• 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 Portable finicky after capacitor replacement

androda

Well-known member
My friend asked me to fix up his old Portable and I'm giving it a shot.  Done capacitor replacement jobs before (several), and replaced the capacitors with very few issues.  There were a few traces where the cap goo was clearly under the soldermask, so I scraped off the mask, cleaned up the goo, plated with solder, and then covered with new mask.  Things generally look good, though I haven't done an ultra-close inspection on the entire board.  Some pins on the video chip (top center pins) needed rework because of corrosion, and a few other things here and there like those tiny transistors.

Now, on to the behavior.  Often, the Portable starts without a problem.  Happy mac, successful boot.  This is both on battery and on battery + power adapter.  I am using the correct Portable power adapter, the battery is fresh, and the switch by the battery is pressed down.  Oh, and it has a 9V battery too.

5v "always on" rail is solid at 5.2 volts.  12v line is also solid, at 11.85 volts.  Power manager chip is getting its 3.9 MHz clock signal.  Hybrid ADC line varies between 1.76 volts and 4 volts, with battery voltage being 6 volts and 7.1 volts respectively.

Sometimes, the Portable does not boot up, and just freezes at a screen of nothingness or garbage lines near the top of the screen.  Other times it has looked like the screen image here: 





There have been some sad mac codes here and there.  A few examples (first line / second line):

05000706/00001FA7

01000700/00001FFA

01000314/00001FF4

A few times on startup the floppy drive has made the classic 'disk read' sound with nothing in the drive, and a failed boot.  Hard drive never starts up, either nothing or random garbage on the display.

Based on the thread I linked above, it could be the SWIM chip that's going out.  Are there any good methods to check this, other than pulling it off the board?

What other power manager signals or voltage rails should I watch?  I was planning to monitor the 68000's clock signal, halt and reset lines, IWM_CNTRL, SCC_CNTRL, STOP_CLOCK, etc.  Speaking of these signals I've noticed some have a dot next to them (on the schematic).  Does this mean the signal is 'on' or 'active' when the voltage is high or low?

Often, when the system fails to boot, pressing the reset button a few times (press, wait a few seconds, press again...) will end up clearing the problem and I get happy mac successful boot.

Other than that, it seems the system struggles to start after it's been sitting for awhile.  Going from no batteries at all, put in the batteries and cover, it starts right up.  Wait 6 hours, crash.

Thanks for all your help, this seems to be "the place" to come for the Macintosh Portable.  I'm hoping to get this old mac up and happy, it was apparently the first computer my friend ever purchased with his own money.

 
Last edited by a moderator:

LaPorta

Well-known member
A few questions:

You say the HD "never" starts up...then how is it you get the machine to start successfully sometimes then? I would disconnect the HD from the logic board. My failing HD in the Portable threw all kinds of error codes until I replaced it with a SCSI2SD.

 

androda

Well-known member
The hard drive works and the Portable will boot from it.  I also thought the hard drive was the primary failure source until I noticed that the vast majority of the time right after power is applied, it boots up fine.  So generally it's happy and fine when you go from no power sources to batteries, power adapter, and startup.  And something is unhappy after the system has been sitting for awhile.  So, start it up and it's fine and working.  Shut it down and sit for 6 hours, often refuses to boot without the reset button trick.  And sometimes that reset button trick doesn't work.

In some cases where the system fails to boot, the hard drive does not start up.  And it stays not spinning indefinitely ("never spins").

In other cases where the system fails to boot, the hard drive does spin up.  But nothing happens, like those funky lines stay on the screen and the system never boots up.  Eventually the hard drive does spin *down* in this case, probably some sort of hardware no-access timer.

 

androda

Well-known member
My next round of checking resulted in an interesting power on failure.  Start the system, it displays that checkerboard background for a moment, and then usually goes on to mouse cursor and happy mac.  This time it did pattern, followed by the screen sort of 'bleeding away' or 'fading away' slowly.  No sounds of hard disk activity, and the screen just sort of 'faded out' over time.  Using the reset button a few times will get it booting normally again.

Now for my signal readings.

When the system is off:

CPU 16MHz not present, sys_rst is 0, stop_clk is 5.17, iwm_ctrl is 5.16, scc_ctrl is 5.16, floppy and 12v rails 0.

When the system is on and happy (booting, booted):

CPU 16MHz present, sys_rst is 5.14, stop_clk is 0, iwm_ctrl is 5.14, scc_ctrl is 0, floppy rail 5.2, 12v rail 11.86.

When the system is on but doing that strange screen bleed behavior:

CPU 16MHz not present, sys_rst is 5.18, stop_clk is 0, iwm_ctrl is 0, scc_ctrl is 0, floppy and 12v rails are fine.

So this is pretty interesting.  Clearly the CPU clock is stopping when the display starts to do that fade-out thing.  iwm_ctrl drops to 0 from the usual 5.  And sys_rst is high meaning the system is on.  It's like the system passes the power on self test, goes on to try and boot, and then just turns the CPU clock off.

Is it the power manager that tells the CPU glue chip to send on the 16MHz clock signal?  The CPU glue is where the CPU clock comes from.  None of the marked pins on the schematic really jump out at me as signals that are good to watch for this sort of thing.  Something's obviously getting into a bad state, I just don't know what yet.

Also, there's an electronic oscillator sound coming from the hybrid area when the power adapter is plugged in.  Sometimes higher pitch, sometimes lower.  Unplugging the power adapter made the sound go away, and it doesn't seem to happen with the adapter unplugged.  Some buck regulator is getting noisy I guess.

 

techknight

Well-known member
ahh yea this is common. 

the Hybrid or the PMGR IC is giving your trouble. common problem and has a low success rate of repair. 

one of these days, the hybrid needs cloned and new PCBs made up as replacements. 

 

techknight

Well-known member
Also, there's an electronic oscillator sound coming from the hybrid area when the power adapter is plugged in.  Sometimes higher pitch, sometimes lower.  Unplugging the power adapter made the sound go away, and it doesn't seem to happen with the adapter unplugged.  Some buck regulator is getting noisy I guess.


You try another adapter? the original adapters need recapped by this point, and there is no buck regulator on a Portable. Just a 12V Boost on the other end of the board, and a negative voltage inverter on the Hybrid. 

 

androda

Well-known member
You try another adapter? the original adapters need recapped by this point, and there is no buck regulator on a Portable. Just a 12V Boost on the other end of the board, and a negative voltage inverter on the Hybrid. 
I replaced capacitors in the portable AC adapter too.  It was awful inside, totally coated with cap electrolyte and a bunch of soldermask eaten away.  But has a nice steady 7.5 volts after cleaning and replacing the capacitors.

No buck regulators?  I was wondering about that, because of battery voltage behavior.  With the unit plugged in, voltage across the battery (measured at the brass-looking battery terminal points) doesn't really go above 7.1 to 7.2 volts.  Makes sense, that's a good charge voltage.  But if you press and hold down reset/nmi buttons the voltage across the battery will spike up to full adapter voltage (7.5).  I was assuming there was a buck regulator somewhere being used to provide the 7.1-ish volts for battery charging, and didn't specifically look on the schematic for something like that.

Looking at the schematic now, is there really just a 1 watt rated 100 ohm resistor between charger positive and battery positive?  About as basic a charge circuit as there can be.  Does this mean that holding reset/nmi buttons causes Q1 to get turned on?  That would bridge out the 100 ohm resistor and apply full 7.5 from charger across the battery as my meter was telling me.

And I totally agree, we need a solution for the failing power manager chip / hybrid combo.  If this was my personal Portable, I'd consider taking the hybrid apart and measuring the values on those funny laser cut resistors.  This hybrid chip seems to be sending a good AD signal to the power manager, so all these symptoms together probably point to a bad power manager IC.

 
Last edited by a moderator:

androda

Well-known member
Thinking on the subject of buck regulators some more, there has to be at least one, and likely more.  How else do you take 6 to 7-ish volts from the battery and turn it into the 5 volt power rail?

Looks like  Q16 is the switching mosfet for the 'always on' 5 volt power rail buck regulator.   You have to turn that mosfet off and on to keep a steady 5 volts, and the little ferrite and 470uF capacitor help smooth things out.  Switched by pin 8 of the hybrid with voltage feedback to pins 11 and 18 of the hybrid.

Disk +5 is another buck regulator, switched by Q20.  This one is a really funky op-amp circuit that has a reference to pin 51 of the hybrid and the 5v rail voltage.  This implies that some hybrid voltage rail is being used as a reference to this external op amp to switch Q20?  This one looks like it has to care about more noise, with the bulk 470uF and 0.1uF filtering capacitor.  So HD_PWR* turns on Q25 which enables Q24 which lets U7M switch Q20.

Maybe I'm totally wrong (or partially wrong), but it sure seems like there have to be buck regulators on the separate 5v rails.

 
Last edited by a moderator:

androda

Well-known member
Ah, so those mosfets are used as linear regulators thanks to their variable resistance.  Being used to more modern electronics, I just expected these to be buck regulators.  Makes me wonder whether an LDO buck would work here.  So maybe instead of reverse engineering the hybrid's design we replace it with a newer equivalent.

 
Last edited by a moderator:

techknight

Well-known member
Ah, so those mosfets are used as linear regulators thanks to their variable resistance.  Being used to more modern electronics, I just expected these to be buck regulators.  Makes me wonder whether an LDO buck would work here.  So maybe instead of reverse engineering the hybrid's design we replace it with a newer equivalent.


This was too far into the future from which point this machine was engineered, or the experience of the particular engineer. 

This is why Apple went back to the drawing board, and Sony designed the PB100 based on the portable. 

 

androda

Well-known member
Seems like the consensus is basically that the power manager chip is having issues then.  I had the thought "too bad I don't have a logic analyzer, that would help me characterize exactly the difference in behavior of the chip between a good boot and a failed one".

So I hopped on my local classified ads site, and luck was on my side.  I just picked up an HP 1660C 136 channel logic analyzer *for free*.  It came with all the cables populated and a few accessories, but is missing most of the signal acquisition ends.  I'm in process of finding decent prices on the acquisition ends and buying them.  Soon (probably January after the Christmas shipping backup) I'll be able to record the behavior of every digital pin on the Power Manager chip (and more).  Then the tedious process begins of figuring out what differences actually matter, and how to interpret the data which is being sent on those data lines from the power manager to U14H (VIA?).

Would be really nice to replace the power manager chip with an FPGA.  Seems like its functions are very basic, although it does do a bunch of things.  Enable and disable chips based on certain states, signal the system if battery voltage is too low, report on battery voltage level in general, stop the system if things look bad, probably more.

 

techknight

Well-known member
Would be nice to get down to the bottom of actually what goes on with these chips. 

My though its bad logic drivers on the MCU's GPIO, but thats just a wild hairbrained guess. 

 

androda

Well-known member
Today I spun a little adapter PCB that converts the 2x25 1.27mm pin headers for the EPWRMGR test connector out into two 1x25 2.54mm pin headers.  Also ordered some 1.27mm male and female connectors for the PCB and board.  Figure I'll solder a female connector on the board (gotta clean the solder out of the vias first) and plug the adapter into that with some header rows.

Theoretically this will make it easier to debug with the power manager test port, since my logic analyzer only supports 2.54mm headers with its acquisition connectors.  Still waiting on the Christmas and New Years shipping backup to clear, so who knows when my adapter board and other stuff will show up.

 

androda

Well-known member
Another issue found.  C116 (by the power manager side keyboard/trackball connector) was very nearly electrically isolated from the traces it is sitting on.  Visually the traces look fine.  Green, not dark or discolored or anything.  Pads look fine, nice and shiny tin.  But when I checked with my multimeter, it read 14 MegaOhms resistance between the capacitor and the physically connected traces.  Crazy what capacitor fluid can do.

So I'll be double checking all the other caps I replaced to see if they have similar problems.  I did verify several right after the replacement, might as well check again.

 

androda

Well-known member
No other capacitors found with this issue.  Guess I just lucked into it while poking around at the power manager test port.  After patching up the connections to C116, no difference in behavior.  Starts right up after power is connected, but fails 3/4 times after sitting for awhile with batteries, battery cover, and power adapter connected.

I read in another thread that some logic chips can get warm / hot and that would mean they're bad.  Next up while waiting for my adapter PCB to arrive is letting the board sit with battery voltage and checking for warmness of the various chips.  Here's where I wish for a thermal camera, because the desire for more test equipment never ends...

 
Last edited by a moderator:

androda

Well-known member
Tested the theory of chips getting warm/hot today, with nothing getting warm on the motherboard after three hours.  Touched every chip, every mosfet, just plain nothing.

And my pin headers finally arrived in the mail today, so I finished soldering my portable power manager connector stuff together.  Now I get to sit down and learn how to use the logic analyzer.  Once I have that all set up and get a trace I'll post it here.  Hopefully there will be obvious differences in power manager behavior between the successful and failed poweron attempts, so I can keep debugging.

 

androda

Well-known member
This morning I got a trace from successful powerup, then let the system sit for awhile and got one from a failed powerup.  The output files are larger than I expected, need to figure out how to network my logic analyzer to get them onto my laptop (won't fit onto a floppy).  Then I'll make a listing of which pin on the LA maps to which pin on the power manager connector and post it all up here.  'transitional timing mode' is nice for this, only records data when there's new data to record.

 
Top