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

[joevt]

Ah, no, you don't need anything other than some traces and a single diode to adapt from PowerMac video to VGA. The adapters with switches are only needed for older macs, and even they just have like, two diodes in them, unless you're messing with sync and auto detection. Here is the required circuit:

A diode from pin 10 (SENSE2) to pin 7 (SENSE1) with ground on pin 4 (SENSE0) is for MultiScan 21" displays.
VGA uses a wire instead of a diode between SENSE2 and SENSE1 with SENSE0 not connected.
#41 #42

VGA and MultiSync 21" are probably the two best choices for an adapter since they give the widest range of display modes.
They may have a slightly different list of default modes but it depends on the graphics controller and its drivers. A program like SwitchRes may be able to enable non-default modes.

The 21" list of modes includes some Mac specific modes. The VGA list includes PC specific modes.

You can check the list of modes using an emulator like DingusPPC which allows specifying the Apple sense code for the display. Does the Power Mac 6100 graphics driver read EDID? DingusPPC needs more work regarding EDID reading. DingusPPC only emulates the default graphics controller of the Power Mac 6100 which is missing the higher resolutions (1024 x 768 and higher).

VGA
 640 x  480  60  Hz
 640 x  480 120  Hz
 800 x  600  60  Hz
 800 x  600  72  Hz
 800 x  600  75  Hz
1024 x  768  60  Hz
1024 x  768  70  Hz
1024 x  768  75  Hz
1280 x  960  75  Hz
1280 x 1024  75  Hz

Multiscan 21"
 640 x  480  67  Hz
 640 x  480 120  Hz
 832 x  624  75  Hz
1024 x  768  74.9Hz
1152 x  870  75  Hz
1280 x  960  75  Hz
1280 x 1024  75  Hz

 

[phipli]

A diode from pin 10 (SENSE2) to pin 7 (SENSE1) with ground on pin 4 (SENSE0) is for MultiScan 21" displays.
VGA uses a wire instead of a diode between SENSE2 and SENSE1 with SENSE0 not connected.
#41 #42

VGA and MultiSync 21" are probably the two best choices for an adapter since they give the widest range of display modes.
They may have a slightly different list of default modes but it depends on the graphics controller and its drivers. A program like SwitchRes may be able to enable non-default modes.

The 21" list of modes includes some Mac specific modes. The VGA list includes PC specific modes.

You can check the list of modes using an emulator like DingusPPC which allows specifying the Apple sense code for the display. Does the Power Mac 6100 graphics driver read EDID? DingusPPC needs more work regarding EDID reading. DingusPPC only emulates the default graphics controller of the Power Mac 6100 which is missing the higher resolutions (1024 x 768 and higher).

VGA
 640 x  480  60  Hz
 640 x  480 120  Hz
 800 x  600  60  Hz
 800 x  600  72  Hz
 800 x  600  75  Hz
1024 x  768  60  Hz
1024 x  768  70  Hz
1024 x  768  75  Hz
1280 x  960  75  Hz
1280 x 1024  75  Hz

Multiscan 21"
 640 x  480  67  Hz
 640 x  480 120  Hz
 832 x  624  75  Hz
1024 x  768  74.9Hz
1152 x  870  75  Hz
1280 x  960  75  Hz
1280 x 1024  75  Hz

I tend to go for the 21" MultiScan option because it gives you 1152x870 - a lot of the older macs don't support any VGA resolutions above 1024x768, but my VGA monitors support 1152x870. So it makes more sense unless you're using a PCI era mac that can do 1280x1024... at which point I'd use a PCI card with a VGA port.

Like you say, you could use SwitchRes.

 

[phipli]

What kind of diode would go between pins 7 and 10 to tell the sense lines it should be VGA? And is it really needed here

Yeah, you do need it to get all the resolution options. Apple ran out of combinations, and so used diodes that meant you could detect not just a connection, but the direction of flow. What diode you use isn't too picky, just use a silicon rectifier diode like a 1n4007 if you already have one, or probably some kind of signal diode like a 1N4148?

 

[wottle]

Been looking at this design for flaws. Short "throw" of logic board installation is my main concern. It's tight, but probably long enough front to back for careful installation with adapter installed using double sided tape. Velcro might be a better option? How long are the pins on Apple's adapter cable?

Printed case offers some interesting possibilities involving (long throw) strap hinged design with built in strain relief for the cable. Kiss compliant option would be mounting a printed case bottom attached with beaucoups double sided tape for rock solid interface. A boltdown, removable top holding PCB would allow straight vertical removal/installation. Better yet might be firmly attached base with boltdown PCB and snap on or single bolt cover?

7100 has plenty of "throw" for far easier logic board installation and a convenient "lockdown" hole. Only the RA Cable would be blocking fan vents.

6100/7100 machines require model specific PCBs. Don't have an 8100 to examine, anyone got a nice pic of its ass to post?


edit: love the notion of a left angled VGA cable, the OEM adapter cable's wall standoff distance is downright ludicrous.

Yeah, I was so focused on the 6100 (and its lack of supported resolutions, I forgot about the 7100 and 8100. A universal solution would be fantastic!

The 8100 is going to be problematic because of the face that the port section is recessed. I actually have one, but never had to worry about the HDI-45 port because it had a video card with DB15 on it.



It's going to be a challenge getting the header pins deep enough into the port, I think, and the board is going to be mounted at an angle. when I get my test board, I'll see how it does. Would it make sense to basically have two sandwiched boards connected by headers on the sides that would allow the smaller HDI connector board to be recessed down against the read plate, and the breakout board with the VGA cable directly on top of it.

 

[phipli]

Yeah, I was so focused on the 6100 (and its lack of supported resolutions, I forgot about the 7100 and 8100. A universal solution would be fantastic!

The 8100 is going to be problematic because of the face that the port section is recessed. I actually have one, but never had to worry about the HDI-45 port because it had a video card with DB15 on it.

View attachment 19424

It's going to be a challenge getting the header pins deep enough into the port, I think, and the board is going to be mounted at an angle. when I get my test board, I'll see how it does. Would it make sense to basically have two sandwiched boards connected by headers on the sides that would allow the smaller HDI connector board to be recessed down against the read plate, and the breakout board with the VGA cable directly on top of it.

7100 is also recessed and the end of the socket is flush (it seems to protrude slightly on the 6100 - so while the opening is recessed, the end of the socket isn't).

 

[Trash80toG4]

Yeah, I was so focused on the 6100 (and its lack of supported resolutions, I forgot about the 7100 and 8100. A universal solution would be fantastic!

Below is a two-fer solution, 6100/7100.

The 8100 is going to be problematic because of the face that the port section is recessed. I actually have one, but never had to worry about the HDI-45 port because it had a video card with DB15 on it.

Yeah, not sure I'd worry about the 8100, did any config ship w/o either A/V or HPV cards installed?

When I misquoted the Sundance Kid about the amount of DS tape on the right angled VGA Cable contraption cases' backside, figured I could go back to my original notion for a vertical VGA port:

The red pins are reserved lines, intended to stabilize the unit for insertion of the PCB, They'd be long, stacking headers, so length's no problem at all. Just solder 'em on at the proper depth and nip 'em off on the component side.


edit: spelling

 

[phipli]

did any config ship w/o wither A/V or HPV cards installed?

Yeah, there was a an option that had a PDS terminator and a Apple Display 24AC Nubus card. Also the 8150 shipped like that.

 

[Nycturne]

The red pins are reserved lines, intended to stabilize the unit for insertion of the PCB, They'd be long, stacking headers,. Do length's no problem at all. Just solder 'em on at the proper depth and nip 'em off to proper length.

I'm curious, why not not use additional pins for stability and keying the headers so it's harder to insert it into the wrong pins, or are they just not shown?

 

[Trash80toG4]

Good suggestion, wondered if having n.c. pins installed to flesh out the interface might be problematic for any reason? I'm amazed at the lack of non-signal specific grounding pins!

Was also wondering if adding a "shroud/surround" to the printed case bottom might be useful for alignment? It would be a plug to fit space around the black plastic, inside the metal shroud.

Just trying to get the basics down so folks can understand where I'm headed with this.

 

[Nycturne]

Good suggestion, wondered if having n.c. pins installed to flesh out the interface might be problematic for any reason? I'm amazed at the lack of non-signal specific grounding pins!

The original problem I saw is that the headers "fit" in 3 different positions, one correct, two incorrect. I don't think anyone wants to be trying to debug an adapter where a user plugged it into the wrong set of pins. 1, 2, 19 and 36 are all shield/ground pins, and spaced in such a way that they help eliminate those two incorrect positions. That would be enough, IMO. But the spacing is different there.

The only real problem I can see with extra pins is that every pin you add which isn't ground/shield now exposes that signal on your PCB. They are still N/C at the end of the day. This can be good if your goal is to debug things without removing the adapter. since you now have an additional point you can probe. But it's now one more thing to be careful of, as a solder bridge across 35 and 36 will blow your Mac's ADB fuse, for example.

Was also wondering if adding a "shroud/surround" to the printed case bottom might be useful for alignment? It would be a plug to fit space around the black plastic, inside the metal shroud.

Absolutely would be useful. It would help protect the pins from bending, and further remove the possibility for someone to insert it into an incorrect position.

 

[wottle]

The original problem I saw is that the headers "fit" in 3 different positions, one correct, two incorrect. I don't think anyone wants to be trying to debug an adapter where a user plugged it into the wrong set of pins. 1, 2, 19 and 36 are all shield/ground pins, and spaced in such a way that they help eliminate those two incorrect positions. That would be enough, IMO. But the spacing is different there.

The only real problem I can see with extra pins is that every pin you add which isn't ground/shield now exposes that signal on your PCB. They are still N/C at the end of the day. This can be good if your goal is to debug things without removing the adapter. since you now have an additional point you can probe. But it's now one more thing to be careful of, as a solder bridge across 35 and 36 will blow your Mac's ADB fuse, for example.



Absolutely would be useful. It would help protect the pins from bending, and further remove the possibility for someone to insert it into an incorrect position.

I was certainly planning to put pins in the “extreme” positions. But he outside pins with different spacing are standard 2.54mm spacing, so it should be easy.

Originally, I was very concerned about the pin alignment when I was considering a 1-for-1 replacement of the current adapter. Meaning, it would be a dongle that your stick out and would be transferable to another machine easily. With this adapter, and the fact the plan is to adhere it to the back of the machine, less concerned about that and think the shroud (for alignment and stability) is less of a concern. As long as car is taken when inserting the one and only time you do so, I’m comfortable just having enough pins to ensure you’re not shifted by one, potentially shorting things you don’t want.

I actually realized I need to create a second, DB-15 version. I agree few people would prefer DB-15 output on their 6100, but then I just realized that I have a handful of DOS compatible versions, and the Y cable only allows for DB-15 output. Should be easy, though, as it’s simply removing the diode and rerouting to the slightly different connector.

 

[ClassicHasClass]

Yeah, there was a an option that had a PDS terminator and a Apple Display 24AC Nubus card. Also the 8150 shipped like that.

So did the 9150 (PDS terminator), though it has a regular DA-15 port for video.

 

[Trash80toG4]

Don't think a DA-15 version of my bodge would be worth the effort. Very few to none would be interested in it to be worth the effort. There are pricey OEM cables available for purists with Apple monitors in need of one available.

I wouldn't want to see a VGA adapter hanging off the case in any event.

Back to KISS principle compliance:
Shroud's not a good idea on second thought, it makes for designing/printing the case far more complicated. Much better to use a flat backside as base during printing would Keep It Simple Simple, Silly me.

Next version of my graphic will include a hole to be drilled (tapped?) so a machine screw added as a stud for alignment purposes during installation.



I really hope someone decides to sell these, I need three PCI risers designed for the Alchemy/Gazelle project. Any straight up volunteers for that? At 70 with retirement postponed, life's too short to await that to begin to climb the KiCAD learning curve.

 

[wottle]

Don't think a DA-15 version of my bodge would be worth the effort. Very few to none would be interested in it to be worth the effort. There are pricey OEM cables available for purists with Apple monitors in need of one available.

I wouldn't want to see a VGA adapter hanging off the case in any event.

Back to KISS principle compliance:
Shroud's not a good idea on second thought, it makes for designing/printing the case far more complicated. Much better to use a flat backside as base during printing would Keep It Simple Simple, Silly me.

Next version of my graphic will include a hole to be drilled (tapped?) so a machine screw added as a stud for alignment purposes during installation.



I really hope someone decides to sell these, I need three PCI risers designed for the Alchemy/Gazelle project. Any straight up volunteers for that? At 70 with retirement postponed, life's too short to await that to begin to climb the KiCAD learning curve.

That’s why I’m proposing two. Simple to have two versions of the Gerber’s, with the board being basically the same, and even possibly you could have a 3-D case that would work for both, with just a wider and taller opening to accommodate both the DB 15 and the HD 15 ports.

getting the PCB designed for the HDI 45 input is the harder part, and I think I’ve already got that figured out and designed. Doing slightly different trace routing and removing the diode is simple.

 

[Trash80toG4]

Thinking my AI playtime input has about run its course here, I'll finalize the graphic and sit back to enjoy the show.

 

[wottle]

OK, so my initial test board to confirm my spacing was correct came in. My original plan to use headers with the plastic spacers in place didn't work because there simply isn't enough room. I could use the spacers on rows 1, 3, and 5, and then insert bare pins into the spots needed in rows 2 and 4.

Also, I was going to use standard 2.54mm headers for the wider spaced ones, but that won't work because apparently the holes I used for the pins are too small. Given the lack of room for traces around the pads, I don't want to go any bigger, so I'll likely leave it as is, and just plan on doing what I did, which was using all 2mm pitch header pins, and for the wider spacing ones on the side (they aren't actually used at all, so in theory I could leave them out), I just soldered in individual pins.

Unfortunately, I apparently cannot count and somehow made my PCB for the even more rare HDI-55 connector, because I somehow added two extra pins on each row... This was easy to work around by simply not using two holes on the side with consistent 2mm spacing on each row.

So, I added some standard length 2mm pitch header pins, sticking out as much as possible, and then used them in my detached HDI-45 port and it seems to fit perfectly.

Please, no commentary on my soldering, I was just trying to get the headers connected and verify they'd make it in the holes on the female HDI-45 port. With those standard length pins, they can reach, and I did verify continuity with all the pins, but the PCB is basically flush with the metal shielding. I have some 19mm long 2mm pitch header pins on the way that will allow it to be more forgiving. I'm a bit worried that the longer the pins are, the more difficult it will be to get them soldered in place straight and plumb, but we'll see.

The good news is that the odd spacing of the pins on the far side mean if you add 4 of those pins, it basically prevents it from being inserted wrong, without really forcing it and bending those pins. This was confirmed when I went to test the connection again and failed for a minute or so because I was attempting it with the board upside down. So, with this layout confirmed, my prototype board with the actual wiring to a VGA connector is on the way. Unfortunately, that one is missing the diode, so my test run boards are going to be 640x480 only. I have the new design with the diode in place that should allow for more resolutions...

 

[Trash80toG4]

Looks great, considering my current skills with the iron I can't comment any other way.
I'd be using wire wrap headers for prototyping, but I'm a lurking mososaur.

You set the pins into the connector as a jig for soldering them, no?
You can file down the sides of the packs to fit all five rows w/o resorting to soldering bare pins.

 

[joevt]

OK, so my initial test board to confirm my spacing was correct came in. My original plan to use headers with the plastic spacers in place didn't work because there simply isn't enough room. I could use the spacers on rows 1, 3, and 5, and then insert bare pins into the spots needed in rows 2 and 4.

The diagram at #31 says the spacing between rows is 3mm so 2mm plastic spacers should have fit. The fact that the plastic spacers don't fit means the spacing between rows is less than 2mm (which is obvious looking at your PCB). In that case, you could try pin headers of < 2mm, 5 pins per column, 11 columns = 55 pins (10 are not used).

 

[Trash80toG4]

Has anyone put calipers on the pins of the OEM cable? Wondering about the proper diameter?

 

[wottle]

The diagram at #31 says the spacing between rows is 3mm so 2mm plastic spacers should have fit. The fact that the plastic spacers don't fit means the spacing between rows is less than 2mm (which is obvious looking at your PCB). In that case, you could try pin headers of < 2mm, 5 pins per column, 11 columns = 55 pins (10 are not used).

Yeah, that diagram was for the PCB pins on the board, not the external connector. The connector is actually less than 2mm between pins.

Has anyone put calipers on the pins of the OEM cable? Wondering about the proper diameter?

I just did that real quick. The pins on the adapter appear to be ~0.6mm round pins. The pins I used were 0.5 square pins.

You set the pins into the connector as a jig for soldering them, no?

I did not. The pins were such a tight fit that I had to push them in with pliers. And the PCB thickness, along with the tightness of the fit made it very easy to have the pins plumb to the board.

You can file down the sides of the packs to fit all five rows w/o resorting to soldering bare pins.

The header spacers are about 2mm and there’s only about 1.6mm of space. So you’d have to remove quite a bit. Given how few pins are needed, I don’t find it a problem to place 4 individual pins for rows 2 and 4.