PDA

View Full Version : S u p e r b r a i n



JonB
April 7th, 2017, 11:52 PM
Time for an update, first of many.

After a 4 hour drive, it's here. Superbrain. Wow, big. Yes, it has a hard drive. Probably from the Jurassic period. The interface card is attached to the right of the drive frame, where the S100 expansions normally go. It's not an S100 card.

JonB
April 7th, 2017, 11:57 PM
Inside, it is predictably filthy and the boot ROM's window is not covered. I imagine it is dead. There are genuine computer "bugs" lying in the dust. And let's not dwell on the keyboard. Eurgh.

37626376253762737628

I have the Y key, it has broken off at the stem. Can be glued back on I think, if I can extract the top part of the plunger from the key.

Now for some dusting.

JonB
April 8th, 2017, 12:08 AM
POWER UP

The hard disk spins up, sounds a bit sick. Nothing else happens. I don't know if there is power to the main board yet. Te hard drive appears to have its own power supply, a toroidal transformer no less.

JonB
April 8th, 2017, 12:32 AM
Signs of stress on the power board. Keyboard plate is bent, and there's an odd board under the motherboard which is part of the HDD interface.

37629376303763137632

This is most weird.

JonB
April 8th, 2017, 12:53 AM
Something bad has happened here, that's for sure.

3763337634

There's a lot of carbonised components here. Look like diodes.

Under the monitor plate, there appears to be another power supply, Astec model AC8233. 240v in, +12v, +5v, -5v out, 50W. There is only one output connected, it goes to the drive power socket and I assume it's the +12v. So I guess the toroidal transformer has been added to supply + -5v to provide more oomph for the HDD interface, or the original supply failed.

I wonder if this led to the failure of the power distribution board?

JonB
April 8th, 2017, 02:29 AM
Hard drive controller is "act 5MSDC REV-1" and its daughter card is labelled "act HOP-0". Plot thickens.

JonB
April 8th, 2017, 06:27 AM
Tasty Tandon drive is double sided. :)

HDD is anonymous.

37635376363763737638

Looks very ST-506? The crackle finish was flaking off the faceplate so I removed it.

JonB
April 8th, 2017, 06:35 AM
Now we can see how the factory power supply has been modified. Thick brown and blue are the 240v connections. Thin blue is the +12v (assumed). They used automotive crimp connectors, what a bodge.

3763937640

The ribbon cable between the serial / power distribution board and the motherboard is soldered in. And it looks like it is going to break any time soon.

37641

It's a swine, because I would like to unplug it to concentrate on the power supply, but that is not possible unless I desolder the thing.

[edit: Actually, it isn't soldered in. I was desoldering it from the motherboard and it came out (of its low profile SIL socket).]

JonB
April 8th, 2017, 09:22 AM
The power supply board is now out and I am examining it. It's an Intertec board with part number 5002020 Rev 3. Naturally a Google search turned up nothing.

The PCB is burned through and there are missing tracks. I think the transformer is salvageable but the two components that caused the damage (a ceramic disk capacitor which must have exploded from the look of it and what appears to be a resistor or diode, except it is just a melted blob of carbon now) are utterly toasted.

There is a third part, a resistor, which survived albeit in a blackened state and I think the colour code is Black (impossible?) Brown Orange. I think it is toast, it is measuring as 129 Ohms! It's big-ish, probably 2 or 3 Watt.

Close up of the site of the explosion. Looks bad, but there are only three components to wire up plus the transformer.

37651

Removed parts: Top: transformer, middle: blob of carbon, bottom: exploded ceramic capacitor.

37652

JonB
April 8th, 2017, 10:40 AM
37653

The mystery solved, after a good look at the schematics.

I'll put R2 back in as it is in tolerance at 129ohms. R1 and C6 on order! Then I can rebuild the circuit and test it.

JonB
April 9th, 2017, 02:08 AM
Carbonised board cut away.

37679 37680

..and repair board in place. Good old Veroboard!

37681

Just need to fit R1 and C6 - then I can test it.

JonB
April 9th, 2017, 02:56 AM
Amazing what you can bodge up with parts..

37687

The yellow box is an X2 mains smoothing capacitor I pulled from a dead Chinese power supply, .1uf. The resistors are a network of 1/4W 10 ohm resistors. By parallelling them up like this, we get the desired 1/2W rating, but we half the resistance; therefore we need another parallel pair in series to bring us back to 10 ohms. Due to the size of the repair board (less than a postage stamp) and capacitor, this scheme works for me because I can build a bridge over the capacitor to make the connections for the resistor.

My main concern here is that "1/4W" is supposedly optimistic and they will run hot. Probably burn out, too. So when the parts I ordered turn up, I'll do the same thing to make a 1W 10 Ohm bridge, and this should prevent another meltdown.

In the meantime, I can test it.

JonB
April 9th, 2017, 03:04 AM
Close up of the underside of the board.

37688

To protect the fibreglass strands that were exposed when I scraped the carbon out of the way, I have coated it with Araldite (2 part epoxy resin). That hole is big enough to get my finger in!

SteveH
April 9th, 2017, 11:58 AM
Amazing what you can bodge up with parts..

<snip>

In the meantime, I can test it.

Wow, does it actually power up now?

Oscar
April 9th, 2017, 01:51 PM
Hard drive controller is "act 5MSDC REV-1" and its daughter card is labelled "act HOP-0". Plot thickens.

Wow! I think I have one of those HD controllers in my Superbrain. I tried to get information on it for years, but never got anywhere. I do know that the importer of Superbrains in Holland used another approach, but this one remained a mystery. Sadly, my hard drive crashed a few years ago - one more project!

Kind regards,

Oscar.

alan8086
April 9th, 2017, 02:32 PM
376923769337694376953769637697

alan8086
April 9th, 2017, 02:34 PM
37698376993770037701377023770337704377053770637707 37708

Just in case you still need...

JonB
April 9th, 2017, 11:48 PM
Thanks Alan. You seem to have some heat damage to the underside of your power board..

I think I have it though (tentatively). The 5v line is the only fully regulated one and needs some load on it before the other rails will read true. I was advised on another forum (http://www.stardot.org.uk/forums/viewtopic.php?f=45&t=12871) to try loading it with a 12v car bulb (a small 5W one) and the voltages changed to something more sensible. So, I decided to plug it into the Superbrain's motherboard to see what happened.

Lo and behold...

37710

So, I think the power supply is OK, although there is a lot if interference as we can see from the 'scope traces I posted on StarDot. On the pother hand, there is an enormous mains filter that I have not connected yet - this may help.

I suppose that the line of weird characters is indicative of a memory problem. I consider that if the ROM was bad there would be nothing on the screen at all, because the screen cannot be initialised if there is no ROM to run the init code. But this thing doesn't have a more modern CRTC so I am not sure that is a fair assumption.

Out with the Z80 ICE methinks!

JonB
April 10th, 2017, 02:19 AM
Setback... I rewired it properly with the ASTEC 12v supply, too (had to, it is hard wired into the 240v loom) but now I am not getting anything on the screen. I can hear that the screen is powered up though. Wondering if I have broken the motherboard further. ROM perhaps? Voltages all look reasonably OK. I'm going to take a look at it with the ICE.

alan8086
April 10th, 2017, 02:22 AM
No Problem - prettys up the thread if nothing else. The burning on that spare psu is from a IN4004 diode tucked away between that row of larger electrolytics. I tried replacing it not long ago to see if the board would work but no go. I have spares of most of the main SB QD boards if anyone needs future pics.

JonB
April 11th, 2017, 02:13 AM
While I am wondering what to do with the blank screen, I thought I'd fix the Y keyswitch stem, which was broken off.

First thing to do was to extract the remaining part of the stem from the inside of the Y key cap. Not easy. I ended up drilling a small hole dead centre, then winding a very small self tapping screw into it. This had just enough grip to allow me to pull the tiny + shaped piece of plastic out.

Offering it up to the remainder of the switch stem, I was disappointed to find that it didn't match up properly, so there was no easy way to ensure that when it was glued together it would be straight and upright. However, I came up with a scheme.

I found that the ENTER key (the red one from the numeric keypad) has three female receptacles to take the key stem (only one of which is used), and the outer two have the same spacing as two adjacent keyswitches. So I could use it as a sort of jig to hold the broken stem upright if I put the snapped off part into the left hand receptacle and fit the right hand receptacle to the key to the right of the broken switch. I put a little bit of epoxy glue on the switch plunger and fitted the ENTER key to the right hand key switch, pressing it down (not fully). I fitted some key caps around the enter key, taped a bar across the top (made of a piece of IC holder tubing) and stuffed foam in the gaps above and below the enter key. The purpose of all this is to ensure the broken stem stays square and in the centre of the plunger while the glue sets.

The whole setup looks like this:

37725

It's a bit Heath Robinson (https://en.wikipedia.org/wiki/W._Heath_Robinson), but it should work. Hopefully the stem will be strong enough by then to allow me to remove the ENTER keycap, then I can swap the switch to a position on the keyboard where it will be less used than Y.

JonB
April 28th, 2017, 06:18 AM
Well, it didn't work at all. The stem was sort of fixed firmly, then got bashed accidentally, and came apart again. I tried with Superglue but no luck, so in the end acquired a spare switch from a BBC Master computer. These switches are made by SMC and are straight stemmed, type 4 linear. I now have a proper repair, and two spare switches.

Meanwhile, I have been trying to trying to fix the mainboard and it's been a long effort that I documented over at StarDot: http://www.stardot.org.uk/forums/viewtopic.php?f=45&t=12875

Head on over there for more news.

JonB
May 22nd, 2017, 08:55 AM
And we're up! (http://stardot.org.uk/forums/viewtopic.php?f=45&t=12875&p=170420#p170420)


Well, sort of. I need a proper keyboard encoder (or the carrier board to be delivered by OSH Park!) and a suitable 8002 character ROM (-003 variant, with the 5x7 character set). Also, a new FD-1791 FDC, because right now it is running on a borrowed chip that came out of my TRS80 Model II. I also need a 50Hz CP/M boot disk so that the 60Hz mode does not get set when it boots.

I seem to be getting close now!

JonB
June 14th, 2017, 07:10 AM
After a humongous effort, documented here (http://www.stardot.org.uk/forums/viewtopic.php?f=45&t=12871) and here (http://stardot.org.uk/forums/viewtopic.php?f=45&t=12875), the Superbrain is working properly!

39086

Quite a few parts were replaced during this repair:-


Power supply rebuilt and recapped.
KR3600-PRO keyboard encoder carrier board and ROM to replace custom Intertec part.
FDC (floppy disk controller IC).
CRT8002-003 (video character generator IC).
All 32 4116 DRAM chips!
Sundry 7400 series logic ICs (about 10 of them, some unnecessarily).
The chassis fan (dead 115v noise maker replaced with a 12v quiet fan).



Conspicuous by their absence are the two Z80s, the UARTs, the CRTC and PPI chips. For some reason these main ICs survived whatever catastrophe befell the machine (my guess: the blown out power supply).

Along the way I implemented EagleTG and dmemphis's ROM reader and EEPROM carrier boards, although it turned out my TMS ROM was OK. Good thing too, as it carries the ACT hard drive's boot code! The hard drive itself is in unknown state. I managed to get the Superbrain to talk to it (meaning by that, issue head seek commands), but it won't read data at all, just times out. I will see if there is any data to be had from this drive; if not, I will attempt to reformat it. The ACT CBIOS code is, I believe, available on one of the Maslin images, so I should be able to SYSGEN it if need be.

Next step? Implementation of uIDE Superbrain drivers (http://www.vcfed.org/forum/showthread.php?57196-uIDE-on-the-Superbrain), of course.

;)

tingo
June 18th, 2017, 01:03 PM
Well done!

JonB
June 20th, 2017, 05:07 AM
I completed the "motor off" modification for Drive A so it stops spinning when not active, fitted a quiet fan (not that quiet actually - I'll have to go and buy another one) and spent quite a bit of time trying to get the drive and screen properly aligned in the Superbrain's front bezel. Not easy! Also, adjusted the monitor for a straight, sharp image.

All it needs is (either) floppy drive B: or the MFM HDD repaired. Speaking of which, I'm not sure I can see the point of trying to fix it when I will have uIDE to play with. I suppose this means I need another Tandon drive to go in the B: drive bay...

zippysticks
November 22nd, 2017, 04:52 AM
Fascinating read - well done. I have have just acquired a Superbrain myself. It works -at least to a boot prompt but I don't have any fboot disks - have ordered some from retro system rescue.

Mine is missing two keys - 2 & 3 from the top row. Interesting to read that you replaced one of yours from an old BBC - they are the wrong colour though ?

JonB
November 30th, 2017, 07:47 AM
I replaced key switches only, not the key tops, as I had those. SMC Type 4 vertical stems, came from a BBC Master, just to avoid confusion. The more common BBC Model B switches are different.