Conversation

Notices

1. Embed this notice
   root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:14 JST root42

   in reply to

   The M4 screws look like they are a bit large, but I think they will do nicely to hold the regulators in place.

   • James Morris likes this.
   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:15 JST root42

     in reply to

     With the heatsink removed we can see that the top pads are not connected.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:17 JST root42

     in reply to

     One of the 7812s took a pad with it. However on the top side it’s merely structural. The pad on the bottom side is still there.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:19 JST root42

     in reply to

     Removing the regulators requires careful drilling of the rivets. Cost saving at its best.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:21 JST root42

     in reply to

     After lubricating drive 1 and adjusting its speed using the potentiometer on the analog board, it is running reasonably well.
     Next up: replacing the voltage regulators with ultra low noise switching ones. The old ones are fine but the cases have a bit of corrosion and they produce much more heat than the new ones. The new regulators also have protection should something go haywire.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:23 JST root42

     in reply to

     Drive 1 has some startup issues. Drive 0 seems fine though. I think either the motor or one of the bearings is still gummed up. After a few tries it starts working just fine. I mean, come one! If you spent the last thirty years in a damp basement you would need a couple of minutes to get up to speed right?
     I can HEAR how it is turning too slow!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:24 JST root42

     in reply to

     • Stefan Vogt
     • Elosha

     Well, if you plug the IEEE-488 plug all the way in, things suddenly start to work! @8bitgames 's Hibernated works, and PETSCII Robots, too!
     I think this was a resounding success! From literal trash to treasure. I still have to do something about the rust on the top cover, as well as perhaps replacing the regulators with @elosha's low noise switching regulators, to protect the mainboard from future overvoltage events, and make it run much cooler. But at the start I never thought to get this far!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:25 JST root42

     in reply to

     The DIRECTORY command does move the heads and spins the disk, but ends in an error LED lighting up. Error status in BASIC is still 10 / empty.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:27 JST root42

     in reply to

     Sadly, we still have issues. The HEADER command does nothing at all. Drive doesn't even go into error mode. No spinning and no head movement. Error channel lists error 10, no error string. That error isn't document in the manual...

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:28 JST root42

     in reply to

     Drive unit 0 is attached and motor is spinning when closing the latch.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:29 JST root42

     in reply to

     The board is in, and first power on test is successful.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:30 JST root42

     in reply to

     The voltage without load looks good. I think it's safe to reinstall the mainboard and power it up. If it comes online, we can attach the drive mechanisms and cross our fingers! How are your bets?

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:31 JST root42

     in reply to

     Next step is now to reactivate the power supply. I removed the mains filter and rewired all the connections using fresh solder and WAGO clips. I also cleaned the power cord. Afterwards I tested that GND is connected to all metal parts of the case, N and L have good continuity and low resistance and that the power switch works and properly opens and closes. I think this should be safe to turn on!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:33 JST root42

     in reply to

     We have life! The machine boots!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:34 JST root42

     in reply to

     Oh wow. Suddenly we have 2Mhz and 1MHz signals!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:35 JST root42

     in reply to

     Out with the old, in with the new.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:36 JST root42

     in reply to

     I got a hint that the 74LS74 flip flop divides the clock twice. First to 2MHz then to 1Mhz. But both outputs look misshapen and measure at 4MHz.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:37 JST root42

     in reply to

     The schematics are there but not REALLY helpful as some of the numbers are totally splotchy! Will have to measure each step. My hope is that some logic chip is borked and the clock is not divided down to 1MHz.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:38 JST root42

     in reply to

     Also good news: RESET goes low, then high and stays high on the first 6502.
     However: the clock signal is 4 MHz!
     The CPUs are supposed to run at 1MHz. However IIRC the clocks are supposed to be phase shifted so that the 6502s take turns in accessing the memory, as the 4KB are shared.
     So something is iffy with the clock generation.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:39 JST root42

     in reply to

     Let us have a look at the mainboard first. Bench power supply is set to 8V and 2A and feeding into Pin 3 of P4 (+8VDC) and Pin 9 (GND). The board draws 1.8A. So far so good. Nothing gets super hot.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:40 JST root42

     in reply to

     And here ist the culprit. A mains filter to avoid RF feeding back into the grid. It used to be mandatory back in the day, but isn't used quite that often anymore. This specific type of filter is very hard to find nowadays, and IIRC my other CBM machines don't have one either. So I guess I will skip this and wire up N and L as shown in the schematics.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:41 JST root42

     in reply to

     As a quick sketch: What Commodore did was to put the filter in front of the fuse and switch. Had it put the filter behind the fuse, the fuse would have blown, and no smoke... Or at least LESS smoke.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:42 JST root42

     in reply to

     The fuse is still intact, as the machine wasn't turned on. Also the two huge capacitors seem to be fine, but probably need a bit of reforming (not due to this event, but because they are decades old).

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:43 JST root42

     in reply to

     So I went to fetch the oscilloscope to start probing. Suddenly -- POOF! Smoke, and the lights in my workspace went out. The mains filter blew -- while the machine was switched off! The mains filters on PETs and CBM Floppy Drives are between live and neutral BEFORE the power switch, so ALWAYS connected!
     I quickly pulled the plug (not needed thanks to my fuse panel) and inspected the damage...

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:45 JST root42

     in reply to

     After I mounted the mainboard back into the case, and verified that the correct voltages were still on the transformer output, I attached one drive unit and switched the device on. LEDs came on, but drive LED didn't turn off! Okay, that is not good, but we can work with that! Motor spins up when closing the latch -- that is excellent!
     ...to be continued...

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:46 JST root42

     in reply to

     By the way: I found the RRIOT! So there are three RIOTS: two regular ones and one RRIOT!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:47 JST root42

     in reply to

     Next up: all socketed ICs get reseated and the sockets get deoxit applied. The sockets are dual wipe AMP branded. None of that single wipe rubbish Commodore would later use!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:48 JST root42

     in reply to

     Now the case is to be cleaned! Here are some before and after pictures as well as the cleaning cloth…

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:49 JST root42

     in reply to

     Yeah, I think I've had it wrong. The RIOTs seem to be attached mostly to the IEEE488 port, and the 6522 seems to drive the floppy mechanisms.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:50 JST root42

     in reply to

     There are also two MOS6532A chips, also known as RIOT or RAM IO Timer. The same chip was used in the Atari 2600. Each comes with 128 bytes of RAM. I guess these are used for controlling the floppy drives and act as working memory/buffer for the CPU‘s GCR encoding and decoding. But this is just a guess. I would have to check the schematics…

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:51 JST root42

     in reply to

     Now that we know that we have good voltages let’s have a look at the mainboard. The CBM floppy drives are peculiar in that they are multi CPU systems. The 8050 and its relatives have two 6502 CPUs. One, together with some SRAM, ROM, and a 6522 drives the IEE488 bus, also known as GPIB.
     The other 6502 drives the floppy mechanisms and decodes the GCR data.
     The ICs have a bit of corrosion on some pins. But luckily all the chips can be replaced if necessary.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:52 JST root42

     in reply to

     The transformer—rusty though it may be—is humming away quietly and outputting the correct voltages.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:53 JST root42

     in reply to

     The motor of the 101SM mechanism is a slightly different variant. All the motors have tachometers, so they have four wires: two for the actual motors, two for the tacho. For the drive unit this black piece fell of, it was glued to the top, same as visible in this pic for the other motor. I wonder what it's for. First I thought maybe it's a magnet for the tacho, but it seems to be a solid plastic piece, not magnetic.Maybe some kind of protection for the cables?

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:54 JST root42

     in reply to

     And another before and after picture for the metal frame. The white crust leaves behind dark splotches. I wonder if this can be treated in some way to avoid future corrosion? So far I have used deoxit, which lubricates with some kind of synthetic oil, so I hope this will keep corrosion at bay.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:55 JST root42

     in reply to

     You know the drill... Before and after. This time it is not quite as visible. The plastics were already relatively clean.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:56 JST root42

     in reply to

     Another interesting difference compared to the 8050 with Micropolis drives is this weird contraption at the back of unit 0. It is there to support the drive PCB! On Micropolis drives the PCB is simply mounted on top of the mechanism. But the MPI mechanisms only allow for a half length PCB! So the Commodore built PCB hangs over at the end and has to be supported!

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:57 JST root42

     in reply to

     Interesting! The two mechanisms are slightly different! Drive unit 1 is a 101S and drive unit 0 is a 101SM. I wonder what the differences are, except for the slightly different motor.
     To me the motors look exactly like the tacho motors used on the Micropolis mechanisms. I do have one or two spares should the need arise...

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:58 JST root42

     in reply to

     This is drive unit 0, which I still have to clean. It looks better AND worse at the same time as drive unit 1. The corrosion on the die cast metal looks worse, but the microswitch looks better. The spindle motor looks slightly different...

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:50:59 JST root42

     in reply to

     Now everything is put back together. All things that I could reach are lubed. I am slightly optimistic...
     To do:
     * one more drive mechanism
     * testing the transformer
     * maybe getting rid of the mains filters...?

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:51:00 JST root42

     in reply to

     The read/write head is freshly cleaned, the rails have their dust removed and a thin coat of lithium grease has been applied to all surfaces that slide around.

   • Embed this notice
     root42 (root42@chaos.social)'s status on Sunday, 26-May-2024 01:51:01 JST root42

     And these are the before and after pictures of the cast metal frame. I used deoxit and an old cloth and cotton buds to remove the dirt and cover the metal at the same time with the deoxit, which usually gives a very thin coating that will protect the metal. Smells horribly though. :)
     I think I will need to lube the top and bottom ball bearings of the spindle. The top one I can reach easily. For the bottom one I probably have to unmount the flywheel, but I don't want to...