Only DEC could make a PDP-11 PSU this insane…

In a haze of jet lag, I woke up with a burning desire to get hands on with my PDP-11/44. It’s been sitting on a back burner for an intense amount of time, but even still, it hasn’t been a high priority item. But, sometimes, you can’t help wanting to work on certain things, so in this episode, I give in and have a bit of fun tooling around with the PDP-11/44!

If you want to support the channel please hop over to Patreon:

Also, we now have some epic shirts for sale!

Come join us on Discord!

Intro Music adapted from:
Artist: The Runaway Five
Title: The Shinra Shuffle

Thanks for watching!

Chapters
0:00 Who needs responsibilities, let’s work on something fun
1:00 The who, what, where, why and when
3:51 Taking things apart
5:09 Even at first glance I can tell this PSU is gonna whip my butt
6:50 Initial testing
9:06 Fan cleaning and testing one more time
10:44 Okay, it’s broken, but how broken?
14:55 Hands on with the PSU modules
16:38 Swapping different modules in and testing again
21:39 Losing the battle, but still hoping to win the war
23:10 Another rescue?!

[mitchwright]

Hi David, Mitch here, you have the schematics, I'm certain I loaded them in your truck, they're also up on bitsavers. The 11/44 PSU was notorious for failing, the DEC guys joked that 11/44 PSU repairs was a very profitable business. It's most likely the oscillator, for that voltage rail convert, is not starting up. Great to see you working on it, will make an awesome system paired with that drive. Ping me if you need ideas.

[neilbarnett]

Many DEC PSUs had an overvoltage crowbar circuit, if the 5V rail went above 5.2V, it would fire a relay that could short out and disconnect the 5V rail. Then, of course, it wasn't overvoltage, so it allowed the rail to rise again and then back to start. I had an instructor at Systime who would deliberately overdo the 5V rails and let us chase that fault. The ticking of the relay was a dead giveaway.

120 amps was fairly normal for TTL in those days.
The VAX750 had a 5V/120A PSU and a 2.5V/240A PSU, separate little 600W units.

[isettech]

I remember a PSU failure in 1978, in a PDP 11/35. The DC fuse is after the main bridge rectifier. A rectifier failure resulted in enough heat to vaporise the epoxy under the bridge rectifier. When working on it and found the fuse after the bridge, gave me a what were the engineers thinking moment.

[bobdinitto]

I was a PDP-11/40 technician at Digital's Marlborough facility where we made DECSystem 10/20 computers. It was the front-end of the Dec 10/20. The PDP-11 was one of the first computers I ever worked on. It has my favorite machine instruction set. The unibus architecture was novel and ahead of its time. My most hated component was the TU-56 tape drive. So many problems! People would come to me and say "Will you fix my TU-56?" They just couldn't do it!

[andljoy]

DEC design meeting:
" guys should we use off the shelf 120v AC fans"
" no way that is exactly what people would expect us to do, we have to do better"
" I know lets build a load of logic to run some oddball 35v AC fans that is way more DEC"
" yeh good call lets do that "

[wtmayhew]

I have worked on 1970s era switcher PSU, but not this specific DEC unit. The designs are quite different from modern, where they typically switched an intermediate voltage at a much lower frequency of maybe 10 KHz or so. You still had the huge magnetics and soup can size capacitors on top of the complexity of the switcher.

I did FORTRAN programming on an LSI 11/03 in a two-drawer cabinet with a walnut Formica top. That system had a very loud switcher PSU. So loud in fact that I used ear plugs even though it was in a carpeted office. I was mostly trying to have the whine not drive me crazy, but I believe it was actually loud enough that hearing protection was probably called for.

[rhodaborrocks]

I worked for an outfit that used a lot of DEC gear, one centre was still using a pool of 11/35s and 11/45s with RK05 disks, other areas had 11/34s, a couple of 11/24s with RL01/02 disks, and there where 11/03s dotted all over the place. Over the years I probably had to repair every type of card and disk drive installed in them but I don't recall ever having to repair a DEC PSU, they seemed to be very well engineered. I was mighty pleased about that because switcher power supplies make me nervous ever since I had one blow up on the bench after a failed repair attempt, it was so violent that I was still pulling bits of transistor etc from my hair when I got home that evening. Needless to say I toggled the switch with a broom stick after the 2nd and thankfully successful repair attempt !!

[jeffl]

Those cables at about 6 min are called FFC, or Flat Flex Cables. They are basically a PCB made with Mylar, with the copper etched, plated and sandwiched with more Mylar. More modern variants can have parts mounted on the FFC,

They are a great way to simplify and make robust connections, while greatly reducing size and cost.

[baxtermullins]

I worked with 11/45’s, 11/55’s, 11/60’s and VAX machines. I own an LSI-11. I only programmed in assembly language for radar, missile, AAA, aircraft real-time. Everything was tied together by busses with up to 128 processors. One of the best machines I ever worked with, tied to hardware and human interfaces. We wrote our own OS, editors etc. Beware of the Master Interrupt card and location on the buss. It must be in the correct slot or another board could try to take control and lock the machine until a reboot!

[NullReference]

So... it might be time to consider picking up or making a few pieces of test equipment for working on analog things.
1. a non-contact signal tracer, when working with things that oscillate or have RF this allows you to quickly trace how far the signal is getting. Given that this PSU uses oscillators... that could be useful in figuring out if the signals are getting where they are supposed to go.
2. A low voltage curve tracer. Just because the multimeter says the part is good doesn't mean it is. A curve tracer lets you see how it responds over a range really quickly, and if it's being suspect. You'd be surprised how many 'working' parts look really funny when you check their curves. You may want to get the type called a 'step' tracer or make your own, there are kits. Or you can just head to your nearest Ham radio swap meet and see if someone has one or can help you make one. These can be much better than oscilloscopes in some cases at finding faults.

[topgazza]

I was a mobile engineer with Xerox in the UK in the early to mid 80s on their successful and brilliant 9700 large laser printers. They were made and used in marketing companies and in companies mainframe setups as high speed double sided laser printers. At 7 double sided prints on A4 they were rapid. At £250, 000 plus, a fortune in 1982, they were not cheap but companies loved them. They had reel to reel tapes but were powered by the PDP 11/34 and on later models the PDP 11/44
Proper computing days

[bborkzilla]

My first computer job was in the 1980's looking after a bunch of Sun 3 computers. The servers were Sun 3/280 each with two Fujitsu Eagles. If nothing else those computers instilled in me an undying love for the 68000 architecture.

[tonyfremont]

3:29 that wire wrapping looks incredible.

[geoffpool]

The Fujitsu Eagle! 300 Meg that sounds like a airplane taking off. Awesome to see!

[f.k.b.]

My dad was an electronic technician at Schlumberger back in the early 80's. This thing reminds me of things I saw on his work bench. No clue what he worked on but brings back quick flashbacks of memories!

[actually_it_is_rocket_science]

Now this makes a father's day all the better

[musiqtee]

Wow, this PSU alone seems to be as complex as CuriousMarc’s Apollo units… Time is your friend, not “the enemy” - take all you need, as being a mere passenger on your journey is a wonderful experience… 👍

[RobSchofield]

@ 14:17 - if you look at your close-up @ 6:50, there's a couple of unsoldered pins on those supply voltage pins.

[TheHylianBatman]

For me personally, I say throw the priority list away!
I don't care what you work on, or even if you work on anything at all; getting to share some of the time you spend on these computers is joy enough for me!

[Orzorn]

Playing Lunar Lander on the Centurion at VCFSW was awesome! It was great to talk to you and hear all the trials you went through, as well as hear the gentlemen nearby share their knowledge about all the arcane workings on the machine. It was a blast, and I'm just in awe of how you got such a huge and complex machine there and kept it running.