CHI-2130 anyone ?

[Alex D]

Good day folks,

the people at the vintage computer forum sent me here - so i am shamelessly copying me message from there to here ;)

Is here anybody with experience on a CHI-2130?

We acquired two systems, one complete and one for spares and the system is not turning on. Only the relay that is triggered directly by the power power button klicks but nothing else. (Power control K11)

The switched contacts on it do not seem to get power.

The machine is wired for single phase 220v

We needed to bridge the 4 big input filter things that posed a big enough capacitance to protective earth that the fuse triggered, should not be a problem at least for testing (?)

Anybody with ideas is welcome.


Btw we also acquired a ibm1130 system, but we first want to get the chi going.


Regards
Alex

[John Pierce]

On Mon, Aug 14, 2023, 11:17 PM Alex D <rou...@gmail.com> wrote:

...

Is here anybody with experience on a CHI-2130?

I actually worked for CHI in 1975, but I was purely software, and I've not seen one since so I can't be of much help.

If the inrush current of the DC filter capacitors is tripping your AC breaker, then you need a higher current AC circuit.  Bypassing the capacitors will just cause noisy DC power which will cause all sorts of digital logic gremlins.   If you are blowing DC fuses, it's quite possible those capacitors are leaky, and drawing too much current.

[Alex D]

Good day,

i am sorry if i was unclear, what is triggered is the ground fault fuse, not the AC overload.

I am talking about the one that triggers at 30mA. It triggers because the caps leak to their chassis.

Man it would be easier if i knew what is the correct term in english :D

I was debating with the colleagues, me wanting to disconnect the safe earth and grounding the machine chassis internally to earth, but they wanted to bridge the caps.

I guess (?) as long as it is not even powering on we will not have to worry with logic faults :D

Regards

Alex

[John Pierce]

That sounds like those big filter caps need replacement then.   Not surprising considering the age, large electrolytic capacitors definitely fail with age and these are pushing 50 years old.

If I recall correctly, the 2130 was almost entirely made from 7400 series TTL logic so almost all of the power is 5 VDC, but I don't know how the power distribution worked on the 2130....   There were plenty of systems that distributed a higher unregulated DC voltage like 8VDC and had distributed 5V regulation on the various circuit boards or modules, S100 stuff worked that way.

The 2130 was a cool piece of kit...   I think it's base clock speed was 3-4 x faster than the 1130, they commonly had 32-128k words of RAM, where memory over 32k was configured as selectable banks, that allowed the CYTOS timesharing system to act like up to 4 1130 systems each emulating DM2, and they supported a variety of larger and faster disk drives 

--
You received this message because you are subscribed to the Google Groups "IBM1130" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibm1130+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ibm1130/fcf92a8a-677f-438c-b29b-aa169c4e6c08n%40googlegroups.com.

[Al Kossow]

On 8/15/23 1:01 AM, John Pierce wrote:
> That sounds like those big filter caps need replacement then.

Old machine bringup 101
Don't just power on an old system and expect it to work.

Basic system restoration calls for the reforming of old electrolytics
https://caps.wiki/wiki/Capacitor_Reforming

When we restored the CHM PDP-1, we spent a few months doing this. Very
few needed to be replaced.

You also bring the power supplies up disconnected from the logic and
test them under load until you are certain they are working well.

[gah4]

The US term is GFI, Ground Fault Interrupter.

That should be on the primary side of the transformer, and not notice leakage

current to ground on DC side filter capacitors.

There might be RF bypass capacitors on the primary side, though.

Usually they connect to ground, but I suppose there could be leakage.

GFI will usually also trigger on excess current just like usual circuit

breakers. 

-- glen

[John Pierce]

On Tue, Aug 15, 2023 at 9:23 AM gah4 <ga...@u.washington.edu> wrote:

The US term is GFI, Ground Fault Interrupter.

That should be on the primary side of the transformer, and not notice leakage

current to ground on DC side filter capacitors.

the chassis is probably grounded to the AC ground, and those big caps, if they are the classic metal case kind, could well be clamped to the chassis, and have separate + and - electrical terminals which are supposed to be insulated from the metal case.     I dunno, pure conjecture, I haven't seen a CHI since I left their employment in 1975, almost 50 years ago.

What Al said about gradual bringup reminded me of the big Variac my friend who did old audio had on his bench.   An old amplifier would always be brought up slowly on the Variac with volt meters connected to the DC supply side of the amplifier.   His bench Variac had AC volt and amp meters built in.

[Alex D]

Dear all,

I am on my mobile so bear the shortness.

Thank you for your suggestions, the 4 big metal case things are labeled fl1-fl4 in the schematics. I will upload a picture as attachment of this Mail if this works. 

As said they are metally screwed to the chassis and thus connected to protective ground.

I second that we should gradually reform the caps and power up the single supplys. I am a bit worried though if it is okay to power them without load. Those are only big linear psu‘s so I guess it is okay(?)


Did it work, I attached pictures here 😌

[Alex D]

[Alex D]

[Alex D]

[Alex D]

[John P. Doty]

GFCI's trip at 5 mA nominally, so if you have a leakage impedance <44k ohms, (220V)/(5 mA), you'll trip. So, get a multimeter. Unplug the computer. Turn on the power switch. Measure resistance between the hot and the ground on the plug. Anything less that a few 100k ohms is a sign of trouble.

But another possibility the EMI filter on the input power. A 70 nF capacitor has 44k ohms impedance at 50 Hz, so anything larger could be trouble. I've seen larger capacitors on old equipment (and gotten a modest shock from one improperly grounded example in the days before GFCI!). Most multimeters these days can measure capacitance, so get one that can, and measure. If it's too high, you'll have to avoid using GFCI or re-engineer the filter.

--
You received this message because you are subscribed to the Google Groups "IBM1130" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibm1130+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ibm1130/CAJnkzXKx_fpoji2e2qq2BChYwGOZVqePK%2B4J0LHNxaSJOztUCA%40mail.gmail.com.

-- 
John Doty
j...@noqsi.com

[Alex D]

Dear John,

We already measured capacitance to case, and although I do not remember how much it was I remember one of the old(er than me) guys immediately stating „no wonder the fuse blows“ … so yeah 😄

As stated, maybe we could ground the protective earth inside the machine.

None the less I hope somebody who knows those boxes will chime in to our round here and help understanding the Power up Troubleshooting. We have tons of schematics but no real manual sadly 😌

Our gfci is 30mA for sure. 

I can not thank you guys enough for your thoughts and comments 😊

[gah4]

As far as I know it, not actually living there, in Europe it is 30mA.

But yes, 5mA in the US.

OK, 0.01uF capacitor at 220V and 314 rad/s is 700uA.

(I am not sure now which capacitor they use.)

-- glen

[John P. Doty]

On 8/15/23 13:58, Alex D wrote:
> As stated, maybe we could ground the protective earth inside the machine.

Sending the leakage to ground won't make the GFCI happy. It's *intended*
to trip on leakage to ground, as that might be flowing through you. You
have to block the leakage.

--
John Doty
j...@noqsi.com

[John Pierce]

I'm not sure those K1-K4 are capacitors, at least I've never seen capacitors that look like that.     Traditionally, K is the designation for a relay or contactor.

[Alex D]

John, the K was written on there by one of us. 

In the schematics it’s labeled fl1-fl4

Filter line ?

[Alex D]

John D:

The idea is to cut the protective earth and ground the machine to the live earth. This way the power will go back the correct way. Of course this will probably lead to further problems once we connect other devices 

[John P. Doty]

By "live earth" do you mean the *neutral*? Using that as ground illegal
and dangerous, since faults that make the neutral hot are common. Your
GFCI is telling you you have an issue: for safety you should track down
what that issue is and fix it. Kludging around it could kill somebody.

On 8/15/23 15:03, Alex D wrote:
> The idea is to cut the protective earth and ground the machine to the
> live earth. This way the power will go back the correct way. Of course
> this will probably lead to further problems once we connect other devices

--
John Doty
j...@noqsi.com

[John Pierce]

On Tue, Aug 15, 2023 at 12:00 PM Alex D <rou...@gmail.com> wrote:

John, the K was written on there by one of us. 

In the schematics it’s labeled fl1-fl4

Filter line ?

FL is "Filter" ...  any passive filter circuit I'm familiar with that's more complex than a bypass capacitor has at least 3 connections, input, ground, output.    Usually they are a mix of capacitors and inductors.

finding FL1-FL4 on the diagrams, they appear to be in SERIES with the incoming AC connections, and are on the exit side of an 'RFI Enclosure", so these would pretty much have to be some sort of inductor.

btw, are you trying to power this with 3-phase or single phase?   The diagrams show both configurations, but they don't make it clear if the 3-phase is 'delta' or 'wye'.   It looks like there's 5 wires, so that implies a Wye system, 3 phases plus neutral plus ground.

[gah4]

While neutral faults on low(er) current outlets are not rare,

it seems that they are on higher current outlets.

In the US, until somewhat recently, the usual outlet for

over/stove/range (10-50), and for electric dryers (10-30)

has a common ground/neutral wire. 

The heating elements run on 240V, but motors and light

bulbs run on 120V, using the neutral/ground.

New installations use the 14-50 and 14-30 outlets, but

there are still plenty of the old ones around.

It seems that problems are rare enough.

(No-one has suggested trying to replace all the old ones.)

That said, we have no idea what the rules are where said machine is.

Also, rules are usually stricter for home use than industrial use.

-- glen

[Alex D]

Dear all,

the Machine was already wired for 1 phase operation,  i checked it and it looks okay as per the manual and my own logic.

In fact, the conversion is just to wire 2 phases to one, plus ground and protective ground. The third phase is only used for powering the external disk subsystem and just left disconnected.

The Machine is located in Germany/Switzerland and running in an industrial complex. In the old days here where there was no GFCI every outlet was what we call "classical earth" where the protective

ground was directly wired to ground/earth in every outlet, and every outlet only connected by 2 cables from the breaker box.

So by doing the same procedure in the machine (or in its own plug) i was under the impression i may be half way safe by 1966 standards, at least for testing.  The Power connector is indexed of course so that live and earth can not be switched.

But as for now i won't do that since you guys are in line with what the others say and we just bridged the filters. if the machine ever decides to turn on its PSUs we will need to source something.

Ah, i wanted to remove the Filters and try some reforming but removing them means complete disassembly of the AC cage and currently its SUPER HOT here. I have a big input filter of an old frequency converter that may be used, too.

It is discrete built with inductors and capacitors and most importantly rated to 3.5kW.

I appreciate for you guys being a voice of reason :) Still hope that somebody that knows the PSUs and the rest of those machines well may chime in.

Regards

Alex

[gah4]

On Tuesday, August 15, 2023 at 11:48:15 PM UTC-7 rou...@gmail.com wrote:

(snip)

 

Ah, i wanted to remove the Filters and try some reforming but removing them means complete disassembly of the AC cage and currently its SUPER HOT here. I have a big input filter of an old frequency converter that may be used, too.

It is discrete built with inductors and capacitors and most importantly rated to 3.5kW.

Reforming is for electrolytics, not usual for power line filters.

I suppose they can still fail, though, especially if overheated.

-- glen

 

[Alex D]

Ah, i was under the impression that a filter circuit consists of a L and C component, where some of the C are connected to ground. The measurements of a capacity to the chassis of the  filter is fueling that believe.

Is it electrolytic though?  I guess not 😇
I know I have seen old bipolar electrolytic caps though, 90% sure

[John Pierce]

An LC lowpass filter has the L in series and the C to ground.    The schematic shows no ground connection suggesting these are purely series inductors

--

You received this message because you are subscribed to the Google Groups "IBM1130" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ibm1130+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ibm1130/29361257-ec43-4ef2-8051-18a4489ca27en%40googlegroups.com.

[Alex D]

Hey John,

i know, but - 

there is a measured capacity to the chassis of the metal cups. is there any logic reason why if there was only an inductor - and further - there should not be any reason why the GFCI triggers when those are in place.

Please correct me if i am wrong, am here to learn :)

isnt it at all possible that the machine was built without GFCI in mind, pretty sure there were none at the time, so having that capacity to chassis ground was not a problem at all by 1966 standards.

have a nice day!

[Alex D]

Maye the schematic is just incomplete here. I am having a hard time figuring out the logic flow of what turns on what somehow, have repaired quite a bit of electronics analog and digital though, but this schematics is giving me a headache. would be SO NICE to have a true service or engineer manual that describes the operations. At least i "repaired" the first fault - the power switch was un operational on BOTH machines. This is what you get for wiring an inductive load directly to a switch i guess..

[John Pierce]

On Thu, Aug 17, 2023, 2:11 AM Alex D <rou...@gmail.com> wrote:

Hey John,

i know, but - 

there is a measured capacity to the chassis of the metal cups. is there any logic reason why if there was only an inductor - and further - there should not be any reason why the GFCI triggers when those are in place.

Please correct me if i am wrong, am here to learn :)

isnt it at all possible that the machine was built without GFCI in mind, pretty sure there were none at the time, so having that capacity to chassis ground was not a problem at all by 1966 standards.

The logical place for an AC RFI filter to be grounded would be neutral, not ground.

Btw, the CHI 2130 was mid 70s, not 1966.

[Alex D]

I know, just attributed this to another design oddity that is out of my ability to understand.

There are at least two other places on the front panel where capacitors are directly connected to the chassis, so this seemed somehow realistic.

[Alex D]

Sorry for all the noise from me.

I have been digging a bit - what is missing is a signal from SAC EPO Interlock / J3 (see schematic)

Anybody has a clue what this could be? Would need to dig through the machine where this goes, could this possibly be "Emergency Power Off" and a hidden switch somewhere.

Not that there is a keylock somewhere i have not found. Will post back as soon as i get back to the machine again.

[John Pierce]

On Thu, Aug 17, 2023 at 3:29 AM Alex D <rou...@gmail.com> wrote:

Sorry for all the noise from me.

I have been digging a bit - what is missing is a signal from SAC EPO Interlock / J3 (see schematic)

Anybody has a clue what this could be? Would need to dig through the machine where this goes, could this possibly be "Emergency Power Off" and a hidden switch somewhere.

Not that there is a keylock somewhere i have not found. Will post back as soon as i get back to the machine again.

EPO also could be 'external power on' which would be an optional external power switch or sequencier...  say you had a datacenter with 6-8 of these in a row, you would not want them all to turn on at once, so you would have them wired to a power sequencer that turns on one, waits N seconds, turns on the next, repeat...

[Alex D]

Success! It seems the machine is supposed to run only with a storage attached. Bridging that jields it functional. I deployed some wait opcodes and it looks promising!

https://youtu.be/SBu3cDoJAEY

[Alex D]

Even more success. After repairing most of the psu‘s the machine is running nice 

https://youtube.com/shorts/Mw55HNUFe9s?si=Kl3md8SorC1bDk4I

[Alex D]

Scanning our Documents is coming along. I modified the CHI with a modded version of Carl Claunchs 1130 Loader to easily load the Diagnostic Cards into it.

The CPU has at least 1 problem, this will make it easier to switch between software for debugging.

[Alex D]

Yesterday all cpu single card tests ran okay. Still I have problems with the terminal / emulated printer. I can single step output characters, no problem, but a soon as it gets faster the machine is completely locking up immediately, stopping with errors of all sorts. 1/10 it keeps running and printing random characters.

What is also not working is the automatic indentation the manual talks about (see screenshot). So either there may be a problem with the interrupts or the interface card itself i guess.

Ps: if I am generating too much noise here please give me a hint. No hard feelings.

Regards 

Alex

[Alex D]

[Alex D]

So, yesterday we traced it down to a specific issue that i fail to understand with the XIO IOCC and our test program and mine.

when i output a char with XIO, in pseudo code, the IO Word looks like

10: 0012    // ref to Char Location

11: 0900    // write to printer 

12: 3C00   // character 

and it works beautiful. I can loop around the XIO instruction and output perfect characters all day long.

but, the program zcrdumpc is using 0901 instead of 0900 @ location 4F after the whole self-modifying (confirmed in emulator and the real machine).

What happens is, on the real machine, there is only 1 char output and then it is jumping into nirvana, halting with operator check.

When i replace 0900 with 0901 in my test output program, the output character is changed.

Can anybody explain what the (undocumented?) last set bit in the write IOCC is supposed to do? Is the zcrdumpc program supposed to be also running on a real machine?

Regards

Alex

[Alex D]

Small progress - 

turns out the chi has an 1130 compatible mode, that converts illegal instructions to wait, and an advanced mode that has special instructions and a floating point unit.

When you have a program like the zcrdumpc that is double using a illegal instruction as buffer and wait for interrupt - this will obviusly only run in 1130 mode.

Once this was fixed, the program is running, but the output is only okay when i modify the statement in in 004f from 0901 to 0900.

https://youtube.com/shorts/UjKMmJukB2o?si=kYNGr_Tu9YQa23d3

But with 0901 all characters are garbled with a constant value. For Memory 4321 it is \X!P!\ .

So i am still looking for a clue, wonder if the console card is defective or if the one set bit is triggering something special.

Can hardly believe though, the machine was advertised as highly compatible

[Alex D]

So,

let's say the gate that is placing the the IOCC status word in the accumulator is defective and also doing this when bit 15 is set in console write, i think this had potential to destroy our output character...

So this will be the first thing i will try to figure out this evening. The logic that does this for me is not to be found in the CPU Cage so i will dig myself in the Interface Card schematics this evening.

Expect pictures of me showcasing my Logic Analyzer and/or Digital Oscilloscope ;)

[Alex D]

After digging my head into the schematics of the machine it is simple.

The "highly compatible" Machine where according to the advertisement "every piece of software will run unmodified" is not compatible with a XIO write with bit 15 set.

This triggers a line that reads "Enable Ascii" where it disables the Character Roms and outputs Bits 1,7 and some status Lines instead of Ascii characters to the rs232 transceiver.

The Machine is so highly compatible that literaly the first program i tried to run does not run unmodified. 

lol.

not lol.