What's inside of it?
What part is directly affected by the Heat ?
How it works?any diagram?pics of disassembled one?

TKS...

I did once depackage it, it it works out to about $50.00 worth of parts, couple of op amps, drive transistor pair, big resistor, pwm controller. thats about it..

its the transistors that fail.. I did do up a schamatic because i was thinking about redesigning it with higher current IGBT instead of the BJT that they used for drive.. but got busy and never went anywhere with it..

i will look around my stuff and see if i can dig it up..

Vin


What's inside of it?
What part is directly affected by the Heat ?
How it works?any diagram?pics of disassembled one?

TKS...

If you'll post it, we can compare notes - I've dabbled in it a little, meself.......:cool:

ok you got me thinking today, so i took one of my old dead (left me on the side of the coquhalla at 2:00am) FSD's and reopened the research i began on FSD failure that i started 4 years ago.

at the time, i was curious what this $500.00 item was made of.. so appying some basic reverse engineeriing techniques i began to reverse engineer the brick known as a FSD or PMD to some...

essencially its a current driver, that has some suverisory circutry to detect when the solonoid has moved off the seat. theis is done through a concept known as "back emf". As the coil de-energizes, it induces a current and voltage in reverse as the magnetic field reverses.

from what i could tell.. the FSD was monitoring back emf to determine when the field has collapsed to a certan point. and then sends the ECM a signal that the valve is off the seat.

if it takes too long, i.e the valve gets stuck, the metal core of the solinoid maintains the magnetic field for longer than normal, this way it can detect if the valve has actually moved or not. conversely if the core gets stuck outside the core, i.e stuck open, the field will collapse faster.

so what we really have here, is a driver with some small amounts of monitoring of coil currents.. (i think it also monitors for shorts and opens as well)

what kills these FSD's? well. heat.. but not the way most people think. it is the way it was packaged that kills them. Standyne in there infinite paranoia, packaged them using a rigid epoxy, this allows no thermal expanson in the components. because most of the components are SMT (surface mount technology)it is very easy to fatigue the solder that holds them to their pads which is the only mechanical strength.
as any first year electronics tech will tell you.. solder aint all that strong. So after repeated cycles of heating and cooling, because they cant move due to the epoxy potting compound, they fatigue and stress crack develop in the solder bond layer.

i so far have had 3 FSD failures, and have now reverse engineered every one from a failure analysis standpoint..

here is how it breaks down..

ALL have complete failures of the bond between the large driver transistors and the board, this is due to the very rigid way they are mounted. (big chunck of steel) and the fact that the board cant flex either (rigid epoxy) and the rather long length of the leads for the base & emitter pins (1/2")

this gives large amount of thermal "E" (expansion) on each pin along the axis of the pin. this puts stress on the only thing that can give, the solder joint and over a few thousand cycles, bingo.. fatigue failure.

only one had a failure of one of the pins on an opamp. this was on a high milage FSD (~100,000 miles)

lastly one failure of the solder joint on a large resistor used to measure the current to the coil. same failure mechinism as the drive transistors..

so.. if one could remove all the epoxy and put in a flexable product. i suspect that we would see alot more miles on these things.

one idea im toying with is to "CAREFULLY" scrape down through the epoxy to the circut board, and resolder the fatigued joints..

i started this on one of my totally dead FSD's that left me on the road even after the retorque trick (which i think probbly works because it moves the fatigued joint just enough to make contact)

I will post pics to this thread as i proceed with the project..

vince

As usual I found some good info. Check the second page for pics

http://dieselplace.com/forum/printthread.php?t=11490

Interesting - thanks.

I had independently duplicated BETA's findings in Sweden, where heat-related compression\expansion resulted in 'settling' of the epoxy thermal potting compund, resulting in loss of tension at the nut to driver case interface.

The screws, kep nuts and case are stainless steel, and that loss of tension allowed chromium oxides to reform, increasing electrical resistance at the interfaces.

We determined that some FSD failures could be 'recovered' by burnishing the points of contact - this involved first fully loosening the nuts off the case, then re-torqueing to 5-7"lbs, which is just over finger-tite with a nut-driver.

10"lbs on a #6 SS screw provides about 3000lbs compression - this is aided by the thermally activated adhesive in the white insulator pad between the driver and the heat sink.

Ignoring this torque-setting serves only to rupture the ceramic hermetic seal used at the driver case-to-lead interface.

The Base and Emitter leads are of tinned steel, which has reduced thermal expansion property compared to brass or copper, intended to prevent lead-to-pcb solder rupture due to thermal expansion where the drivers are bonded and under sufficient compression.

I found no problem in the lead-to-pcb solder connection

My company has since eliminated using the black thermal epoxy compound, now using a two-part silicon-based flexible conformant compound to reduce the component\pcb disruption in our potted assemblies, which are under no compressive tension.

I also de-potted a module, getting a good idea of the circuitry minus some of the really small smt items, which popped outta sight during the depotting procedures.

There are several type replacements available, with some of the same problems, but I just wanna know every jot and tittle of this one's operation, every minute nuance and nano-amp.

The FS armature moves only 0.005" to close the valve, so the size of the solenoid coil indicates the inherent power, and the rev emf is fairly powerful signal on an oscilloscope, even tho actively damped.

Interesting - thanks.

My company has since eliminated using the black thermal epoxy compound, now using a two-part silicon-based flexible conformant compound to reduce the component\pcb disruption in our potted assemblies, which are under no compressive tension.



I gotta sound stupid here, but your company is making the GM/Stanadyne PMD's or is there another alternative available?

No cause for alarm - that ambiguous statement is just an affirmation of the epoxy problem and the solution - we build industrial process computers.

Very interesting. The epoxy problem makes a lot of sense.

I get the feeling there has been much more research done here among the forum members as to the cause of PMD failures than either GM or Stanadyne has performed. :cool2:

Yes out of self interest it's been studied some.

Not in GM's or Stanadyne's interest to fix something basically no longer a production item, warranties have pretty much expired for most, so least painless way for them was the "special policy" roll the $$$ dice and hope for lower cost to deal with it by attrition as fails occur (some never fail) rather than a wholesale campaign to fix them all.

It's a shame NTSB did not pick up on this as a safety issue, I guess since no-body as of yet has died from this, other than heart attack having your truck die while loaded climbing or going down a grade, or in rush hour traffic.

The FSD applies power to the Fuel Sol to close the valve at each Injection Event.

The parallel PNP collectors are connected to the Fuel Sol, the Emitters are tied to +12 thru a couple BMF laser-trimmed steel resistors and a large Schottky diode, all in a darlington drive configuration.
Emitters are actively snubbed by large pair of driven transistors to damp rev emf from the FS coil.

Essentially, engine running - FI Control goes hi +5v, Closure Signal +5v goes low, FS Drive goes hi +12v, overshoots with ringing

FI Cont remains hi for 2ms, Closure Signal stays low until FS Drive stops ringing -100usec - then goes hi, remains hi until next Inj Event

FI Control goes low, FS Drive goes low with heavy negative ringing, end of Injection Event

All signals vary in duration with varying rpm

Whatcha got, hrjack?

Uhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh... .........

Can somebody toss me a beer? My brain hurts after reading all that.

Actually, it is quite apparent that there are a number of very "electronics savvy" folks in here and have figured out what the problem might be. BUT.....knowing what caused the problem doens't help someone sitting on the side of the road with a dead truck. Is is possible to build a prototype or clone that is more reliable? Could one be built on a regular board like inside of a gutted stereo amplifier with cooling fans instead of packaged up in a tiny no-tolerance box? And if that is possible, could it then be mounted inside of the cab like up on the firewall behind the pedals where it could utilize some of the cooler air inside of the cab rather than the hot air under the hood?

I know that I'm just a dumb ole redneck, but could this be a common sense approach to a high tech problem?

Rite now not many options, replace with new in OEM spot, new on cooler on intake or other underhood location (IMO this location only serves to gain better access when it fails again), remote out of bay on cooler (seems to work for most I think key here is to start with a new driver and large heat sink), Heath's driver kit large & remote out of bay $$$ but with 7yr warranty, and REMAR-Q total redesign with some growing pains still $$$ and only 5 yr warranty.

It's a simple but not so simple circuit for what it is being asked to do.