• Please review our updated Terms and Rules here

Fun stuff, just got a new old PDP-8/M chassis

G

gwiley

Experienced Member
Joined
Nov 12, 2021
Messages
250
Location
San Diego, CA, USA
I’m excited to now have a PDP-8/M chassis for my system which was previously open-frame with a home-brew backplane and powered by lab supplies. I thought it might be interesting to share my experiences and benefit from the knowledge of experts in this group.

PDP-8/M chassis, freshly unboxed. Also in the box but not shown here is a bezel, two plexi panels and an extra power supply. It’s a bit dusty inside but the system components that I was missing are all here.
8M chassis 01 box unpacked.jpg

Chassis on the bench with cover removed and ready for disassembly, cleaning and testing.
8M chassis 02 open box.jpg

Key items are removed and most of them cleaned up. The backplane is only part way done. I need to borrow an air compressor to blow out the card slots. Bottom left is the spare supply board. Lower right is my present system with one of the new plexi panels resting over the KC8E programmer’s console. Interesting challenge here: KC8E is for an 8/E and I have an 8/M chassis. The KC8E that I have has incandescent bulbs but the chassis is wired for a panel with LEDs. I currently have a DC-DC module with digital meter to convert +15 to +8, and might stick that inside the chassis to power the lamps (the “yellow wire” to the KC8E) unless a better solution is possible.

Side note: I’m really more of a craft beer guy, but this is where we keep the wine. LOL
8M chassis 03 parts cleaned.jpg

Measuring the power supply enclosure for new foam. The original foam had turned to dust. The old foam adhesive came off quite easily with a good coat of “Goo Gone”. Same with the foam on the inside of the top cover. A new roll of adhesive foam from Amazon is arriving today.
8M chassis 04 measuring foam.jpg

Reforming the electrolytic caps. This is the first time I’ve ever done this and have not had any experience using 45-year-old electrolytics. I searched quite a bit but was unable to find what I thought were rigorous technical papers on electrolytic reforming. However, there’s a lot of practical material available that was quite helpful. Present-day data sheets do provide formulas for leakage current as a function of capacitance and working voltage, so I had a ballpark target for the expected leakage. This was also an opportunity to characterize what happens in this process. I’ll share the data soon. I wish I had thought more about the reforming process in advance to be more methodical with data recording. This experience shows that the reforming process reduces leakage a lot.
8M chassis 05 reforming caps.jpg
1. On the left is 6,000uF 10V (C7), connected through 8.2K to the 10V supply on the left.
2. Next is 4,500uF 25V (C14), connected through 8.2K to the middle supply.
3. 3 and 4 are 24,000uF 50V (C2 and C1), each connected through an 8.2K resistor to the two 25V supplies that are connected in series to get 50V.

There’s a tape-and-reel strip of 8.2K resistors in the garage, so that’s the reason for the value choice. It’s close to 10K which I was looking for.

I’ve now put the two output caps (the smaller ones, C7 and C14) back on the power supply board and started testing it under load. This preliminary test is powered by a lab supply; haven't yet connected the transformer. So far so good. It seems pretty robust so far. The -15V and +15V are solid up to 8A and 1A, respectively, and a little beyond. I’ve only loaded the +5V output with a few amps so far. After watching Jerry Walker’s video on this topic and it’s clear that the +5V output needs to be fully loaded before victory is declared. I can share the load regulation and efficiency data soon.
 
Your power supply has the ECO applied to stop the PCB traces from burning near the capacitors.

I have an LTspice model of your power supply that makes it easier to understand how it works. It is a very early design switcher, even though it looks like a linear.
 
ftcnet said:
Please remind us what the 'gray handle' Omnibus module is or does?
Click to expand...
It's a secret module ;-) JK, this is the M8357R RX01/RX02 floppy interface designed by Roland. I had JLCPCB build a small run of these using his Gerber files. I've been testing it with the RX02 Emulator but it fails the DIRXA maindecs. It's most likely a problem with my assembly or testing because this module and emulator should pass. I'll have to get back to this after the system is running in the 8/M chassis.

The gray handles were designed by Vince and I had some printed at JLCPCB. I learned later that nylon with black finish is smoother and looks better. The STL file is available either on Vince's website or github, can't recall which.
 
Roland Huisman said:
Congratulations! A little hint for the power supply... The connector on the regulator board is a bit underrated. I've added an extra set of wires (GND and +5V) from the +5V capacitor directly to the backplane. On the backplane are a few extra connectors which are unused.
Click to expand...
Thanks for the tip! I'm copying what you did, down to the last detail... double 12 AWG stranded wires run separately from the single connection through the Mate-N-Lock connector, and even the red and black shrink tubing around the screw lugs.
 
gwiley said:
Lower right is my present system with one of the new plexi panels resting over the KC8E programmer’s console.
Click to expand...
That's a pretty nifty system. For those of us with a boardset but no backplane or chassis, do you happen to have the design files, or even just gerbers, for the backplane?
 
dfnr2 said:
That's a pretty nifty system. For those of us with a boardset but no backplane or chassis, do you happen to have the design files, or even just gerbers, for the backplane?
Click to expand...
Thanks. I could make the Gerber files available, but there are some issues that should be fixed first. The LINK L signal is shorted to GND, which I fixed with a cut and jump on the backside. Also the distance between the B and C connectors is too close by 0.75mm, so it takes a bit of care to insert the boards properly. I can probably fix it without too much difficulty. (However, I often say that, and find that one thing leads to another... etc.)

Connectors are available from a couple of sources on AliExpress. The sides of the connectors need to be cut. Maybe it would be good to clean up this design. Someone asked me about it in the Facebook group too.
 
Capacitor reforming of electrolytics in the H740 PDP-8/M supply.

The test was conducted from about Mar 29 through April 3; sorry for the slow reporting of data. Prior to testing the H740 supply, all of the output filter caps were reformed to avoid any potential issues with shorting electrolytics. I didn’t quite know what to expect, and as a result, the data capture process was somewhat lacking at the beginning where I wish I had more data. It was only after I started reforming that it seemed like a good idea to log the capacitor leakage current.

The capacitors were removed from the supply and connected in the reforming circuit. The reformed caps were:
* C1, 24,000 uF 50V, filters +39V after the bridge rectifier after the transformer.
* C2, 24,000 uF 50V, filters -39V after the bridge rectifier after the transformer.
* C14, 4,500 uF 25V, filters output of -15V regulator
* C7, 6,000 uF 10V, filters output of +5V regulator
* Later also reformed C16, 22 uF 35V, filters output of +15V regulator, just unsoldered one end and connected the reforming circuit using 25V through 8.2K.

The reforming circuit schematic is shown below. The rather odd configuration is because I have three bench supplies, each can be adjusted up to 30V. I needed 50V, 25V, and 10V and have 4 capacitors. Searched for guidance for the ideal reforming current, but could find no scientific consensus. I chose about 5 mA which is only 1/4W at 50V, thinking I probably won’t have exploding capacitors at such low power. Also, power dissipation in the resistors at time=0 is within spec. This would be 10K with the 50V supply, reforming at the rated voltage of the capacitors. I happened to have a bunch of 8.2K 1/4W resistors (close enough to 10K) so that’s the reason for that value. Probably could have chosen smaller values to control the C16 and C7 current since the voltages are lower.
Capacitor Reforming Circuit v01.png
The setup looks like this. The blue "resistors" on the left terminal of each capacitor are 0 ohm and on the right terminal are 8.2K.:
8M chassis 05 reforming caps.jpg
At the beginning, occasionally checked the drop across the 8.2K resistors. At some point I started thinking how to determine whether the measured current is leakage or if it’s just charging the capacitor. When the capacitor voltage got close to 50V, I replaced the 8.2K with 200 ohms to individually charge the capacitors (C1 & C2) very close to 50V, and then put the 8.2K back in the circuit. The capacitor voltage started to drop from 50V due to leakage. From this point, after the capacitor voltage stabilizes a bit below 50V, it’s safe to assume the drop across the 8.2K resistors is all leakage current.

Unfortunately, I didn’t record the first reading (current and time) after it stabilized, but I recall it being about 2 or 3 mA after about 12 or 24 hours from the very start of reforming. That’s about 40x more than the current when I stopped reforming after 4 or 5 days. After reforming, I disconnected the capacitors and checked the voltage occasionally. They held a charge for a long time. Caps were discharged through the 8.2K, then progressively lower values, before connecting them back in the H740.

Leakage current when I started measuring carefully is in the chart below. I have much less data for C7 and C16, so it didn't seem interesting enough to share.
Capacitor Reforming graph v02.png
 
gorgeous gorgeous gorgeous system! very cool find, excited to see what you do with it!
 
You must log in or register to reply here.
Back
Top