iMac G3 mod - Video Connector

[ym58]

Hello everybody,

Attention : This post is a dupe post from the iMac section coz I have been advised there to repost in the "PowerPC Macs" section. Thanks for your understanding.


I have been trying for the last few weeks to include a mini ITX-PC board in an iMac G3 slot-loading while keeping its PSU and CRT (sort of nostalgic 'vintage' thing , I guess ....).

I have made a short custom cable to internally connect the VGA output of the PC board to the HE20-pin video connector but it looks like I am still missing some pins.

Even when I short the +5V-trickle (T5V) to PFW, I cannot hear the "PONK" sound of the CRT charging up ... (needless to say that the CRT works quite right with the original iMac logic board).

(...)

Then, I decided to make a test board to parse every single signal that comes out of the mother board and connects to the video card.

What I found is that, besides the "classical" red/blue/green, hsync/vsync and ground signals, most of the other signals are NOT used BUT the two SDA and SDC signals (respectively pins 1 and 2).

When connected to an oscilloscope DURING THE BOOT, these two signals show that the motherboard raises them to +5V then real quick ( after a few 100th of msec ) applies a particular pattern sequence (logical states +5V and 0V) then finally leaves them to +5V all through out.
From this point on, the "PONK" of the CRT is heard, and everything works out right !

Then, you can even drop the two signals (SDA and SDC) to 0V, no matter what : the iMac is definitely ON and so is the CRT.

My deal now is to find how to replicate those TWO short sequences (they are both different, I mean SDA seq is not the same as SDC seq !) WITHOUT the main board.
And then, I should be ready to get rid of the iMac motherboard and install a PC board in the iMac G3 !

Any hint, idea, suggestion, reflection whatsover is VERY welcome ...
ym58

 

[harrymatic]

This is a really great project you're undertaking - going much further than just putting a horrible LCD in that doesn't match up with the bezel at all.

I imagine it would be possible to capture the signals with a PC logic analyser. Hopefully they're slow enough that a basic microcontroller would be able to spit them out. You might have better luck asking an electronics forum - try the Hackaday forums at http://forums.hackaday.com/viewforum.php?f=4&sid=b6757610cdd9751ec0ad13fa15e164c1

I really hope you can get this working.

 

[ym58]

Thanks for the encouragements, Harry !
I knew it would be tough but not as much ...

Actually, I am now up to the point where I don't want to program a microcontroler or whatsoever logic gates array to perform this stupid sequence !

There are two good reasons for this :
#1. the two SDA/SDC lines are unidirectional lines only (MotherBoard ***TO*** Video PCB)
#2. even if there was a bi-directional handshake (by ways of this sequence) between the MB and the Video PCB, that would be totally irrelevant to try to reproduce it since there will no longer be any Apple MB in the final project !

So my reflection now is the following : since the iMac MB sends during the boot a certain sequence to the Video board to trigger it ON, I must find WHERE (on the video board) is this trigger (be it a logic or analogic level) and FORCE IT ON as soon as the P/S is on ... as simple as that.

What is less simple though is that I have searched FOR HOURS the internet to find a schematic or even a block diagram of the iMac Video board ... to no avail, so far

But I will find out ... with a little help from Mac friends
ym

 

[AmestrisXServe]

Before anything else, please read this article, that explains the refresh rates of the CRT.

(I have a feeling that Apple may have designed it this way to prevent viewing actions between POST and the actual login display.)

My expectation is that the signals peaking at +5VDC and dropping to +0VDC are no more than a power-on signal; you may need to create a manual trip to emulate this, with a live +5VDC feed, that switches that between n/c and the desired pin, for a similar duration to that of the in-built timer circuit. (You could do this with a physical switch, or with a PIC.)

To my knowledge however, the display will power-on whenever it receives a signal at the correct horizontal refresh rate. You may wish to try setting your video output to a supported rate, and verify that this is correct, ere worrying about the undocumented connections, and then follow up on them later.

Alas, I do not have a full system schematic for these units, nor do I have an original iMac, on which to run any tests.

 

[ym58]

@amestris,
Yes, I had read this article

Reason why I set-up an old DELL laptop @1024x768-75Hz and tried to connect its VGA output to my test board.
But this test aborted because I could not wake-up the CRT.
An appropriate horizontal refresh rate is, according to my feeling, ***NOT*** sufficient to power the video board on.

I also tried your 'manual switch' suggestion and alternate +5V and 0V states to the SDA or SDC pins, but no avail ... the CRT never woke up

My oscilloscope showed that the iMac MB sends (via SDA&SDC) very specific sequences to the Video board, and my guess is that :
#1. trying to replicate such sequences 'manually' is impossible
#2. the video board perhaps needs ***both*** signals (SDA&SDC) to be involved in the routine

I haven't try to play this game with ***two*** 'manual' switches, though ...
I wish I was Tommy, the pinball wizard !!!

----
Once again, my best deal is to do something directly on the video board to have it automatically switched ON when power is applied to the G3.
The best info that I could get on the net concerning the G3 video board is here (sorry, it's in French), but that's not enough for me to go ahead at this stage.

 

[AmestrisXServe]

Have you tried to trace, on the video board side, where those signals are routed? You may be able to isolate the circuit(s) that determines the wake/sleep mode, and bypass it, forcing the display to be in an always-on state; then add a power-off button somewhere, to be able to manually disable it.

Considering that others have managed to do this, I find it odd that you are having this degree of difficulty, but I have never attempted to use an iMac CRT on any other logic board, and as I said, I haven't any CRT-based iMac systems to inspect.

It's certainly odd that Apple used such an elaborate method of setting CRT wake status.

If you run the Apple mainboard connectors that you have isolated as controlling this, with jumpers, to those on the adapter, use the Apple mainboard to send those wake signals, and use your new logic board for the rest of the VGA signals, do you see video as appropriate from the new system?

 

[ym58]

Have you tried to trace, on the video board side, where those signals are routed? You may be able to isolate the circuit(s) that determines the wake/sleep mode, and bypass it, forcing the display to be in an always-on state; then add a power-off button somewhere, to be able to manually disable it.

I haven't disassembled yet the video board coz (I confess) I was expecting someone having a good knowledge of the PAV to hint me precisely on the PAV's on/off-circuit area ...
But if I don't eventually get any diagram, I will definitely have to tackle this

Considering that others have managed to do this, I find it odd that you are having this degree of difficulty

Actually, not so many !
I have put in my favorites numerous links pertaining to the G3 mod and very few of them did exactly what I am attempting to do, that is ***keeping the CRT*** !
The most relevant link is those of Steve Ferris but the G3 that he used was a ***drawer*** G3, not a ***slot*** G3 as mine, and the drawer G3 has an internal Mac DB-15 video port.

(quote from Steve Ferris) : Since the rev. A through D iMacs use a standard Mac db15 video connector for the internal monitor, connecting it to the VGA output of the ITX motherboard is a simple matter of using an adapter. I had three of these adapters lying around, but if I used one, I would have this big thing sticking out the side of the computer. I didn't want that, so instead I made my own adapter that would contact the pins of the VGA connector that stuck out underneath the motherboard (end of quote)​

It's certainly odd that Apple used such an elaborate method of setting CRT wake status.

Agreed.

If you run the Apple mainboard connectors that you have isolated as controlling this, with jumpers, to those on the adapter, use the Apple mainboard to send those wake signals, and use your new logic board for the rest of the VGA signals, do you see video as appropriate from the new system?

Unfortunately, so far I have only wired certain signals to my dip-switches, but not all (only those that I had wanted to test for their supposed actions to the PAV)
But you're right, I will have to rewire them ***all*** if I want to make an ACTUAL test of connecting a PC board 1024x768@75Hz to the iMac's CRT.
---
Once again, I was expecting some feedback from MacRumors friends before taking the next steps on my side.

 

[AmestrisXServe]

That, I'm afraid is the price to pay, for doing something original.

If there is no previous public, documented record of doing this, then you will need to document it on your own, and I hope, make the information available.

Few people desire to keep the original display, due to its limited modes, so it is oft the first thing to go, and thus, the necessity of documenting the pinout for that connector has not been in demand.

Apple also don;t make system schematics public, even to authorised repair centres I can give you every Apple Service Source document for the iMac series, and not a one will have that connector diagram that you need.

I tried to source one for you, but all that I have found, relates to the iMac G4 series; not the G3 series, and is thus, useless to you.

There may also be changes between the various iMac models, as you have mentioned, and the Bondi G3, the 'Fruits' G3, and the 'slot-loading' G3 units could have different video sub-boards. (Apple are notorious for incompatibility across models in a series.) I also recall the Bondi iMac having both an internal VGA header, and an ADB header; and that both were removed in following revisions.

The thins is, that display sleep is not normally so exotic. I can;t even fathom why it is set up this way, as it seems both counter-intuitive, and unnecessary, and in engineering, we try to eliminate necessary expenses, not accrue them. That's one good reason to seek out not only what is happening here, but also why it is happening; and to try to understand the logic behind it.

I wish that I could be of greater help, but I think that once you fully wire your switches, between both video controllers, and are able to analyse the signals more thoroughly, you will be the foremost expert in this particular topic, outside of Apple.

I have had to do this with various devices in my lifetime, where no documentation exists, and in the process, became the authority in working with them, as the manufacturer provided no information to the public, and no-one else wanted to spend the time to do a proper, and full analysis.

In my case, many of these things were profitable, as I became one of the pew people that could repair them; but some were purely for my own interest. I don't know how many other people will want to use the original CRT in an iMac, and to be honest, it may be easier to use an alternative CRT of the same dimensions, that will give you more display modes. The only problem I foresee with that is that the cage/enclosure will not be identical, and mounting it may be a bloody pain.

It's somewhat akin to converting arcade cabinets, save that you have far less room for variance. You may have to make a bridge between your new logic board, and the apple display board, to provide the missing signals, and the only way to do that is to analyse them, and try to replicate them.

If all that you need is a +5VDC pulse for a specific duration, on two or more connections, it should be fairly straightforward. If there is some kind of modulation involved, you will need to replicate that; but I don;t understand why anything that elaborate would exist in this case. I would certainly be interested in the outcome, to understand what reason lies behind Apple making this...

(I am curious what you see, if you connect the video signal of the normal mainboard to another video source: It may be possible that some signal sent by the mainboard turns video on/off to hide video display during POST.)

Regarding the problem itself, I suspect that the enabling conundrum it is merely a timing issue, and that you need to provide the voltage on each line, for a specific duration. That probably trips some kind of circuit, that has a logic of (disable display if on; enable display if off). That may explain the multiple lines carrying the current, of which, I recall you said there were two.

The main thing here, is to trace the routing of those signals, and to analyse what the circuit that receives that voltage is doing. That may make it easier to enable, or disable the undesired behaviour, or to replicate what causes it.

 

[ym58]

@amestris,
Thanks for all your interesting clarifications and explanations.
I now understand that, as Mr Livingston, I am just alone there out in the jungle facing a fierce predator called Apple !
I will definitely disassemble the remainder of my G3 to access the PAV and try to locate the circuits that are involved in powering the CRT on.
I will indeed let you all know the ongoing of my exploration !
In the mean time, should any of you all come across any documentation concerning the G3 video board, let me know !
Thanks.

 

[ym58]

iMac G3 (slot) mod - PAV wiring diagram

I was able to get hold of a wiring diagram of the PAV (analog video board) of the iMac G3.

I am sharing it with you :

Apparently, ***MY*** concern (see previous posts) lies around the P506 connector area ...

 

[Intell]

That diagram is for the tray loading G3 iMac. The slot loading is very different inside and not comparable, they don't even have PAV boards.

 

[AmestrisXServe]

That diagram is for the tray loading G3 iMac. The slot loading is very different inside and not comparable, they don't even have PAV boards.

I believe you have those two models transposed, if you mean that the slot-load does not have the P/A/V board..

The Summer 2000, Early 2001, and Summer 2001 iMacs, such as the 'Indigo iMac', are a slot-loading units, with a P/A/V board. Two, incompatible versions of this P/A/V component were used on this grouping, one of which has a toggle switch on it. The Summer 2000, and Early 2001 units have a VGA connector on them; while I believe that the Summer 2001 units may not.

The iMac DV series, that was released in late-2001, and is also slot-loading with a P/A/V board.

It is the tray-loading model that lacks the P/A/V, and has a separate analogue/video sub-board, on the side of the CRT.

Reference: Apple Service Source Manual, 'iMac (Summer, 2001)'; revised edition from 11th July, 2005, page 9 (VGA port), page 11 (slot-load), and pages 81-97 (P/A/V and other Video Logic).

 

[Intell]

All slot loading G3 iMacs have the same internal layout. The only difference in them is the logicboard. They have a down converter board to power the hard drive, but do not have any form of a PAV board that is comparable to a tray loading iMacs. The OP's diagram is still for the incorrect model as it is for a tray loading where he wants the slot loading.

 

[AmestrisXServe]

All slot loading G3 iMacs have the same internal layout. The only difference in them is the logicboard. They have a down converter board to power the hard drive, but do not have any form of a PAV board that is comparable to a tray loading iMacs. The OP's diagram is still for the incorrect model as it is for a tray loading where he wants the slot loading.

Sorry, I misread your response. I thought you were stating that the slot-loading systems lack the P/A/V board. Now that I have re-read it, I see that you werer discussing the diagram.

I have never seen a schematic of the 2001 iMac, available anywhere. I would have thought that after thirteen years, one would have been leaked from Apple, but if so, it has never been published anywhere.

 

[xcalibre]

I'm part-way through the same mod, I'm trying to use the DVD/HDD cage to house a Mac mini etc, but getting stuck at the video led me to this thread. Any luck so far?

 

[ym58]

I'm part-way through the same mod, I'm trying to use the DVD/HDD cage to house a Mac mini etc, but getting stuck at the video led me to this thread. Any luck so far?

Well, I'm close to drop the whole project.
It really pisses me off too that no schematics ever leaked from Apple or else for all this time and, to be honest, I feel very LAZY now to disassemble and study the PAV layout so to find and mod the hardware section that drives the CRT on.

I will let a little while for other Mac addicts to maybe hint me or indicate me where they think I could find the proper information from, but in case I don't get any positive encouragement on this thread I will simply give up.

So bad indeed coz I think that a PC-based board in an all vintage iMac (with its CRT !) would have been quite a funny thing to build and make work !

 

[xcalibre]

Sorry to hear it, if I find anything of use I'll be sure to post it here.

 

[turnkit]

Hope you don't quit. I have some of these systems -- more of the tray loader though actually -- and would enjoy reading up on how to make the original CRT work properly, preferably even with sleep modes.

Do you have access to the Apple Service discs? Not sure what's in that collection but I have seen it online. Perhaps some info in there?

Well, I'm close to drop the whole project.
[...]
So bad indeed coz I think that a PC-based board in an all vintage iMac (with its CRT !) would have been quite a funny thing to build and make work !

 

[ym58]

Do you have access to the Apple Service discs? Not sure what's in that collection but I have seen it online. Perhaps some info in there?

No, I don't !
Should you have any connections to get some, please don't hesitate to PM me !
Thanks for your support.

 

[blackburn]

The SDA (Data) and SDC (Clock) pins are most likely the I2C / TWI interface that somethings use to communicate (this case the motherboard and the crt controller board).

You should get an micro controller and try to read out what the motherboard sends to the crt, them make the micro controller send that to the CRT for it to come on.

Edit: Indeed they are check this https://forums.macrumors.com/archive/index.php/t-53733-p-2.html

And btw the vga also has those lines for monitor identification.

 

[oxydenz]

Hi everybody...
I'm in the same adventure, trying to replace the apple MB by a PC one, and keeping that old CRT as the main screen...
I'd like to say to ym58 "don't give up !" (lâche pas l'affaire !)
the fact is that i cannot be of any help i guess... cause i lack of electronical knowledge...
well, if someone think I or my iMac DV can be of any help, don't hesitate...
Good luck !

 

[Nick Gillard]

It might initially sound odd but had you considered leaving the original logic board in just so it can activate the CRT? If the original logic had no speakers or hd connected, it should activate the video and sit it the boot selection screen. If you left the 'control' lines connected to the original logic but drove the VGA lines from the new logic, this should at least confirm whether the video signal has anything to do with activating the crt. If it doesn't, but you still can't replicate the control sequence to turn on the CRT, you could actually leave the original logic in the final design. The remaining problem is then much simpler, i.e. how to pick up a power line from the original logic and get that to trigger the power on of the new logic (or viva-versa).
Or a less elegant solution would be to add a small push button switch to the side panel so you have one switch to power on the new Mac logic and one to power on the original logic and hence the crt.

It would not be energy efficient as the original logic would have to always be on but a crt based computer is hardly Eco friendly anyway.

 

[moicez]

Is anyone still out there? So I want to give this a try, I have a G3 that's been sitting around and I would like to put in a Raspberry Pi in there.

So my idea is to try and create the SDA and SDC signals with an FPGA. So the FPGA would be a video driver for the Pi.

Does anyone happen to have the sequence?

Hope someone can help, otherwise I would have take out my flimsy oscilloscope.

 

[cablemunkey]

If anyone is still interested, I currently have an iMac G3 working as a monitor for a separate PC, with the original iMac motherboard removed. Apart from the motherboard being removed, the rest of the original iMac hardware is intact. It was a bit complicated (there's an Arduino sitting in the iMac making everything talk nicely), but works well. If there's any interest in it I can pop up some details.

 

[moicez]

If anyone is still interested, I currently have an iMac G3 working as a monitor for a separate PC, with the original iMac motherboard removed. Apart from the motherboard being removed, the rest of the original iMac hardware is intact. It was a bit complicated (there's an Arduino sitting in the iMac making everything talk nicely), but works well. If there's any interest in it I can pop up some details.

Yes yes! I am still very interested
Thank you!