Adventure in fixing a Lisa DataPower PSU

[sigma7]

T4:5 - T4:6. in ohms mode 0 ohms
T5:5 - T5:6. in ohms mode 0 ohms

Ok, good.

Do I connect CR18 again?  I would presume so.  I also took out the heatsink and screws for T3, and T4.  And should the daughter board be placed back in before testing?

You can reconnect CR18 or leave it disconnected at your convenience.

The daughterboard should be in place.

T3 and T4 don't have a heatsink or screws. If you mean Q1 & Q2, at a minimum, the screws should be removed, if the heatsinks are then loose, removing them too is wise.

[berskyboy]

Normally, with Q1 and Q2 out of circuit (either completely, or with the 4 phillips head mounting screws removed such that their cases are disconnected), the main transformer will not receive any switching current.

In this state, the only output voltage will be +5Standby.

With the high power circuitry inactive, we can (hopefully) check some of the control circuitry without having a fault in the high power circuitry blowing the fuse.

Double check that Q1 and Q2 are disconnected (4 screws removed) or completely removed from the circuit board.

Ensure a good fuse is installed.

Connect the test jig load circuit to the card edge, and turn its small switch "off".

The small switch should still be in the OFF position, and the power strip will be on. Check the DC and AC voltages from pin 3 of the daughterboard connector to ground. This should be about 16 to 18V DC, and should have an AC voltage of less than 1V; if the DC and AC voltage are in that range, continue with the next step. If not, stop here, remove power, and report what you observed. If the DC voltage is about half or double, investigate whether the 115/230V jumpers are not installed correctly or damaged or not making proper contact. If the DC voltage was zero, check the fuse.

-DC: 17 V
-AC:  0.111 V

-Fuse OK

Remember to set the meter back to DC volts after measuring the AC volts.

Check the DC voltage across R33 (from one end to the other, not to ground)

-161V

 & across R2.

-161V

(These are two large resistors beside the large black capacitor C8.) They should be about 150-170 V each (positive or negative, depending on which way you probe them): if so, continue with the next step. If not, then the primary rectifier, line filter, and associated components are suspect - stop here, remove power, and report what you observed.

Check the DC voltage at pin 12 of the daughterboard connector to ground. This should be about 5V:

-4.97V

if so, continue with the next step. If not then Z2 on the daughterboard is suspect - stop here, remove power, and report what you observed.

Make sure the interlock switch has been engaged. Set the small switch (on the test jig connected to the card edge) to the ON position. There should not be any smoke, flashes, odours, blown fuse or other dramatic evidence of something happening. If something does, stop here, disconnect the power and report what you observed.

-ok

Collect measurements from the following steps A - H and report what you observe. As before, stop, disconnect power and report observations if anything happens that isn't expected.


These measurements are much more easily done if the black lead of the DMM can be secured to ground, freeing up a hand to operate the switch. eg. clip the black lead to the grounded side of a load resistor. If nothing else is available, inserting the black probe into the grounded mounting hole near the card edge connector and leaving it leaning against the edge may work. Be careful to not dislodge the probe while power is on; you might want to tape it in place. Turn off the power strip if you need to fiddle with this connection.

On/Off in the following refers to the small switch on the test jig attached to the card edge. The power strip can remain on (the behaviour of some of the following measurements is different if the power strip is turned off/on instead of the test jig's switch).

A. Turn off, locate pin 13 of the daughterboard connector (at the CR15 end), then while watching the voltage at pin 13, turn on: the voltage should rise from a low voltage about 0.5V to about 4.8V over a few seconds.

-0.388V to 4.64V (then stays at 4.6V)

B. Turn off, then while watching the voltage at pin 8 of the daughterboard connector, turn on: the voltage should rise briefly, from near 0 to near 5V, then return to 0. This takes less than a second but should be visible. If it is stuck at 0, double check the interlock switch has been bypassed.

-0.01v to 4.6V then down to 0.025v

C. Turn off. Carefully (be careful the probe does not slip and cause a short circuit) hold the red probe to the lower end of R30 on the daughterboard (R30 is at the left edge/corner). The voltage should be near 4V.  Then while watching the voltage, turn on: the voltage should drop immediately to near 0. If the voltage remains high (around 4 VDC) then the soft-power circuit is suspect.

-no 0V, but I might not be putt the probe correctly

D. Turn off. Carefully (be careful the probe does not slip and cause a short circuit) hold the red probe to the + (plus) side of C14 on the daughterboard (not far from R30). Then while watching the voltage, turn on: the voltage should rise over a few seconds from near 0 to near 4.8V. If stuck (around 0 VDC), then the brownout circuit is suspect.

-4.63V

E. Turn off, then while watching the voltage at pin 10 of the daughterboard connector, turn on: the voltage should drop from about 2.4V to about 1.4V for a noticeable but brief period (less than a second). If you don't see the brief drop, turn off, switch the DMM to AC Volts and when you turn on again, you should observe the AC voltage briefly rise from 0V to perhaps 1V (the value will depend on the DMM), then return to 0 before you turn off again. If no activity is observed, then Z2 on the daughterboard is suspect. Set the DMM back to DC Volts.

-goes 2.25v to 1.43v then back to 2.25v

F. Turn off, then while watching the voltage at pin 11 of the daughterboard connector, turn on: the voltage should drop from about 2.4V to about 1.4V for a noticeable but brief period (less than a second). If you don't see the brief drop, turn off, switch the DMM to AC Volts and when you turn on again, you should observe the AC voltage briefly rise from 0V to perhaps 1V (the value will depend on the DMM), then return to 0 before you turn off again. If no activity is observed, then Z2 on the daughterboard is suspect. Set the DMM back to DC Volts.

-2.25v then 1.84v then 2.25v

G. Turn off, locate R7 on the main board. R7 is on ceramic standoffs beside a silvery heatsink plate. One end should measure around 16-18V. While watching the voltage at the other end, turn on, the voltage should rise briefly, from about 1.6V to 5-15V (varies depending on the speed of your DMM), then return to 1.6. If no activity is observed, then Q4 or T5 is suspect.

-1.621v to 11.3V then 1.63V (resistor to ground)


H. Turn off, locate R5 on the main board. R5 is on ceramic standoffs near the center beside R4. One end should measure around 16-18V. While watching the voltage at the other end, turn on, the voltage should rise briefly, from about 1.6V to 5-15V (varies depending on the speed of your DMM), then return to 1.6. If no activity is observed, then Q3 or T4 is suspect.


-1.631v to 9.72V then 1.63V (resistor to ground)

[sigma7]

-1.631v to 9.72V then 1.63V (resistor to ground)

I'm assuming the "-" prefix in these readings doesn't indicate negative voltage... if it does then let me know.

The measurements E-H suggest the SG3524 on the daughterboard is attempting to drive the power circuit, so it is probably good. The other measurements suggest the ancillary circuitry is also working.

We haven't thoroughly checked the secondary side of the PSU, but since you have replaced the output capacitors, and we confirmed the rectifier diodes look like diodes, there is a good chance it is ok.

I suspect we found and corrected the original fault (a broken solder joint at the trimmer potentiometer R11), and the fault that caused Q1 & Q2 to fail, solder bridge(s).

So I suspect that replacing Q1 & Q2 (and ensuring there are no solder bridges remaining) is likely to be sufficient to repair the fault that blows the fuse.

The original 2N6308 transistors are no longer made, so finding a substitute is the usual option, although one could look for old-stock originals on eBay etc.

A few posts back, you referenced a Newark/Element14 cart with BUX48A transistors in it. Did you find references that indicate this is a suitable substitute?

Does anyone else have a recommendation for a currently available 2N6308 substitute that works in this PSU?

[berskyboy]

I'm assuming the "-" prefix in these readings doesn't indicate negative voltage... if it does then let me know.

that denoted a measurement.  Sorry I wanted to separate  it.

But what about this value?:

C. Turn off. Carefully (be careful the probe does not slip and cause a short circuit) hold the red probe to the lower end of R30 on the daughterboard (R30 is at the left edge/corner). The voltage should be near 4V.  Then while watching the voltage, turn on: the voltage should drop immediately to near 0. If the voltage remains high (around 4 VDC) then the soft-power circuit is suspect.

-no 0V, but I might not be putt the probe correctly

[berskyboy]

So I suspect that replacing Q1 & Q2 (and ensuring there are no solder bridges remaining) is likely to be sufficient to repair the fault that blows the fuse.

The original 2N6308 transistors are no longer made, so finding a substitute is the usual option, although one could look for old-stock originals on eBay etc.

A few posts back, you referenced a Newark/Element14 cart with BUX48A transistors in it. Did you find references that indicate this is a suitable substitute?

Does anyone else have a recommendation for a currently available 2N6308 substitute that works in this PSU?

This is what I found.

BUX48A
TRANS, BIPOLAR, NPN, 450V, 15A, TO-3

https://canada.newark.com/multicomp-pro/bux48a/trans-bipolar-npn-450v-15a-to/dp/23AJ5969

from AI or google:

Yes, the BUX48A can replace the 2N6308 in most high-voltage power supply, inverter, and switching applications. It matches the package type, exceeds all critical maximum voltage and current limits, and provides equivalent or better switching speeds. Solitron Devices, Inc. +3

https://solitrondevices.com/datasheets/2n6306_8-ds.pdf

[sigma7]

But what about this value?:

C. ..

-no 0V, but I might not be putt the probe correctly

The fact that E - H show the SG3524 attempting to run indicates the soft-power circuit is working, so I think it doesn't matter if either you weren't able to probe it or my predictions were incorrect.

This is what I found.

from AI or google:

Yes, the BUX48A can replace the 2N6308 in most high-voltage power supply

Neither AI nor google have any direct knowledge, they've only collected bits of information from elsewhere, and sometimes combine them incorrectly, so they can't be considered an authoritative or reliable source. eg. AI was previously telling you to replace a part that isn't there.

So I'm hoping there is a better source, if not, we will compare specifications in an attempt to confirm the suggestion is reasonable.

[berskyboy]

I'm found and bought from this site the 2N6308

https://www.tedss.com/2028023156

[sigma7]

found and bought from this site the 2N6308

Good choice!

It looks like there are very few TO-3 transistor packages still being made. The plastic case alternative will not be trivial to retrofit in the DataPower PSU.

[sigma7]

found and bought

While you're waiting for those to arrive, anticipate if you will need to procure some additional items to install them.

If you can remove the mica insulators from the old transistors without breaking them, you can re-use them.

In addition, thermally conductive grease/compound is needed with mica insulators. In a pinch, if the old parts have plenty and it is clean, you can re-use the grease, but cleaning it off and using new grease can help avoid issues. The fancy heatsink compound used for computer CPUs/GPUs is the same type of material (but more expensive) and may be easier to get locally, or you may already have some.

Watch some videos to see how much grease to use... the final layers should be very thin (eg. almost translucent), too much will make thermal transfer worse. You may need slightly more for broad contact if the heatsink surface isn't very flat. Keep grease away from where the leads enter the transistor case.

There is also the option of replacing the grease and mica with a TO-3 shape silicone insulator pad. I've not read confirmations that this works ok in this PSU, but I expect so. The silicone pad option does not need grease; clean off all the old grease if switching to a silicone pad.

If any of the TO-3 mounting screws were difficult to remove, the solder joint between the threaded mounting post and the circuit board may have broken. This is best resoldered without the screw and transistor in place. On the bottom of the main board, carefully check for cracks in the solder around the outside of the threaded mounting posts and resolder if the joint is suspect. Avoid putting more solder in the hole (and remove solder in the hole if possible).

Important: When re-installing the TO-3 transistors, you must complete the physical mounting before soldering the leads. ie. clean and grease the mating surfaces, assemble, and fully tighten the phillips head screws before soldering anything. Soldering too soon introduces mechanical stresses that can damage the transistor and circuit board.

The DataPower PSU was originally wave soldered after the TO-3 transistors were installed. As a result, the transistor mounting screw holes have solder in them. Check carefully that the transistors are rigidly clamped down against the heatsink; if not, the solder in the screw holes may be preventing full engagement of the screws. If you have appropriate equipment, you may be able to remove the solder from the screw holes, or you can use a 6-32 tap or something similar to clear solder from inside the threaded hole. You do not need to re-solder the mounting screws once re-installed.

Once the screws are tightened, check that the insulator has not been compromised by measuring the resistance between the heatsink and the case of the transistor. It should have a very high resistance. The black coating on the heatsink is non-conductive, so you'll need to pierce it, scrape a bit off, or find a spot that isn't black. Measure the resistance from one spot to another on the heatsink to confirm you've established good contact.

Double check that all other components (such as CR18) have been re-installed to return the PSU to the fully assembled state.

[berskyboy]

While you're waiting for those to arrive, anticipate if you will need to procure some additional items to install them.

If you can remove the mica insulators from the old transistors without breaking them, you can re-use them.

I think I can use them and wipe them off

In addition, thermally conductive grease/compound is needed with mica insulators.

I have ordered this.

https://canada.newark.com/wakefield-thermal/120-sa/thermal-grease-compound-packet/dp/00Z1244?CMP=e-email-sys-shipping-GLB-Product

[berskyboy]

Watch some videos to see how much grease to use... the final layers should be very thin (eg. almost translucent), too much will make thermal transfer worse. You may need slightly more for broad contact if the heatsink surface isn't very flat. Keep grease away from where the leads enter the transistor case.

would this be a good video for that?

https://www.youtube.com/watch?v=ufTi-lH4nw4

[sigma7]

Watch some videos to see how much grease to use... the final layers should be very thin (eg. almost translucent), too much will make thermal transfer worse. You may need slightly more for broad contact if the heatsink surface isn't very flat. Keep grease away from where the leads enter the transistor case.

would this be a good video for that?

https://www.youtube.com/watch?v=ufTi-lH4nw4

No.

Edit to add some details:

To be fair, if it is done that way, it might be fine, and probably is in many cases, especially if you are doing service for a fee and the time consumed to do the job is a concern. But I think the video is probably not considered a demonstration of ideal technique by many, and we can hope the Lisa PSU might be useful for another 40+ years, so taking the time to do a good job is warranted.

It is a good point that if the old material is dried out, re-using it would be a last resort.

My criticisms of the approach shown:

1. Not keeping the transistor leads clean: Apparently the silicone oil can migrate past the glass insulator around a lead and contaminate the transistor inside the case. Supposedly this takes a long time, so probably only an issue for ultimate longevity. I don't know what consequences there are to contaminating the inside, perhaps not relevant here to a switching/power transistor. Regardless, aside from saving the time it takes to be careful, there isn't any advantage to putting grease on the lead.

2. Using a power tool to attach the hardware: the time saved over dozens of repairs is completely wiped out when you have to fix a stripped/damaged tapped hole in a heatsink or threaded standoff. There are specific circumstances where a power tool might be used, but it should not be implied as standard procedure.

3. Not starting with clean surfaces: the best thermal contact is when the surfaces are as close together as possible. This can easily be compromised if the surfaces are not clean.

4. Using more grease than necessary: As above, the best thermal contact is when the surfaces are as close together as possible. The grease is there only to fill the gaps (since air is a good insulator), so there should be as little as possible. This is not a case of "the bigger the gob, the better the job" as AvE sometimes says tongue-in-cheek when tearing down equipment. (If you're not familiar with AvE, you might find him amusing, although some of the vernacular may be a bit opaque if you're not Canadian.)

Not specific to the video is checking the quality of the mating surfaces. Quality-control cutbacks are sometimes revealed as non-flat surfaces, raised dings, as well as burrs around holes. These prevent close contact and result in poor heat transfer (which results in failed transistors as they overheat). At minimum, check for and remove burrs around holes in heatsinks.

I may add more thoughts/opinions later.

[berskyboy]

To be fair, if it is done that way, it might be fine, and probably is in many cases, especially if you are doing service for a fee and the time consumed to do the job is a concern. But I think the video is probably not considered a demonstration of ideal technique by many, and we can hope the Lisa PSU might be useful for another 40+ years, so taking the time to do a good job is warranted.

Good point.  40+ years, yes a definite goal to aspire to.  I'm in your debt for all the help you have shown me, and even these tips for success.  (ie Goal of 40+ years), which is "inline" with how they designed and made the Apple Lisa (from my exposure it looks like they didn't cut corners and did the best job to last 40+ years). 

I will do more research on the topic and find other videos.

[stepleton]

did the best job to last 40+ years

except for the I/O board battery though 

(Could they have known?)

[sigma7]

did the best job to last 40+ years

except for the I/O board battery though 

(Could they have known?)

I think someone expected as much - witness the 16 year range of the RTC.