Apparent Regulator Failures

[lragan]

Nice patch job, Guy.  Hope it works for you.  Keep us informed here. 

Thanks for the pictures.  The metal strap may provide some thermal benefits as well as a case ground.  If it was thicker, it would conduct heat better.  Is it made of aluminum?

You may encounter an electrolysis issue where the aluminum strap (if that is what it is, actually) contacts the steel frame.  A little moisture, significant electrical current, over time = corrosion.  There are conductive greases designed for copper "pigtailing" of alluminum wire which can effectively exclude the oxygen from the joint, obviating corrosion.  My last bottle finally dried up, and I haven't looked for it lately.

Do you still have the ground wires hooked up too?  If so, they may carry most of the current, I don't know.

[Magnum Magna]

Well I will look at checking my voltage tomorrow to see if it is 14V at 3000 RPMs. I am glade you got it fixed. 

[Magnum Magna]

I check the voltages with no engine load in neutral the effective voltages were 14 volts at idle ( 900), 3000 and 4000 and 5000 RPMs
   
When they design heat dissipation they normally put thought and hopefully testing for verification of effectiveness.   The fins run from side to side not front to back.  I wonder if the fins would work better if the air would flow across the fins or does the side to side layout catch the upward draft hold the air it for a second then let the heat off with the slower action of the wind in the fins. 

[lragan]

   
When they design heat dissipation they normally put thought and hopefully testing for verification of effectiveness.   The fins run from side to side not front to back.  I wonder if the fins would work better if the air would flow across the fins or does the side to side layout catch the upward draft hold the air it for a second then let the heat off with the slower action of the wind in the fins. 

I did speculate that fins running parallel to the bike would achieve better cooling, but that was based on a naive assumption that airflow at this point was along the axis of the bike.  The point is, I don't have a clue what the airflow pattern looks like under the battery box.  

Short of mounting a video camera and some small strips of thin plastic under there, and running the bike at speed while recording, I don't know how to find out.  I don't plan to do this, by the way.

I am still waiting for the rest of my instrumentation package to arrive to run dynamic temperature tests.  I will be away on business for a couple of weeks starting next week, so it may be a while before I get to dofurther work on this problem.

[Charles S Otwell]

With the rear tire so close to the regulator I wonder if there is a wind flow direction as much as just wind turbulance?

[guywheatley]

Iragan:
I'm not real sure what the metal is. I'd assumed it was aluminum. It came out of a corrosive environment in a film processor. It had grounding wires attached to it from the main electronics board. The surface had whitish blemishes I'd assumed was aluminum-dioxide. But when I started working with it, I was surprised to fine it mailable. It bends as easily as copper. I did check it with a ohm meter to be sure it really was conductive.
I'll be keeping an eye on that strap for corrosion.

[lragan]

Guy, thanks for your quick answer.

Aluminum (pure) is quite soft, and very malleable.  I have never encountered pure aluminum.  It almost always comes as an alloy.  The most common iare called "6061"  or "6063".  Once long ago I think maybe I knew what these numbers meant, but that knowledge has escaped me.  I found this link on Wikipedia:  http://en.wikipedia.org/wiki/Aluminium_alloy

As they allude to there, aluminum will "cold flow" -- meaning that extended time and pressure will cause it to "squish" out of shape.  Probably not a problem with the strap you have, but back in the 60's, during a copper shortage, a lot of homes in the Dallas area were wired with aluminum -- and caused a lot of fires due to cold flow and dissimilar metals corrosion issues. 

I owned two homes that were so wired.  One could alleviate the problem with tight spring wire nuts, a short length of copper wire, and some of the conductive grease that I mentioned earlier.  Every connection to a switch or socket had to be patched in this manner.  Big pain.  Interesting side note -- when I finished with the home that we were living in with our young children, I called a friend and asked him to help me remember if there were any boxes I hadn't opened.  After reflecting a moment, he recalled the one in the AC closet that disabled the indoor unit.  When I went to patch it, it was very hot.  That phone call and friend probably saved my home, and possibly the lives of my children.  Whew!!

[guywheatley]

Iragan:
Thanks for the wiki link. I too lived in a house with aluminium wiring. I remember be surprised at the thought of wire made from what I'd always thought of as a very brittle material.  I assumed (falsely it turns out) that the wire was alloyed for malleability.
I never had the entire house reqwired because of cost, but I did run several copper runs, and of course all new wiring was copper going back to the breaker box.
The Wiki link was both interesting and instructive. It gives particular warning to galvanic reaction with aluminium and stainless steel. I'm sure the maggie frame is steel, but not stainless. I'll really keep a close eye on it for the next month or two.

[The_Raven]

Howyadoin,

I've been bitten by the regulator bug on my '94.  I was wondering if anyone has any tips on removing the battery box to get to the regulator, or is it such a PITA that I should take the wrench 'n socket through the top approach that I've seen?

Thanks!

-Mark

[TLRam1]

Mark,

I think you have to remove the Rear Fender and it slides out the back.

[Sledge Hammer]

As they allude to there, aluminum will "cold flow" -- meaning that extended time and pressure will cause it to "squish" out of shape.  Probably not a problem with the strap you have, but back in the 60's, during a copper shortage, a lot of homes in the Dallas area were wired with aluminum -- and caused a lot of fires due to cold flow and dissimilar metals corrosion issues.

In that application, would the problem be "cold-flowing" or electromigration? In discussions I had with engineers at Siemens, they told me that research had shown that electromigration was the culprit, which was why the problem was most prevalent at high-current/high-voltage connections in panels and at junction boxes for heavy loads such as A/C. All conductors carrying current are susceptible to forming voids in the material over time, with current density being the greatest contributor to the problem and energy per charge carrier being another significant factor. (Note that this isn't the same mechanism for failure as resistive heating effects, so just keeping the conductor cold doesn't help much.) Aluminum and its alloys are simply more susceptible to this effect than copper.

[lragan]

The screws and plates on the switches and outlets were supposedly "safe" for use with aluminum wire, meaning that someone somewhere had determined that the dissimilar metals would not cause electrolysis issues.   Many of the connections were loose, as the wire had deformed and relaxed the pressure between the screw, wire, and back plate the screw went into.  I have been told that it was a "cold flow" problem, and finding these contacts loose seemed to confirm that.  As I recall, a temporary fix was to open every box and tighten all the screws, which I chose not to do.  Instead, I installed copper "pigtails" consisting of a short (3" or so) length of wire which went under the screw, and a wire nut with some conductive bearing grease in the wire nut.

The really large connections, as from the power meter to the distribution panel, could not be fixed in this manner.  For these, I opened the panel every three or four years and retightened the screws -- and I was always able to tighten them, even though I had pulled them as tight as I could get at the last inspection. 

This problem did not seem to show up until ~10-15 years after initial installation.

I suspect that alloys which reduce this cold flow property are less conductive, and certainly less malleable, making them unattractive for wire.

Electromigration is another issue.  We encounter that in very thin metals, especially aluminum, in semiconductor interconnects.  Inspection with a scanning electron microscope show pieces of aluminum that shear out and pop up from the surface, little "hillocks".  It is dependent on temperature (exponentially), current (linearly) and time (linearly).  Copper interconnect of the same thickness are much more robust.  I have never seen the problem in copper interconnects.  In fact, just a bit of copper (like 3%) in the aluminum sputtering operation drastically reduce the phenomenon in aluminum interconnects.  Almost all aluminum interconnect (and most interconnect in integrated circuits is still aluminum) use 3% copper.

I suppose one could settle the question if there was a box that was never used -- such as a socket behind a dresser for 15 years or so.  I am sure I had those, but this many years after the fact, I cannot remember which connections were looser.  All I can say is that the wire visibily flattened to a large degree -- much more so than copper.

[lragan]

Well, I fiinally got around to taking temperature data today. First, an explanation of the procedure.  I obtained a cheap ($22) DMM at the local Harbor Freight tools which has a temperature probe included, consisting of a type K thermocouple.  Not terribly well calibrated, as when it is immersed in ice water, it shows 2^oC. (Henceforth noted at 2C, etc., to avoid finding the superscript button every time.)  Nonetheless, it will give more accurate results than an infrared sensor, which is based on an emisivity assumption. 

After about 30 minutes trying to tighten one of the mounting screws over the thermocouple, I gave up and mounted it against a fin using the lock washer shown in the first photo.   (yes, I know it is out of focus, but there isn't much room between the regulator and the pavement!)

I ran the wire up the right side of the bike, as shown in the second photo, to stay away from the chain on the left.

Then I mounted the DMM in the map compartment of my tank bag, as shown in the last photo.

Today, I ran the test.

Before startup -- 29C
After 3 min at 3000 rpm or less -- 57C
Riding at 55 mph in 5th gear stabilizes at 66C
Turned on 100 watts riding lights -- stabilizes at 65C
When I slowed down to street traffic speed -- 73C
Parked at the bank -- just stopped -- 76C
After 3 minutes or so -- returned to bike - 57C

I rode in and out of traffic, at various speeds and gears, keeping engine generally 3000-5000 rpm, and never saw temps above 75C while in motion.

Conclusions:
1) My assertion that running extras reduces the dissipation may be correct, but just barely.  Other factors, including bike speed and engine speed have greater effect.
2) While case temperature in the range of 75C is troublesome, it is not disastrous.  Depending on how effectively the heat is removed from the circuitry to the case, reasonable junction temperatures are possible, even likely.  If we were running case temperatures nearer 100C, we would probably see a much higher failure rate.

It is not clear how orientation of the fins would effect the outcome, but it seems to me, given the short time it takes the case temperature to stabilize, and how quickly it is affected by motion of the bike, even orientation to the wind at lower speeds, it doesn't seem to me to be a change that would make much difference.

Thermal rise is a linear phenomenon, so on a hotter day, the difference between 29C (84F) and, say, 38C (100F) would add directly to the above readings.

One could argue that I would get a higher reading just under the bolts, as the edges of the fins are already cooler than the body of the case.  This is most certainly true, but I doubt the difference is more than 3 - 4 degrees C.

So, where does all this leave us in the struggle to keep the regulators operational and the batteries from boiling out?

I think we need either a regulator with a more efficient and robust design, or a method to get a good thermal mount of the units we have to the frame of the bike.

It is not clear to me that the development effort to design a more efficient and robust regulator is justified from the size of the available market.  If some of you mechanical types could design a way to get a good hard mount to the frame, that seems the easiest approach.  But then, the easiest approach is always to get someone else to do the work, eh??



[attachment deleted by admin]

[lragan]

Howyadoin,

I've been bitten by the regulator bug on my '94.  I was wondering if anyone has any tips on removing the battery box to get to the regulator, or is it such a PITA that I should take the wrench 'n socket through the top approach that I've seen?

Thanks!

-Mark

Mark, I didn't take mine out, but was able to get to the mounting nuts from the bottom with a short 3/8" ratchet and a 10mm socket.   Don't know if I could have extracted the unit, but it looks like I could take it out over the swing arm after detaching the leads. 

Good luck with it.

[TLRam1]

Converted to Fahrenheit.

Before startup --------------------------------29C = 84.2 F
After 3 min at 3000 rpm or less ------------- 57C = 134.6 F
Riding at 55 mph in 5th gear stabilizes at----66C = 150.8 F
Turned on 100 watts riding lights stabilized- 65C = 149.0 F
When I slowed down to street traffic speed --73C = 163.4 F
Parked at the bank -- just stopped ----------- 76C = 168.8 F
After 3 minutes or so -- returned to bike ----- 57C = 134.6 F