Serving the automotive and truck parts remanufacturing industry since 1941

about APRA

join APRA

contact APRA

online Forums

shop APRA

APRA home

Serving the automotive and truck parts remanufacturing industry since 1941

Frameless JavaScript Tree
Driving Quality


Things Are Not Always as They Appear: Diode Trio Reliability
by Bill Bowman, Technical Consultant

Some alternators use an integrated diode trio while others use three individual diodes to supply current to the field. Have you ever wondered which one is the better architecture? The answer might surprise you. This column has looked at a number of commonly held beliefs in the past and found some to be lacking. In this edition, we will examine a premise whose conclusion might at first seem counterintuitive, or at least it was to me. We will utilize the same analysis technique employed in the past, that is to use logic rather than strong mathematic arguments to look at the “big picture.” Let’s examine the premise that the superior trio configuration is implemented with discrete diodes rather than the integrated diode trio.

In both configurations the anodes of the diodes are connected to the stator windings and the cathodes are connected to the brush to supply current to the rotor. These three diodes switch the currents from the appropriate stator winding into the field. In each configuration, the diode currents are approximately the same. The only difference is how the diodes are packaged (integrated versus discrete).

A previous article reviewed the rationale behind the decision to eliminate the trio in the CS alternators (“Why Are There So Many Different Voltage Regulators? July, 2006 “Driving Quality”). To summarize, in the days before the CS, a significant contributor to SI alternator warranty cost was diode trio failure. The thinking was: “when the trio goes, so will all its warranty.”

After the manufacturing defects were removed from warranty, the primary cause of trio failure was electrical overstress due to excess heat. For that reason we will restrict this analysis and discuss only the variables that affect the diode temperature.

If you look at the design of the integrated trio, it is not hard to figure out why it gets so hot. The diodes and the heatsink on which they are mounted are potted with plastic, epoxy, or some other molding compound. In other words the diodes are insulated, thereby making it difficult for the heat generated in the diodes to get out.

Thermal conductivity reflects the ability of a material to transport heat. In the case of the integrated diode trio, the heatsink will carry the heat from the silicon power diodes through the frame to be ultimately passed on to the air. Certain materials, such as copper, have high thermal conductivity. Steel, on the other hand is a poor thermal conductor. In fact, the thermal conductivity of copper is nearly six times greater than steel.

If you want to keep the silicon chips as cool as possible, use material with high thermal conductivity to take the heat away from them. Did you ever wonder why the Nippondenso rectifiers are so reliable? One of the reasons is they use massive copper heatsinks which “suck” the heat out of the diodes and pass it quickly to the outside air.

Next to the engineering lab and the production floor, one of my favorite places to work is in the kitchen. So let’s go there to see what we can learn about thermal conductivity.

When you blacken fish, you want an even high temperature in your pan that doesn’t drop when the meat is put on it. That way you get the outside seared quickly with the rest of the meat uniformly cooked. The temperature of a copper clad skillet will drop quickly when fish is placed on it, efficiently sucking the heat from the surface. The temperature of a cast iron skillet is slow to drop because of the high thermal resistance (low thermal conductivity) of the material.

Have you ever seen a copper coffee cup? Because copper has such a high thermal conductivity, it will quickly remove the heat from the coffee. Most coffee cups are ceramic which is a poor thermal conductor that will keep the coffee hotter longer. Thermos bottles keep the contents hot or cold because they use air to insulate the contents from the outside. Air has an extremely low thermal conductivity. The reason down jackets keep the wearer warm is the feathers hold so much of the thermally resistive air.

Higher thermal conductivity of the heat sink means the diodes operate at a lower temperature, which leads to longer field life. Now, let’s move from cooking to the clean-up phase to develop a model of another variable that relates to diode temperature. Let’s build a model based upon the kitchen sink to better understand the thermal heat sink.

Like with many models, they do not have to be 100% technically correct to facilitate an understanding. Given that, let’s draw an analogy between the water (in the kitchen sink) and the heat (in the diodes). Both flow. The sink itself is a storage vessel for the water and the drainpipe carries the water through the drain. In a trio, the heat is generated in the diodes by the flow of current and stored in the heatsink waiting to be evacuated into the surrounding air.

When washing the dishes, the flow of water is variable. During the operation of the trio, the current is variable. If the water flows into the sink faster than it can be carried away through the drainpipe, the sink will overflow making a mess in the kitchen. If the heat builds up faster in the diodes than it can be evacuated, it will overflow making a mess, catastrophic failure.

There are several things we can do to prevent the sink from overflowing. One is to use a bigger sink. Another is to use less water. Still another is to use a bigger drainpipe.

With the trio, we cannot really use less current or the alternator output will suffer. We can use a bigger sink, but that has some inherent problems. The most efficient approach is to use a wider drainpipe.

With the diode trio a bigger kitchen sink is analogous to using bigger silicon diodes. They will handle more current and spread the heat out over a larger area than a smaller diode. Silicon is expensive, so we need to be careful with going to bigger diodes because of the cost implications.

A larger diameter drainpipe is analogous to using a higher thermal conductivity material to remove the heat from the diode. In the case of the kitchen sink, it would be a more practical solution to use a bigger drainpipe as opposed to a much larger sink. The same thing is true for a trio. You can use much larger diodes or go with a better heatsink.

On the diode trio, the silicon chips are mounted on a plated steel frame. If the trio function is implemented using discrete diodes the mounting structure is different. In a discrete axial leaded diode, the silicon chip is mounted between two large copper slugs at the ends of the copper wires—so the heat path is from the silicon to the slugs and out through the highly conductive copper wires to the outside air.

When I performed this analysis I looked at three axial leaded 5-amp diodes. Comparing the different diodes, the silicon chips in the integrated trio were 50% larger than with the discrete solution. Using the kitchen analogy, the sink is 50% bigger with the integrated trio. But the drainpipe carrying the heat away from the silicon chips is almost six times larger on the discrete diode trio due to the higher thermal conductivity of copper versus steel.

Note: Thermal conductivity is a function of the crossectional area of the heatsink. More area yields a higher thermal conductivity. The crossectional area of the steel frame was 20% larger than that of the wires connected to the discrete diodes examined. But the thermal conductivity of copper is many times that of steel, so the slight difference in crossectional area makes a minimal impact on the overall thermal conductivity.

Therefore, due to the fact that the drainpipe carrying the heat away from the discrete diodes is copper and is steel in the integrated trio, the discrete solution is thermally superior. If that wasn’t enough, there is another more subtle factor working that gives the discrete solution another edge over the integrated trio thermally.

See Figure 1 for a pictorial of the trio. From the diagram you can see that the heat generators (silicon chips) are insulated (potting compound) from the outside air. So most of the heat must flow through the frame to get to the air. Figure 2 shows a blow up of one of the diodes inside the trio. Figure 3 shows a blow up of an axial leaded diode used to implement 1/3 of the trio function.
 

Figure 1: Integrated Diode Trio

 

Figure 2: Trio Diode

 

 Figure 3: Discrete Diode

The junction of the diode is not in the center of the silicon chip, rather it is located near the topside surface. So with the trio, the heat has to travel from the junction through the silicon then the solder onto the steel frame. The thermal conductivity of silicon is approximately 1/3 that of copper. In other words, in the integrated trio, the heat has to travel first through the bulk of the silicon in order to get to the poor thermal conductor (steel) to the outside.

Now look at the axial leaded diode. Positioned on both sides of the silicon chip are large copper slugs. The slug closest to the junction pulls heat directly from the junction. The rest of the heat is pulled through the silicon to the copper slug on the bottom.

In other words, with the integrated solution the heat is predominately pulled from one side of the silicon by a poor thermal conductor. The discrete solution pulls heat from both sides with an outstanding thermal conductor. So from a thermal perspective, the trio implemented with three individual diodes is superior to the standard integrated diode trio.

Following are some notes to consider regarding this subject. Some manufacturers mount discrete diodes inside the trio onto the steel lead frame. This is not a good solution, because the heat still has to be transported through the steel frame and the diode chip inside the package is generally smaller than the chips mounted directly to the frame.

Since the primary cause of failure with trios is excessive heat, this article discusses only the thermal considerations of the different architectures. Suffice to say that whatever approach is used, discrete or integrated; the proper mechanical design rules need to be followed in either configuration to make a reliable trio. (e.g., sufficient stress relief, proper soldering/welding process, etc). If you do decide to implement the discrete solution, ensure that you have sufficient stress relief in the diode leads.

With the discrete solution, there are more interconnects. A trio has 4 mechanical interconnects while the discrete version has 6. More interconnects, by definition, means lower reliability. But the cooler running diodes will yield a more reliable solution even though there are more interconnects with the integrated solution.

The question regarding the trio came from a client. He gave me permission to publish the findings in this column. Different editions of Driving Quality are aimed at different audiences. If you have suggestions or questions that will appeal to more readers, please let me know and I will try to address them in a future column.


Diverse Enterprises is responsible for technical consulting, development of electrical test equipment, etc. Specific problem solving is done by PROBLEMETRICSSM. … Problem Solving by the Numbers. You can reach Bill Bowman at bowman.we@att.net


APRA Global Connection / January 2007



Additional APRA Supported Websites

utoBulletins.com
it's as easy as your A, B, C's


AutoElectricPro.com
 


Heavy Duty Remanufacturers Group

 Automotive Parts Rebuilders Association

Note: This site uses PDF files, which require Adobe's Acrobat Reader for viewing. Click the Adobe logo to download this free browser plug-in.

Automotive Parts Remanufacturers Association | 4215 Lafayette Center Drive, Suite 3 | Chantilly, VA 20151-1243
703-968-2772
| Fax: 703-968-2878 | mail@apra.org | www.apra.org