This is a DC circuit. Fuse F+ on left and F- on right are supposed to carry equal currents.

Initially as shown ion slide 1 we found F- was in the neighborhood 20-30C hotter than F+ (depending on where you do your comparison... I realize we didn't choose the best points for comparision / reference on attached images).

The connection was disassembled, cleaned, and retightened.

Then in slide 2, the difference is smaller, but F- still hotter than F+.

Better visual images are shown on slides 3, 4, 5

The plan is to disassemble again and this time reassemble with new hardware. There is not enough slack to relug the lead (and it doesn't particularly appear there is heat coming from the lug crimp).

Then we will do another post-repair test.

At that time I will request to check current to verify they are the same (they should be the same by circuit analysis... no other path for return current). I will also ask to do a wider-angle view of the lead so we can check for tailing off which is not obvious in the above pictures.

What does the source of the heat look like to you?
Any other suggestions for repairs or things we should check?

PPT.ppt (3,878 Kb, 115 downloads)

Pete,

I look like only the connection labeled "another" has no-ox between what appears to be aluminum lugs and copper bar. If the second set of images from F+ F- were after reterm, why wasn't the copper bar burnished, and why isn't there no-ox applied?

Rich Wurzbach
ASNT PdM Level III
Maintenance Reliability Group, LLC

Pete,

Another thing I would look at is the flatness of the connection. The resistance across that connection is a function of the surface area of contact between the bar and the lug. If either mating surface is curved or has a high-spot, it will affect the total surface area for current transmission, even at proper torque. Also, if the studs are crooked, pulling up on the nut can cause the surface to bow a bit and reduce the contact area. But first I'd look at the burnishing job on both mating surfaces and on the use of no-ox.

Rich

Thanks Rich. The story has grown by many twists twists and turns. We did changeout the hardware (bolt, washers) on this one lug – no significant change.

=========================
VISUAL INSPECTION OF LUGS
==============================

Let me start with the visual inspection:
Attached slide 1 shows 5 lugs from left to right: F+, F-, X1, X2, X3.

F+ and F- are the dc output of this voltage regulating circuit. They do in fact carry the same current as we verified by measurement. F- running hotter than F+.

X1, X2, X3 are the three-phase ac inputs to this regulated dc module. Not particularly important.

One thing you can notice by looking at all five together is that F- has the most bend. It has the least slack and that is tugging the lug toward the left.

There is a heat-shrink tube over the lead/lug interface that is supposed to provide mechanical support at this point of stress concentration. You'll notice it is completely overlapping the top circumference of the lug on all 5 lugs except for F-, where it is starting to slip off (more detail on slide 2). Perhaps this mechanical force is causing our temperautre difference (?).

Now some other visual observations which I don't think are as important, but you be the judge:
1 – As Rich said the greasey compound is not restored on F+ and F- as was originally present and still present on X1-X3 (I did notice that, which is why I included the photo). That compound has been determined to not be needed. Same condition on both F+ and F- so does not seem a contributor to temperature rise.
2 – The silver plating is partially removed from F+ and F- during the recent excercize of removing legs, cleaning with scotchbrite and reassembling. (similar to burnishing that Rich had suggestted). We should have replenished that silverplating using coolamp (we intend to go back and correct that condition). However, again it is very similar on F+ and F- and does not appear to explain the temperature rise.
3 – The hardware is different on F-. That is because we recently replaced it.
4 – There is very slight brownish discoloration on the left side of lug F- visible in slide 2.

There is more to come. Notice while you're looking at these slides that the lugs are mounted to busbars which are mounted to heatsinks (aluminum I think). Those heat sinks do have very low emissivity. Down below are other components mounted to these same heatsinks which I will describe in more detail in later posts.

This message has been edited. Last edited by: electricpete,

post2.ppt (1,980 Kb, 40 downloads)

OK, the next part of the story.

==========================
HEAT SINKS AND DIODES
=========================

Slide 1 shows an overveiw of what's below F+ and F-. These lugs are bolted to bus bars which are bolted to heat sinks. The heat sinks for a part of the electrical circuit.

Mounted to the heat sink below F- are diodes CR6, CR4, CR2.
Mounted to the heat sink below F+ are diodes CR5, CR3, CR1.

The arrangement is SIMILAR TO (but not exactly) a full wave bridge rectifier. The ac comes through the diodes to arrive at the dc output (F+ and F-).

Now the thermography slides 2 and 3 shows CR6, CR4, CR2 hotter than CR5, CR3, CR1. At first this raised a concern for the diodes. However check of the current reveals that CR6, CR4, CR2 carry twice the current of CR5, CR3, CR1 (even though the outputs F+ and F-) carry the same current. The reason has to do with the "similar but not exactly" part. In addition to those diodes shown here are SCR's in another part of the circuit that regulate the voltage. An engineer familiar with the circuit said the current measurements we saw were exactly as expected. He said something like: the negative diodes also carry a current which is shunted back to the power supply (or something like that). I would like to look at the circuit to understand that better, but for now I'll take his word for it.

So the question is: Are F+ and F- at different temperatures because of a high resistance at F- termination, or because they are being heated from below by the hotter diodes CR6, CR4, CR2 (running 85-90C which is hotter than the termination F- appears).

Either way we will go back and do some rework on the lugs based on the visual findings. However, this question has some impact on how we view the urgency of that repair.

I did note with my carmera that the right side (negative) heat sink looking at cavities appears a few degrees hotter than the left side (positive) heat sink when looking in cavities. But it was only a few degrees, I could not find when comparing cavities in heatsinks the 20+ degree C difference we're seeing at the terminations F- and F+.

I think it is shown best in slide 4. In the center of the slide is F- and associated heat sink. To the left is F+ and associated heat sink. The heat sinks are highly reflective (low emissivity). I think the yellow color in both slides represents the actual temperature seen in the cavities looking down the crevices. The green (on left) and red (on right) represent the reflected temperatures (heavily influenced by the hot lug/bus on the right). We can confirm taht the green/left and red/right are the reflected temperature by looking at the solid diagonal portion of the heat sink which is not ribbeed. That diagonal rib is clearly reflecting and it shows the green (left) and red (right). So again the yellow is the cavity (true) temperature for both heat sinks. The green is the relfected on the left and red is reflected on the right.

I have added a boatload additional images after that. Sorry - couldnt' get any good overveiw shots due to very close quarters. Will probably overload you with data. But perhaps something in there may shed some light.

You may notice slide 20. There is a horizontal insulating strip supporting termination point in the middle, bolted to F- heat sink on right and F+ heat sink on left. The bolting point looks much hotter for F- than for F+. Similar pattern evident in a few other slides as well. At first glance it suggest the F- heatsink is much hotter. But I believe it is very localized in the area of the diodes, and all of these mounting strips are very near the diodes.

What do you think? The temperature difference at F+/F- is caused by connection problem at F- or by heat conducted from the known hotter diodes attached below F-?

This message has been edited. Last edited by: electricpete,

Post3small.ppt (1,506 Kb, 51 downloads)

No response? I would be interested to hear any opinion, even if only a gut feel.

Is it
A - the discussion of the physical layout and thermal observations is too convoluted to follow?
or
B - the discussion of the physical layout and thermal indications are is clear, but the symptoms don't point strongly toward one or the other scenario ?

EPete

Silly question coming up, when you alter the span on your image, what is the last thing to disapear, i.e the hottest spot?

Epete,

You mentioned that you did not feel that the lug crimp was the source due to no localized heating pattern. From what I can see it is a low emissivity surface and would therefore be very difficult to isolate the crimp point. How about temporarily hitting the spot with some flat paint? It would definitely improve contrast. How much current flows through those DC connections? A slightest increase in resistance in that lug crimp could easily add up to 10s or 100s of watts of dissipation.

How about temporarily removing the connection and replacing it with a jumper wire which would then be connected at the next available termination point. I think one of the images shows a terminal strip possibly before the wires leave the electrical cabinet.

Is it realistic at some point to remove the heat shrink tubing which I doubt provides much mechanical support anyway? Maybe it will slide back to reveal if there is any corrosion down inside the lug crimp. That "grease" if allowed to wick its oils down into the lug crimp might have initiated a chemical reaction that would not easily be seen.

Anyway try bypassing that piece of cable temporarily to verify the cable as the source.

Another idea would be to install a foot long piece of cable between the current Lug to Busbar connection. This would really help to isolate the heat source as to whether it is from the heat sink or the lug crimp.

Just some ideas....

Dan

EP-
My hesitation in responding is that this is a complex situation, difficult to get one's hands around, even with you excellent documentation; so it is really BOTH "A" and "B!"

I agree with Dan's suggestions EXCEPT would not recommend using spray paint on energized components. Clearly you have many components with different emissivities and widely varying reflected backgrounds. Other testing is probably indicated.

Indications seem to point to heat in F- that then transfers by conduction to other parts of the equipment with high enough emissivity that the signatures can be seen, but, again, without being there, I'd rather not say. All I can be certain about from this vantage point is that the heating will follow the path of least (thermal) resistance.

Do keep us informed please!

I for one find this difficult to follow.

If you suspect the diodes are at different temperatures, why dont you measure them?
Why should the diodes be at different temperatures, if the load is (as you suggest)the same?

Post repair failures are more common than some would believe. If the connection is hot, it is loose, dirty, corroded, burned, oxidized, or annealed. The only way to correct this is to get back to new shiny metal surfaces and proper contact pressure. In the case of cables and wires this may require the conductors be cut back or replaced. If you take the shortcut and just tighten then you should do a follow up inspection to see if it helped (as you have done, but it suggests it has not really helped). You should clean or cut back the cable conductors and clean or replace the lugs. When cleaning the cable, don't forget the individual strands. If there is resistance between strands then the outer layers will carry all of the current and will be hot from current density.

Also, looking at the temperature in your original image, I would suggest it has reached a temperature over the insulation rating (assuming it to be standard PVC). This means the cable has degraded, and has to be replaced, not just cleaned and retorqued. Once this failure begins it cannot be reversed, and will cause the copper itself to deteriorate. This is what I believe has happened.

To cure this, I suggest you replace the cable, and the lugs. Forget about cleaning it, just replace it.

Thanks for all the comments. I do appreciate your patience to read through all that info wwhich was not well organized (My excuse is that there have been several twists and turns in reviewing the problem).

Let me clarify the current readings:

F+ 148.7 A dc
Diodes feeding F+
CR1 62.7Aac 48.2Adc
CR3 64.6Aac 50.5Adc
CR5 66.2Aac 50.5Adc

F- 148.6Adc
Diodes feeding F-
CR2 139.2Aac 111.2 Adc
CR4 132.6Aac 104.3 Adc
CR6 141.5Aac 112.5 Adc

SCR's connected to F- bus (upstream of F-)
SCR1 82.9Aac 59.2Adc
SCR2 88.9Aac 65.9Adc
SCR3 85.6Aac 63.1Adc

So, the bottom line: the currents through F+ and F- are balanced. The currents through the diodes are not balanced. And these are the expected current readings As I mentioned before, the circuit looks like a full wave rectifier except that in addition to diodes connected to the negative bus, there are also SCR's connected to the negative bus (not shown in the images) associated with the voltage regulating function. Schematic is attached showing F- and F+ terminals at the top near the left. If you can figure it out, please explain it to me. Note that the positive diode dc readings to in fact sum up to the F+ dc reading. The negative diode dc readings are higher by an amount which is approximately the same as the current in the SCR's.

quote:

when you alter the span on your image, what is the last thing to disapear,

I did not try that technique. But I have the raw-data images (Mikron SIT files) saved and can manipulate it adjust the span to check for the precise location that indicates hottest (will try it if I get a chance). But the emissivity complicates the picture.

quote:

If you suspect the diodes are at different temperatures, why dont you measure them?

The heat sinks are energized to the voltage of F+ and F- (as you can tell from the photo's) and in fact form part of the circuit current path. The diodes are recessed within the heat sinks. I don't know any safe easy way to do the measurement. I am reasonably sure that the diodes CR2/CR4/CR6 are hotter than CR1/CR3/CR5 as expected based on the higher loading. I am not concerned about the temperatures of the diodes because they have a temperautre rating of 170C. I am trying to figure out if the known difference in diode temperautres is causing the observed difference in lug temperatures.

I am still thinking about some of the other comments. We will be replacing the cable, simply because there is not enough slack to relug. If we had slack, we would cut of the lug, strip back the insulation, inspect strands and relug (It would never occur to me to recommend replacing a whole cable because the termination had been overheated... should it?).

FIGURE1.ppt (886 Kb, 19 downloads)

Wow John, I guess I didn't think I needed to mention that one should not spray paint on an energized circuit. But they have a saying about "ASSuming" and it should be headed I suppose. Anyway several of the tests had already indicated that power had been de-energized and I guess I was just going with that.

Please folks... if you are going to attempt to modify the emissivity of a surface then by all means "Lock it Out and Verify the circuit is de-energized"

I took a very quick look at the schematic. It looks like a typical variable output bridge with a "little wierd stuff" worked in. I would have suggested that CR1, 2, 5 and SCR 1, 2, 3 are forming the bridge and should carry approximately the same current. The diodes CR2, 4, and 6 are in the typical snubbing position to assist with the commutation of the SCRs. The "wierd magic" is why there is a CT between the SCRs and the Snubbers.

Anyway you peeked my interest and I have saved the image and your voltage/current readings. I will try to sit down in the near future and do the math on it.

Good Luck,

DanS

Hello, everyone !
I’m electrician and like Dans, I took quick look at your schematic and current readings. Pete, for me, it looks like full wave bridge with six diodes and CT for current readings for main control regulator with 3 SCR *weird *connected. Their cathodes are connected with negative bus and when they are opened make some kind of short circuit with diodes CR2, 4 and 6. For example, the current through diode CR2/4/6 is sum of currents through CR1/3/5 and SCR1/2/3 and I see that in your current readings but, if the scheme is right, current through SCR1/SCR2/SCR3 doesn’t goes through field of motor!
This think is strange for me because, if I’m right, this regulator has big electric loses and I didn’t see something similar in the past.
Good hunting!

This message has been edited. Last edited by: Panta,

quote:

Wow John, I guess I didn't think I needed to mention that one should not spray paint on an energized circuit. But they have a saying about "ASSuming" and it should be headed I suppose. Anyway several of the tests had already indicated that power had been de-energized and I guess I was just going with that.

Please folks... if you are going to attempt to modify the emissivity of a surface then by all means "Lock it Out and Verify the circuit is de-energized"

Well, yes, most of us would never try something like this but I know of one company that advocates using spray paint in a hot stick live. Another puts paper stickers on hot components using a pair of chop sticks.

We can only hope the Darwin Theory is really hard a work here!

One word of caution, if you are using paint, either spray or liquid, to increase the emissivity, beware the stuff moving up into the threads of connecting bolts/nuts by capillary action. One of our customers had this problem.

Of course, there are no "standards" for doing any of this, so check it out with your safety team and use common sense. The attached shows the use of Glyptol painted on prior to start up in a data center; a "find" shortly after energizing probably would have gone undetected without it.

We had an opportunity to survey an identical sister panel today. Currents were roughly the same pattern as the first panel. Thermal pattern was roughly the same as the first panel(including twice the current through the F- diodes as through the F+ diodes). Temperatures at the F+ and F- had a similar pattern: F- was about 23C more than F+.

So, unless there is a bizarre coincidence of two bad F- connections (and I don't much believe in coincidences), I'd say the theory of heat conduction to the F- lug from the hot diodes below (as opposed to loose connection) has been proven.

It leaves one question in my mind, how do we interpret my slide 4 of my 02 March 2008 11:07 PM post? It seems to show both heat sinks of similar temperature if we look at the cavity areas.

I did finally figure out more about how the circuit works and why those negative diodes CR2/4/6 carry more current than the positive diodes CR1/3/5. The regulating feature is as follows: there are inductive impedances in series with the input ac power X1, X2, X3 (not shown on this drawing). For max voltage output, all the SCR's are not gated (remain open-circuited) and the circuit acts like a full bridge rectifier. To bring the F+/F- output dc voltage down, the SCR's are closed during portions of the ac cycle. When those SCR's are closed it causes a short across the ac inputs (for portion of a cycle), causing hi current to flow through the series input impedances, causing an ac voltage drop at the input to the rectifier, causing an output voltage drop.

An example is shown in the yellow highlighted path attached. It shows that when SCR3 is gated closed, it short circuits that ac inputs at X2 and X3 for a portion of the cycle (while SCR3 is gated and while X3 voltage above X2). That short circuit causes high current and voltage drop accross the external impedances and decreases the ac voltage seen at the bridge inputs X2 and X3. But more importantly the current in diode CR4 goes up during this portion of the cycle when X3 voltage is above X2 (the yellow path). Likewise, with SCR3 gated, the current in diode CR2 goes up when X3 voltage is above X1 (the blue path).

Likewise, gating SCR2 can increase current in diode CR2 and CR6, while gating SCR1 can increase current in diodes CR4 and CR6. All of the extra current associated with gating SCR1,2,3 to short circuit the inputs flows through diodes CR2/4/6, causing them to have higher currents thant CR1/3/5.

FIGURE1_markedup.ppt (869 Kb, 13 downloads)

quote:

So, unless there is a bizarre coincidence of two bad F- connections (and I don't much believe in coincidences), I'd say the theory of heat conduction to the F- lug from the hot diodes below (as opposed to loose connection) has been proven.

It does seem strange that in the photos of the heatsink that there is not much heat being measured from the cavities of the heatsink. The aluminium heatsink is surely of low emissivity but typically a hot heatsink will offer a peek of its true temperature down in the cavities between the fins.

Just a thought...

DanS

quote:

quote:
Well, yes, most of us would never try something like this but I know of one company that advocates using spray paint in a hot stick live. Another puts paper stickers on hot components using a pair of chop sticks.

We can only hope the Darwin Theory is really hard at work here!


John Snell
The Snell Group

This has to be on of the most repugnant comments that I have ever seen. Injuries due to electrical accidents are some of the most horrific that anyone could ever imagine. They are not the stuff of humor.

Anyone who would suggest that others should sustain an electrical injury or be sorted out by the Darwin Process should be ashamed of himself.


Jim Seffrin, Director
Infraspection Institute

This message has been edited. Last edited by: Jim Seffrin,

Thank you, Jim. Clearly my intent was not to wish harm on anyone. I appreciate your comments.

I think we have all known John long enough to recognize the intention was purely humour, nothing more sinister.

Another update - we analysed a third panel last night - the results very similar to the previous two - this time F- around 20C hotter than F+. That pretty well seals the deal for me. I don't believe three panels all happen to have a bad F- connection by coincidence (and those second two panels didn't have the same lead bending as the first). And the heat conduction explanation is perfectly logical.

I think it makes an interesting case study of a false alarm. It looks like a hot connection based on temperature difference between terminations carrying supply and return current in a dc circuit. Current measurement confirms the current in both F- and F+ cables is the same. Only through further investigation can we see there is an assymetry in current flow within negative/positive circuits in a portion of the circuits that doesn't include through those F+ and F- lugs up top, but apparently the heat makes its way up there by conductive heat transfer. The heat conduction is very effectively obscured by the low emissivity heat sink.

Who'd of thunk it!

This message has been edited. Last edited by: electricpete,

have you ever check the hormonics effect on conductor?
i am not sure but i am interested to know wheather harmonics affect in such type of condition , if yes then How?( rectifier circuits )

regards