Phipli
Well-known member
You're changing the capacitance and inductance of the circuit.
-
Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.
Macintosh SE/30 logicboard recreation (thread revival)
- Thread starter Bolle
- Start date
croissantking
Well-known member
How could I replicate this electronically? I’m thinking to fit a bodge modification to get this working as a temporary fix.
I wonder if having a PDS card installed would change the symptoms as well, but I don’t have one to test.
Phipli
Well-known member
Try putting something like a 10uF capacitor between the pin and gnd.
That's the signal to indicated an ongoing access is in the 'nubus' area of the memory map ($6000_0000 and up). It should be an output from GLUE to the PDS. It should be a combinatorial function of the four upper address lines, nothing super sophisticated.
It is explicitly marked as "this signal is active when the CPU accesses the on-board video display", but it really shouldn't mess with the data part, or any part really.
As mentioned by @Phipli, adding a capacitor could help. If on the other hand there's already too much capacitance and it causes voltage drop during transition, try with some weak pull-up (and/or pull-down) resistor as an alternative (1 to 10k perhaps ?). If you don't intend to use the PDS slot for now, another option would be to unsolder this specific GLUE pin and isolate it from the PCB to see if without any load it makes things better or worse.
Phipli
Well-known member
@croissantking : What @Melkhior said, I wasn't really paying attention and very literally answered "how do I replicate this" rather than, how should I bodge this.
In all truth though, you should use this as a clue to work out what's wrong. It suggests some signal in that area is floating that shouldn't be. Or something doesn't have a good connection to ground.
In all truth though, you should use this as a clue to work out what's wrong. It suggests some signal in that area is floating that shouldn't be. Or something doesn't have a good connection to ground.
croissantking
Well-known member
This prevented the onboard video from functioning, I got a simasimac screen (but a normal chime/boot).
Phipli
Well-known member
Yes, did you see Melkhior's comment?
croissantking
Well-known member
I sure did, I began testing with a pull up/pull down resistor but then started to have issues with my ATX extension cable not making a good connection to the boards anymore. I’m waiting on a new one to arrive before I resume.
croissantking
Well-known member
I’ve been doing more poking and prodding on my graphical issues.
What I have done is to isolate the ‘Nubus’ pins on the GLUE and PAL chips from the logic board, and run a bodge wire directly between the two. Like this, there are no graphical problems.
What this tells me is that there is something interfering with the Nubus circuit on the logic board that shouldn’t be. Is it possible that one of the adjacent tracks is introducing interference? I am confident nothing is bridging this circuit.
With this pin disconnected the onboard video does not function.
What I have done is to isolate the ‘Nubus’ pins on the GLUE and PAL chips from the logic board, and run a bodge wire directly between the two. Like this, there are no graphical problems.
What this tells me is that there is something interfering with the Nubus circuit on the logic board that shouldn’t be. Is it possible that one of the adjacent tracks is introducing interference? I am confident nothing is bridging this circuit.
croissantking
Well-known member
This is what I find interesting - it seems to be.
hi all, just wanted to share my (by all appearances) successful build. I got my first successful boot (off of a real floppy, no less) last week after spending a couple of frustrating weeks tracing out address and data lines trying to figure out why i was still getting a Simasimac pattern on startup.
I still had some hardware issues at this point (FPU connectivity, mostly) but the feeling i got the first time i flipped the switch and got a proper gray background and cursor made the previous frustrations totally worth it
at some point i will probably write up some of the debugging steps i went through to get here in more detail in case it would be helpful to others.
Also wanted to say thank to everyone who has posted in this thread previously -- while i didn't post here looking for help myself, i don't think i would have been able to figure this out without the previous discussion here.
I still had some hardware issues at this point (FPU connectivity, mostly) but the feeling i got the first time i flipped the switch and got a proper gray background and cursor made the previous frustrations totally worth it
at some point i will probably write up some of the debugging steps i went through to get here in more detail in case it would be helpful to others.Also wanted to say thank to everyone who has posted in this thread previously -- while i didn't post here looking for help myself, i don't think i would have been able to figure this out without the previous discussion here.
croissantking
Well-known member
An update with my graphical issue saga.
I cut the Nubus line and ran a bodge wire in its place. This completely cleans up and alleviates the issues, until a PDS card is installed.
An interesting experiment, but not a solution. I'll need to get my hands on an oscilloscope to get further into this.
I cut the Nubus line and ran a bodge wire in its place. This completely cleans up and alleviates the issues, until a PDS card is installed.
An interesting experiment, but not a solution. I'll need to get my hands on an oscilloscope to get further into this.
croissantking
Well-known member
Hello CircuitBored! Is that an oscilloscope you're using there on your iBook? I see you also have a real, old-school oscilloscope in the background.
CircuitBored
Well-known member
Hi there!
That's a cheapo logic analyser plugged into my trusty unibody MacBook. The analyser uses the Salae Logic program.
The oscilloscope is a mighty old HP system. It's actually a very early sampling scope - with a 68000 inside, no less! With a 200MHz range it is still more than capable of doing what I need. The traces and pads on the logic board are all gold plated so it looks fantastic inside. It has an HPIB (GPIB) port that I will one day get around to hooking up to the GPIB card in my Quadra. I just love that green phosphor...
croissantking
Well-known member
Would the cheapo logic analyser be a good starting point for me, in learning to use an oscilloscope, and for working on my Reloaded board interference issues?
CircuitBored
Well-known member
Logic analysers are different to oscilloscopes. They're sort of two sides of the same coin. Logic analysers decode digital signals (ones and zeroes displayed as "high" and "low" on a graph), whereas oscilloscopes capture an analogue signal and display a waveform that represents its amplitude over a specified time. 'Scopes don't store anything, they just display the instantaneous signal that they are fed. Only "sampling oscilloscopes" like my HP can capture and store waveform data. Bear in mind that digital signals are still transmitted by an electrical signal that can be displayed on an oscilloscope. The same cannot be said for analogue signals on a logic analyser. I believe you can now also get hybrid scope/analysers.
There are both analogue and digital signals in most old computers so having a 'scope and an LA comes in very handy when troubleshooting hardware. I'd recommend getting both if you can. If you're not pressed for space, the older and larger scopes are often better value when it comes to specifications but you obviously sacrifice some modern perks and compactness. A lot of the time you don't need perfect precision, just a waveform, so an old scope that's slightly out of calibration will often serve you perfectly well.
A good way to learn how to use an oscilloscope is to buy one of those DIY FM radio kits and use your scope to follow the radio signal through the circuit, observing how it is manipulated to become an audio signal. You need to practice holding the probe steady if you want to be probing big QFP and PLCC packages.
Scopes (old and new) tend to be a lot more expensive than LAs. For reference: my HP scope was £100 and the logic analyser I use was £8, both acquired from eBay.
Last edited:
Seems I looked at the schematics too quickly. As you mentioned yourself, the /NUBUS signal is also connected to a PAL according to the schematics I have... The PAL includes connection to various video-related signals, so yes it could have an impact on the video. And in fact, it's right there in DCDMF3:
EDIT: actually I quoted that once already, old brain forgets the specifics of the issue
It should stops things working, but not some part of the data and not others - mmm, unless it's a timing issue and the signal rise too slow so some stuff doesn't register yet some other does...so...
(a) the connection between the GLUE and the PAL must exist for the video to function, as you noticed (sorry!)
(b) any unexpected behavior of that signal will cause issue with the video and/or machine.
Specifically, if the signal is not active when it should be, the video won't work (including the video ROM); if the signal is active when it shouldn't be, the video might interfere with other parts of the system. The signal is active low.
You said "I can make contact at any of these locations and it fixes the problem for as long as I hold down on it." The analog part I don't know much about, but it sounds to me like for some reason, the signal doesn't go low enough for the PAL to register it as 'active' and so the video doesn't work unless something helps with the grounding / lowering the voltage somehow when the signal is meant to be low...
You also said "I cut the Nubus line and ran a bodge wire in its place". Did you cut it at one end, the other, in the middle, or everywhere ? Phrased differently, do you still have 'stray' wire(s) connected to any of the 3 connections point? (GLUE, PAL, PDS - assuming I didn't miss another one in the schematics...). Those could still pick up some stray voltage from somewhere in the system. Adding a PDS card would add capacitance to the line, and could cause the issue to worsen.
EDIT2: if it works with the bodge wire and no PDS, but fails when you add a PDS card, you could try with a weak pull-down on the line (a resistor between ground and the /NUBUS). Something like 10kohm at first, and if it doesn't help be bold and go down to 4.8kohm. Going lower is also possible, but might entually overwhelm the drivers and force the signal low (active) all the time which doesn't help you. If the pull-down helps, then you know that there's some form of 'leak' between +5V (or a often-high signal) and the /NUBUS line.
Last edited:
croissantking
Well-known member
Cool, I will try to get both then. Might start with the LA followed by an oscilloscope. Which LA did you purchase - do you happen to still have a link to it?
croissantking
Well-known member
Hi @Melkhior ! Always a pleasure to hear your wisdom.
I only have one PDS card I can test with at the moment, Bolle's combo ethernet card. And it actually works without the Nubus line connected at all. About other cards, I am not sure. I'm not really sure of the function of the Nubus line (it tells the machine which device is active on the Nubus bus? - but the SE/30 doesn't have Nubus? Confusing).
That makes sense.
Everywhere; nothing stray is connected. So the three pads/vias you listed are severed from their traces directly next to them and then linked up with my bodge wire.
Makes sense.
I only have one PDS card I can test with at the moment, Bolle's combo ethernet card. And it actually works without the Nubus line connected at all. About other cards, I am not sure. I'm not really sure of the function of the Nubus line (it tells the machine which device is active on the Nubus bus? - but the SE/30 doesn't have Nubus? Confusing).
I'll give this a go and report back with my findings.
Last edited:
AFAIK, it only combines the address signals (upper 8 bits, A[31:24]) into a single signal telling you "the current access is in the NuBus range". All NuBus and pseudo-slot devices (on PDS) needs to know if an access targets them, and it helps reducing the amount of decoding on the boards. It's just a 'shortcut'. Apple calls it 'NuBus', but that area of memory covers all non-onboard I/O in '020/'030/'040 Macintoshes - NuBus & PDS.
If a device does all the decoding by itself from A[31:24], it doesn't need the /NUBUS signal. My own IIsiFPGA doesn't use it, because it"s not an issue doing the decoding completely in the FPGA (vs. using discrete logic and/or PALs in the mid 80s!).
And yet you still have issue if you use a PDS card? Strange. IIRC, that recreated board is two layers only so it's not an issue with an internal layer... I concur with other than if you want to know exactly what's going on, you'll likely need an oscilloscope telling you the shape and voltage on that signal (and its relationship to at least the clock signal).
I presume the card does all the decoding onboard, so doesn't need that signal at all. In fact, I would expect most device to ignore it, it might be easier to just decode everything by yourself anyway as you still need to figure out if it's your slot. The Asante MacCon for instance doesn't use that signal either.
And just to be clear - if you don't connect the NuBus line to the PDS, everything works fine, both the PDS device and the internal video? If so, then while it's interesting to know why the signal has issue, the "don't connect it" solution might be a viable workaround, at least until you find a PDS device that relies on /NUBUS...
