VOGONS


First post, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

I was not able to find online any information for the aforementioned in the subject IBM PS/2 model floppy drive interface pinout. It's 40pin connector, but here:

https://ardent-tool.com/floppy/Pinouts.html

model 8556 is not even mentioned. So, I can rule out this one:

https://ardent-tool.com/floppy/Pinouts.html#40p_Planar

because I measured and there are no +5V on pin38 and +12V on pin40, that's why I presume it's more similar (or even exactly the same) to this one:

https://ardent-tool.com/floppy/Pinouts.html#40p_Planar_MS

So, here is what I did - attaching pictures on which I also added some comments:

A. confirmed there is +12V on pin6 and cut the corresponding wire on my regular 34pin floppy cable (see "picture 1")
B. confirmed there is +5V on pin3 and cut the corresponding wire on my regular 34pin floppy cable after the middle connector(see "picture 2") . this way +5V doesn't go to the connector after the "cable twist" at the end of the cable where my floppy drive will be connected, but I still have +5V on pin3 of the middle connector, which I will use to pull-up all necessary signals (that are all open-drain and need to be pulled-up on PS/2).

after A. and B. I tested the floppy cable on PC and it is still working, because those wires are not used on PC with 1.44MB floppy drive.

C. further "picture 2" shows exactly how I pulled-up the necessary signals - essentially using the middle connector as a breadboard. plus it shows that I cut pin4 wire, which is "Drive Select" in order to be able to pull-up/pull-down that signal, which is supposed to select that way between 720KB and 1.44MB floppy drive type. IBM document "ps2_fdd_trm_s42g2194_00.pdf", which I am attaching to this post, says "Drive type signal, when Active 1.44MB drive has been selected" - I guess that means pull-up.

D. "picture 3" additionally shows just how I pull-down the pin4. I can in similar way pull it up, but in both cases I cannot get proper work.

Finally, the problem and my question:

on startup the drive sounds very healthy - in fact watching videos it sounds exactly like original working IBM drive sounds when PS/2 is powered on, making the exact same series of sounds. However, when it's time to boot from the drive, it makes few sounds (I presume reads) and it gives up. I was thinking that is indication the "Drive Type" is not properly recognized, because on PC I can kind of simulate similar behavior when select in the BIOS 1.2MB drive and try to boot from that same 1.44MB drive. However, as I mentioned in point D. - no matter if I pull-up or pull-down pin4 "Drive Select" signal - nothing changes.

So, any ideas what I am missing and/or doing wrong? Any information how to make regular PC 1.44MB floppy drive work in IBM PS/2 model 8556? thanks!

Attachments

  • Filename
    ps2_fdd_trm_s42g2194_00.pdf
    File size
    430.96 KiB
    Downloads
    38 downloads
    File comment
    IBM PS/2 3.5-inch Diskette Drives Technical Reference
    File license
    Public domain
  • 001.jpg
    Filename
    001.jpg
    File size
    145.45 KiB
    Views
    1366 views
    File comment
    Picture 1
    File license
    Public domain
  • 002.jpg
    Filename
    002.jpg
    File size
    172.46 KiB
    Views
    1366 views
    File comment
    Picture 2
    File license
    Public domain
  • 003.jpg
    Filename
    003.jpg
    File size
    129.61 KiB
    Views
    1366 views
    File comment
    Picture 3
    File license
    Public domain

Reply 1 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

wow, did so many experiments, but nothing helps. so, it ain't pretty anymore and by that i mean i cut so many wires on the cable and splice them back together without any real results. during those experiments only 2 things cause difference:

1. cutting pin9 wire (by default it's shorting to GND via the floppy drive connector). that pin9 at least according to:

https://ardent-tool.com/floppy/Pinouts.html#40p_Planar_MS

is "Drive Type ID0" - that makes PS/2 behaves like no floppy drive is attached to the system, i.e. no attempt to boot from the floppy is made. Interestingly, pulling it down or up via 2.2kOhm resistor makes no difference, i.e. in both cases PS/2 behaves like there is no floppy drive attached to the system and no attempt to boot from the floppy is made. So, it seems the only "good" connection for that pin9 is direct short to GND (without pull-down resistor in between)

2. disconnecting the +5V power supply to the 5 pull-up resistors from picture2 in my previous post - that causes:

Non-System disk or disk error

which I guess is correct and actually is confirmation for signs of life. That also brings me back to my initial hypothesis - that it cannot determine the type of the floppy drive and that's why it hangs when trying to read, i.e. trying to read it as 1.2MB instead 1.44MB drive. Unfortunately, pulling up/down pin33, pin17, pin4 - nothing made it work. I don't know and I am totally lost...

Reply 2 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

Oh no, 4 years ago - exactly the same problem:

Trying to adapt a standard 34-pin floppy drive to 40-pin IBM PS/2 edge connector, won't read 1.44MB disks!

and no solution seems to be found - those adapters suggested in the end of that old thread are useless in the case, because they are for IBM PS/2 Model 50, 50Z and Model 70, which is different connector pinout, i.e. +5V on pin38 and +12V on pin40 and so on. Now, I am really discouraged - it seems no one has succeeded 🙁

Reply 3 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

ok, I think I failed and other obviously failed in the past as well, because contrary to other PS/2 models and to PS/1, in this case the drive type is no longer determined just by 1 bit status, i.e. pin4 be pulled down or up, but by 3 pairs of bits:

Drive Type ID1 (pin 4) / Drive Type ID0 (pin 9): (0,0), (0,1), (1, 0), (1,1)

Media Type ID1 (pin 17) / Media Type ID0 (pin 27): (0,0), (0,1), (1, 0), (1,1)

Data rate Select 1 (pin 2) / Data rate Select 0 (pin 33): (0,0), (0,1), (1, 0), (1,1)

that means each of those wires needs to be cut and manually set to proper value. Also, I tried already many of those, but no wonder I had no success, because brute-forcing it there are 4*4*4 = 64 possibilities. That's why it was bad strategy, but now with that new hypothesis, I can research further and narrow down for 1.44MB drive what each of those bits value should be set.

Reply 4 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

according to "IBM PS/2 3.5-inch Diskette Drives Technical Reference" document that I attached to my first post:

Drive Type ID1 (pin 4) / Drive Type ID0 (pin 9): (0, 0)

Media Type ID1 (pin 17) / Media Type ID0 (pin 27) : (1, 0)

are for sure the correct settings of those bits for 1.44MB floppy drive. Also, few floppy disk controllers chip datasheets gave that:

Data rate Select 1 (pin 2) / Data rate Select 0 (pin 33): (0, 0) or 500kbps

is correct setting for 1.44MB floppy of those 2 bits. Unfortunately setting all those as the above using pull-up/pull-down resistors- give no any change and it doesn't do anything differently. So, it's not working and I am now really out of any ideas - I tried several more settings than the above, but nothing changed as well.

Reply 5 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

OMG, it turned out 5 months ago adapter PCB was designed by June Tate-Gans from Texas, the design was published on Github:

https://github.com/jtgans/PS2FF

Based on the attached pictures of the PCB, it seems I was on the right track with "Media Type" and "Drive Type" bits. Unfortunately, I still cannot compare the schematic of that PCB to what I did, because it's made in KiCad and they don't support Win7, which is the newest Windows I have. Anyway, now it's obvious all the usual information found online about using regular PC 1.44MB floppy drives with PS/2 doesn't apply for PS/2 models newer than model 55, because those models use that "media sense" feature.

Attachments

  • pcb_top.png
    Filename
    pcb_top.png
    File size
    85.3 KiB
    Views
    1287 views
    File license
    Public domain
  • pcb_bottom.png
    Filename
    pcb_bottom.png
    File size
    51.37 KiB
    Views
    1287 views
    File license
    Public domain

Reply 6 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

hmm, I am not sure if the above board actually works. in the last 5 months since its release maybe it's still untested, because the README file says:

This is the first prototype release of the PS2FF board. This release is totally untested and unassembled, so don't get your hopes up much.

Especially, considering that today I made test with all "Media Type", "Drive Type", etc. signals pulled down/up to correct values without any success. Now, it's possible with all experiments I did, I fried the FDD controller in the PS/2, the README file of PS2FF board warns about the same:

In theory if you set all four DIPs on, we'll dump around 20mA into the /DRIVE_SELECT line, which might cause magic smoke to be released from your host controller chip. You have been warned!

However, I think this topic is kind of hugely uncharted territory - at the moment I don't believe I fried anything and at the same time I am sure I made test with all "Media Type", "Drive Type" signals pulled down/up to correct values without any success and considering PS2FF board is deemed as untested I am not sure what to think - total confusion mode?! After all, I was also sure after realize about the "Media Type", "Drive Type" signals all will work in practice when I set them, but - it didn't. So, if PS2FF is based on untested in practice theory, then it will not work as well. Again, I really don't know what to think...

Reply 7 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

Finally - Success! I really cannot believe it!! the explanation is below - I am not posting pictures, because after so many experiments my 34-pin regular PC floppy drive cable is an ugly mess that will confuse you and that's why I will explain in details with words only.

It was entirely brute-force approach, i.e. doing countless experiments and observe the results from them, because as you will see, when you read below what solved it - that all information online including PS2FF adapted made by June Tate-Gans that I mentioned in the few previous posts are wrong! (at least wrong for my PS/2 model 8556, which I believe is also called Model 56SX).

So, it's amazing how simple it is when you figure it out - no any resistors or circuitry is involved - hence making adapter PCB makes even less sense. What is most confusing of all is that IBM own "PS/2 3.5-inch Diskette Drives Technical Reference" document (the "ps2_fdd_trm_s42g2194_00.pdf" attached to my 1st post) is wrong - my best guess is that document is made before those models like 56, 76, etc were released and hence the information there doesn't apply for them. (I say IBM docu is wrong, because if we use the information in that docu then setting the motherboard to 1.2MB 5.25" drive made my 1.44MB 3.5" drive work - clearly that documentation cannot be right for those newer PS/2 models like 56).

Initially, when I made it work I was thinking it's a fluke and maybe it will work just with that particular regular PC 1.44MB drive I was using at the time, but luckily - it wasn't - I've already tested with 3 different regular PC 1.44MB drives - Sony, Teac and Chinon - they are all working!

Here is the minimum working configuration - one more time no resistors, no circuitry, all modifications are done on a regular PC 34-pin floppy drive cable:

1) cut pin3 and pin6 wires (see my first post pictures of those cuts and why those pins corresponding wires need to be cut). So, I cut pin3 wire after the middle connector on the cable (for considerations I explained in my first post, which are no longer relevant, but anyway - it needs to be cut - where is the best place on the cable to cut it - you decide). pin6 wire I cut close to the end of the cable that goes to the motherboard and that is the best place as it's +12V that will most likely fry your PC 1.44MB drive if it's not cut there. Nothing new here - those cuts are done for all PS/2 models when you want to mod them with regular PC 1.44MB floppy drive.

2) cut pin4 wire - again see my first post about that

3) the New finding and Critical for the Solution: cut pin9 corresponding wire - it's the wire just before the "cable twist" part of the cable starts - that's why I cut it there, i.e. after the middle connector on the cable. That pin9 is shorted to GND via the floppy drive connector - that's why it needs to be cut, because when it's shorted to GND makes 1.44MB drive not possible to work.

4) on the middle connector of you regular 34-pin floppy cable for PC, use jumper wire and short pin4 to pin5, i.e. short pin4 directly to GND (pulling it down via resistor - I used 2.2kOhms one - is not working, needs to be directly shorted to GND).

That's it - now my recently acquired IBM PS/2 is free of IBM custom parts - I removed their custom (and now dead) power supply and modded it to work with modern regular PC ATX power supply (if anyone needs guide/howto for that - let me know, it's super easy when you acquire proper parts and materials for doing the ATX mod - I did it with all off-the-shelf parts). now it's free from IBM very special floppy drive as well (dead in my case as well). In other words now it uses all standard parts - my RAM is also regular PC one that I modded to be compatible with PS/2, using this:

https://ardent-tool.com/memory/Mods_02.html

the HDD is SCSI, but standard one without any IBM "customization" and thus no changes are needed there in order to acquire cheap available replacements (if needed).

I hope the above will help other PS/2 users - as far as I can tell based on PS/2 Floppy Drive Interface Pinouts here:

https://ardent-tool.com/floppy/Pinouts.html

the solution should work with PS/2 models 56, 57, 76, 77, 90 and 95, etc, i.e. all models that have 40 or 44 pin floppy drive connector and are "media-sense" type.

Reply 8 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

Warning: Model 90 and 95 users - I read here:

https://ardent-tool.com/floppy/floppy.html

that it's very dangerous to short the "security features" pin (I believe that is pin14) to GND - it burns the FDD controller, here is the exact words from the above link:

If you really like your 95 you *do not* try it in there. A team mate once did it ... and it took us some days to solder in a new FDD-controller ... (main problem was to find one at first)
I cannot figure out *exactly* what caused the mess, but it has to do with the "security features" available on the 95 - and the corresponding pins on the 56 - 77 being not present and set to GND.

So, Models 90/95 will need some additional modifications on the cable, i.e. make sure those critical pins are not shorted to GND via the floppy drive. I don't have such model and thus what I described in my previous post is most likely valid only for models 56/57, 76/77 and it was tested by me only on model 56.

Reply 9 of 34, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member

I just now saw your thread and it was very interesting read, like pretty much everything with these IBM machines. IBMs design shenanigans can make these challenging to work with, but otherwise they are beautiful machines. Floppy issues are a major problem getting these machines to work, because of the requirement for setup floppy. There is three threads about PS/2 floppy issues hovering high on the Vogons general hardware currently! ESDI and MCA are just an icing on the cake.

I recently acquired model 35SX. It is an easy one (and the reason I got it): it has ISA, standard IDE and PSU with molex connectors. Pretty much only strictly IBM stuff is the floppy, which was not working and parity RAM. I got the floppy working after recapping it, but I was thinking about replacing the drive with standard one, but decided to go with recapping as the number one option. And here’s the reason:

I looked at the ardent tool pinouts for 34 pin media sense drives and compared them to standard floppy, and I couldn’t figure out how to approach making the adapter. Non media sense is straight forward and there are several instructions for that, but for media sense the signal types of PS/2 controller seem very different compared to standard floppy. With the help of your work and documentation here, I might now figure it out as there is very little information about adapting this sort of controller to standard floppy. As I said, most instructions deal with the early non media sensing type and afaik newer PS/2 floppy controller, at least in many cases, require media sensing drives.

My computer came with 4MB RAM installed and I should have already few compatible sticks on the way so I should be able to max my machine out (16MB). I got genuine FRU sticks for a few euros a piece, so no harm done, but did I understood correctly that using those instructions you can take practically any industry standard FPM sticks with parity and modify them like shown and they work just fine?

I’d really like to hear more about your ATX mod. I have no immediate need for it, but it could be an interesting project at some point.

Reply 10 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie
CharlieFoxtrot wrote on 2023-03-11, 09:13:

I’d really like to hear more about your ATX mod. I have no immediate need for it, but it could be an interesting project at some point.

no problem, I am going to write it here (depending on my free time in the next week or two), I won't open new forum thread, even it's off-topic, I think it's better to keep all the information in one place, because it's very likely someone that needs floppy replacement, needs power supply replacement (and/or more ram) as well.

In fact, regarding my model 56SX, member here @SSTV2 (who helped me a lot on IBM PS/2-related questions - thank you!), pointed me out that the power supply in 56SX is known to be kind of impossible to fix after it fails, the info is here:

https://www.ardent-tool.com/35-77/x6_Power.html

based on that the same applies not just for 56SX power unit, but for your model 35 as well, because the entry in the aforementioned link is called "Repairing a 35 / 56 /76 PSU". So, I guess it's the same issue for all 3 models: 35, 56 and 76.

Regarding the RAM mod - it's ultra easy, I find the instructions here:

https://ardent-tool.com/memory/Mods_02.html

kind of hard to understand and maybe even confusing, but essentially it's as easy as buy any FPM stick that has on its PCB spare space for 4 x 0 ohms resistors (some could be populated, other not). Basically those 0 ohms resistors are jumpers based on which the system IDs the RAM type - put them in position that is compatible with PS/2, i.e. set to 70ns (even if the RAM in reality is 60ns) and set to correct size. Of course, it has to be FPM stick with 36bit and not 40bit parity (those are rare, but I saw some).

Reply 11 of 34, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member

Yeah, based on the images the PSU on my system is exactly the same unit, so PSU mod may be something I need to do at some point. I did open the PSU when I cleaned the system and checked it visually and it looked brand spanking new (even very little dust, machine was overall very clean). I wonder if some preventive maintance would make any sense here, since mine is still functioning just fine…

But thanks already, I’ll be anxiously waiting your PSU mod posts then! You might consider sending that info to Ardent Tool too, so they can update it on the proper system page.

Reply 12 of 34, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member

I checked the PSU label from my 35, and interestingly date code was 9726, so that is definitely not the original PSU and has been replaced by sometime after that. This also explains, why there was so little dust inside the unit when I opened it, it probably has seen very little hours after replacing the unit. It is also made in Taiwan, so it isn’t at least those most problematic Magnetek units mentioned in the Ardent Tool.

Reply 13 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie
CharlieFoxtrot wrote on 2023-03-11, 15:10:

it probably has seen very little hours after replacing the unit. It is also made in Taiwan, so it isn’t at least those most problematic Magnetek units mentioned in the Ardent Tool.

you're lucky then, my model 56SX was like some Frankenstein-build of dead parts. it's also missing the speaker, the plastic that hold it, the hdd caddy/slide, etc. but at least the planar/motherboard are in excellent and 100% working shape (also the case outside look is in good shape). even my IBM floppy cable isolation is brittle and falling off exposing the wires - another reason why I needed to go for PC floppy and PC floppy cable.

Reply 14 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie
mattw wrote on 2023-03-11, 11:48:

no problem, I am going to write it here (depending on my free time in the next week or two), I won't open new forum thread, even it's off-topic, I think it's better to keep all the information in one place, because it's very likely someone that needs floppy replacement, needs power supply replacement (and/or more ram) as well.

OK, I had some unexpected free time today and thus here is the ATX guide:

0. I kept the original IBM power-supply intact - some people might prefer to cut the original IBM connectors from it and use them, but my guide uses replacements for those connectors - in any way, essentially nothing much changes if you decide to reuse the original IBM connectors and cut them from the original power-supply, you just will need to buy less materials in such case, the connections - what goes where stay the same.

1. Necessary materials:

* A) 1x "24/20Pin ATX Adapter PCB Kit" - there are many such kits available from China, attached Picture1 is of the kit I used.

* B) 1x "7pin 2.54mm XH cable" - see attached Picture2 of 10x such cables. of course, you need just one such cable. make sure the wire is at least 26AWG (0.129 mm²), because original power supply is 3.6Amps@12V carried by 2 wires, i.e. calculators tell me the minimum for 2Amps@12V is 26AWG

* C) 1x "pieces of AC power cord", I used one with 3x 0.75mm² wires, because that is the maximum the kit from A) PCB holes can accommodate, I used that for the +5V power rail - I got really paranoid, because IBM original wires are insanely very thick - almost 4mm, which makes no sense to me, based on the current they suppose to carry, the original power supply is marked as "16Amps@+5V"

* D) 2 x Wago 221-412, 1 x Wago 221-413 (or just use 3x Wago 221-413), I used those to attach C) to the motherboard - when you see the pictures of the finished ATX adapter, it will become clear

I am not mentioning obvious things, like soldering iron, solder, flux, etc, but costs for the above are: A) is about 5 bucks, B) depends on how many you're forced to buy, but just one is like 1-3 bucks, C) is free (part of the cord from an old AC power outlet), D) you can get 10x Wago 221-413 for less than 5 bucks. So, all and all, the total is less than 15 bucks.

2. the Guide itself:
since a picture is a thousand words - I will try to do it with adding comments to the attached pictures itself, but just want to mention that the wire colors doesn't matter and could be different in you case depending on what you buy for cable B). So, I will use both wire color in my case and what is the voltage it carries. Also, the main idea behind the adapter is 3.3V will be "removed" and made to be "-5V" and 5VSB will be made "PWR_GOOD" signal, all other voltages are basically directly send from the ATX power supply to PS/2 motherboard. So, you need to use ATX power supply that has "-5V" - if you want to use such without -5V then you need to add to the adapter "LM7905 Voltage Regulator" to provide -5V from the +12V rail (I will not cover that, but you can easily find schematic how to wire LM7905 to +12V in order to get -5V).

* Picture3: General overview of the connections, all connections are to "P2" connector on PS/2 Planar (motherboard), except the 3 explicitly stated that are to "P1" connector

* Picture4: how to prepare the ATX connector before solder it to the A), i.e. what pins to cut and extract from it, i.e. to remove

* Picture5: the 2 only modifications made to the PCB of A)

* Picture6: Top view of the ready solution

* Picture7: Bottom view of the ready solution

I guess I don't even need to mention that - but be very careful and of course before attach anything to PS/2 measure all voltages are correct!

Attachments

  • Picture3.jpg
    Filename
    Picture3.jpg
    File size
    176.26 KiB
    Views
    1101 views
    File comment
    P3: ATX Adapter connections to PS/2
    File license
    Public domain
  • Picture4.jpg
    Filename
    Picture4.jpg
    File size
    195.06 KiB
    Views
    1101 views
    File comment
    P4: ATX Adapter Connector Mods
    File license
    Public domain
  • Picture5.jpg
    Filename
    Picture5.jpg
    File size
    258.25 KiB
    Views
    1101 views
    File comment
    P5: ATX Adapter PCB Mods
    File license
    Public domain
  • Picture6.jpg
    Filename
    Picture6.jpg
    File size
    634.28 KiB
    Views
    1101 views
    File comment
    P6: Top view of the ready solution
    File license
    Public domain
  • Picture7.jpg
    Filename
    Picture7.jpg
    File size
    580.43 KiB
    Views
    1101 views
    File comment
    P7: Bottom view of the ready solution
    File license
    Public domain
Last edited by mattw on 2023-03-12, 16:44. Edited 15 times in total.

Reply 15 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

Ops, it's not allowed more than 5 pictures per post, that's why Picture1 and Picture2 are to this post. See the previous post for Picture3 to Picture7. one small, but very important correction - on Picture7 I made an important typo - it's not "P2", but of course it's "P1" connector bottom side.

Attachments

Reply 16 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie

ok, because it's kind of critical, I made new Picture7 to address the typo I made in the original one - also added some additional information.

Attachments

  • Picture7_correction.jpg
    Filename
    Picture7_correction.jpg
    File size
    444.87 KiB
    Views
    1071 views
    File comment
    P7: corrected and more info added
    File license
    Public domain

Reply 17 of 34, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member

Great stuff, thanks! One thing though, about the final modification. I didn’t take any measurements about the original PSU vs ATX, but apparently normal sized ATX PSU fits nicely in place of the original PSU?

Reply 18 of 34, by mattw

User metadata
Rank Oldbie
Rank
Oldbie
CharlieFoxtrot wrote on 2023-03-13, 05:11:

normal sized ATX PSU fits nicely in place of the original PSU?

normal size (by that I assume from desktop computer ), I doubt, but ATX power-supply from 1U rack-mountable old server machines - fits and there is even plenty of space left.

there are many options when it comes to ATX nowadays - there are even some very-small-sized ATX power supply that works with 12V DC input (not 100V-240V AC input). for that last option you will most definitely need to add LM7905 to make -5V from 12V, but as you can see on the ATX Adapter board there are plenty of pads available where to solder very easily the legs of LM7905 between +12V and +3.3V rail (that with the jumper wire added on Picture5 is reworked in our case to be -5V). In fact if you sacrifice the original IBM power supply (especially if it's dead), you can use not only the original wires/connectors, but fit everything inside its metal housing. you can see such approach here:

https://youtu.be/PRNL7kthiO0?t=1870

for IBM PS/2 Model 70 PSU - the guy there used 12V DC input ultra-small-size ATX Power supply and put everything inside the housing of the original dead IBM PSU. you can see the insane amount of available space left. So, really plenty of options and fitting it is the easiest part of the job when you made yourself working adapter from ATX to IBM PS/2 PSU connectors for your model.

Reply 19 of 34, by CharlieFoxtrot

User metadata
Rank Member
Rank
Member

Yup, I meant desktop size. And yeah, I thought how the heck you have fitted a regular ATX PSU there 🤣

Those small pico PSUs are indeed neat, I installed one similar small PSU in my Dreamcast as a replacement for the original internal PSU couple of years ago. Of course the external AC/DC PSU does big part of the job while converting from mains voltage to 12V DC, but it is still quite amazing something that small can output like 160W worth of power. Really shows how much PSU efficiency has progressed.

Here is a near example:
https://www.mini-box.com/picoPSU-160-XT