Limited resources. A tale of retrocomputing

Having unlimited access to the resources of a computing history museum has diverted my free time activities to different kind of equipment: (retro)computers and (retro)computing. I do miss the hiss of SSB or the smell of soldersmoke, but I don't mind following unexplored lands of lost knowledge. Regardless, there's plenty of stuff to fix, too!

I was going through a stash of 5.25" floppy disks when I found a textual adventure in Italian language for DOS from 1986. On modern Internet I found a person looking for a copy of the game since he had played it back then: why not sharing the joy of my discovery with him?

I had to transfer the file from 5.25" floppy to 3.5" floppy on a 80386 Olivetti machine (circa 1988). Then the floppy went into a modern high-end Compaq Presario 2100 laptop (2003) to be transferred on a USB stick. Finally the USB stick went into an Internet connected machine (2023) for a final backup and delivery to my new friend.

How cool is that? I've used some 40-year old media and time-traveled its content ~20 years ahead, twice.

Since the 5.25" floppies were DS/DD with 360 kB capacity, I could fit 4 of them in a 1.44 MB 3.5" floppy. Why not? So while I was at it I copied another disk with Epic Megagames shareware game, a Tetris from 1986 and a Z80MU (Z80 emulator).

When the USB stick hit the modern computer, the antivirus detected the Junkie virus in TETRIS.COM.

HORROR!

The problem is not the single infection, but all the write-enabled floppies I read on the Olivetti machine once I had played a bit of Tetris myself. And, worse, the infection in the Olivetti machine! I am not sure what was infected first: the PC or the .COM program. Nevermind, now.

While I could reinstall the 80386 computer (DOS 6.x), I chose to try to preserve its content and see what can be done. This means traveling back in time in order to have a functional antivirus software on a single 1.44 MB floppy disk. That was the preferred portable media, so a time-correct antivirus had to fit on a single disk.

Everyone agrees that the free-for-personal-use F-PROT antivirus in 1980's and 1990's was the best choice. The latest version 3.16f from 2009 is just too large at 9.2 MB. Reader, I'll boot the machine off a clean floppy disk and run the antivirus from another floppy. No fiddling with multi-volume .ZIP files, especially since it's not needed.

The biggest challenge for the unexperienced retrocomputing guy that I am, was to locate an F-PROT version released after Spring 1994.

While browsing old software on archive.org I remembered that computer magazines (yes, printed on paper) usually came with a CD full of shareware software. Then in a matter of minutes I have downloaded the .ISO image of a CD and finally found F-PROT 2.24A from August 1996.

When I finally saw the folder with the wanted piece of software (in a .ZIP file, of course) I realised why most of my searches failed. In DOS days, filenames followed the 8.3 format convention. In 8 characters you had to fit both a mnemonic for your product and a version number. So it is not F-PROT_224A.ZIP (11.3) but rather FP-224A.ZIP. "FP" reader, "FP"!

Directory listing of Pegasus 5.0 CD, 1994.

At last the unzipped antivirus went in the USB key and on a 3.5" floppy. Next step will be to scan and clean as many floppies as possible, while I came up with a safe procedure to deal with the hundreds of potentially infected removable media in the warehouse. 

Modern AV detecting 30+ years old threat!

Apologies for the text-only post. At least you know I'm alive and kicking.

Fault on Atmel AVR ATmega

For the blinkenlights project two posts behind I chose to use an ATmega48 because the firmware is very small, while all my other (clock) projects would not fit. I noticed that it was not scanning all neon lamps and that the current consumption was periodically jumping about +100mA.

I determined that the problem was in the ATmega48 and probed all output pins with an analog voltmeter, looking for a swing that matched the current variations.

I found an output pin reaching +3.5V in the HIGH state, and it was in sync with the extra current draw. That digital pin measured 64 ohm to ground even with the uC out of the circuit. Culprit located.

I cut off the pin with broken control circuitry and used a different one with a simple jumper on the PCB. It's just a matter of updating the firmware.

Problem solved: no more I variations and all lamps light up as expected.

Why did I use an analog voltmeter? Because I could visually see the voltage change and estimate its value. A DVM would take too long to detect the change and I would not be able to correlate it to the current flow. I could have used an oscilloscope, of course.

Scheda supplementare solare Fondital Pegasus KC

ATTENZIONE. LE INDICAZIONI RIPORTATE DI SEGUITO SONO IL RESOCONTO DI UNA OSSERVAZIONE PERSONALE PER ILLUSTRARE UN POSSIBILE GUASTO, NON SONO DA INTENDERSI COME ISTRUZIONI PER RIPARARE QUESTO O ALTRO DISPOSITIVO ELETTRONICO E DECLINO QUALSIASI RESPONSABILITÀ PER EVENTUALI VOSTRI TENTATIVI DI RIPRODURRE QUANTO ILLUSTRATO.

Mi è stata portata la scheda supplementare di una caldaia Fondital Pegasus KC che andava in errore E41: "Mancata comunicazione fra schede supplementari".

La scheda del solare è stata sostituita dal tecnico (150€+IVA, marzo 2024)  e l'impianto ha ripreso a funzionare senza errori. Ma perché non provare a capire dov'è il guasto?

Blocchetto nero al centro: fusibile.

Il circuito è semplice: un alimentatore 220 V > 5 V lineare (con trasformatore, ponte diodi e 7805!), tre relè, un microcontrollore. Ho iniziato la caccia del guasto individuando l'ingresso 220 Vac. Il tester non mi dava continuità sul blocchetto nero visibile in foto. Stando alla mia ricostruzione (e alla serigrafia) si doveva trattare di un fusibile ad azione ritardata da 6.5 A.

Parentesi sul fusibile. Questo fusibile ritardato non salta quando si superano i 6.5A (1430 W!), ma ha un tempo di risposta che dipende dalla corrente. Per ottenere l'interruzione immediata devono passare 65 A (sessantacinque A). Non ci credete? Controllate il datasheet dei fusibili Bussman! Non so quanto tempo abbia impiegato ad aprirsi, ma la scheda non presenta segni di bruciature come mi aspetterei da un circuito del genere che gestisce 1500W o più.

Ho sostituito il fusibile con quanto avevo in casa ed ho alimentato il circuito con un trasformatore d'isolamento che eroga al massimo 30W, poi "si siede." Il risultato è stato promettente, con i relè che si sono mossi e il LED con lampeggio rosso e lento. Il LED comunica che il microprocessore ha rilevato l'assenza di comunicazione (RS-485) con l'unità centrale (corretto, dato che non ho tutta la caldaia a disposizione).

Il guasto è stato molto probabilmente causato da una perdita/infiltrazione di acqua dall'impianto solare dei vicini(!!) con corto circuito o simile. Proveremo ad inserire la scheda riparata per capire se il guasto era veramente così limitato, ed avere un backup in caso di necessità futura.

La scheda supplementare Fondital.

PS. Mi piace l'elettronica Fondital perché è ancora semplice e, in alcuni casi, riparabile. O quantomeno "comprensibile" senza uno schema elettrico a disposizione.

Blinkenlights project

I spotted the dusty board shown in the picture on the right, carrying a keypad with large buttons and a good deal of neon lamps. Considering the 29 lamps it might become a weird clock, but I chose to make a blinkenlights gizmo. 

The circuit is simple, the board was certainly hand drawn and it has a comfortable 25+25 pins 0.1" pitch edge connector.

I chose to reverse engineer connections with my picture-over-picture method rather than probing with the continuity tester:

 

Front, back and numbered neon lamps.

Each lamp leg was mapped to a connector pin and everything went into a final table that looks like this:

The result of reverse-engineering the PCB. Click to zoom-in.

Two lamps ("q" and "1") go directly to the connector, probably because who drew the PCB didn't have time to look for a better routing layout. I think their "anodes" (neon lamps don't have proper anode and cathode) were tied to another "anode" pin, or at least that's what I did when looking for a connection layout that makes sense. In the end I came up with four groups almost ten of lamps, marked as A_ in the table. Lamp cathodes are connected to K_ pins.

Wait a minute. 4 anodes, 10 cathodes: that's the layout of a 4-digit multiplexed Nixie clock! I can drive the blinkenlights gizmo with one of my Nixie clock boards and a bit of custom firmware! How cool is that? It will need a heavier multiplexing strategy than 1:4 of a clock because in each "configuration" there can be more lamps "ON" for each anode. And you cannot have two lamps in parallel "ON" at once for some physics of Neon gas :)

Some cathodes "cross" just one anode, others three or four anodes. Well, that adds some extra randomness to the light show.

UPDATE 2025-04-13. I did make it light up with the Nixie clock board. Unfortunately many bulbs were flickering (an effect of age/use) and the driver board misbehaved. I have abandoned the project and recovered neon bulbs.

Tektronix VX1405 power-up sequence

Could I refrain from powering up the Tek VXIbus mainframe? I do have a rather equipped "lab" and this was the chance to use several power supplies at once.

I needed 5V at few Amps (estimated 3 to 6), +12V at 2A and -12V at 0.5A. Without rearranging the whole desk I had to limit the +12V at 1A, but that's easy: unplug fans!

Since I cannot get an automatic proper power-up sequence, I opted to implement the most logical one: 5V goes first, then +/-12V and as last the second (internal) power supply with the remaining "service" voltages.

At 5V only without any plugin it draws 3A. With the command module that increases to 3.7A. 

Without fans the +12V sits at 0.5A and -12V a fraction of 0.1A.

If the command module is installed, at this point it shows two red dots. Then I pressed the power button on the rack that controls the left over PSU and in a couple of seconds I got a green "Ready" sign. See the video below, you can hear me pressing the rack button.

At this stage it is possible to keep the rack powered with a retrofitted external power source and test the command module and plugins. This test would require too much room for my small lab, so it will remain documented for the future repairman.

VXI bus, Tektronix and more

If you could publish just once a month, would you miss the opportunity to post on the unique day of a leap year? Here's my February 2024 sum up.

Tektronix VX1405 and plugins.

It turns out that I spent the whole month after a new "toy" I stumbled upon. It's a technology from 1990's that puts together many computer-controlled instruments into a rack. They are controlled via HP-IB (GPIB) and connected with a VXI bus. You can mix the plugins depending on your needs: signal generators, digital (storage) oscilloscopes, voltmeters, ... The configuration shown here has: arbitrary function generator (HP E1445A), two channel oscilloscope (HP E1428A), waveform analyzer (Tek TVS625) and a command module (HP E1406A).

First of all: the thing shown in the picture weights more than 20 kg. It is bulky, heavy, noisy and power hungry (500W). The whole troubleshooting endeavour is described on eevblog, so I'll try to summarize what I've done.

The baby did not give any sign of electronic life. I initially checked the command module and removed the NiCd battery pack with the obvious leak, fortunately contained within the pack. Still no signs. So I checked power supply voltages and +5V, +/-12V were missing.

Therefore I moved my attention to the power supply unit in charge of those voltages and tried to troubleshoot it: a switch-mode high-current configuration without a schematic diagram. It is a very packed circuit. I understood that there are some components that are failing since increasing the temperature the PSU tries to start up. I located the fault in the AC side of the switcher, most likely in a custom part/board so there are very little chances of a succesfull repair. Even the service manual says that PSU are black boxes that must be replaced as a whole.

In the end I gave up: I have learned enough of switch-mode power supplies that now I want to see a working unit on the oscilloscope. I might try to retrofit an external PSU to confirm the plug-ins respond to commands.

Also in February 2024 I received a bunch of Yaesu FT-23R and alike handhelds. Out all of them I got one working with the battery eliminator. I've always dreamed of owning one and now I feel 17 years old when I hold it. It's amazing how heavy and current hungry it is!

Aging carbon resistors

4k7 ohm 5%, right?

Here's a post to mark January 2024. I've been doing very little in the lab because of work, flu and mostly cold temperature in my cave.

Today while trying to tidy up the desk I picked up a bunch of new old stock resistors I bought together with a drawer box at a flea market. They have a brown body and they are marked as 4k7 ohm 5% tolerance (yellow, violet, red, gold bands).

In order to sort them out I did as usual: confirm the color code value with a DVM reading. They all returned a high value, well above the expected 5% tolerance range, actually more between +10% and +20%.

I have read warnings about carbon resistors increasing their value with time, and I have met some while restoring valve equipment. But I never had "NOS" specimen.

4k7 now is 5k4!

These resistors cannot be trusted and will be used for artistic projects. I got quite a few of them with the drawer box: I will save time and avoid trying to sort them.