Another IIci ROM hack

[ojfd]

dougg3, you mean the whole top layer is connected to + 5V? Not that it's wrong.... it's rather unusual. Normally one dedicates the whole layer to VCC when there are 3 or more layers available.
When there are only two layers availble I usually route +VCC as traces, dedicate one side to GND and connect all large copper islands on the other side to GND with several vias evenly spread around that copper island. That's when working with clock rates up to 25 MHz.

P.S. I know, giving advice is soo easy... Good luck!

[dougg3]

Yes, exactly, ojfd. Everywhere you see big red areas (except around the crystal) its connected to +5V. I do have thick +5V traces routed to everywhere inside the copper fill (the copper fill just covers them up), so it won't be a big deal to change the top copper fill to GND if needed.

I guess I was under the impression that a ground pour on one side and a VCC pour on the other side was common. I'm definitely not opposed to changing it so the top is ground too though I'm new to this PCB design stuff so I appreciate the help from more experienced people such as you

Would you mind looking at my +5V traces after I change the copper fill? I'm a little worried because I have no idea what I'm doing when I route the +5V traces and the copper pour kind of masked my concerns for example, does it matter how I do the traces as long as they are thick and touch each part that they are supposed to touch? Or are there more complicated rules about how they are supposed to be connected?

Thanks for all the tips/suggestions you've been giving me!

[ojfd]

Would you mind looking at my +5V traces after I change the copper fill?

No problem, but try not to overdo your design, if it works, then it is ok as it is.
I just thought that, if you route your VCC as traces and leave GND issues out for a moment it might turn out as simpler design.

Or are there more complicated rules about how they are supposed to be connected?

There are some, but I don't think these apply to your board. I'll see what papers on this subject I can dig up.( if I could only remember their names and where they are...)
One good idea is to separate ICs by using ferrite beads in their VCC lines and capacitors to GND after them. As bonus you gain space underneath those beads to route traces there, in cace where they will otherwise cross. I also don't see any larger (10uF +) filtering caps on your board. At least one would be good to buffer VCC that's coming from USB. SMD or 'thru hole', whatewer works.

EDIT - here's one by respectful Dutch engineer who also deals with mixed signal stuff, just like me:
http://www.tentlabs.com/InfoSupport/Technology/page35/files/Supply_decoupling.pdf
Quote from that paper:

Never use power planes. It is not needed, as dc-supply currents only need small traces. A power plane may resonate with the groundplane: your PCB will act as dipole antenna!

I knew, it was something like that, nice to have it confirmed by someone else. But as I said before, it might work OK on your board, so don't worry too much about it

EDIT 2 - see end of the paper Appendix 1:
http://www.micrel.com/_PDF/HBW/App-Notes/an-01.pdf

[ojfd]

dougg3, to confuse you even further, here are some more papers on power supply bypassing.
http://www.ti.com/lit/an/scaa048/scaa048.pdf
http://www.ti.com/lit/an/scba007a/scba007a.pdf
but that's ENOUGH!
TIP: try using different size of capacitors - i.e.1206 instead of 0805 if you need space to route trace underneath them. Ferrite beads could be1806 or even 'thru hole'.

[bigmessowires]

When there are only two layers availble I usually route +VCC as traces, dedicate one side to GND and connect all large copper islands on the other side to GND with several vias evenly spread around that copper island. That's when working with clock rates up to 25 MHz.

That's good info! I've made several custom PCBs over the past two years, all of them with GND and VCC pour on the top and bottom like Doug did, just because that's what seemed intuitive to me. But I don't have any formal training in PCB layout, and have just been winging it. I wish there were some kind of "PCB layout for Dummies" to learn from. I've been adding bypass capacitors, and trying to minimize trace lengths, but otherwise it's all just seat of the pants engineering. Fortunately everything I've designed so far has worked.

[Trash80toHP_Mini]

p.23 & gaining!

JT, the interrupt for each slot has support in hardware. So even though all the ROMs for both three and six slot models may have code to poll six slots when an interrupt occurs, the interrupt signal line(s) from the slot need to be connected properly.

I don't know how the 68030 models handle the hardware end -- I think it involves the VIA registers -- but in the PowerMac x500 models, the Grand Central chip has a pin for each interrupt and the interrupt wire from each PCI slot must be connected to a unique interrupt pin on Grand Central. Since the PCI slot IDs are coded into the ROM/Open Firmware Device Tree in those models, I suspect that the interrupt used needs to match the PCI slot address.

Anyway, the point is, that expanding the number of slots supported goes beyond providing support in the ROM and hooking up the expansion bus. The interrupt line connection can be pretty complicated.

Thanks for the informed input, trag. Your mastery of hardware/firmware is much envied by this technically untrained comrade.

I was ASSuMEing that the NuBus Bridge Chipset handled the Two Slot (DuoDock's Gemini Slot Splitter Card w/Slot Assignments ____ and ___ ), Three Slot & Six Slot NuBus implementations, which are ASSuMEdly Buffered by that ChipSet from the rest of the I/O structure of the Mac Hardware.

The evidence that leads me to that conclusion would be that the NuBus Slots on the DuoDock remain active even with the Dock's DeclROM removed/disabled. In this configuration, all related Dock Services (and all the Dock's remaining circuitry?) are disabled. This was the basis for my plans to create the NuBusMiniDock™ hack for the DuoBoomBoxDock™ hack back in the day. There was a prize of a Mac offered by ________________ , one of the mods over on 'fritter, for the first one to hack some kind of Stereo System ten or so years ago.

The DuoDock Gemini Card uses the same _________________ Chipset as the IIsi's NuBus adapter, while the Dock+ uses the NuChip (____?____) Bridge Chipset, not that that matters. Considering that the Expanse Boxes made it possible to use extra PseudoSlot (my guess) NuBus IDs to be addressed by the Mac, I was hoping that a Six Slot ROM would Make the Extra IDs available in Firmware, while changing the NuBus Slot ID glue on the IIsi's NuBus adapter would give me a fourth PseudoSlot (NuBus) ID freeing up the IIsi's PseusoSlot ____ for Three PDS Card/NuBus slung underneath the MoBo, addressed in the IIsi's stock Memory Map. That, along with allowing a full compliment of RAM to be addressed in the unbuffered Bank A.
No Video Sense Lines/Cable polled on the Vampire Video Connector at startup = No buffer enable = freeing up the "normally" buffered Bank A for HACKING . . .
. . . maybe! It seems as though I've gotta study a bunch of Block Diagrams again . . .
. . . not to mention the NuBus Spec for identifying the digital glue that determines the Slot ID! FEH!

Let me know if you think I'm heading for a dead end or three!

Thanks to ojfd's help in my IIfx/CD 5 year tune-up . . . thread, I've now got a chance to determine new info on Slot Id's!



< . . . reminds self to change name of thread to ___ & NuBus Card Madness or some such . . . >

__________________________________________________

I'll edit in all the blanks left for Slot IDs, Chipset and ASIC info when I get a chance.

[dougg3]

One good idea is to separate ICs by using ferrite beads in their VCC lines and capacitors to GND after them. As bonus you gain space underneath those beads to route traces there, in cace where they will otherwise cross. I also don't see any larger (10uF +) filtering caps on your board. At least one would be good to buffer VCC that's coming from USB. SMD or 'thru hole', whatewer works.

Hey ojfd, thanks for all the detailed info! Wow! I really appreciate all of the info you dug up. I didn't even think about the benefits for routing by creating places that can be crossed. I saw that ferrite beads are a good idea (I think in the AVR documentation?) but I have been skipping them because I wasn't sure if they mattered for my application. If I added beads to my microcontroller would I need a different set of them for each VCC/GND pair on the microcontroller?

I actually do have a 10uF capacitor just to the right of the USB connector, on the bottom side of the board. I actually read that the value of that capacitor should not be any bigger than 10uF for USB (otherwise you get too much inrush current when it's first plugged in?).

Never use power planes. It is not needed, as dc-supply currents only need small traces. A power plane may resonate with the groundplane: your PCB will act as dipole antenna!

I knew, it was something like that, nice to have it confirmed by someone else. But as I said before, it might work OK on your board, so don't worry too much about it

OK, thanks! This is really good info to know

I am using smaller bypass capacitors (particularly on the microcontroller) because the bigger they are, the more routing space they take up, making it harder to route the traces for pins directly next to the VCC/GND pins.

I wish there were some kind of "PCB layout for Dummies" to learn from. I've been adding bypass capacitors, and trying to minimize trace lengths, but otherwise it's all just seat of the pants engineering. Fortunately everything I've designed so far has worked.

Haha, exactly! I'm in the exact same boat. I'd love to read a book like that too. The whole PCB layout is a lot tougher than it sounds. I have a lot of respect for people who design PCBs!

Edit: Here is what the PCB looks like with VCC as traces. All of the copper pours are GND. I have highlighted VCC so it's easy to see:

I had some isolated areas that didn't have a fill before, near the middle of the board. Some of them overlapped with areas on the other layer that did have ground available, so I used vias to connect them and fill in the empty spaces. In some cases, this new fill then overlapped with an empty area on the other layer, so I added yet another via, and now a bunch more of the board has a copper pour. Is this OK/advisable to do, or am I creating some kind of crazy antenna or RF emitting? Or does it not matter at all?

[ojfd]

Hey, guys, RELAX! As long as you don't have to deal with high clock rates, transmission lines, ground bounce, mixed signal, clock coupling into sensitive analog nodes, thermal drift and all that sort of stuff, I think your boards will work just fine the way you've designed them and it will not matter whether they have power plane or not.
And, btw, I have great respect for you, programmers. That stuff is mostly double dutch to me (although I've been successfull with ResEdit on a couple of occasions )

[ojfd]

Is this OK/advisable to do, or am I creating some kind of crazy antenna or RF emitting? Or does
it not matter at all?

It is OK - I would do it. I come from school where more ground is always good, but, as I've said before, it might not matter at all in your particular case
Board looks nice, btw.

[dougg3]

Thanks ojfd!

Now I'm busy separating the traces from each other a bit. I just realized my design rule check was set at 6 mils for clearance. Seeed Studio claims they can do it, but I don't trust that they will be able to do it reliably. I'm trying to get 8 mils of clearance at minimum...

[Trash80toHP_Mini]

I worked a few tricks I've learned over the years doing analog photography.
Basically, I hoodwinked my very first digital camera into taking some halfway decent pictures of the Jolly Roger in action!

IIsiHackKluge:


Close-up of the Rev.1 Jolly Roger in action:

p.s. Tweaking in GraphicConverter helps a lot, but there's NOTHING like dodgin'-n-burnin' print effects in an analog darkroom!

[dougg3]

Very nice jt! It's awesome to see the Jolly Roger happy inside a Mac! Cool pics!

Latest update: I have a headache now, but 400+ DRC errors fixed later, I now have a board that passes the test with 8 mil spacing...

I think I'm going to pass on the ferrite beads on this board, but I'll definitely keep them in mind in the future (or if I start having hardware problems...)

[Trash80toHP_Mini]

Thanks, doug! Although, at this point, I think I'd describe it more as the Jolly Roger resigned to being in the middle of a Mess!

BTW, I discovered something quite by accident while I was taking those shots. When there's no monitor attached to the MoBo and no Monitor attached to the Liberty Adapter for the Radius Color Pivot II/IIsi Card, I get a five note death chime a few seconds after the boot chime every time! Plug in the 16" Sony Trinitron to power up the sense lines on the IIsi's MoBo and it boots up like a champ.
The results are repeatable, therefore I just have to test it out with the MoBo ROM and both the IIsi ROM SIMM configurations to see what's up.
Simply maaaaaahvelous!
Though it normally ROCKS, serendipity apparently gots its compliment of Yang to balance the Yin . . . like everything else!

__________________________________________________________

I just had a thought about the placement of your header rows . . . for clearance . . .
. . . how complicated would it be to move them another 10-15 mills farther from the SIMM Socket?

[dougg3]

Interesting! I don't think that would matter as far as the SIMM is concerned, but I guess it depends on which ROM image you're booting from.

It is definitely possible to move the headers for some clearance. Good idea--there's plenty of board space available, even more if I move the RS-232 chip up. I'll do that! Shouldn't be a difficult change.

[dougg3]

Moved it back as far as I could with the size of the board being what it is...sorry if these PCB design updates are getting annoying, I'm almost done!

[ojfd]

I have a headache now, but 400+ DRC errors fixed later, I now have a board that passes the test with 8 mil spacing...

Congratulations, dougg3, very nice job! Fixing 400+ errors manually indeed takes some time.

I think I'm going to pass on the ferrite beads on this board, but I'll definitely keep them in mind in the future (or if I start having hardware problems...)

Yes, skip them. I am 100% sure that your board will work as it stands now.

...sorry if these PCB design updates are getting annoying, I'm almost done!

I thin'k it's very good and very educational to that you're publishing progress of your design, especially for casual reader. Next time people, who thought that it's all automatic, 5 minutes, auto-router does it all, will (hopefully) aprreciate the amount of work needed to create a finished product and (hopefully) will stop whining when asked to pay fair price for it.

Go dougg3!

[Trash80toHP_Mini]

I'll second that motion, I think it's great that you're documenting the process of creating, not just grounbreaking 68k firmware hacking, but the actual hardware to make it practical for others to try, and now, the tools required to get involved for others to play around with old Macs at an affordable price.

I'm continually amazed at your three level hacking thread, bbraun's incredible ROM Expansion achievements and all the technical expertise you've brought out of the woodwork from so many comrades. The tangential projects under development, like BMOWs unprecedented work and his involvement here in this thread, have been a boon to all.

Your introductory thread has become a treasure trove of content and demolished windmills, DQ!
Go, doug, go!**
jt

p.s. you moved the headers much farther back, great job. I've got pictures of triangular adapter cards dancing around in my head!

** silly literary reference for anybody who's read to kids

[dougg3]

Congratulations, dougg3, very nice job! Fixing 400+ errors manually indeed takes some time.

Thanks!!! It took a ridiculous amount of time Although sometimes fixing one error would make 10 of the DRC errors go away, but still...it was ridiculous. I think if I had set the auto router correctly for 8 mil spacing I would have been better off. I agree totally with what you said. The auto router is nice but you really have to do a lot of cleanup by hand, and it ain't easy!

I thin'k it's very good and very educational to that you're publishing progress of your design, especially for casual reader.
...
Go dougg3!

I'll second that motion

Thanks guys! I appreciate all the help and suggestions and everything. This has been a cool way for me to learn a lot about hardware Not to mention I'm incredibly impressed by all the other cool stuff that has happened in here, like olePigeon's graphics and bbraun's ROM disk!

I've got pictures of triangular adapter cards dancing around in my head!

** silly literary reference for anybody who's read to kids

LOL...or for anyone who has ever been a kid


OK, so here's the deal. I think I'm pretty much done with the board. I added a few labels for the pin numbers and stuff. I'll still go over the layout and schematic with a fine-toothed comb a few more times, but it's pretty much done. In celebration, I used a cool script for EAGLE to render an (almost assembled) PCB in 3D! (Sadly, I don't think it has a 3D model of the SIMM socket)

[Trash80toHP_Mini]

Go, Dawg, GO! VERY, VERY
That rendering is beeeaaaaauuuuutifuuuuul!
Now that I've seen the connectors and header pins, I've got a couple of packaging considerations for you.

If there's any way to move the USB connector over so that it hangs over the edge just about the thickness of a piece of thin plexi (actual thickness to be determined) and the DB-9 out to just about the same amount, I can see your board enclosed in a clear box with a clear plexi hinged lid or a rabbeted pop-top!

I know the crystal can positioning may be problematic, doug, but it might be worth a bit of rework. Some metal screening material inside the plexi for RFI reduction purposes and ESD event prevention might push the packaging parameters out a little farther than I first thought.
ojfd, what do you think?

edit: a bent or mitered aluminum angle exoskeleton with a rabbeted clear plexi pop top and clear fixed base plate/chassis might be better!

[ojfd]

If there's any way to move the USB connector over so that it hangs over the edge just about the thickness of a piece of thin plexi (actual thickness to be determined) and the DB-9 out to just about the same amount, I can see your board enclosed in a clear box with a clear plexi hinged lid or a rabbeted pop-top!
<snip>
ojfd, what do you think?

I can't speak for dougg3, but for me that would mean total re-design. Time=$, you know.
Besides, you can make your plexi box just a tad larger than PCB, so that USB and DB-9 will line up with the outer surface of the walls nicely and it will look just fine. Others might want to put the PCB in different box and they might have different opinions about preferred look of the final product.
Just let the dougg3 do his job and finish his PCB.
Otherwise, you know what they say about the camel? It is a horse designed by the committee.

[dougg3]

That's kind of what I'm worried about too. At this point moving the microcontroller out of the way to make more space for the USB connector would require many changes...

I agree with ojfd, I think it would be easier just to make an enclosure bigger on that side so the USB connector would be flush. I knew the USB connector would overhang over the side of the board--I did it on purpose to give myself more board space. I have a microcontroller dev board that does it, too.

The other option would be to make the board a bit bigger, but unfortunately the board height is already at the limit for EAGLE light edition

Anyway thanks for the idea jt, but I think I'm going to have to leave that alone at this point. I'd be happy to do it if it was as simple as moving the DB-9 would be, but it's not

[Trash80toHP_Mini]

I meant to the right and up, no moving of the microcontroller, just the crystal can a smidge.
I know I'm nit-picking after the fact, but it pays to ask, the only stupid question . . .

It's still as all getout, DQ!

Otherwise, you know what they say about the camel? It is a horse designed by the committee.

I've heard that one, but I prefer R.A.Heinleins quip: An elephant is just a mouse built to government spec!

[dougg3]

Thanks jt

It's hard to see in the 3D model, but I have the crystal positioned in such a way that both wires to it are the same length, and moving it around would be problematic. Also I have other wires routed where the USB connector would go. I think I'm going to have to leave it as is...like you said, definitely doesn't hurt to ask though

[techknight]

Whats the Microchip IC for? I understand what the AVR is for, but the microchip IC? just curious. it was probably explained several pages back, but going through a several page thread is not easy.

[dougg3]

It's a 16-bit SPI GPIO expander chip that I used because the AVR does not have enough IO pins to handle all the data and address lines on the SIMM. I wanted to make sure I used something that has both input and output capabilities as well as pull-ups so that I can also use the board to run some basic tests on newly-assembled SIMM PCBs to find any shorts

All the lines attached to the SIMM (on both the microcontroller and the GPIO expander) have optional pull-up resistors, so I should be able to check if they are shorted to ground by setting them all as inputs, turning on the pull-ups, and making sure none of them read a 0. I should also be able to test for any shorts between pins by setting one pin as an output outputting a 0, and everything else as an input with the pull-up activated. If anything reads back as a 0, it would mean it's shorted to the same line I set as an output. As far as I can tell the only short I won't be able to detect will be a short to VCC, since I don't have pull-down resistors available. Anyway, I wouldn't have thought to do this crazy stuff, but after assembling the SIMMs and fighting shorts, I'm thinking it will be a useful tool to have.

Status update: The OCD in me just finished fixing all traces that were not exactly a multiple of 45 degrees. I'm thinking I'm going to send the first batch off to Seeed Studio tonight...

Status update 2: The gerber files have been sent to Seeed Studio. I went for UPS shipping again Can't wait!