Reparing an old DEC Rainbow PC100

[brendan_orr]

Hello, I've been a long time viewer of Dave's videos on Youtube but on the forums this is my first post.  Hello all!

When I was at my parents I dug out my dad's old DEC Rainbow 100B (with a massive 20 MB hard drive!) and tried to power this computer up.  Made sure the voltage was set to 115V and hit that switch...nada.  I yanked out the PSU to see if there are any blown caps, fuses, etc anything that an amature like me could easily diagnose and replace.  They used Nippon Chemi-con caps as I gather from Dave's videos as being 'the ducks guts' and they looked a ok.   But working with power supplies is a bit of a step up from what I'm used to and I don't know fully everything there is to know.   I've check continuity across the fuse and its ok.  There is a 3.3? 400V power resister that looks alright.  No magic smoke holes or obviously blown parts.

Anyway,  any pointers when dealing with an older PSU?  Where to go next to finding the fault if there is one?  The documents on archive.org that I found about it didn't list a troubleshooting flow chart if it doesn't power on, just to remove and get another "Field Replaceable Unit" part from DEC.

By the way, a picture: https://goo.gl/photos/tJU5c6UmbxihkECo7  And yes, the transformers are a bit rusty.

[tec5c]

Are there any markings for DC voltage rails on the PSU anywhere? You need to check if the whole thing is dead or if there are any signs of life (voltages) at the output in an attempt to isolate the fault.

Edit: Found this; http://bitsavers.trailing-edge.com/pdf/dec/rainbow/EK-PC100-TM-001_Rainbow_100_Technical_Manual_May84.pdf

Not sure if this is what you already have. But it appears that there should be a +5Vdc, +12Vdc and -12Vdc rails on the PSU. Check to see if any of these are there.

Also, this block diagram should be of some assistance. Check the outputs of the rectifiers.

[station240]

Almost certainly the problem is in Control Module, or the V-bias power supply that feeds it. If the supply is OK, then replace any small electrolytic caps in and around that module.

The other thing that can happen, is the capacitors in the output go bad/leaky and shut down the PSU.

[Urs42]

Did you check the voltage selection switch for bad contact?

I have a working Rainbow 100B, i had to replace some capacitors in the monitor. I still have to find a way to reformat the disk in my system.

[edpalmer42]

According to that block diagram, the fan is AC powered.  If 'nada' happened when you hit the power switch, the only things that were involved were those few items on the AC side of the circuit.

Ed
 

[NF6X]

Welcome to the forum! In case you weren't already aware of it, I'd like to point out that the Vintage Computer Forums is another place where you might find enthusiasm for your Rainbow project:

http://www.vintage-computer.com/vcforum/forum.php

[brendan_orr]

Thank you all for your replies.

tec5c:
Yup, I downloaded the whole collection of rainbow-related documents.  Interestingly enough the monitor document has a full schematic while the tech does not.  Perhaps its in another binder yet to be scanned?  I'll be having a probe around to find voltages and ground when I get home from work.

station240:
I'll have a go at that too.  I'm guessing this board is the Control Module?  I've no idea as there isn't a whole lot of silkscreening etchings to identify other than digital part numbers that google's not finding that easily.


NF6X:
Thanks, I'll check it out when I get home.

Urs42:
Thanks, I'll check that one out too.

edpalmer42:
Well, if thats the case then, then that's a bit more manageable than figuring out the dc portion.


Thank you all for your help and input.  I'll give it a go.

[SpidersWeb]

Usually I've found it's a case of an output being shorted out. Check the inputs to the motherboard, if they (except the earth leads) measure near 0 ohms then that'll be the issue. That situation is common (in general) and will cause the PSU to not start up or spin it's fan and appear dead.

I'm not sure what it is with tantalum capacitors, but you sure seem to find a lot that measure 0 ohms on 1980's motherboards- usually easy to find as they seem to bridge a power line to ground, but the decoupling caps next to ICs never seem to fail?

The Rainbow is a very special machine, good luck. Once you get it started I would recommend heading to VCF too.

[brendan_orr]

Well, I think I found (one of the) problem(s).  The power switch isn't making connections between the blue/blue of the SPDT switch.  Popping open the fan/switch assembly I saw the crimped end of one of the blue wires separated from the quick disconnect end.  Whether or not I did that while opening it, I don't know.  I was measuring using the molex connectors so I think if I didn't do it recently it might've gotten jostled a bit some time in the past.  Makes me worry about the hard drive.  Then again, if the hard drive and/or dual floppy drive are inoperable then it still would function as a perfectly viable text terminal!

Unfortunately, I've looked through my parts and junk bins and could only find spade connectors so I'll have to grab some tomorrow.  I mean, yeah I could solder the wire directly to the switch but I want to try to keep it as stock as original as possible.

For now I bring out the isopropyl alcohol:

SpidersWeb:
Thanks for the heads up.  Hopefully this doesn't take me down that rabbit hole.  And yes, I'll definitely be sharing my story at VCF if this is at all successful.  This really is a neat little machine and got quite a bit of use at my dad's old floral business.  It was used to send to other florists in the FTD network so it also has a modem and printer (and a little bit of burn in on the monitor).   And then there is the LK201 keyboard.  Amazing if it really was the first keyboard to feature the now standard arrow keys.  And you can't forget, it was "Dual Processor" before it was cool. 

[NF6X]

Before any more application of power, I suggest that you look inside the power supply for any paper-dielectric EMI filtering caps that look like this:



These capacitors have a nasty habit of quickly looking like this after application of power, with an exhilarating release of smoke and plasma:



If you find any, replace them right away with modern poly-film safety-rated caps with comparable ratings. I don't know if the Rainbows used these caps, but a lot of other DEC (and other) machines from the 70s and 80s used them. They are known for bursting into flames when reapplying power after many years nowadays. My hypothesis is that after a few decades, the plastic outer casing develops cracks which allows atmospheric moisture to seep into the paper dielectric insulation, causing it to break down under bias. The case cracking is typically visible, just like the cap on the left in the top picture. I don't expect poly film caps to exhibit the same failure mode, since the polyethylene or polypropylene dielectric films are much less prone to absorbing moisture even if the outer case seal is compromised.

These caps have specific ratings related to how they fail, with safety implications. Modern caps with the same safety ratings and poly film dielectrics are easily available, and we can help you pick out the right ones if needed.

Pictures are random internet pictures found in a Google image search. Many other such pictures, both before and after smoke emission, can be found by searching for "rifa capacitor". After experiencing the exciting 60 Hz arc myself in a TRS-80 Model 12, I now replace those Rifa caps on sight before applying power to new acquisitions.

Identical replacements are out there, but I think that it's better to use poly film replacements instead.

[brendan_orr]

I don't appear to see any.  Is the telltale sign is a transparent (translucent?) outer casing? As for as packages do you mean ones similiar to the burgandy one to the right?  Or also including smaller packages such as that inline one to the left.  It doesn't look like a diode.  At least there are no polarity markings.  The lead that goes out of the frame to the smaller on goes to something with quite a small trace.  I also see those bodged in elsewhere like at the right angle pins on this daughterboard.  Most are marked as '50V' and some 4 digit number.  It almost looks like a SMD cap is inside the glass(?) surround.

[NF6X]

The telltale sign is the rectangular translucent yellowish plastic body like in the pictures I shared. The components shown in your last post are not likely to be problematic. The glass ones that look like they have an SMD cap inside are exactly that. They are ceramic caps, sealed inside a glass tube. The brown one looks like a dipped silver mica cap to me.

[tec5c]

According to that block diagram, the fan is AC powered.  If 'nada' happened when you hit the power switch, the only things that were involved were those few items on the AC side of the circuit.

This is actually a very good point.

Did the fan operate when you applied power? If not, then it's highly likely that the connection to the switch that you showed us was there before you opened it up.

Quite sure that once you reconnect the blue cable you'll see at least some signs of life. Once you've fixed that up, let us know how it goes and we can go from there...

[SpidersWeb]

If the fan is AC powered as per the diagram, then yeah the switch was probably an issue. I have actually had this happen a couple of times myself, but for me it's usually been badly oxidised contacts and a few good flicks eventually got it coming right.

As for the paper filter caps - you'll probably find them when you power it up 

[SL4P]

While you're chasing the mains switch connections - I used to work for DEC, and it wasn't unusual to replace the IEC input socket / filter module (even on relatively young machines)

[tec5c]

As for the paper filter caps - you'll probably find them when you power it up 

Cheeky!   

[brendan_orr]

Unfortunately the connectors I bought were too small so I'll have to wait another day to get the connectors.  In the mean time I trimmed and tinned the ends and connected them together to a permanent "on" state for now.  But my question is this:  Must there be a load on the power supply to test that the supply itself operates?  Or even if it does operate without being connected to the mainboard, is it foolish to do so?

I know if I were to probe the pins' voltages they should all be greater than rated but still around the intended voltage without the load to bring it down.

SL4P: Thanks for the heads up.  That IEC connector/EMI filter is one beefy mother.  Probing the IEC connector to the switch-bound molex connector resulted in continuity.  More importantly the ground pin and the chasis was a good connection too.

While I was at it I probed continuity on some other connectors and I found an open loop at the fan   I'm guessing the PSU will not operate without the fan?

SpidersWeb:  I've got a few box fans to evacuate the smoke and a fire extinguisher nearby (and a glass-top coffee table that my fiance doesn't care too much about!)

[tec5c]

Must there be a load on the power supply to test that the supply itself operates?  Or even if it does operate without being connected to the mainboard, is it foolish to do so?

This is highly dependent on the design of the SMPS. Without a schematic it's hard to say, thus, it'd be best to not power it up without a load.
The width of the PWM pulse train is determined by the load, in order to determine how much power needs to be delivered.

While it appears that there is some prevention circuitry present in the design (crowbar protection, antisaturation cct) I still wouldn't advise powering it up without a controlled load.

[Rasz]

old car bulb(12/24v) is all you need to load it

[edpalmer42]

old car bulb(12/24v) is all you need to load it

I never use a light bulb as a test load for a power supply.  When a light bulb is cold, its resistance is low.  As it heats up, the resistance rises to its normal value.  So when it starts, the power supply sees a load that could be an overload and it might not start up or it could even be damaged.  I just checked standard 60W and 100W 115VAC bulbs.  They go from 17 to 220 ohms and 10 to 130 ohms from cold to hot.

It's not worth the risk.  Use an appropriate power resistor instead.

Ed

[brendan_orr]

Thank you all again.

I'll probably wait then until I get all the components put back in place until I attempt to fire this thing up.  I found a 119mmx38mm AC fan at Jameco (Part #  2125931) for only ~$14.  I've got other things to get as well and will probably shelve the electrical side for about a week.  Although I did have a slight concern.  I slid the mainboard out to see what extension boards were installed.  I saw the RX50, a winchester interface, and a memory module with one other spot open for presumably another memory module.  I did see a little corrosion on a screw near the front that held the plastic runner on the bottom of the board.  But what worried me the most was what looked like a peeling/cracked green laquer on the top side:

Luckily, the traces I probed were all ok.

In the mean time I'll clean the yellowed plastic as much as I can and pop the floppys into an old IBM compatible to see the state of the floppies.  I think I remember there was software that allowed for translation between DEC's format and IBM's

[Rasz]

60W and 100W 115VAC bulbs.  They go from 17 to 220 ohms and 10 to 130 ohms from cold to hot.

and 10 ohm at 12V is what?

[tec5c]

That crack just looks like some type of formal coating that has split. Doesn't look like the actual PCB is damaged, so should be okay?

[SpidersWeb]

My two cents, from a kiwi collector point of view, not a museum curator or EE point of view [so it's what I'd do, rather than best practice] and it's not specific to the Rainbow 100. I'm the type of guy who just plugs it in and runs away.

- paper filter caps, if present and dry, lots of smoke will come out and it will be hilarious, I've had them blow on UNIX systems and taken the time to shutdown the OS correctly before powering the unit down. My wife makes a  face and life goes on.

- load - most power supplies I've seen will have internal resistors that are suitable loads, I've never had any computer power supply become damaged from not having a load, although I respect with some designs it could be possible (I just haven't come across it)

- if load is needed - I use the motherboard. I've found the protection circuits in most of these big 80's supplies to be very reliable. The only exception I've had to this is small mini supplies - like the Commodore 64 which like to go 18V and blow up your favourite DRAM. [Reminder: this is me being me, you can use resistors or light bulbs if you want to do this properly]

- check the resistance on the mothebroard lines before powering up, there is nothing to be gained from testing the overcurrent function.

When you power it up, given it's age, it's also quite likely the hard drive is starting to lose it's original factory format, so if it has errors on startup be aware that you may need to perform a low level format on the media (it doesn't mean the drive is toast). As long as it doesn't sound like it's ripping itself apart and spins up to speed - you're usually golden.

With floppy disks, rotate the centre donut, and shine a light off the surface - if you see any spots, do not put it in a drive. You can buy new sealed media. I'd also recommend not using a valuable disk the first time you try the drive - if the heads are dirty, it'll tear up the disk.

Not sure if that's useful or not, but thought I'd share just in case.

[brendan_orr]

Well, she lives!

Off the bat there is a Message 8 - Drive A error after boot-up and a Message 27 - Memory Board error appearing additionally after running a self test.  I'll muck around with those later.  The whole thing is a bit noisier than I remember it, but then again I did replace the fan (with one that has a little over twice the airflow)

Thank you all for your input and help.  I'm glad it was actually something trivial and not anything too involved.  I was actually fearing the worse seeing that I think it was out in a outdoor storage facility for a while.

Edit:  It makes a very nice console for my Dell PowerConnect 3024:

I have to say that smooth scroll is quite buttery smooth!  I need to open up the monitor to fix the margins

Removing the memory module drops the availbable memory down to 127k.  Thankfully the RAM ICs themselves are socketed and populated with MB8264 and HM50256 modules so I may be able to get away with replacing those.  Can't seem to boot to the hard drive, it was probably used only for storage.  I'll mess with that after I can figure out what's wrong with the floppy drive(s).  For now I will bask in its amber glow as a terminal.