PMD(FSD) ? Why Fail?

[rjwest]

Disected 2 FSD (PMD's) :

Thought I would tighten the nuts on the transisters to see how they made contact to the power transister. Surprise, 1 nut broke off, with very little torque. Took out screw and examined, The screw appeared to be Aluminium, non magnetic and soft. Next I examined the surface where the screw goes through the component side of the FSD.
There is a circular printed circuit 'land' that appears tined, as in circuit board manufacturing.

The underside of the screw head is the electrical
contact with the circuit board.

Also the transister case ( which is an electrical part of the circuit ) can only contact the circuit board through the screw which holds down the transister.

( Note: The alu block that the Transister mounts on is covered with a non-conductive material.)

Therefore: Electrical path for the power transisters is: Case , to nut, through Alu ? screw to Tinned LAnd on circuit board.

OK, Bottom Line: Design is failure prone ( like we did't know this already ) as the Heat cycles allow the screw/contact point to Fail

This design " would not fly " IE: Not an acceptable design for the appplication.

Also : In removing the transisters ( destructive process ) It appears the transister lead to circuit board solder joints may also be suspect ( cold solder )

In my opinion, there is a design problem, which GM should still address, regardless of years/milage.

New parts should have at least the same Warrenty.

I have one more failed FSD, Will try to repair and validate my conclusion.

Electrical Experts: any comments:

[MTTwister]

rj - Aluminium screws can't be the best conductors in the first place.
I would think Aluminium expansion in the heat cycles would cause more flex in the solder joints, causing them to crack.
Does it look like the screw heads are digging into the circuit board enough to make decent contact?
And, I assume that there's not a snowball's chance of retro fitting some decent screws without destroying the FSD.
Can you melt the potting mix out of the body? turn it upside down, and apply various heat sources to it to drain that junk out of the way without destroying the circuitry? ( didn't think so). Thanks for digging into this.

[Hye]

There have been enough failures to prove that the PMD isn't reliable. Reducing the PMD's maximum temperature by putting it on a heatsink mounted away from the engine can improve its reliability. (Good reason #1 to use a cooler, even if you haven't had problems, yet.) I don't think any more proof or opinion is needed.

But since you asked, IMHO, what you found would contribute to poor reliability. PMD reliability would be better if the screw heads were soldered to the circuit board. Soldering improves the electrical connection, duh. And without the solder or some other restraint, the screw can turn even if the nut and lockwasher don't. Then the nut looses contact with the transistor, double duh. Neither aluminum nor stainless can be soldered using the same methods used to solder electronic components to a board. That is probably why they aren't soldered, triple duh.

I found, when I "fixed" my pump mounted PMD symptoms on the '94 six weeks ago, that the nuts and screws looked like 6-32 thread, stainless steel, but I did not check them closely. Without a lock washer, #6 stainless 6-32's should be tightened to around 9 pound inches. When a lock washer is added, torque values found in published tables may not mean much. I just tightened it by feel and assumed I could retighten anytime since the PMD was now easily accessable on a cooler. (Good reason #2 to use a cooler.)

I didn't try to get inside the PMD. Just tightening the nuts fixed it. If I had one that I knew was truely dead, I'd do some more investigating. I did look up the transistors' data sheet. Assuming that they operate in parallel, its likely that if one transistor lost contact, the other would keep the I/P working. But it might not last long in hot weather if you don't fix it soon after symptoms appear. (Good reason to use a cooler #3: You'll have a better chance of getting to the loose nuts before it fries.)

I've used plated, mild steel, pressed fit "screws" to connect TO-3 transistors to a PC Board. They look just like mini wheel lug studs. When pressed into a plated thru hole and then soldered to the board, they make a reliable electrical connection to the board. Like wheel lug studs, they won't turn after they are pressed into place. Its easy to twist a 6-32 screw in two. A torque screw driver is almost essential to get maximum torque without destroying the screws. Properly tightened star lock washers make a good connection to the transistor case IF they don't corrode.

One other note, the transistors have a date code printed on them. Mine read 9618, meaning the 18th week of 1996. So the PMD was built some time after that. GM replaced the I/P at 116K miles. The PMD had to come with that pump, not my '94 truck.

If someone has a way to fix 'em, permanently, we'd all love to hear about it.

Hye, BSEE '70

[rjwest]

Transisters are ( I think ) in some series arrangement, as a Fail safe mode, ( can't short Solenoid "Open"

Removed Material over screws,( can not melt ) and removed screws, They are in a printed circuit land and do not show any real sign of good contact( do not dig into land ). But the Screw heads are held secure by the covering material. This may not be good, as expansion could move the screw ( loosen ) and than suspend it from the printed circuit.
The cases are joined electrically, and the screw/
contacts are in parallel ( IE all for screws are common point )
The nuts have a Locking washer arrangement, But the screw heads do not, the screw heads are the only contact point to complete circuit.

I think the screws are ALU ( not sure ) cut easily with a side cutter, May be " Just Junk ".

I am going to get some decent screws , with lock washers under the screw heads, and re install.
See if the PMD will run, Use to fail within a week.

One other note, Tryed to remove Transisters( Had quite a few bad one's ), Can not do it without " destruct " the PMD,
In one instance I was not confident in the power transister " Lead " solder joints, looked like cold solder.

[Hye]

I think your plan will work. Zinc plated steel hardware is OK. Adding some locktite or fingernail polish to the threads will help hold it all together after you get it tight.

[gmctd]

Anybody done a 'search' on FSD\PMD, here?

[DmaxMaverick]

^

:D

[moondoggie]

Good Day!

Massive amounts of FSD/PMD information was summarized in FSD, PMD, PMD, FSD -or- It's 10 pm - where is your FSD? (http://forum.thedieselpage.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=1;t=005199#000000) (Click on the colored text.) This topic was preceded directly by FSD\PMD - BFD (http://forum.thedieselpage.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=1;t=005013#000003) & two other long topics, which was preceded by years of trying to figure these things out.

Blessings!

Brian Johnson, #5044

[rjwest]

FAILURE: Installed new screws, on clean contacts, retorqued , PMD is still DEAD!!!!

OHM'd the Power transistors, appear ok!

May try to remove transisters, But loosing interest again: as I have at least 2 good spares.
1 still on the pump and 1 in the glove box.

Read all the old posts: ( Thanks ). Think I am fighting windmills.

[tom.mcinerney]

Good effort, RJ. Others had speculated , and discovered bad/cold soldered leads.
I think you are first to have found aluminum screws , they'd have buch better electric conduction than the SS. But weak, fatigueable...

I'm betting the Remarque SolD is the answer. Only i can't justify purchase when my OEM Stanadyne is fine so far!
But Remarque will be my replacement.