pentiumspeed wrote on 2023-02-13, 01:05:
Your scope is set up wrong. It is displaying two waveforms from two channels this is confusing. You need to separate them o […]
Show full quote
Your scope is set up wrong. It is displaying two waveforms from two channels this is confusing. You need to separate them or probe with one waveform on and turn second channel off by turning off the channel 2 switch. Or rolling because trigger is set up wrong. Turn the trigger frequency (1/t) or change trigger on or off.
Where is your grounding clip is located when probing this? Be very cautious with this. High frequency measurement requires you to have ground clip about a inch away or so and is short wire from the probe's body.
What does multimeter say to you on voltages everywhere on the motherboard? The jumpy frequencies is indication of unstable board, the voltages could be jumping too or board is in boot looping.
Used known good memory module like Micron branded, Samsung or Hynix?
Known good CPU?
Good power supply (did you check it with scope and multimeter?
1; nope. single channel.
2; the screw hole via/ground adjacent to the RTC and cache sockets. bit over an inch from the nearest cache socket.
the rest of my probing shows no overlapping signal issues, now that i've managed to get my pll freqencies (aside from 2xcpu, but i don't know if thats supposed to bounce from 66 up to sometimes 100) stable. voltages are all within tolerances, give or take a few .01's. all of those signals are clean.
the _only_ place i'm seeing this issue atm the moment is the cache data lines. and its 3 full waves overlaid on each other, i've counted.
all 3 cpus i have handy here for this work in other boards, no issues. i've tried every stick of ram i have, and tested them all in two other boards with memtest.
i've tested the psu with other boards as well, and probed the voltages with a dmm under load. zero real issues that i could find. a bit of drop, but thats to be expected (couple mv).
aside from the visual setting (the boxes below the x/y grid, and getting rid of a couple i didn't want), its the same general settings aside from the horiz div to show the actual waveform. still rolling, still going. channels 2, 3, and 4 are _off_. it doesn't show here, but they're dimmer than 1, the lights are 'off', the traces aren't there. they're also different colored traces.
the problem is with the board, i just need to either pick the brain of someone who knows where to poke to find out, or stumble across wherever that point may be.
if the 2xcpu is supposed to be stable, that may well be whats causing it, but i also go hrmmm, because i have no idea what would be using that. it has a dedicated isa speed output, and the pci clk output should run off the same clock signal as the 1xcpu line. and a quick confirmation shows that yes, the pci clk signal is a flat stable 33mhz, and traces back to the 1xcpu pin. no 'ghosting'.
someplace between the cpu and the cache or ram, there appears to be something dead/malfunctioning? except for most of it, its just straight traces with no passives, and no visible damage to said traces at all. they also test out for continuity.
and hitting single for giggles did nothing but give me a slice of the same nonsense, vs the rolling constant read.
so again, not sure what the issue is, but its not the test equip from what i can tell.