TwiSlot PCI Riser and ComSlot2 NIC repro in widened CS riser form factor

[Trash80toG4]

This insanity started just about 5 years ago:

Twin PCI Riser Project for TAM, 6360, 5x00 Single Slot PowerMacs.

My 63060 and its 5400/5500 low drawer brethren are fairly low hanging fruit as far as this project goes, so I'm taking on the TAM challenge first. Cutting that riser down and removing the rearward offset is not biggie, but fitting a second, low profile PCI Card into available CS2 cubic is a...

68kmla.org

Best explanation: TwinSlot Riser for TAM & version for 6360.

Looking at having a USB connector and a 12v in jack for powering the two PCI Slots with Buck converters on board, that I might be able to handle.

Developing a repro 10/100 NIC in the form factor of the TAM's ComSlot Riser and powering it from same would be oh so cool, but beyond my ken.

So we've got squat for power in the TAM, might feeding 12V into the backplane power the three expansion cards?

Table 4-11 Pin assignments for the PCI-bus communications slot connector
Odd-numbered pins Function Even-numbered pins Function
1 /DCD ____________2 /DTR
3 /CTS_____________4 /RTS
5 RxD _____________6 TxD
7 IN_SENSE _______8 SCC_ENAB
9 INT_MIC _______10 MIC_SENSE
11 MIC_RET______12 EXT_AUD_L
13 Reserved _____14 EXT_AUD_RET
15 GND__________ 16 +12V
17 -5V___________ 18 +12V
19 SYS_WAKEUP _20 Trickle +5
21 GND __________22 GND
23 A1 ____________24 A0
25 A3 ____________26 A2
27 +3.3V _________28 +3.3V
29 A5 ____________30 A4
31 A7 ____________32 A6
33 +5V ___________34 +5V
35 A8 ____________36 C/BE(0)~
37 A10___________ 38 A9
39 GND __________40 GND
41 A12 ___________42 A11
43 A14 ___________44 A13
45 C/BE(1)~ ______46 A15
47 GND __________48 Gnd
49 SERR~ ________50 PAR
51 PERR~ ________52 SBO~
53 LOCK~ ________54 SDONE
55 +3.3V _________56 +3.3V
57 DEVSEL~______58 STOP~
59 IRDY~ ________60 TRDY~
61 +5V___________62 +5V
63 C/BE(2)~ ______64 FRAME~
65 A17 ___________66 A16
67 GND __________68 GND
69 A19 ___________70 A18
71 A21 ___________72 A20
73 A23___________ 74 A22
75 GND __________76 GND
77 C/BE(3)~ ______78 IDSEL
79 A25 ___________80 A24
81 A27 ___________82 A26
83 +3.3V _________84 +3.3V
85 A29 ___________86 A28
87 A31 ___________88 A30
89 +5V ___________90 +5V
91 REQ~__________92 GNT~
93 +5V____________94 +5V
95 INT~ __________96 Reserved
97 Reserved _____98 RST~
99 GND _________100 Reserved‘
101 CLK ________ 102 Reserved
103 GND ________104 Reserved
105 Reserved ___106 Reserved
107 Reserved ___108 Reserved
109 CommGnd __110 RefGnd
111 AudToSlot __112 AudFromSlot

to be continued . . .

 

[Trash80toG4]

First, is the notion of powering expansion cards from an external 12V source sound like a possibility?
Reports in MLA thread indicate TAM lacking power to run G3, Tango PCI and a 10/100 NIC in combination, but the Apple NIC works.

That's why I'm looking at powering the twin PCI Slot card from an external source in the TAM. Thoughts?
Lacking a TAM, I've no need of such tomfoolery, but it would be cool to push TAM I/O over the top!

Found some strangeness in the first pinout I found for PCI.

Universal PCI Bus Pinouts Rear of Computer

__________________ :------:------:
_____________ -12V |- B1 A1 -| Test Reset
________Test Clock |- B2 A2 -| +12V
___________Ground |- B3 A3 -| Test Mode Select
_Test Data Output |- B4 A4 -| Test Data Input
_______________+5V |- B5 A5 -| +5V
_______________+5V |- B6 A6 -| Interrupt A
_______Interrupt B |- B7 A7 -| Interrupt C
_______Interrupt D |- B8 A8 -| +5V
_________PRSNT1# |- B9 A9 -| Reserved
_________Reserved |- B10 A10 -| +V I/O
_________PRSNT2# |- B11 A11 -| Reserved
__________________ :------:------:
__________________ :------:------:
________Reserved |- B14 A14 -| Reserved
__________Ground |- B15 A15 -| Reset
____________Clock |- B16 A16 -| +V I/O
__________Ground |- B17 A17 -| Grant
_________Request |- B18 A18 -| Ground
___________+V I/O |- B19 A19 -| Reserve
______Address 31 |- B20 A20 -| Address 30
______Address 29 |- B21 A21 -| +3.3V
__________Ground |- B22 A22 -| Address 28
______Address 27 |- B23 A23 -| Address 26
______Address 25 |- B24 A24 -| Ground
____________+3.3V |- B25 A25 -| Address 24
___________C/BE 3 |- B26 A26 -| Init Device Select
______Address 23 |- B27 A27 -| +3.3V
__________Ground |- B28 A28 -| Address 22
______Address 21 |- B29 A29 -| Address 20
______Address 19 |- B30 A30 -| Ground
____________+3.3V |- B31 A31 -| Address 18
______Address 17 |- B32 A32 -| Address 16
___________C/BE 2 |- B33 A33 -| +3.3V
__________Ground |- B34 A34 -| Cycle Frame
__Initiator Ready |- B35 A35 -| Ground
____________+3.3V |- B36 A36 -| Target Ready
___Device Select |- B37 A37 -| Ground
__________Ground |- B38 A38 -| Stop
_____________Lock |- B39 A39 -| +3.3V
______Parity Error |- B40 A40 -| Snoop Done
____________+3.3V |- B41 A41 -| Snoop Backoff
____System Error |- B42 A42 -| Ground
____________+3.3V |- B43 A43 -| PAR
___________C/BE 1 |- B44 A44 -| Address 15
______Address 14 |- B45 A45 -| +3.3V
__M66EN/Ground |- B46 A46 -| Address 13
______Address 12 |- B47 A47 -| Address 11
______Address 10 |- B48 A48 -| Ground
__________Ground |- B49 A49 -| Address 9
__________________ :------:------:
__________________ :------:------:
_______Address 8 |- B52 A52 -| C/BE 0
_______Address 7 |- B53 A53 -| +3.3V
____________+3.3V |- B54 A54 -| Address 6
_______Address 5 |- B55 A55 -| Address 4
_______Address 3 |- B56 A56 -| Ground
__________Ground |- B57 A57 -| Address 2
_______Address 1 |- B58 A58 -| Address 0
___________+5 I/O |- B59 A59 -| +V I/O
_____ ACK 64-bit |- B60 A60 -| Request 64-bit
_____________+5V |- B61 A61 -| +5V
_____________+5V |- B62 A62 -| +5V
___________________ :------:------:
___________________ :------:------:
________Reserved |- B63 A63 -| Ground
__________Ground |- B64 A64 -| C/BE 7
___________C/BE 6 |- B65 A65 -| C/BE 5
___________C/BE 4 |- B66 A66 -| +V I/O
__________Ground |- B67 A67 -| Parity 64-bit
______Address 63 |- B68 A68 -| Address 62
______Address 61 |- B69 A69 -| Ground
___________+V I/O |- B70 A70 -| Address 60
______Address 59 |- B71 A71 -| Address 58
______Address 57 |- B72 A72 -| Ground
__________Ground |- B73 A73 -| Address 56
______Address 55 |- B74 A74 -| Address 54
______Address 53 |- B75 A75 -| +V I/O
__________Ground |- B76 A76 -| Address 52
______Address 51 |- B77 A77 -| Address 50
______Address 49 |- B78 A78 -| Ground
___________+V I/O |- B79 A79 -| Address 48
______Address 47 |- B80 A80 -| Address 46
______Address 45 |- B81 A81 -| Ground
__________Ground |- B82 A82 -| Address 44
______Address 43 |- B83 A83 -| Address 42
______Address 41 |- B84 A84 -| +V I/O
__________Ground |- B85 A85 -| Address 40
______Address 39 |- B86 A86 -| Address 38
______Address 37 |- B87 A87 -| Ground
___________+V I/O |- B88 A88 -| Address 36
______Address 35 |- B89 A89 -| Address 34
______Address 33 |- B90 A90 -| Ground
__________Ground |- B91 A91 -| Address 32
________Reserved |- B92 A92 -| Reserved
________Reserved |- B93 A93 -| Ground
__________Ground |- B94 A94 -| Reserved
__________________ :------:------:

I've never seen these before. logic level references vs. power supply?
+V I/O
+5 I/O

12Volt input jack to power cards yields
+12V
+5V via stepdown converter
+3.3V via stepdown converter

These can't be derived from 12V input jack?
-12V
-5V - don't see it on PCI, was that on ComSlot for the MoDem only?

 

[Trash80toG4]

Parallel Developments:

TAM - Internal PicoPSU - AUX Power for Expansion Slots
TAM - Fatback - Printable copy & FatterBack maximum expansion Mod of baseline model?

Cardboard template development, close approximation achieved I think:

Baseline TAMtent w/ Rage128/DVD Daughtercard

TwinSlot adapter card at standard PCI interslot spacing. Rage128/DVD Daughtercard + MacallyUSB installed, plenty of clearance"

Physical design test: clearance of USB Card at standard PCI interslot spacing. Rage128/DVD Daughtercard + Tango2 installed:

Top, setback slot can be spaced closer to standard riser slot than PCI spec for additional clearance for combo card. Depending upon actual spacing or inner surface it may fit with standard interboard spacing. There looks to be more height available at the inner surface level than the template would indicate?

 

[Trash80toG4]

Last pics for the day:

Mock-up in my 6360 with Sonnet Tempotrio in the top/setback slot. Card isn't supported in the funky Gazelle architecture of the TAM/5500/6500, but docs say Alchemy Macs are supported. USB, Firewire and faster ATA/133 bus all on one card.

Slot connector is spaced .2" lower than the standard 2cm PCI slot spacing. There's some wiggle room to increase the spacing a bit.

Top slot is set back far enough for cables to be hooked up internally and snaked wherever for whatever. In my 6360, FW/USB will likely be routed out the vestigial PDS Card opening or wires terminated to connectors on a PCB in that location.

It's too bad this puppy won't play nicely with the TAM, ATA/133 and SSD would make it scream. Wondering if 6360 will outrun the TAM despite its slower bus, given faster disk I/O?

Tempo Trio ATA 133/FireWire 400/USB 2.0 PCIe Computer Card - Sonnet

 

[misterg33]

Couple of things -- I wasn't sure from the other thread if you were planning to use an external power supply for one of the cards. I'd be worried about pushing the TAM power supply (located in the subwoofer module) too far by installing so many cards, since it wasn't designed for this much power draw.

I also noticed you might be using a Sonnet accelerator *and* a USB card (or USB & firewire card) in one of the pictures? I didn't have much luck with the G3 accelerator in my TAM. It made it lock up a lot. Sonnet made a firmware patch you can apply that's supposed to help, but I didn't notice much difference. If you reset the pram, or your pram battery goes dead, you have to reapply the patch (it's an app you run).

There's also some weird EFI patching you can do to try to get the firewire port of the Sonnet Tango card to work in the TAM. If you do some online searching, you can probably find the thread about it. If not, I can dig around in my bookmarks to see if I can find it. I've never tried it. It's something to do with modifying the device tree in open firmware.

 

[François]

AFAIK the PSU of the TAM is the same PSU used in the Power Mac 6500 and you can install two PCI cards in the 6500.

Or you can always replace the TAM PSU with a modern one, surely you can find one providing more power for less volume (and heat).

BTW, I think @Action Retro had already done that, with a PCI riser from a 6500 in his TAM, check out his YouTube channel!

Here’s the video:

 

[Trash80toG4]

Thanks for the input. I saw a similar TAM upgrade using the 6400/6500 riser almost twenty years ago. Interesting that he's using USB and SATA cards. I've got a SATA card ready to go, but I've been using a USB card for the diagrams as that's what folks expected to see five years ago.

Ideally, I'd like to explore removal of the HDD/bracket for employing something along these lines:

https://www.newegg.com/kingspec-1tb-m-2-sata3-2280-ssd/p/0D9-000D-00136

Dunno, finding and filling available cubic in Mac cases has long been an obsession.

I don't need USB for my purposes, I want to put a high performance VidCard in the stock slot position, which will put a lot more strain on the TAM PSU than what his setup did. Love that big fan***

I've addressed the power issue by using a PicoPSU for aux power in this parallel thread:

https://tinkerdifferent.com/threads/tam-internal-picopsu-aux-power-for-expansion-slots.3612/

Hope to have the heat issues discussed and ironed out there. Currently planning a customized, repro CSII riser for the TAM where I'll have provision for the PicoPSU, leaving the connector thruholes unimplemented. Leaving off the PicoPSU socket and populating the edgecard connector should be helpful for those who've said they lack the CSII riser.



tangent

*** This blurb is for yet another parallel thread:

I've been searching for a bigger fan than his Noctua already. A clear plexi FatterBackwould seem to be an interesting upgrade, those too are missing from many a TAM or so I've read. This one may be a bit too wide, but it or a smaller fan mounted into a circular well at the bottom of the vacuum formed clear plexi coverplate would be cool . . . so to speak. Running it off 5V or even 3.3V should be a very quie solution. How wide are the TAM backs?

Amazon.com

Hard to imagine Apple thought removing the stock back's fan in the implementation of FatBack for adding an expansion card could work out:

/tangent

 

[Trash80toG4]

I also noticed you might be using a Sonnet accelerator *and* a USB card (or USB & firewire card) in one of the pictures? I didn't have much luck with the G3 accelerator in my TAM. It made it lock up a lot. Sonnet made a firmware patch you can apply that's supposed to help, but I didn't notice much difference. If you reset the pram, or your pram battery goes dead, you have to reapply the patch (it's an app you run).

Methinks the problem lies within the funky Gazelle architecture, not power issues, hence that firmware upgrade. Problems with FireWire bridges and my no longer supported TEMPOtrio abound, the Alchemy architecture was solid at its 40MHz bus speed. G3 L2 accelerators and combo cards were devekoped for Alchemy.

 

[ClassicHasClass]

Yeah, my TAM runs just fine with an L2 G3. Perhaps there was a difference in submodels or boards.

 

[Trash80toG4]

According to everymac L2 G3 cards were released between 1/4/99 thru 11/1/01 for different architectures.

January 4, 1999 - https://everymac.com/upgrade_cards/sonnettech/crescendo_g3_l2/crescendo_g3_l2_225.html
January 4, 1999 - https://everymac.com/upgrade_cards/sonnettech/crescendo_g3_l2/crescendo_g3_l2_300.html

May 15, 1999 - https://everymac.com/upgrade_cards/sonnettech/crescendo_g3_l2/crescendo_g3_l2_250.html

May 17, 1999 - https://everymac.com/upgrade_cards/sonnettech/crescendo_g3_l2/crescendo_g3_l2_400.html

September 15, 1999 - https://everymac.com/upgrade_cards/sonnettech/crescendo_g3_l2/crescendo_g3_l2_350.html

November 1, 2001 - https://everymac.com/upgrade_cards/sonnettech/crescendo_g3_l2/crescendo_g3_l2_350.html
Two versions of this one. Blooper calling Alchemy models Gazelle at the beginning, but specifically stating support dropped for
Gazelle models later in the description for second(?) version.


(c) 200 G3 l2 Installation guide - https://www.sonnettech.com/support/downloads/manuals/crescendo_l2_manual.pdf


After work tonight I'll try to edit in the Sonnet info for each model.

Interesting tangent!

 

[joevt]

Methinks the problem lies within the funky Gazelle architecture, not power issues, hence that firmware upgrade. Problems with FireWire bridges and my no longer supported TEMPOtrio abound, the Alchemy architecture was solid at its 40MHz bus speed. G3 L2 accelerators and combo cards were devekoped for Alchemy.

The Sonnet Tempo Trio problem was mostly solved at:
https://68kmla.org/bb/index.php?threads/the-great-gazelle-pci-hack-thread-part-2.38360/post-467825
There's a patcher:
https://68kmla.org/bb/index.php?threads/the-great-gazelle-pci-hack-thread-part-2.38360/post-467844
And a video:
https://68kmla.org/bb/index.php?threads/the-great-gazelle-pci-hack-thread-part-2.38360/post-471344

 

[Trash80toG4]

Thanks! Forgot all about that fix! Saw it and commented on it in January '23 and then promptly forgot. I fell off both forums for quite some time, just getting back into things here and there and digging back into this project for R&R.

Interesting commentary in that video:

The lack of slot support jankiness he's talking about when using the 6400/6500 riser is due to it being incompatible with the slot in the backplane (sideplane?) panel of the FatBack. It's missing from his setup in the video because of physical interference from the inappropriately positioned cards.

Such won't be a problem for my TwinSlot Riser. The main PCI card's bracket bolts right up where it belongs. This in turn supports the second slot which is set back from the first by a minimum of 2cm so connections can be made to its backplane connectors.

The clear FatterBack fabbed in plexi I mentioned in a parallel thread is no big deal. I'll take that up again in there if someone wants to scan the four sides of the baseline model.


BTW, what's the deal with using the crappy 10bT NIC from Apple when USB is available for networking? OS/driver limitations?

 

[joevt]

BTW, what's the deal with using the crappy 10bT NIC from Apple when USB is available for networking? OS/driver limitations?

I don't know. USB has more CPU overhead than a PCI NIC?

Old World Macs had 10Base-T.
New World Macs got 100Base-T.
Power Mac G4 (Gigabit) and later got 1000Base-T.

 

[Trash80toG4]

Hrmmm . . . how much does the CPU actually do while communicating with a network?

Your posts, especially that video, have me energized to take the physical experimentation cardboard mockups to fruition. I'd love to send that guy a riser that interfaces correctly with the fatback slot cover's support and lid.

 

[Trash80toG4]

Can anyone take some pics from the "inside" of the FatBack with a card and its PCI riser bolted up to the slot cover for me?

Clearance measurements between the top of the stock riser and inner surface of the lid would be a huge help. This is about the only time I've ever been interested in having a TAM in my collection.

Developing a printed sideplane has been a casual interest in this five year backburnered project.

 

[Trash80toG4]

Is anyone in the U.S. interested in doing the physical/fitment testing the cardboard prototype for me? No harm can be done by my paper/cardboard doohickey.

pm if interested.

 

[ClassicHasClass]

Can anyone take some pics from the "inside" of the FatBack with a card and its PCI riser bolted up to the slot cover for me?

Can you be more specific about the angle you want?

I'm in the US with a TAM, but I may not be able to do this for a bit, since I have a family issue to deal with first.

 

[Trash80toG4]

Thanks, I'll mock it up and post pics.

If any TAMs reside within an hour or so from Raleigh, NC or are in the D.C vicinity between June 12 and June17th I'll be up there for a bit.

 

[bakkus]

Nothing to add here, other than that I always enjoy the heck out of @Trash80toG4's musings about Apple futures we never had

 

[Trash80toG4]

AFAIK the PSU of the TAM is the same PSU used in the Power Mac 6500 and you can install two PCI cards in the 6500.

Looks like you're probably right. Development was hurried, so they must have used the already UL approved 6500 PSU? What a kluge!

My 150 W 6500(?) PSU looks a lot like that picture. Only the A/C passthru for display connection is missing. Looks like the metal housing is enclosed in a plastic cover?

From the service source specs:
6400 - 220 W maximum continuous
6500 - 140 W maximum continuous
TAM - Max Watts: 140

Is there a teardown/repair video of the TAM base unit? It's service source is the only one I can recall without the exploded diagram. Apparently Apple was too embarrassed about how friggin' ugly the TAM insides were to provide one, someone might even take it apart! Yep, can't judge a book by its cover.

Entire PSU transplant or transplant of the guts of the 6400 PSU into the TAM can/base unit would appear to be the easiest way to get the additional power required. Methinks the LCD and its craptastic controller board likely devours more than an L2 G3 and second PCI slot's share of 140 W?