Yellowstone Glitch, Part 8: Down the Rabbit Hole
This glitching investigation has taken a turn down a strange path. Although the Yellowstone board seems to be working, I wanted to further reduce the nasty-looking signal spikes seen on the oscilloscope. Series resistors on the data bus seem like a good way to do that. So I performed some more board surgery, and separated the output pins of the 74LVC245 from the pads on the PCB. Then I used 16 fine-gauge wires to insert a socket between the ‘245 and the PCB pads, so I could experiment with resistors of different values in the socket.
I tested it with 100 ohm series resistors, and the steepness of the edges on data bus transitions were noticeably more gentle. I measured about 0.15V per ns, instead of 0.4V per ns before. But I was very surprised that the weird spikes I’ve been chasing on D0, A1, and the 3.3V power supply didn’t improve at all. As far as I could tell, series resistors didn’t help even a little bit. They were a total dud, at least for 100 ohms.
My attention then returned to something I’d noticed before. There were also spikes on the signals at other times, not during the Yellowstone bus cycle. What was that about? Is Yellowstone doing something odd from an electrical standpoint, even when it’s not driving the bus or receiving a value from the bus?
Then I got to thinking: three of the four signals that I’ve been measuring in each scope experiment are just Apple II bus signals, not Yellowstone signals. So what would happen if I completely removed the Yellowstone card from the computer, and just measured those signals? I took a blank unpopulated PCB, and soldered a few wires to points on the board so I could measure some voltages. Then I ran a test program on the Apple IIgs that attempted to read values from the non-existent peripheral card.
- Blue (Ch4) is address line A1. It’s a 5V input from the Apple II
- Yellow (Ch1) is the 5V supply voltage, measured at the VCC pin of the PCB.
- Cyan (Ch2) is IOSTROBE, and marks the boundaries of the bus cycle.
All those ugly signal spikes are still as bad as ever when the Yellowstone card isn’t even present. Cue disorienting sound effects, flashing lights, evil laughter, fade to black…
What have I even been looking at for these past few days? Either it’s some noise inherent in the Apple IIgs itself, or more likely it’s just operator error and an artifact of my crappy probe setup. But it’s not a problem with Yellowstone.
I’m still confident there was initially a signal integrity problem with Yellowstone, probably multiple problems, caused by high current flows through the 74LVC245. But my earlier attempts to address the issue may have been more successful than I realized, and now I’ve been chasing phantoms for several days.
Read 4 comments and join the conversation4 Comments so far
Leave a reply. For customer support issues, please use the Customer Support link instead of writing comments.
Most of the spikes seem to be moving in unison. How is your ground lead connected? Sometimes it can be useful to put a scope probe on ground and see whether you get noise on that which matches noise elsewhere.
If you were having behavioral problems before, and aren’t now, that’s a pretty good sign that there was a real problem which has been to a considerable degree fixed. It would be nice to be able to get visual confirmation on the scope, but that may not be practical. On the other hand, if adding bus capacitance to a card other than Yellowstone card under test doesn’t adversely affect operation when precharging four pins one way and four the other, that would be a good indication that the design is robust.
I don’t know if my experience is even related, but the IIgs bus seems a bit noisy in my limited experiences with it. I have a ReactiveMicro Phasor clone in my IIgs. And it picks up bus noise and turns it into sound out the speaker connected to the Phasor. I think it happens mostly during drive access. But this was a while back I was messing with it and don’t recall exactly the trigger for the bus noise. I found a video on YT of someone else working with a IIgs and the Phasor clone, and he also commented on the bus noise issue. I’ve recapped both the PSU and main board, which should eliminate bad/old caps as the cause of the noise. So it may just the nature of the IIgs bus.
There may be a little IIgs bus noise, but I’m pretty sure it’s 99 percent my fault due to flawed measurements. I didn’t appreciate how important it is to have a good, short ground lead connection for the scope probe when measuring signals with fast edges, even though nice people keep trying to explain it to me.
My ground connections are *ahem* poor. I’m using the ground leads that are included with the scope probes, which are six inch wires with an alligator clip on the end. The alligator clips for all four probes are clipped together. One of them is also clipped to another wire, about 12 inches, that is connected to a ground point on the IIgs motherboard. That point is about 8 inches from the Yellowstone card’s peripheral slot.
I’ll try getting some new scope traces using a short ground spring to a PCB point very close to the signal I’m measuring. That will be trickier than before, and I’ll probably need to hold the probe in place by hand while taking measurements, being careful to avoid accidentally shorting something. I’ll also try adding more bus capacitance, by loading up all the other peripheral slots with 1980s vintage cards. I agree that’s a good stress test.
The Apple IIgs is notorious for having noisy power. If you want to see some fun, try running a Transwarp GS at 12+Mhz on a ROM 01 machine (ROM 3s are less noisy by design) with the stock power supply.
Regarding your testing. Did the Yellowstone fail at all installed in an Apple IIe? Also, what happens in the IIgs when the system speed is set to “Normal” 1Mhz? The slowdown logic on the board is supposed to slow the entire machine down to 1Mhz whenever their is expansion bus activity, but that could expose flaws in other areas.