy-wheel printer that you could purchase for $1995. Gave it a thorough clean out and oil up, at the same time. It has the optional fanfold paper tractor on the back also.

Desktop computer, circa 1983

It is noisy and slow, and with a change of cable will even talk to an IBM-PC, despite adding extra line-feeds in where they need not be. The print quality isn’t bad either, considering it’s just a glorified typewriter.

Printer test

As it turns out, the controller chip on the floppy board has gone bad. It no longer outputs the signal to hunt the heads one track over- applying 5V to the pin causes the drive to hunt, so new chip needs to be obtained. Unfortunately, Western Digital’s popular FD1793 chip now commands a high price on most popular bidding sites and I honestly don’t have the patience to be ripped off for one ($45 at current market prices).

TRS-80 Floppy drive controller board

Typically also the ribbon cables fell to pieces so several hours rebuilding them meant that I was no further forward with the floppy drives than I was to begin with. However, the optional add-in RS-232 serial port card is fully functional.
With a little bit of BASIC programming, 300 baud serial communication is possible. Here it is, attached to my terminal server, displaying the output from an apt-get upgrade action on Ubuntu.

RSS feeds on TRS-80

That’s about as far as I’ve gotten with the thing so far but there is light on the horizon for this one. I do plan on getting another controller chip for the floppy drives. Those 5.25″ ones are going, to be replaced by 3.5″ ones and possibly a hard-disk emulator board that runs off Compact Flash.

Fun piece of equipment. I can see why it was America’s best selling computer of the early eighties.

VTVM redux (incl. amplifiers)

Thoughts on amplifiers, vacuum tube meters et al.

I was looking at the circuitry for this meter when I was troubleshooting it. As an amplifier, it is fairly simple. It’s a class B device.
Class B? Might have heard that before, or more possibly A or AB1 when audio amplifiers are concerned.

What does that mean? Took me a while to figure out the ones I do know.
There are a few types commonly found on the market today.

A, AB1, B, B1 and D.

I like to think of these in terms of tug-o-war teams, where you have a group (or two) of people on either end of a rope, and a flag in the middle to determine who’s pulling whom.

I’ll start with class B, as it’s common in really cheap radios. You can make one from a couple resistors, a transistor and a loudspeaker.
In terms of the tug-o-war though, the principle is very simple. Fix a big spring to the wall, and have the team pull against the spring. The flag will move as the group is instructed to pull. Note that all they can do is pull. They have no means of pushing, using a rope.
This was all good and well using vacuum tube technology, as vacuum tubes used for this type of amplification have a fairly “flat” response curve. That is, they begin responding from a tiny voltage on their input and continue to amplify the signal in a stable ratio.
Transistors made from silicon do not share the same characteristic. In general, silicon transistors do not begin to do anything until their input reaches or exceeds about 0.6 Volts. Therefore, if you have a little signal that is below 0.6 Volts, a transistor simply ignores it. Good for computers, bad for audio.

This ushers in class B1, for more than a few reasons.
First, an audio signal is recorded to provide a “full” waveform. That is, negative energy where the wound wave is pulling away from the listening person/microphone and positive, where it’s pushing toward.
This is demonstrated almost universally on an oscilloscope to draw the waveform in terms of sound pressure, with the wiggly up/down line in a sine shape. The center line is a zero-energy point. Essentially silence.
Class B can only provide one direction, that is, pulling. They can’t push the opposite way past not-pulling, so what happens is we tell them to pull halfway to their maximum and call that position “zero”. In other words, we provide a bias.
This works for tube and transistor amplifiers, transistors particularly because it gets us out of the 0.6V “dead” zone where they don’t change their output.
Net result? We can reproduce the peaks and troughs of the original signal by pulling real hard to make a positive, or letting the spring pull back against us for a negative.
This is why cheap radios go POP when you turn them on. They push the speaker out to its’ three-quarters point and call it zero.

Class A amplifiers take two class B amplifier sections and turn them back to back.
Class A is your full, proper tug-o-war crew. One rope, two groups able to pull against each other. There needs be no bias as one team can make the pushing energy by pulling against the other and vice versa. It needs little more explanation than that.

AB1 is a blend of two B1 amplifiers back to back, used for transistor amplifiers. If you place two class B transistor amps back to back you have a 1.2V “dead” zone where the amplifier does nothing in the middle, and will not faithfully reproduce the signal.
Instead, you bias each side to the point where each side is beginning to conduct.. then feeding a tiny signal in makes a change. Imagine we bias to 0.7V, then change to 0.8V.. the output of the transistor will change in relation to the input.
This is why it’s important to set larger transistor amplifiers up correctly, otherwise one “team” will be pulling against the other and you’ll end up with distortion, or worse, if they are in balance but both pulling hard, the amplifier will have to exert that energy and expel it as heat.

Class D amplifiers rely on physics of loudspeakers to operate. They are usually found in sub-woofer amplifiers, because they are not progressive amplifiers. They are either all-on or all-off. Nothing in-between. If you switch on the current to a large sub-woofer speaker, it takes a finite amount of time to react and accelerate in the direction you give it the “shove”. If you turn the current off before it reaches the end of its’ suspension travel, it’ll slow down against the reaction of the coil and springiness of the suspension.
Pulse it enough and you can create very powerful bass notes in this fashion. They are not high fidelity, but at low frequency that is not so much of a problem. Especially if all you wish to do is shake the floorboards.

I think next up will be diodes and rectifiers.

Time to wake up now!


After having given a very brief run-through of my VTVM, I realized there’s a bit of explanation missing.

Why would I want to have a fifty year old piece of test equipment sitting on my bench? Why not a nice new digital multimeter?

Well, there are a few simple reasons, and the’re mostly based upon what a digital meter does.
I do have a digital multimeter, and it’s a very nice, versatile tool. It provides me with easy to read, auto-ranging Volts, Amps, Ohms, capacitance, frequency, diode testing but to name a few functions.

That sounds wonderful, and in all essence it is. However, a trip down memory lane is in order. In days gone by, if you wished to measure a voltage, you’d place a calibrated moving-coil meter across the points to measure, and the deflection of the meter needle against a scale would give you a reading. My VTVM has such a moving-coil meter attached to its faceplate.
However, in order for the meter to move, it must pass a current. My VTVM’s meter is “200 micro-Amps to full deflection”, which doesn’t sound like a lot (and in the grand scheme of things it isn’t really) but that current draw is too much to measure sensitive circuits where 2-3 micro-amps is all that’s being dealt with. Connecting a regular meter across the circuit would present it with an effective dead short, killing what you were trying to measure. Or, at best, it’d make the reading highly inaccurate*.

Examples of such circuits are radio-frequency tuned circuits, where the signals being tuned into are miniscule. What is needed is a method of connecting a meter with a very high resistance- so high that the circuit to be measured hardly “notices” it. But, a very very high resistance meter will pass next to no current, and wouldn’t move. Enter, the vacuum tube amplifier.

(For those of us young enough, vacuum tubes are.. er, a bit like glass transistors. Er, yeah. That’s a poor analogy, but it’ll do for now.)

Vacuum tube amplifiers do what the name suggests. They take a small signal and make it larger. In some cases, they need to pass virtually no current in order to operate. See where we’re going with this? Sure. So, if we connect the vacuum tube amplifier up to the meter, a tiny little signal can be made to make the meter move.

That’s basically what is done. A very high resistance is placed between the testing probe and the amplifier (in this meter, that’s 11 Mega-Ohms, or 11,000,000 Ohms) which is an improvement over the resistance of the meter’s coil (about 200 kilo-Ohms, or 200,000 Ohms).

This will lead on to a little very basic vacuum tube concepts and amplifier theory. But, for now, that’s basically why it’s used. Also, all the old manuals I have are marked out to be measured using one of these devices. If I connect my digital meter.. first, it can only read up to 2.0 Mega-Ohms before it goes off-scale. My VTVM can read up to 1,500 Mega-Ohms. In fact, my VTVM can measure the resistance across my digital meter (about 8 Mega-Ohms).

Hence why it’s not in the trash any more. It’s still very useful.

* Side note: Yeah, Volt-meters actually aren’t. They measure current, through a fixed resistance. Sorry.

Heathkit VTVM

A while ago a friend of mine saw this bring pitched away in back of a goods recycling place (sorry we can’t accept electronics).

It’s a late fifties Heathkit VTVM (Vacuum Tube Volt Meter).
I’d been on the lookout on and off for one for a while, so of course, this was a welcome surprise. It was very dirty- pictured above I’d already cleaned it and polished the face.

The board had flexed with age and some of the traces had come off. Soldered the traces back up and glued them back down onto the board. Pictured here, setting the pilot light back onto the board.

Inside wasn’t too bad really. Gave it a rudimentary clean-up.

Powered up, this is all it would do. Pegged to one end of the scale.

Found a few resistors that had gone high value with age. Shotgunned the lot, and was greeted by a nice zero. But, that’s all it’d do.

Dug about a little more and found a cold solder joint on the amplifier tube. Tubes run warm.. so, gratuitous vacuum tube shot:

With that done, it was tested using a 9V PP3 battery. Right on the money.

Buttoned it up and bought some new leads for it. Nice working meter now. Not bad for half a century old.


Recently picked this beast up- neighbors had decided to junk it.

It’s an X-Treme Scooters X-360. 36V with a top speed of about 15-20 mph, 15-20 mile range. It did have the keys, but my nephew is light-fingered and has a fetish for keys, so those went missing. New lockset and keyswitch required.
It needs new batteries (the ones in it are stamped 2003) and the controller module appears to have burned up. Needs a couple new tyres because it looks like it’s been sat in gasoline and the tyres have swollen up like a prize boxer’s face.

Might try fix the controller, but a completely brand new one is $49. Typically the batteries are going to be about $70 also. It’s not on the high-priority list right now, but it should be fun to tool about on.


The new DUARTs arrived!

Soldered in. Powered up.


Just a plain raster. No beep. Checked the voltages again and everything’s drifted off spec. The +12 was at +12.7 which is worryingly high. The EEPROM is marked to write at 12.7V

I’m hoping it hasn’t wiped the blasted EEPROM of its’ code- if it has, the entire thing’s trash. I need to go through the PSU component by component. Tried tweaking it. I forget which of the voltages you are supposed to set and the rest are meant to follow.

With the +12 set at +12.0 I get +36.3, +5.6, +12.0 and -10.8

That’ll be a fairly epic failure then. Something’s gone all burned up and way out of spec in the PSU.

Not a project for tonight. I’m not in the mood for it right now. It’s in the naughty corner again.

Small beginnings

First post on this site. Like most inauspicious beginnings, this one is somewhat tenuous but serves one purpose:  To remove the default post and title.

Expect more. (A lot of it post-dated, imported from the large scatter of sites I have this posted on).