Xl1201 on Symbolics LISP Engineering

Symbolics XL LAN Challenges

Wed, 05 May 2021 22:26:17 +0000

I recently set up a LAN of my three XL Symbolics machines at home. The XL1201 which was having stability issues has been largely stabilized (thanks to help from David Schmidt). I will post more about this LAN soon, but I thought I would mention an interesting problem.

The machines connect to the LAN via an AUI to 10BaseT transceiver, to my core LAN which is a mixed 10G/1G LAN. The core router is a Ubiquiti EdgeRouter Pro which is attached to UNMS, their central management tool. When I connected these machines to the LAN, I was getting Merlin (the name of the CPU board) Ethernet errors on occasion, and eventually the computer would crash. I put the Symbolics computers on a separate segment connected only to a small Netgear switch and the EdgeRouter, but the computers kept crashing.

Ivory CPU Heatsink and Instability

Sat, 06 Aug 2016 14:13:58 +0000

One of the theories about the instability of my Symbolics XL1200 is that the CPU has no heatsink and so it is suffering from thermal problems after a while. (15-30-ish minutes)

David K. Schmidt was kind enough to send me an old heatsink. It’s 38x38x5 mm and has 11 fins on it, as you’ll see on the pictures; slightly smaller than the Ivory CPU itself.

First I cleaned the old epoxy from the heatsink using acetone, which was surprisingly easy to do.

XL1201 Moment of Truth

Sun, 24 Jul 2016 21:17:33 +0000

UPDATE: No joy. Still self-reboots.

So, I put the Merlin II card back into the XL1201, connected the power and console, and powered it up.

Nothing. Blank screen (which means the monitor was at least getting a signal) but no boot, nothing. I hit the reset button a few times and powered it off and on a few times, and tried to reseat the card. Not a thing.

So, removed the Merlin II and took a very close look at it, comparing my before pictures to everything, to make sure every chip was in the right place and every pin was in its slot.

XL1201 Merlin II Chip Reseating

Sun, 24 Jul 2016 19:48:20 +0000

UPDATE: I clipped and removed the broken capacitor. Replacing it will be the task of another day.

I spent two days removing, cleaning and reseating each DIP chip on the Symbolics XL1201 Merlin II Main (CPU) board. Earlier I closely examined the solder side of the board and found a few minor discrepancies. This time I also encountered three discrepancies:

Bent pin on chip BB27 - unbent
Broken bypass cap next to chip GG7 - I clipped this one off entirely and will replace another day
One long bodge wire running along the entire top has come loose. This could get snagged on something and fall off.

Additionally, I removed and resocketed the Weitek Floating Point co-processor without doing any cleaning. It looked sparkling clean.

Detailed Inspection of XL1201 Merlin II Solder Side

Sat, 23 Jul 2016 19:18:35 +0000

To see if I can isolate the cause of the XL1201 instability, I am going to physically inspect the two main cards with a loupe and flashlight. If I see anything suspect, I’ll take a picture with a (cheap) USB microscope.

I am doing a detailed inspection of the solder side of the Merlin II CPU card first.

Overview picture:

Discrepancies

There are only two that are concerning to me, but several others were found. The two of most concern are:

XL1201 Stability Update

Sat, 16 Jul 2016 15:47:51 +0000

In an effort to rule out the power supply causing the XL1201’s instability, I sent it out for overhaul by Power Clinic. After reinstalling it, I booted the XL1201 and just let it sit in the Document Examiner.

Unfortunately, the computer decided to reboot itself after about 15-16 minutes. So, the power supply would seem not to be the problem.

I will have to disassemble the XL1201 and give it a thorough examination to see if there are any obvious physical defects. If none are found, I will try to remove, clean, and reseat all socketed chips, and see if that helps.

Todd MAX-354-1252 Power Supply for XL1201

Wed, 13 Jul 2016 23:28:33 +0000

So, I learned something about the Todd MAX-354-1252 power supply in the Symbolics XL1201 today that I thought I’d share. Not that I understand it, mind you.

The power supply has six trim pots. Four are obviously labeled for adjusting the four output voltages:

5V at 50A
12V at 8A, 12A peak
-12V at 2A
-5.2V at 10A (for ECL circuits)

However, according to the expert engineers at Power Clinic “The two [remaining] trim pots in question are related to the front-end boost circuits. One is for operation frequency and the other is for boost fold. NEITHER should ever be adjusted.”

More XL1201 Crashes

Thu, 07 Jul 2016 03:56:51 +0000

Well, tonight I had more crashes using the XL1201. Each time the computer crashed after about 20 minutes. This indicates to me that there’s something odd going on somewhere.

Before the first crash, I put the Fluke 289 on 1-second measurement mode of the 5V rail. It was steady the whole time, even after the first crash at 20 minutes.

I then took a few hours break.

Later, I removed both Merlin II and Merlin I/O cards, put contact cleaner on all the socketed chips, and pushed down on them.

XL1201 Instability Next Steps

Wed, 06 Jul 2016 02:44:36 +0000

Well, the instability on the XL1201 has been disheartening. However, I have a few suggestions and plans to see if I can figure it out.

Kalman Reti would like to see backtraces to see if it’s a software problem. I will definitely get these for him; awesome to have a Symbolics expert offering to help - thank you.
I’ll put the Fluke 289 data logger on the 5V power supply rail and see if the voltage has any spikes or “events” when the computer reboots or acts strangely. If not, I’ll try the other three rails too (12V, -12V, -5.2V).
There are a number of tantalum bypass capacitors on the RAM, Merlin II and I/O boards. If any are going bad, they could short, heat up, “heal” and then work again for a little while.
- Test the tantalum caps for resistance (briefly, as the resistance will change as they charge) - anything under 6 ohms might be bad. This will be very difficult to do with the backplane caps, though; I may have to learn to disassemble the XL1201 chassis.
- After the problem happens, quickly power down and remove the boards to see if any components appear overly hot on the FLIR one IR camera.
Remove the cards again and push on all the socketed chips to make sure they’re all seated properly. Possibly apply contact cleaner to the socketed chips. Maybe the whole board - they seem dirty. Not sure if this “whole board” is safe or not.
Remove the cards and all socketed chips, clean the chip and socket, and reseat them. This would take a long time as there are a ton of chips and I haven’t pulled chips in literally decades. I would need to get a chip pulling tool and be super careful as this stuff is, as far as I’m concerned, irreplaceable.

Unfortunately, I don’t have a way to run the Merlin II without it being installed in the XL1201 chassis so I can’t test the caps directly or watch the heat signatures directly without doing a massive disassembly and hack job on the computer chassis, backplane, power supply, etc.

Power Supply, Memory, Crashes

Tue, 05 Jul 2016 04:03:54 +0000

I don’t have the time to post all the details before going to sleep, but I did a few things tonight including measure the power supply voltages, install new RAM, and have a lot of problems and crashes.

XL1201 Power Supply Voltages

I tested the voltages of the XL1201 power supply with a Fluke 269 multimeter and a Rigol DS1000E, while the system was on and sitting at the FEP command prompt.

Clean Genera 8.3 FEPfs

Mon, 27 Jun 2016 11:46:00 +0000

I was asked to provide a full list of all the files on the XL1201 Genera 8.3 hard drive I received from David K. Schmidt. Here they are in screenshot form from the first boot this past Thursday.

Setting Up Genera 8.3, Part 1

Mon, 27 Jun 2016 04:48:00 +0000

I’m setting up Genera 8.3 on the XL1201 using a clean hard drive. Literally all this hard drive has on it is Genera 8.3 and these patches:

Genera 8.3 y2k patches
Copy world patch
Fepfs patch

This hard drive was sold by David K. Schmidt and has had most things set up already, so we’re not actually doing a full installation from CD-ROM.

Let me preface this by saying that I have read a lot of the documents on Bitsavers but have no actual experience with Genera and haven’t had almost any guidance on what I’m attempting to try. My goal here is two-fold:

XL1201 Power Consumption

Sun, 26 Jun 2016 21:32:29 +0000

I was asked how much power is used by the XL1201. Here are a few data points…

XL1201 with only SCSI2SD: 222-233W
Console: 106W

With the XL1201 sitting idle it takes the same amount as calculating (fib 36), so apparently the CPU is running at a constant speed at all times.

When the console goes into sleep mode (after 20 minutes by default), power usage doesn’t appreciably drop.

Console sleep mode: 105W

With the XL1201 running a 15,000 RPM Seagate Cheetah X15 drive (ST318451LW), it doesn’t really take all that much more power.

XL1201 and SCSI2SD v5 128GB

Sun, 26 Jun 2016 20:50:44 +0000

I’m attempting to use the SCSI2SD v5 with a 128G microSD card.

You can find out the size of an SD card attached to a modern Mac with diskutil list and then diskutil info /dev/disk8 (or whatever device it’s listed as). This microSD card is reported as:

# diskutil info /dev/disk8
 Device Identifier: disk8
 Device Node: /dev/disk8
 Whole: Yes
 Part of Whole: disk8
 Device / Media Name: SD USB_3_0 Read

 Volume Name: Not applicable (no file system)

 Mounted: Not applicable (no file system)

 File System: None

 Content (IOContent): None
 OS Can Be Installed: No
 Media Type: Generic
 Protocol: USB
 SMART Status: Not Supported

 Total Size: 127.1 GB (127103139840 Bytes) (exactly 248248320 512-Byte-Units)
 Volume Free Space: Not applicable (no file system)
 Device Block Size: 512 Bytes

 Read-Only Media: No
 Read-Only Volume: Not applicable (no file system)

 Device Location: External
 Removable Media: Yes
 Media Removal: Software-Activated

 Virtual: No
 OS 9 Drivers: No
 Low Level Format: Not supported

So, the microSD has 248,248,320 sectors.

SCSI 50 to 68-pin Adapters

Sun, 26 Jun 2016 04:52:17 +0000

Well, it seems that not all SCSI 50-pin female to 68 pin female adapters are created equally.

I bought several of these from Amazon, made by Monoprice. (Pictures below.) My SCSI chain in the XL1201 was then:

SCSI on Backplane -> 50-68 adapter -> cable -> ST318451LW -> cable (same cable) -> terminator

The terminator is built into the cable and is labeled “Foxconn Ultra 320.”

This was able to boot my SCSI2SD v5 card (with another 50-68 pin adapter) but not the ST318451LW.

XL1201 Mini-benchmark

Sat, 25 Jun 2016 04:09:00 +0000

UPDATED July 28, 2016: More mini-benchmark results added.

I was asked to do a simple benchmark of the XL1201 by Jerome Ibanes and share the results. Here is his requested code:

(defun fib(n) 
 (if (or (= n 0) (= n 1)) 
 1 
 (+ (fib (- n 1)) (fib (- n 2)))))
(compile 'fib)
(time (fib 30))

He asked for a few timings with values from 20 to 38.

For comparison, here is the output from LispWorks Personal 6 on a 2015 MacBook Pro with a 3.1 GHz Core i7:

XL1201 Known Working Disks

Fri, 24 Jun 2016 04:42:55 +0000

Note: After formatting a disk to 1280-byte blocks on Genera, make sure to hard power cycle reboot the computer before trying to use it for a FEP file system or any other purpose.

From my own trials, these drives work:

Seagate Barracuda ST336918N - 36G drive (up to 4,096 bytes/sector)
Seagate ST39173N - 9G drive
Seagate Cheetah ST19101W - Tall 9G drive

Per David K. Schmidt on June 23, 2016:

Seagate 18 GB drives known to work:
- Cheetah X15 ST318451LW
- ST118273N/W
- ST318416N/W
- ST318417N/W
- ST318418N/W

It seems that Seagate’s later SCSI drives will only format to 528 byte maximum block sizes per the PDF manuals. For example:

Large SCSI Disks and XL1201

Fri, 24 Jun 2016 02:57:49 +0000

TL;DR:

SCSI disks with 20M 1280-byte blocks work
SCSI disks with 25M 1280-byte blocks work
SCSI disks with 28M 1280-byte blocks work
SCSI disks with 28,870,145 1280-byte blocks works

Both Narrow and Wide single-ended SCSI drives work - the wide ones with appropriate

Chances are I was having problems with the full ~28.9M blocks initially because I didn’t power-off reboot the XL1201 after the full .

Details

I have a Seagate Barracuda ST336918N, a 50-pin narrow SCSI drive which is formatted at 1,280 byte blocks. Unfortunately, the XL1201 cannot create an FEP filesystem.

Clean Boot of Genera 8.3

Fri, 24 Jun 2016 00:04:00 +0000

The XL1201 now boots nicely from a new ST318451LW Cheetah drive that David K. Schmidt sent (sold) me. It loads FEP version 333 and a clean Genera 8.3 distribution world with just a few patches:

Genera 8.3 y2k patches
Copy world patch
Fepfs patch

Does anyone know what the hold.fep file is?

Broken SCSI Pin on XL1201

Fri, 24 Jun 2016 00:03:00 +0000

Well, I was installing a new hard drive into the XL1201 today and just before I connected the SCSI cable, I noticed a pin seemed to be missing on the 50-pin SCSI male connector on the backplane. On closer inspection, it’s definitely missing.

Fortunately, it seems to be pin 1, which is a Ground pin on both single-ended and differential SCSI. Since half the pins are ground on single-ended SCSI, that’s fine, and there are still quite a few ground pins on differential, so we’re fine. It’s the bottom left pin in the picture here.

Clean Install Genera 8.3 on XL1201

Thu, 23 Jun 2016 02:19:00 +0000

Attempting a clean install of Genera 8.3 on XL1201 using just a CD-ROM drive and a blank hard drive. However, from preliminary results, this seems to require a working Genera to install Genera again.

References:
- Genera 8.3 Software Installation Guide for XL Family Machines
- Formatting SCSI Disks

Format SCSI Disk

I have connected three devices with these SCSI IDs to the XL1201:

0: SCSI2SD originally purchased, with 12G partition - boot
1: Seagate ST336918N (TODO: URL to PDF) to install
3: Second SCSI2SD with Genera 8.3 Install CD
4: Second SCSI2SD with Genera 8.1.1 Install CD

Process: