Need solution for HDI-45 to DB-15 connectors for video out

[Trash80toG4]

Ungupatchka! Just gotta love that beautiful mess. Haven't done the pin count, are those PCI slots?

Back on topic, has anyone worked out the footprint for the required pins for this conversion? If they're not splayed out across two adjacent rows, we might interleave the HD-15 connector pins.

 

[Nycturne]

No problem, I wasn't attacking I'm just... I tend to clarify rather than leave stuff, a bad habit from a character flaw and critically precise working environment where misunderstandings cause major problems. No offense meant.

Thanks for this, seriously. I’m still working on improving this particular habit myself, so I get it. My EE is quite rusty, and one reason I’m messing with older hardware is that I put that part of my degree aside as my career moved further and further from the low-level stuff. I want to shake off that rust.

I was sharing thoughts because it is something I've considered doing to my 6100 and / or 8100. I really hate the HDI45 adapters.

My current bane is the ADC connector, but these sort of projects that help address the expense/difficulty of adapting Apple’s proprietary connectors is great. I was mostly trying to lurk to see where it went to be honest.

But back on topic, I guess the bit I’m having trouble visualizing in my head, is that with the desire to mount this PCB to the through holes for the HDI-45 socket on the logic board, how do we avoid physical interference with the VGA receptacle’s pins? At first glance it looks like there’s likely going to be some overlap, and I wasn’t able to find a version of the VGA receptacle that used surface mount connections for the signal pins in a cursory search.

 

[phipli]

Ungupatchka! Just gotta love that beautiful mess. Haven't done the pin count, are those PCI slots?

They're not... I potentially worked out what they possibly might be from an old apple engineering doc that listed interfaces including one in that era that was never used... I can go and dig again. I think I was talking about it with someone on discord so might have the chat history to save me working out which engineering doc I was in

 

[phipli]

My current bane is the ADC connector, but these sort of projects that help address the expense/difficulty of adapting Apple’s proprietary connectors is great. I was mostly trying to lurk to see where it went to be honest.

I'm not great at the G4 era, I sort of jumped over it mostly other than a Mac Mini at the time - it aligns with when I left home and stopped having as much exposure to macs.

I have a couple of G4s now, but have fitted DVI capable cards to them, I don't have any cinema displays anyway. Not sure what the solution would be there again... potentially hack an ADC cable in half and splice into a DVI cable... but that is nasty. Truthfully I think finding another video card is probably the cleanest solution

Thanks for this, seriously. I’m still working on improving this particular habit myself, so I get it. My EE is quite rusty, and one reason I’m messing with older hardware is that I put that part of my degree aside as my career moved further and further from the low-level stuff. I want to shake off that rust.

Its nice to exercise that part of the brain, especially with how easy getting PCBs made has got and how cheap MCUs are. You can do some pretty impressive stuff from home on a budget these days. If you're not already, buy a couple of ESP8266 based development boards to play with - the little D1 Minis only cost a small number of dollars and give you an up to 160MHz 32bit MCU with integrated wifi. I'd suggest the even more powerful ESP32, but truth be told, I keep having issues with bugs in the supporting software and so haven't been as impressed. I suspect it is just that the software isn't as mature and is still seeing a lot of change.

I'm in the process of a bit of a messy career reset. Couple too many corporate buyouts changed my surroundings beyond recognition.

 

[Trash80toG4]

I guess the bit I’m having trouble visualizing in my head, is that with the desire to mount this PCB to the through holes for the HDI-45 socket on the logic board, how do we avoid physical interference with the VGA receptacle’s pins? At first glance it looks like there’s likely going to be some overlap, and I wasn’t able to find a version of the VGA receptacle that used surface mount connections for the signal pins in a cursory search.

If nobody has done it yet, I'll pull out and adapter cable from the morass and buzz the connections.

If there's physical interference, SMT HD-15 won't work as a substitute, interference would still be there.

Worst case scenario would be a low level interface card with an inverted HD-15 on the bottom side of a standoff card hybrid? Judging from the DB-25 SCSI connector, looks like the HD-15 shroud should clear the top of the backplane opening.

AI8 playtime! I'm pretty sure I've got a spare 7100 or 8100 board to molest.


edit: looks like one of my pair of 6100 boards might be heading under a grinder setup. I can wrap the board up in plastic and do a few passes to remove all but the bottom quarter inch or so of the dratted shield and top two rows of pins. That'll make it very easy to desolder the rest of the abomination.

 

[phipli]

But back on topic, I guess the bit I’m having trouble visualizing in my head, is that with the desire to mount this PCB to the through holes for the HDI-45 socket on the logic board, how do we avoid physical interference with the VGA receptacle’s pins? At first glance it looks like there’s likely going to be some overlap, and I wasn’t able to find a version of the VGA receptacle that used surface mount connections for the signal pins in a cursory search.

f nobody has done it yet, I'll pull out and adapter cable from the morass and buzz the connections.

If there's physical interference, SMT HD-15 won't work as a substitute? Worst case scenario would be a low level interface card with an inverted HD-15 on the bottom side of a standoff card hybrid? Judging from the DB-25 SCSI connector, looks like the shroud should clear the top of the backplane opening.

Yes, it is a mess there isn't it (eBay photo for reference)

Options I'd consider...

1. Unlikely to be enough room, but you have a little flexibility to move the VGA connector forwards and backwards, even left to right, to get small pads between the rows and columns on the HDI45 footprint. Not a fan of this one.
2. Double stack PCBs. Lower PCB for connecting to the footprint, upper for the VGA port. There is enough height in the rear opening on the case.
3. Go back to making an L shape from PCBs, then you don't have the issue.

For low volume stuff, I'd actually consider cutting out a bracket in (1mm? 0.8mm?) steel that went into the four ground / mechanical soldered connections on the PCB and provided a panel mount for a regular VGA port, and then just hand wiring to the small number of pads you need. Makes the whole thing a couple of hours with a grinder, a printed template and a standard solder bucket DSub port.

 

[Trash80toG4]

It'll be very easy to buzz the connections. Dedicated R/G/B ground lines and any others can be ignored along with the other ground lines can be picked up elsewhere. Gives a decent shot at getting something cobbled together I think?

 

[phipli]

Tum te tum...

Edit : an extra fold tab at the top is likely needed for rigidity - it will be harder to include if you don't have the right tools, but worth it. Sorry the folds don't look like proper sheet metal folds, I haven't used this cad software in 10 years and never have much.

 

[Nycturne]

For low volume stuff, I'd actually consider cutting out a bracket in (1mm? 0.8mm?) steel that went into the four ground / mechanical soldered connections on the PCB and provided a panel mount for a regular VGA port, and then just hand wiring to the small number of pads you need. Makes the whole thing a couple of hours with a grinder, a printed template and a standard solder bucket DSub port.

On this vein of thought, I’d be tempted to look at a panel mounted connector with a ribbon cable, and a simple PCB that presents a header for the ribbon cable. Too bad there isn’t a CAD model of the logic board to play with.

 

[phipli]

On this vein of thought, I’d be tempted to look at a panel mounted connector with a ribbon cable, and a simple PCB that presents a header for the ribbon cable. Too bad there isn’t a CAD model of the logic board to play with.

We could potentially use photogrammetry - there is free software for it - it might not be perfect dimensionally, but it would be close for mockups.

I keep meaning to build a stepper motor driven turntable for photogrammetry - possibly with an arc rail for the camera too. Taking the photo takes ages One day I will.

 

[wottle]

Yes, it is a mess there isn't it (eBay photo for reference)

View attachment 19333

Options I'd consider...

1. Unlikely to be enough room, but you have a little flexibility to move the VGA connector forwards and backwards, even left to right, to get small pads between the rows and columns on the HDI45 footprint. Not a fan of this one.
2. Double stack PCBs. Lower PCB for connecting to the footprint, upper for the VGA port. There is enough height in the rear opening on the case.
3. Go back to making an L shape from PCBs, then you don't have the issue.

For low volume stuff, I'd actually consider cutting out a bracket in (1mm? 0.8mm?) steel that went into the four ground / mechanical soldered connections on the PCB and provided a panel mount for a regular VGA port, and then just hand wiring to the small number of pads you need. Makes the whole thing a couple of hours with a grinder, a printed template and a standard solder bucket DSub port.

So with the connector off, and using the HDI-45 to DB15 adapter, I got the spacing, and PIN numbers. This diagram is looking down on the board from the top. It looks like the horizontal spacing matches the actual connector (mostly 2.0mm with some 2.54mm on the right side), and the vertical spacing is a bit higher (3mm vs ~1.6mm). The ground holes are about 4mm out to either side of the first and fifth rows.

As for the pin mapping to the DB15 connector, I believe this is correct:

HDI-45 pin	DB-15 pin	Notes
8	4	Monitor ID sense line 1
9	7	Monitor ID sense line2
10	6	Green ground (shield)
11	5	Green video output (75 )
18	10	Monitor ID sense line 3
26	1	Red ground (shield)
27	2	Red video output (75 )
33	12	Vertical sync signal
34	3	Composite sync signal
42	15	Horizontal sync signal
43	11 and 14	Video sync ground
44	13	Blue ground (shield)
45	9	Blue video output (75 )

I haven't gotten around to trying to make the PCB, but do you think the with a two layer board, and the 3mm spacing I'll be able to route the traces? The problem is for two of the row, the space in between the pins will be taken up with the pins / holes for the DB-15 connector.

 

[Trash80toG4]

Excellent, we can ignore pins 8, 9, 10, 18 and 26!

That yields plenty of PCB real estate for a thruhole HD-15 I'd think. Might be a tad off center, but we're in business. Sense lines can be generated in VGA conversion if needed. Dedicated red and green ground shields can be had elsewhere and GND pin 10 can be picked up wherever.

Great job!

 

[Kai Robinson]

No problem, I wasn't attacking I'm just... I tend to clarify rather than leave stuff, a bad habit from a character flaw and critically precise working environment where misunderstandings cause major problems. No offense meant.

I was sharing thoughts because it is something I've considered doing to my 6100 and / or 8100. I really hate the HDI45 adapters.

Unrelated, but Apple were putting the HDI-45 on the dual 601 CPU prototype that they cancelled - the tesseract project that the PDM / 6100 and family was the fallback for.

View attachment 19332

Wait, three PPC601's on a board??

 

[Kai Robinson]

Tum te tum...

View attachment 19334

View attachment 19335

Edit : an extra fold tab at the top is likely needed for rigidity - it will be harder to include if you don't have the right tools, but worth it. Sorry the folds don't look like proper sheet metal folds, I haven't used this cad software in 10 years and never have much.

View attachment 19336

Could that be 3D printed by PCBWay in metal?

 

[ClassicHasClass]

Ungupatchka! Just gotta love that beautiful mess. Haven't done the pin count, are those PCI slots?

More likely to be PDS. O.G. PCI only dates to around 1992 and Apple had no compatible ecosystem at the time.

 

[Trash80toG4]

@phipli absolutely love your work here, but scale is problematic.



Your lug holes will be nearly tangential to the bend lines on the sides. Nuts won't work and I don't think heavy enough metal to tap would be a workable alternative. Too much work even if possible.

I'm wondering if the shield slots are large enough to heat/moldover/weld(?) pins on the bottom of a printed chassis to the bottom of the PCB? Can't remember the name of the process Apple used to attach heat shields parts to case plastics.

Anyway, drilling the slots out to round and bolting up a printed chassis for the connector ought to work fine?

 

[phipli]

@phipli absolutely love your work here, but scale is problematic.

View attachment 19347

Your lug holes will be nearly tangential to the bend lines on the sides. Nuts won't work and I don't think heavy enough metal to tap would be a workable alternative. Too much work even if possible.

I'm wondering if the shield slots are large enough to heat/moldover/weld(?) pins on the bottom of a printed chassis to the bottom of the PCB? Can't remember the name of the process Apple used to attach heat shields parts to case plastics.

Anyway, drilling the slots out to round and bolting up a printed chassis for the connector ought to work fine?

I'm afraid that I don't quite follow, the issues you raise aren't things that I was worried about and I'm not 100% sure which features you mean - regional dialects can vary a lot for this sort of thing - lug can mean several things, in this context it could mean the threaded parts that go through the VGA port, or the four tabs for the bracket itself - it isn't usually a technical term except in the context of cable mounting lugs perhaps. I'd been worrying about a some difficulties, like corrosion prevention, thermal mass, and order of operations, but the first - you can plate at home, the second, pre-heating and a good iron, and the third I'd worked out a workable order.

Are you worrying about distortion due to the proximity of the VGA ports mounting hole to the fold? That shouldn't be an issue, the face with the openings can be clamped during the fold, or alternatively the holes can be drilled after the folding is complete.

Threading? Nothing on the part needs threading? I'm not quite sure where you mean? It sounds like you mean into the four tabs... but that isn't what they're for - they're for soldering. The thing I'm wondering here is if I'd need to go down to 0.6mm steel to help with the thermal mass. The thinner the metal used, the more critical the extra fold in the third picture I put in the post afterwards becomes, that is for rigidity. I'd fold it with extra length, and then cut it down afterwards - it is difficult to fold things without sufficient material.

Dimensions on the mounting face are vaguely accurate (I measured connectors with a ruler), to build I'd measure accurately and prototype in cardboard. But the tabs that go into the PCB have just been guessed, and the length of the two... "wings" have only been eyeballed. My 6100 is in use so I'm not able to take measurements atm.

Some of the other things you've mentioned, I realise I probably haven't specifically mentioned it, but my intention was to solder the bracket in place, in a similar manner to how most ports are mounted on a mac.

A 3D printed part could potentially be fixed by applying heat to the four tabs like you say, but I don't have a 3D printer - I'm more of a metal kind of person.

Really sorry if I'm misunderstanding, well, I know I am, I don't quite follow your post.

 

[phipli]

Could that be 3D printed by PCBWay in metal?

Should be fine for that, probably a little expensive. Likely cheaper to get it laser cut locally and even folded. Then you'd just need to plate it at home to stop it rusting.

 

[phipli]

Wait, three PPC601's on a board??

Just two - I think you're looking at the prototype chipset chips.

The CPUs are missing - one connector has been torn off and you can see 601 in the silkscreen footprint, the other is still there.

 

[phipli]

More likely to be PDS. O.G. PCI only dates to around 1992 and Apple had no compatible ecosystem at the time.

I believe they are probably what Apple called "BLT" (probably doesn't stand for Bacon, Lettuce Tomato). This is what was used for earlier revisions of this project - Jaguar and Hurricane.

Couple of extracts from Apple docs from the period.