IIsi Repaired By Reheating Solder Joints and Replacing Crystal

David Cook

Well-known member
I recently bought two Macintosh IIsi computers. Neither worked upon arrival. The initial restoration steps are common:
1. Remove old battery
2. Blow out dust
3. Replace the capacitors on the motherboard and power supply. I decided to replace the big capacitors on the power supply as well. They cost $3 each, but my labor is worth more. So, might as well complete the job in a single pass.

Next, I tested the power supply before reinstalling it in the computer.
4. Check for 5.1V (approximately) on pin 10. Hook meter ground to either pin 5, 6, or 7.
IIsi-Power-Supply-Pinout.jpg
5. Using a low value resistor or a piece of wire, briefly connect pin 10 to pin 9. You should hear the relay click.

My relay did not click. There was no reason to try the power supply in the computer until this step works. I checked for blown fuses or other bad parts. I replaced some diodes and two bridge rectifiers, per other threads on this forum. All of the removed parts tested as working -- so they were not the problem.

Let's take a look at how the vertical PWM switching boards looked in each of my power supplies before replacing the capacitors. Gross!
IIsi-Power-Supply-PWM-Switching-Board.jpg
Replacing the capacitors and ultrasonic cleaning the boards removed the cause of the destruction, but not all the damage. Inspecting the chip pins with magnification revealed why the power supply was not turning on.
Tiny-chip-pins-may-look-connected-when-viewed-from-above.jpg
From above, without magnification, I could not tell that the solder had been eaten away by capacitor leakage and that many of the pins (and leads on other components) were either disconnected or intermittently connected. I carefully picked away, cleaned, and resoldered every dull joint on the vertical board. Afterwards, the power supply clicked its relay when pins 9 and 10 were contacted!

You won't be able to test the other power supply voltages until it is attached to the motherboard or other load, because the power supply turns off instantly.

Upon connecting each power supply to a motherboard, I was greeted with one Mac with sad chimes and one Mac that wouldn't power on at all.

My mistake was not inspecting the motherboard for dull joints, flaky solder, or floating pins, as I had found on the power supply PWM board. Pay particular attention to all parts near the old capacitors.

The source of the Sad Mac chimes were floating pins on UE5. This prevented proper access to onboard memory, causing the ROM startup test to fail. Honestly, these pins didn't look bad to naked eyes after the board was ultrasonically cleaned. But, tapping them with a dental pick showed they were completely loose. In other people's photographs I see leakage on UE5. So, you should always check here.
Sad-Mac-Chime-Caused-By-Dry-Solder-Joints-On-Memory-Interface-Chip.jpg

Here's another dull solder joint. Is that leg of Q5 actually electrically connected? Or has the chemistry/bond of the damaged solder been altered? I don't know, but it needed to be addressed by picking, cleaning, and resoldering. By the way, notice this part is near a replaced capacitor.
Dull-corroded-solder-joint.jpg

This fixed the Sad Mac IIsi!

The other Mac IIsi doesn't respond to the rear power button or the keyboard power button. Here is the chip that monitors the power buttons. You can test it without the power supply by simply making sure you have a fresh PRAM battery installed.
UB1-PRAM-Power-ADB.jpg
Pin 14: GND
Pin 13 (V): +V around 3 volts supplied by the battery.
Pin 25 (B): Should be around 3 volts, but drops to 0 when the rear button is pressed.
Pin 24 (K): Should be around 3 volts, but drops to 0 when the keyboard power button is pressed.

You can also check pin 2 (reset) for 3 volts to make sure the 68HC05 isn't being held in reset.

Assuming all of this checks out, test the pad labeled Hz in the above image. See below where an oscilloscope is connected to battery ground and the lower pad for the crystal.
Testing-oscillator.jpg

Even with only a battery installed, you should see a nice waveform on this pad. If you don't have an oscilloscope, a good multimeter should be able to display a frequency of 32768 Hz (32.768 kHz). This is the 68HC05 clock. If the waveform isn't present, then the chip can't run, and thus can't turn on the power.

32768-Oscillator-Waveform-on-Oscilloscope.jpg

On the dead IIsi, there was no waveform! Replacing the crystal fixed the problem and the board powered up and ran nicely.

Hopefully my experiences will help you next time you're restoring a IIsi or other classic Mac. Basically, assume every dull joint is broken.

- David
 

joshc

Well-known member
Nice. When I repaired a IIsi, I removed most of the components from the soft power board in the PSU with hot air and cleaned that area manually with flux/wick and then soldered everything back with fresh solder. I did the same with the entire sound circuit as all of it was flooded with cap goo and was very dirty.
 

David Cook

Well-known member
I repaired another IIsi today. The power supply vertical board was nasty. Several components fell off during cleaning. I ended up removing the main chip in order to clean the pads. Fortunately, after adding missing components, replacing the large capacitors, and resoldering most joints, the power supply returned to life. It seems that these capacitors have electrolyte that primarily rots the solder, not the traces or components.

Nasty-IIsi-PWM-Power-Supply-Board.jpg

Interestingly, I noticed that this board differs slightly from the other boards that I've repaired. One of the capacitors is in a different location. Did they move it away from the chip to increase solderability during manufacturing? Or did they move it closer to the chip to improve electrical performance?

IIsi-Power-Supply-Revision.jpg

With the power supply fixed, I tested the motherboard. Everything on the motherboard worked except for sound. Plugging in headphones verified that the sound chips work. So, the problem is somewhere between the headphone jack and the speakers.

On the IIsi, speaker problems may be caused by dirty connectors. This is the first thing you should check, because they are easy to clean with a pencil eraser and contact cleaning solution.

Clean-the-speaker-contacts.jpg

For my IIsi, that was either not the cause or was only part of the problem. So, I next looked at the amplifier components. The speaker amplifier circuitry is down by the speaker (not near the sound chips like on other motherboards).
IIsi-Speaker-Amplifier-Circuit.jpg

If you look carefully in the image above, you'll notice corrosion caused by capacitor leakage. For example,. the pads on R42 are suspect. And, check out the leads on UK3 in the image below:
IIsi-UK3-Speaker-Amplifier-Chip-Damaged-By-Leaky-Capacitor.jpg

Using a multimeter in continuity mode, I traced the circuit based on the schematic. No traces were damaged. However, there was a short across C41. I removed the capacitor only to discover that the capacitor tested fine and the short remained. I removed R40 only discover it was also fine. So, I removed UK3 and still the short remained!

It turns out that transistor Q11 (2N3904) had failed short. It looked fine on the board. There wasn't even any corrosion on the pads. I successfully replaced it with a 2N2222 (that's all I had in SOT23 surface mount). After cleaning and resoldering, the speaker works fine now.

The lessons learned are:
* Check for both expected continuity and unexpected shorts
* Transistors can go bad
* Don't ignore clean-looking components near gross-looking components

Lastly, I notice that C42 and C43 measure 10 nF not 100 nF when off the board. Here is the marked-up schematic.

IIsi-Speaker-Amplifier-Schematic.jpg

- David
 

joshc

Well-known member
Similar experience when I restored my IIsi. Had to remove a lot of chips from the logicboard, clean the whole area with flux/wick and thoroughly clean the soft power board from the power supply.

DC97807B-2FA6-4720-B729-31879ADC8EDF_1_105_c.jpeg
7EB53BB9-96BD-4E70-A659-83B3016FB58F_1_105_c.jpeg

325E6620-F898-4470-95F8-1DE5053F3081_1_105_c.jpeg

All the brown you can see there is from the crap that was still on the solder pads, the flux will go a dark colour and then all that needs to be cleaned off with isopropyl alcohol.

Afterwards everything looked new again in that area of the board:

1677615814794.png


64858D68-2732-4865-AFE6-3AB73C86B2A7_1_105_c.jpeg


DE738764-5E73-41AF-8823-D013FF568F5B_1_105_c.jpeg

The caps from the power supply were disgusting:
093D2420-E0D1-4C8E-A406-52BFB59F3A40_1_105_c.jpeg

I've had to do similar SMD rework on Classic / Classic II boards, you need those solder pads to be squeaky clean for when you put the chips back on:
5EF5EBA5-F67C-48EA-B317-B7768562F293_1_105_c.jpeg
 

joshc

Well-known member
Josh, you using hot air with paste, or just an iron/dragging the solder for that?
Hot air for removal only, soldering by hand to put stuff back on. Sometimes drag soldering but not for the most part - just lots of flux and a steady hand to touch up each pin with enough solder.
 

David Cook

Well-known member
Beautiful work!
Similar experience when I restored my IIsi. Had to remove a lot of chips from the logicboard, clean the whole area with flux/wick and thoroughly clean the soft power board from the power supply.

Afterwards everything looked new again in that area of the board:

Beautiful work! Much cleaner than my boards.

Any advice for removing grody solder like on that vertical board?

I don't have any good advice. My most recent vertical board took 4 hours to clean!

Here is my pattern:
1. Ultrasonic cleaner
2. Apply flux paste
3. Hit each solder point with a soldering iron
4. Apply fresh solder to anything that looks like it might stick
5. Clean up mess with isopropyl alcohol and a sponge swab (https://amzn.to/3y1EXDR). Wring out the sponge in a shot glass* and reply alcohol. The sponge is good for applying, light scrubbing, and soaking up / removal of the waste fluid.
6. Pick at it with dental picks (https://amzn.to/3IKLMOQ) and poke goop out from behind certain pins.
7. Sometimes apply isopropyl alcohol and then use canned air to spray out goop.
8. Sometimes use flux solder braid to remove solder blobs.
9. Repeat until clean.

I acknowledge that @joshc probably has a better approach to remove chips and parts with a hot air gun, clean the pads unencumbered, and then reapply the parts. I'm not sure how he gets the leads of the dirty parts clean before reapplying.

David

*Do not drink from the shot glass.
 

David Cook

Well-known member
Two follow-ups to my Mac IIsi repairs:

1. My Mac with the replaced crystal worked, but exhibited some strange symptoms:
* Wouldn't work with Turbo Mouse
* Date/time clock ran fast when power was removed.
* Apple Personal Diagnostics claimed a logic board test failed, without telling me which one. Snoops and MacTest Pro passed all tests.

Spoiler alert, the root cause of all three was the replacement crystal. I pulled it out of an unmarked pile in my 32768 crystal box. I decided to try replacing it with a new-old-stock Citizen CFS308-32.768kDZF (12.5 pF tuning fork) crystal. I then cleaned the area very well to avoid stray capacitance. All three issues went away. The ADB timing and power off clock are driven by the crystal connected to U7 (the 68HC05).

2. My Mac with the fixed sound started issuing the death chimes during boot after cleaning with an ultrasonic cleaner. This suggests that a connection was loose or intermittent, and the cleaning put it over the edge.

I spent several days (no kidding) tracing all of the memory pins, connecting to the remote diagnostic mode (http://web.archive.org/web/20210923025242/https://mac68k.info/wiki/display/mac68k/Diagnostic+Mode), and resoldering questionable pins.

The diagnostic mode gave me misleading information at first, as the initial status code (*R) of 1387 was actually a leftover register value from the successful Mod3Test. I think there are two bugs in the IIsi ROM that don't start the initial status message (*5) and doesn't set D7 upon a failed bussize test. When I tested bank A memory (*S *V *A *4 *000000000 *100100000 *T000200010000), the Test Manager indicated no error. So, memory was good. Why the death chimes?

I then reviewed the ROM source code and noticed that in USTStartUp.a, the startup test that follows the memory test is dynamic bus sizing (dynamic_bussize_test). The death chimes were probably happening here. That caused me to look more closely at the CPU. Reheating three CPU pins near C34 caused gray and brown residue to spew out. So, the 'clean' looking CPU pins definitely had some corrosion. With flux, I reheated all the CPU and MDU pins. The Mac booted fine after that.

Leak-from-Capacitor-C34-Affected-Pins-on-CPU.jpg

By the way, the references in the ROM code to 'Ericson' and 'Erickson' refers to the codename of the IIsi.

David
 

cheesestraws

Well-known member
I then reviewed the ROM source code and noticed that in USTStartUp.a, the startup test that follows the memory test is dynamic bus sizing (dynamic_bussize_test). The death chimes were probably happening here. That caused me to look more closely at the CPU. Reheating three CPU pins near C34 caused gray and brown residue to spew out. So, the 'clean' looking CPU pins definitely had some corrosion. With flux, I reheated all the CPU and MDU pins. The Mac booted fine after that.

This is top notch troubleshooting here, nicely done.
 

chiptripper

Well-known member
The other Mac IIsi doesn't respond to the rear power button or the keyboard power button. Here is the chip that monitors the power buttons. You can test it without the power supply by simply making sure you have a fresh PRAM battery installed.
View attachment 52687
Pin 14: GND
Pin 13 (V): +V around 3 volts supplied by the battery.
Pin 25 (B): Should be around 3 volts, but drops to 0 when the rear button is pressed.
Pin 24 (K): Should be around 3 volts, but drops to 0 when the keyboard power button is pressed.

You can also check pin 2 (reset) for 3 volts to make sure the 68HC05 isn't being held in reset.
Hi David, I have a clean, recapped IIsi and recapped PSU. It boots and works just fine, but restarts immediately after I select “shut down”. This thread has been very helpful!

I tested pins 24 and 25 on 68HC05. On 24, I get 3 volts, which drops to 0 on keyboard press. But pin 25 shows 0 volts at all times. So that’s seemingly identified the issue, if not the exact cause.

Interestingly, I have two IIsi boards and they both have this issue. Time to dust off the schematic. :)
 

David Cook

Well-known member
Just repaired another IIsi. Same issues as other machines.

1. Bad crystal prevented the power on switch from working.

2. Bad memory due to corrosion on the transceiver chip UE5. I had to remove the chip and clean the pads to find three broken traces right at the point that they connect to the pads. These are not easy to spot by eye.

IIsi-memory-transceiver-broken-traces.jpg
 

zigzagjoe

Well-known member
FWIW, ultrasonic cleaners are well known to kill crystals and crystal oscillators. Might be that your cleaner is killing the crystals rather than the crystals being defective, especially if the microcontroller is otherwise fine.
 

David Cook

Well-known member
FWIW, ultrasonic cleaners are well known to kill crystals and crystal oscillators. Might be that your cleaner is killing the crystals rather than the crystals being defective, especially if the microcontroller is otherwise fine.

Yes, I also believe that is the cause.
 
Top