Archive of Lee's Wiser Moments

In my meanderings in the EV space, I have met two men who have what I would refer to as 'guru' status. One is a mechanical guru, a practical mechanic with real world experience. The other - well, the other is Lee, and this is his page.

This is where I stick postings and emails from Lee that I think I, and others, can benefit from. I have obtained Lee's permission to do this (I think) so hopefully I won't be being sued any time soon.

Anyway, no words below are mine. They are the words of Lee Hart, EV guru.


Lee on Battery Maintenance

Let's anthromorphize a bit, and consider lead-acid batteries as alive;
like the family dog.

1. They need exercise; it's good for them. You get the longest life
   when they are worked to about 50% of their capacity at moderate
loads.
   After they have been loafing for weeks, you will notice a distinct
   improvement just from giving them moderate exercise. 

2. But don't work 'em till they drop! If you drive an EV until it
   barely moves, the batteries are having a near-death experience!
   This is outright battricide, and a leading cause of early death.

3. They need to be be fed regularly (charged). Feed as soon as possible
   after a workout; they don't like to sit around starving after use.
   Batteries left sitting for days in an undercharged state develop a
   condition called sulfation.

4. Don't overfeed, or they get fat and have cumulative health problems
   and so die early. Chronic overcharging is a major cause of early
   death.

5. Don't underfeed, or they can starve to death. Chronic underfeeding
   also leads to a weak sickly battery and an early death.

6. Batteries can sit unused for months (hibernate) without needing to
   be fed. You don't need to put them on a trickle charger; just be
   sure to feed them occasionally so they stay near full charge.

7. They need fresh, clean water occasionally. Sealed batteries have a
   built-in watering system, but flooded batteries do not. Be sure to
   check water levels, and fill with distilled water as needed (dirty
   water poisons them!)

8. They need to be kept at reasonable temperatures, that you would find
   comfortable. Not too hot, and not too cold. Lead-acid batteries are
   "cold-blooded", so the lower the temperature, the slower they get.
   Likewise, they can't "sweat", so high temperatures cook 'em to death.

9. Batteries can't talk. They won't whine when they're hungry, or cry
   when you hurt them. You have to check their state of health with
   instruments, like voltmeters ammeters and hydrometers.

10. There are different "breeds" of batteries, each with its own good
   and bad points. Slow plodding workhorse floodeds, but long lived.
   Racehorse AGMs that are fast and powerful, but short lived. Using
   the wrong breed of battery for the application, or unrealistic 
   expectations leads to disappointing results.

11. And some is just the "luck of the draw". For no obvious reason,
   identical batteries in the same vehicle will have some die young, and
   some seem to live forever.

The usual reason you see used EVs that say "needs batteries" is because
the previous owner treated the batteries cruelly. Whether by ignorance
or laziness, some or all of the above guidelines were violated. But
batteries are replaceable, and it usually means you can get the EV
"cheap".

But such problems can be cured. A little detective work to fix the
problems, and then some tender loving care will go a long way toward
getting the longest life possible on the next set of batteries.

Lee explains the tao of the fuse

A fuse is just a resistor. It contains a fat piece of metal at each end,
with a skinny piece of metal between them. It "blows" because the skinny
piece melts when the power dissipated in it heats it up to its melting
point. 

The voltage rating is defined by how long the skinny piece is. Low
voltage fuses have a very short skinny part, to keep the resistance down
and so the voltage drop low. But once it has blown, an arc can continue
across the big pieces.

Your fuse's 500v rating is probably for AC. AC arcs automatically stop
120 times a second as the AC line passes through zero, but a DC arc
won't stop until the gap is very large. Unless you have manufacturer's
data to the contrary, assume the DC rating is 1/4 of the AC voltage
rating.

Using a fuse at less than rated voltage has no effect on the current or
speed at which it blows. However, the higher the rated voltage, the
longer the skinny part, and so the higher the resistance of the fuse.

Since it's just a piece of metal, the actual current at which it blows
depends on the exact dimensions of the metal, and the ambient
temperature. There is quite a large variation. The specs will be
something like, "does not blow for 1 hour at rated current, and blows in
less than 1 minute at 200% of rated current."

As the current goes up, it blows faster. A fuse's I^2T is approximately
constant; so if you double the current, it blows 4 times faster.

Lee on surface mount and board prep

Surface mount is cheap, and small. But it is harder to build and less
reliable. You pays your money and takes your choice :-)

I have installed more surface mount parts by hand than I care to think
about. You really don't even need any adhesives at all. I usually work
under one of those big magnifying glasses with a circle fluorescent lamp
in it. The PC board is pre-tinned (pads are solder coated). I use
tweezers to position the part, and touch the iron to one pad. The solder
already on the board and part melt together to hold it in place. Then I
solder the rest of the pins.

On ICs, if everything is scrupulously clean, you can hold the board
vertically, and just slide the soldering iron tip down the row of pins,
walking a little ball of solder. The pins solder beautifully once you
get the knack.

The main problem with surface mount is that there is no component lead
length to absorb stress when the board flexes, or when parts expand and
contract with temperature. So vibration and temperature cycling causes
cracked solder joints.

The second problem is that repairs are likely to destroy the part and
the board. If it fails, you throw it away.

> I really wish someone made a cheap automatic drill press.

Same here. Machine tools are still expensive. Your best bet is to buy an
old one at 10 cents on the dollar.

People have made their own out of drill presses with stepper motors,
etc. 
Another method uses a pen plotter, with the pen replaced by a flex shaft
to a dremel moto-tool. This has the advantage that most PCB layout
software already knows how to drive plotters.

Another uses very thin PCB laminate, the kind used for multilayer
boards. They wrap the laminate around a drum, chuck it in a lathe, and
spin it. A duplicate drum has a printed photo of the desired copper
layout. A photocell and cutter move slowly along the drum as it spins.
The cutter removes the copper wherever the photocell sees light. When
you're done, the PCB laminate is glued to a thicker piece of board
material.

Lee on propiganda (incl. quote from Carl Sagan)

Someone once said, "Be careful what you read, because once you get it into your head, it is damned difficult to get it out." Given bad facts, even the brightest individual will come to bad conclusions. Critical thinking is vital. Fiction is easier to write than fact, and it sells better. *Most* of what you read and hear is likely to be fiction. It is essential to maintain a healthy skepticism, and treat *everything* as unproven until gets past your BS filter. In Carl Sagen's excellent book "The Demon-Haunted World -- Science as a Candle in the Dark" he gives some excellent rules for critical thinking. Given the recent events, it is worthwhile for all of us to keep them in mind. Here is his "BS (Baloney Sandwich :-) detection kit": 1. There must be *independent* confirmation of facts. The same "fact" repeated by 10 different sources doesn't count. 2. Encourage substantive debate, by knowledgeable proponents of all points of view. That means listening to *both* sides of the argument. 3. Arguments from authority carry little weight. It's not true because "X" says so. You want to hear from experts, not authorities. 4. Spin more than one hypotheses. List *all* the possibilities, and then try to prove each one false. The hypothesis that survives is more likely to be right than your first guess. 5. Don't form an opinion before you have the facts. Those that do tend to exaggerate facts that support their opinion, and ignore facts that contradict it. 6. Quantify. Numbers are more precise than words. Many arguments disappear once the protagonists figure out that "17" is being called "many" by one side, and "few" by the other. 7. Faulty logic: There aren't just two possibilities; there are MANY. If there's a chain of arguments, every link in the chain has to work. 8. Occam's Razor: When faced with two hypotheses that explain things equally well, the simpler one is more likely to be true. 9. Always ask whether the hypothesis can be falsified. Is it untestable, and impossible to confirm? Then it is little more than an interesting idea; not something you can base action upon. He also gives some common errors to beware of. 1. Argument ad hominem (Latin for "to the man."); attacking the argurer and not the argument. "The Reverend Dr. Smith is a known biblical fundamentalist, so her objections to evolution need not be taken seriously." 2. Argument from authority, not expertise: "President Richard Nixon should be re-elected because he has a secret plan to end the war." 3. Argument from adverse consequenses: We must do this because the only alternative is worse. "God must exist, because if he didn't, society would be even more lawless and dangerous than it is." 4. Arguments from ignorance: "There is no compelling evidence that UFOs are not visiting the Earth; therefore they must exist." 5. Arguments based on special pleadings to the divine or unknowable: "How could God permit such cruelty? Because God moves in mysterious Ways." 6. Begging the question; i.e. assuming the answer without evidence: "The stock market fell today due to profit-taking by investors." 7. Counting the hits and ignoring the misses: "Our war on drugs is working because we have put thousands in jail." 8. Lying with statistics: "Half of all arabs are of below-average intelligence." 9. Inconsistency: "It is prudent to plan for the worst threat our opponents can offer." "There is no point in spending money on environmental dangers because the threat has not been proven." 10. non sequitor (Latin for "it doesn't follow): "Our nation will prevail because God is on our side." 11. post hoc, ergo propter hoc (Latin for "it happened first, therefore it was the cause") "We never would have had nuclear weapons if women hadn't gotten the right to vote." 12. The fallacy of the excluded middle (allowing only the extremes, with no compromises): "America -- love it or leave it" "You're either with us, or against us." 13. Short-term vs. long-term: Really a variation of the excluded middle but so common that it needs special attention: "We don't have money to fight poverty; it's urgently needed to fight crime on the streets." 14. The slippery slope, also related to the excluded middle: "If we let them win in Viet Nam, they'll take over the world." 15. Mixing up correlation and causation: "Studies show that more college graduates are homosexual than those with less education; therefore college makes people gay." 16. The straw man: Caricaturing your enemy as stupid or subhuman to make him easier to attack: (endless examples of this one in the news lately) 17. Half-truths: Telling lies by selective use of the facts. "I saw him yelling at the woman. She ran away. He chased her down, still yelling, and tore off her clothes. Then he threw her on the ground, and got on top of her and rolled her around." A detail was omitted; her clothes were on fire. 18. Weasel words: "Police action. Armed incursion. Pacification. Safeguarding American interests. Freedom fighters. Collateral damage." Sorry if this got a bit off-topic, but I think is is important. We need to pay more attention to people who show us HOW to think than to those that tell us WHAT to think.

Lee On Battery Balancing
Phil Bardsley wrote:
> I've got a new pack of 24 Optima YTs for my EV and a Zivan NG3 that
> needs to be reprogrammed for them. My concern is finding the best
> algorithm to charge the pack for the most cycles... The advice is
> conflicting. I'm hoping that you can guide me.

Indeed, you have hit squarely upon the problem. Now that you've spent a
fortune for a nice new pack of batteries, what do you do to maximize
their life?

Maximizing battery life is mostly a matter of what you DON'T do to it.
Don't discharge it too deeply, or too quickly. Don't charge it too much,
or too little. Don't let it sit around in a partially discharged state.
Avoid temperature extremes. Make sure connections are clean and tight.
Phil Bardsley wrote:
> I've got a new pack of 24 Optima YTs for my EV and a Zivan NG3 that
> needs to be reprogrammed for them. My concern is finding the best
> algorithm to charge the pack for the most cycles... The advice is
> conflicting. I'm hoping that you can guide me.

Indeed, you have hit squarely upon the problem. Now that you've spent a
fortune for a nice new pack of batteries, what do you do to maximize
their life?

Maximizing battery life is mostly a matter of what you DON'T do to it.
Don't discharge it too deeply, or too quickly. Don't charge it too much,
or too little. Don't let it sit around in a partially discharged state.
Avoid temperature extremes. Make sure connections are clean and tight.
Don't let it run low on water.

The battery is damaged slightly by every mistake. The damage is
cumulative. When the battery has accumulated enough damage, it is shot.

The manufacturer's charging algorithms are chosen to insure that the
battery survives the warranty period; not to maximize life. They
generally err on the side of overcharging. This shortens the maximum
life, but prevents early failures from undercharging.

The manufacturers also assume that you will murder the battery from
other types of abuse before charging abuse becomes a factor. If you
deeply discharge it on every cycle, or routinely draw huge currents, or
use it at high temperatures, the battery will die young no matter how
perfectly you charge it.

But, let's assume you are very careful to avoid all these other types of
damage. How can you minimize charging damage, and so maximize life? I'll
assume we're talking about Optimas, though this information applies to
any AGM.

First, you need a charger that has reasonably good voltage and current
regulation. If the charger can't hold a reasonably accurate voltage or
current, all bets are off.

Second, you need sufficient instrumentation so you know what is going
on. Measure individual battery voltages, either manually or with an
automated system. Then control the charging of each battery according to
its needs. The batteries will NOT be the same!

Third, you will need to alter your charging algorithm according to the
behaviour of your batteries and how you use them. There is no one
"perfect" algorithm that always works. You have to keep tweaking it. (If
there was a perfect algorithm, don't you think it would have been
discovered in the 150+ years we've been using lead-acid batteries? :-)

> The conflicting advice I'm seeing is in how to handle the final
> stage of charging after bulk charging is complete.

Yes. Your charger can do pretty much anything between about 20% and 80%
SOC. Since you're not supposed to go under 20% SOC, that part of the
charging algorithm is less important. Almost all the differences between
charging algorithms is in what happens to take the battery from 80% to
100% SOC.

> Optima recommends a final "conditioning" charge of 2 amps for 1 hour.

My opinion is that this is too much. It amounts to a full equalization
on every cycle. I would only do it if you are deeply discharging on
every cycle. If you are limiting your depth of discharge to prolong
life, then only do this "conditioning" charge as needed (i.e. when you
observe differences in state of charge between batteries). Or, use a
much lower current for a longer period of time.

> It's not clear whether they recommend letting the voltage rise at
> will, but I think that's the case.

You basically HAVE to let it rise at will, because when you apply 2 amps
to a fully charged AGM, the voltage WILL shoot way up. If you don't let
the voltage rise, then you won't get the current, and equalization will
take far longer.

> In battery chemistry, as I understand it, the problem is balancing
> the need for higher voltage to polarize the negative plate (and
> prevent sulphation) against the dual problems of venting and
> positive-plate corrosion that result from higher voltage. For
> equalizing the string, this stage seems to help as well, and in
> fact many people refer to the last stage of charging as the
> "equalizing" stage. Some list members have written about longer
> charging at lower current, often at unlimited voltage, others of
> interrupted-current charging (from the NREL article), and others
> of lower voltage charging (e.g., the Powercheq article).

Look at it this way. If you have no means to apply different charging
currents to different batteries (i.e. no regulators, balancers,
individual chargers, etc.) then the batteries WILL get out of balance.
Even if your load and charge currents are identical (all batteries
always in series), the batteries themselves are not all identical, and
have different charge/discharge efficiencies.

So, SOMETHING has to charge them at different rates. The battery
manufacturer's algorithms assume you will do this by deliberate
overcharging; forcing more amphours into a fully charged battery. Optima
is telling you to put in an extra 2 amphours (2 amps for 1 hour). If a
battery in the string isn't yet ful, it charges an extra 2 amphours. If
a battery is already full, this energy just goes up as heat and gassing.
But they are saying there is enough air space inside to contain this
amount of gas without venting, so it can be slowly recombined back into
water later.

It doesn't especially matter how you put in this 2 amphours: 2a for 1hr,
1a for 2hr, 0.5a for 4 hrs, etc. Since the rate of recombination is a
few tenths of an amp, currents this low can be applied for days without
venting.

The "pulse" algorithms are doing the same things, but applying a higher
current but at a lower duty cycle. It comes back to the same thing in
the end. There might be some minor advantages to it, but the results are
far from conclusive.

> Does the use of regulators change the algorithm for this final stage

Yes! If you clamp battery voltage, then it will take far longer to
equalize. Clamp it too low, and it will NEVER equalize.

But, regulators are a good idea because you no longer have to
deliberately overcharge the full batteries to get the weaker ones up to
"full".

Finally, keep in mind that the "correct" voltage and current for
equalization changes with temperature, and as the battery ages.
And, Lee recounting another engineer's advice about relays:
[Anecdote]

When I first began my career as an engineer fresh out of college, I
shared a cubicle with an 'old master' engineer, Ed Wheterald, that was
near retirement. We had regular design reviews, and he always had good
suggestions and spotted things I never thought of. But he preferred
extremely simple, straightforward designs wherever possible.

One day, after a particularly long debate on whether to use a solid
state relay or a mechanical relay, Ed said, "When I started my career,
we used relays. When they were off they were OFF, and when they were on
they were ON. When you used them right, they lasted nearly forever. And
when they did fail, they were easy to diagnose and replace."

"Then vacuum tubes came along. They were supposed to be better, faster,
quieter, longer-lived. But they had lots of problems, and weird failure
modes, and after a while you couldn't get 'em to fix anything. So they
went away; we don't use 'em any more."

"Then transistors came along. They were supposed to be better, too;
faster, smaller, longer-lived. But they had a whole new set of problems,
and when they failed things were even harder to diagnose and fix. So
they pretty much went away, too."

"So integrated circuits came along to replace them. More of the same.
Lots of promises, but lots more problems. In the end, they were designed
out, too."

"And now you're using microcomputers. Even more promises, but even more
problems, weird failure modes, and stuff that is absolutely impossible
to understand much less fix. They will go away, too; to be replaced by
whatever the next great new idea is."

"But, the relay is forever. Not only are they still being used, their
sales are still going up, year after year. They work today, they'll work
next year, and they will still be working long after your microcomputers
turn to inscrutible junk. Don't design out a relay unless you've got a
good reason -- a *damned* good reason

Lee on MOSFET and other semiconductor limits

(Oh, dear!)

I've been assuming that you already knew a lot about MOSFETs, circuit,
design, and high power construction techniques. I'm beginning to worry
that this assumption is incorrect.

The first thing you must learn is that the published specs for a MOSFET,
(or any other part for that matter) are the absolute maximums. No real
application can ever come *close* to operating at these limits! If you
try, the part will die!

OK; let's say you want to use the IRFZ48 that Joe Smalley suggested.
First, get out the data sheet:

http://www.irf.com/product-info/datasheets/data/irfz48.pdf

Right on the front it says:
Vdss = 60 volts
Rds(on) = 0.018 ohms
Id = 50 amps

But those are the *absolute maximum* ratings (see them in the table
immediately below?). You aren't supposed to operate at these limits; the
part would soon fail if you did.

For example: The absolute maximum operating junction temperature is 175
deg.C. The life expectancy of a semiconductor is very long at room
temperature (25 deg.C); it can easily be a million hours. But its life
goes down about 2:1 for every 10 deg.C rise in temperature (the
Arrhenius equation). So, at higher temperatures its life is:

25 deg.C 1,000,000 hours (11 years)
35 deg.C   500,000 hours
45 deg.C   250,000 hours
55 deg.C   125,000 hours
65 deg.C    62,500 hours
75 deg.C    31,250 hours
85 deg.C    15,625 hours
95 deg.C     7,812 hours (325 days)
105 deg.C     3,906 hours
115 deg.C     1,953 hours
125 deg.C       976 hours
135 deg.C       488 hours
145 deg.C       244 hours
155 deg.C       122 hours
165 deg.C        61 hours
175 deg.C        30 hours (~1 day!)

So, a junction temperature of 175 deg.C is a 'reasonable' operating
temperature only if you expect the part to only last one day! So, as a
rule, engineers limit the maximum junction temperature to 100 deg.C;
this will give you roughly 1 year of life at full power. (This is
usually enough for the warranty to run out :-)

Ok, so how much power can we *actually* dissipate without the junction
exceeding 100 deg.C? The 'thermal resistance' numbers at the bottom of
the data sheet are the key to figure this out.

Rjc = 0.8 deg.C/watt
Rcs = 0.5 deg.C/watt

These numbers tell you how much of a temperature difference there will
be between the junction and case (Rjc) and the case and heatsink (Rcs)
for each watt of power dissipation. There is another number, Rsa, the
thermal resistance between your heatsink and the ambient air that is
missing. This is determined by your heatsink. Let's assume you have a
really good heatsink, whose Rsa = 0.5 deg.C/watt.

If the air temperature is 25 deg.C, and the maximum junction temperature
is 100 deg.C, then we have 75 deg.C across Rjc+Rcs+Rsa. Now we can solve
for watts:

Pd = delta deg.C / (Rjc+Rcs+Rsa) = 75 deg.C / (0.8+0.5+0.5)
Pd = 42 watts

This is the TRUE maximum power dissipation that this MOSFET can stand,
mounted to a 0.5 deg.C/watt heatsink, cooled with 25 deg.C air, so the
junction temperature does not exceed 100 deg.C.

(The spec says 190 watts; but to actually get this, you'd have to
submerge the part in Lake Superior and tolerate a 1-hour life
expectancy.)

Ok; next step. At 42 watts each, how many of these MOSFETs are needed to
dissipate the entire 700 amps at 4.25v?

P = 700a x 4.25v = 2975w
2975w / 42w = 71 MOSFETs

That's a big (i.e. expensive) number. You can see why people prefer to
user resistors instead of silicon.

Ok, so only use the MOSFETs to dissipate part of the heat; put fixed
resistors in series to dissipate as much as possible. When a cell is
dead, it is down around 2.5v. Let's say 2v is dissipated in the
resistors; that leaves 0.5v for the MOSFETs and all the rest of your
wiring. Now the MOSFETs only have to dissipate

P = 700a x (4.25v - 2v) = 1575w
1575w / 42w = 38 MOSFETs

Much better. 38 is still high, but livable.

Next problem; that current rating. Look at the leads of a TO-220
package. What is their equivalent wire size? The leads are
0.020"+/-0.002" thick, and 0.050"+/-0.005" wide where they come out of
the case. That's a nominal cross sectional area of

0.02" x 0.05" = 0.001 sq.in.

That's smaller than #18 wire (0.0013 sq.in.), the size they use for
ordinary lamp cords! UL rates #18 for 3 amps continuous. How long do you
think it would survive at 50 amps? Again, the 50-amp rating only applies
if you have infinite cooling, and don't care about the extra resistance
of the leads.

Here again, you have to seriously derate the maximum current to
something you can handle with those tiny leads. The rule of thumb is
that you can't carry more than 15 amps continuous through the leads of a
TO-220 without drastic measures. So, how many devices do we need to get
the current per device down to 15 amps?

700a / 15a = 47 MOSFETs

Now, for that Rds(on). That 0.018 ohms sure sounds great! But, look at
fig.1 and fig.2. They show the *actual* output voltage drops at various
currents and two temperatures. Sure enough, Vds=0.9v at Id=50a at 25
deg.C, so Rds=0.9v/50a=0.018 ohms. But notice that this is for a 20usec
pulse; quick enough so the junction stays right at 25 deg.C. Look at
fig.2, where the junction is at 175 deg.C. Now Rds=1.8v/50a=0.036 ohms;
it doubled. For our 100 deg.C junction temperature, Rds is about 1.5
times higher, or Rds(on)=0.027 ohms at 100 deg.C.

Luckily, this isn't a limiting factor. The voltage drop Vds is still low
enough to work; Vds = 15a x 0.027 ohms = 0.405v. So if our resistors
drop 2v, at 2.5v per cell we will have 0.405v across the MOSFET, leaving
0.095v for all our other connections (difficult, but not impossible).

Conclusion

So, it looks to me like you would need about 47 of the IRFZ48 MOSFETs to
make your load.
Lee talks about thermal management:
There are lots of ways to get rid of heat. If you plan your thermal
management ahead of time, there are some really neat techniques.

Thermal mass
------------
If your peak power is far higher than your average power (like Otmar's
controller), then you can simply store the heat in some massive object.
Like 12 lbs of copper, or 5 lbs of aluminum, or 1 lb of water. The mass
heats up during power peaks, but then has time to gradualy cool back
down during the off-peak periods.

Since the specific heat of water is so high, you can have a large water
filled space inside the heatsink, and it will hold down the temperature
even if it isn't being circulated with a pump.

Finned Heatsinks
----------------
The obvious way to get rid of heat is with a heatsink. Heatsinks get rid
of heat by radiation, convection, and conduction.

Radiation is only significant when there is a very large temperature
difference between the heatsink and its surroundings. It works great for
things that are red hot, but poorly for things like controllers. People
paint semiconductor heatsinks black to enhance radiation, but this is
largely cosmetic. In fact, it is foolish if it is exposed to the sun;
the sun is *much* hotter than your heatsink, so you will *gain* rather
than lose heat from radiation.

Convection is produced by air movement. Air touching the heatsink is
heated (thus cooling the heatsink). The warm air expands, and so it
lighter, and so it moves upward and away from the heatsink. This
obviously assumes the heatsink fins are oriented vertically, and that
there is a place for the air to rise, and a source for cooler air to
come in from below.

The bigger the fins, the more convection you get. But without a fan, the
rate of convection is very low. The surface area needed without a fan is
huge. Thus, a convection-cooled heatsink is big and heavy.

Conduction
----------
This is when you conduct the heat into something else, which is then
hauled away by a fan or pump. Conduction also requires a lot of surface
area, proportional to the specific heat of the material being used.

First consider air. Adding a fan to a heatsink drastically reduces the
surface area required. It can easily be 1/100th the area with a big
blower. But the fan requires power to operate, so you have lowered
efficiency. There are also issues of fan noise, life, reliability, and
the dirt it brings into the equipment to consider.

Fans are *far* more efficient at full speed, so if efficiency is your
goal, running the fan at low speed is the least efficient approach.
You're better off switching it on/off, so when it runs, at least it is
at its highest efficiency. However, this approach can be noisy and
irritating.

Strangely enough, in PCs where cost and size matter above all else, they
use the tiniest heatsinks possible, and disproportionately huge fans.
They get away with this because the absolute power levels are low, they
are always used indoors, and people junk them anyway in a few years.

If you apply this strategy to an EV, you could have an amazingly small
lightweight controller, but the fan would be huge. Interestingly enough,
there *is* a huge fan available; the traction motor. So, it makes some
sense to duct the traction motor's fan intake air through the
controller, then to the motor (because the electronics needs the coolest
air; the motor hardly cares). Better still; eliminate the fan from the
traction motor, and use a single blower to cool both the motor and
controller.

Now for liquid cooling: Liquids have hundreds of times the specific heat
of air. So, it takes much less surface area on the heatsink to conduct
the heat to a liquid. Liquid cooled heatsinks are hardly any bigger than
the part they are cooling.

But, this just moves the heat to the liquid. You still need just as big
a heatsink somewhere else to transfer the heat to the air. Thus, a
liquid cooling system is often bigger and heavier than an air cooling
system. Frequently they need both a pump *and* a fan to do the job.
Thus, your efficiency can be even lower.

For people like me that live in cold climates, water is a pain. You have
to dilute it with so much antifreeze that it is no longer cheap, easy,
clean, or as efficient at moving heat. One thought I've had is to use
the transaxle oil for coolant. It is already there anyway, and the
transaxle has a lot of surface area and mass to act as a radiator. A
"dry sump" would be more efficient that gears splashing in a puddle of
oil. But to keep the pumping losses down, you'd probably have to run it
with a lighter weight oil or automatic transmission fluid.

Odd Systems
-----------
Designers get very creative sometimes. Here are some examples.

1. Submersion. Fill the controller case with a non-conductive liquid
   that directly cools the parts, or sometimes trivial heatsinks on
   the parts.

2. Phase change. Use a coolant that changes state, such as a liquid
   that boils on the part to be cooled, and then recondense the gas
   back into liquid in a condenser. People have used freons, butane,
   alcohol, special parrafins, and even distilled water this way.
   The phase change absorbs vastly more heat, so you need far less
   coolant.

3. Heat pipes. Basically, a self-contained version of #2.

4. Thermoelectrics, such as Peltier devices.

5. Refrigeration systems, like an air conditioner.

6. Compressed air. Seal the electronics in a pressure vessel, and
   run it under pressure. Higher pressures considerably increase
   the effectiveness of gas as a coolant, and you can pick gases like
   helium that are far more effective than air.

Tradeoffs
---------
This is the hard part. All these systems work, but have different
tradeoffs. How much room do you have for the cooling system? Does noise
matter? How about reliability? Do you have to worry about water freezing
in the winter? And, of course there is always cost.
And, in another email, a heatsink tip: M.G. wrote: I thought heat sinks were rated in square inches of exposed surface.
You can rate them this way, but you'll find that their performance is
not very well predicted by surface area. It turns out that cubic inches
of heatsink (including the air spaces between the fins) is a better
predictor.

Lee philosiphizes about small-scale production and how to be successful at it

Philippe Borges wrote:
> We (EV community) and you (Potential EV constructor) need big
> numbers! for low price and for EV success.

This is mostly true; but not absolutely true. The part people forget is
that a product needs to be designed to suit the numbers that you will
manufacture.

It is easy to look at a mass-produced product, and assume that it
represents the *only* way such a product can be built. Large numbers of
very cheap parts. Many custom parts, used on that product and nothing
else. Manufacturing processes that are very expensive to set up, like
injection molding or stamped sheet metal. Assembly processes optimized
to be done by machines, not people.

When an entrepeneur starts a small company, there is tremendous pressure
to produce a product in small quantities that is still built like one
built in large quantities. This is almost certainly doomed to failure.
The tremendous setup and tooling costs that would normally be spread
across millions of units now must be paid by the sale of mere hundreds
of units. This drives the price per unit way up. The company can't sell
enough at the high price to make a profit. So the company fails.

Instead, if you know you will only be selling 100's your first year, you
have to design a product that is economical to manufacture in 100's.
This means no expensive, custom parts -- use inexpensive mass-produced
parts that already exist. It means no expensive manufacturing processes
-- only those that use simple, inexpensive tools. Since a lot of
hand-assembly labor is necessary (and labor is expensive), it mean you
have to work *hard* to make it go together as quickly and easily as
possible.

These alone *still* won't be enough to keep the price down. The real key
is that you have to be *very* clever and creative! Find new, unexpected
solutions. Do things in different ways. Turn disadvantages to your
favor. Examples:

 - Aggressively minimize the number of parts. Parts that aren't there
   cost nothing to buy or assemble! Do you really need all those
   fasteners? Leave off the paint or trim; choose or style the base
   material to look good as-is. Get rid of assemblies of many parts
   that could just be one.

 - Buy surplus parts at a penny on the dollar. The only successful new
   auto manufacturer I've seen in the past 50 years was Neuman-Altman
   Motors. When Studebaker got out of the business in 1965, they bought
   the molds, leftover parts, plans, etc. and reproduced the Studebaker
   Avanti for another 15 years.

 - Forget the "automotive" way to do things. The "Think" used blow
   molded plastic doors, made like children's outdoor playsets --
   that stuff is amazingly cheap and durable.

 - Learn from related industries. Boats, airplanes, furniture, and
   houses are all built in small quantities and are labor-intensive.
   Look at the tricks their manufacturers are using to get price down.

> I'm just thinking: What a pity all that energy, all that knowledge
> and money put on individual special requirement project that will
> never succeed because they are... individualistic and too focused.

I think it is a "catch 22" that a) only people who have a tremendous ego
can start an automotive company, because they won't listen to the
nay-sayers; but b) this tremendous ego means they won't listen to anyone
who *does" know what they are talking about, either. Since no one can
possibly know it all, they are doomed to make mistakes; and small
companies can be destroyed by just one big mistakes.

> Your project show exactly what i mean, you are focusing a
> 3 wheels EV, good for you but it would be (imho) a tiny market.

Nobody knows what kind of EV would sell, because no one has been
successful marketing one yet. Marketing research is only good for
products that already exist; not products that don't yet exist.

Lacking any good marketing data, most entrepeneurs assume the customer
is exactly like themselves. They build the car they want, and assume
everyone else will want one, too. Since entrepeneurs are frequently
*much* different than the average consumer, they are often seriously
wrong. 

Or, entrepeneurs assume the customer is an idiot -- a fool with money.
So they use wild marketing claims to sell some piece of junk. Zero
repeat business, zero work of mouth advertising.

Neither strategy is going to be successful.
Lee explains about power output in electric motors, and discusses possible strategies for implimenting a hybrid:

> I just read the post where someone suggested using a 4x4 front axle
> and attaching a motor to it to run just the front axle.

Yes; this is a viable strategy for building a simple hybrid.

Attaching an electric motor to the front differential might be hard;
it's pretty crowded up there in most cars. It might be easier to remove
the drive shaft to the rear axle, and use the electric motor to drive
it. 
> What would be the minimum motor size and voltage to be able to get
> say a 45 mph top speed and 20 mile range?

Electric motors don't work like ICEs -- they don't have a peak HP
rating. The more electrical power you feed in, the more mechanical
horsepower you get out NO MATTER HOW LARGE! What basically limits an
electric motor's size is heat. It works sort of like this:

 - a 10 lbs motor can make 100 horsepower for 1 second
 - a 20-lbs motor can make 100 HP for 10 seconds
 - a 50 lbs motor can do it for 1 minute
 - a 100 lbs motor for 10 minutes
 - a 200 lbs motor for an hour
 - a 400 lbs motor can do it continuously

So, with electric motors, you have to decide how much power you need,
and for how long you need it. In most EVs, you don't have enough battery
power to run at high power for more than a few minutes; thus, the motor
can be considerably smaller than you'd think.

In some hybrids (called serial hybrids), the ICE drives a generator
which in turn drives an electric motor to drive the wheels. If you build
a hybrid like this, then the ICE can deliver high power for a long time,
so you need a much larger electric motor and generator to handle the
power continuously.

In other hybrids (called parallel hybrids), the electric motor only runs
for very short times, or runs at very low power levels. For example, it
might be used only for accellerating or passing, or creeping around at
low speeds. In this case, the electric motor and generator can be much
smaller.

> I assume an E-Tek is too small and would be toast in short order.

It mould work for a parallel hybrid, but not a series hybrid. Still,
it's a pretty lightly-built motor. There are better choices.

> How would you control it?

Control strategies are the bane of hybrid designs. There are ENDLESS
ideas for the best way to do it. Everything depends on what you are
trying to get it to do.

I suppose the simplest strategy would be to have two accellerators; one
for the ICE and one for electric. You could choose which one (or both)
you wanted to use. Once you have figured out when and how you use each
one, maybe then look for schemes to combine them so the car
automatically uses the "right" one.

> How tough would it be to charge the pack from the alternator while
> in ICE mode?

Not hard at all. Basically, you would have a battery pack of some size
(how big depends on how far and how fast you want it to go on pure
electric). Then put an alternator or generator on the ICE set up to
charge this pack.

> Do DC to DC conveters go in the other direction, 12v to 72v or 96v?

Yes; you can get DC/DC converters for any input and output voltage.
However, a given converter only works one way -- 12v to 72v, but not 72v
to 12. Thus, if you want power to go both ways, you need two DC/DCs.

> What happens to a DC motor when it freewheels? Can anything be
> damaged?

It acts as a generator. You can either ignore the power it generates, in
which case it just freewheels and consumes essentially no power. Or, you
can use the power it generates to charge the batteries (regenerative
braking) or just burn it up in resistors (dynamic braking).

> I'd like to make my pick up (98 Nissan Frontier) as fuel efficient
> as possible by possibly making it a hybrid. I would like to be able
> to make it run under battery power, with the engine at idle so I'll
> have steering and brakes, when accelerating from a stop or backing
> up.

That's a good simple plan. It won't give you the ultimate in gas
mileage, but is simple to implement.

> One, put hub motor wheels on the front

Hub motors are essentially "unobtainium". In other words, you can't get
them.

> and batteries under the bed.

That's fine. For a hybrid, you don't need all that big a battery pack,
since EV-only range can be quite short.

> Is it possible to machine a cover that would replace the cover
> on the back of the rear differential that would have a motor

Yes, it's possible. It adds a lot of weight to the rear axle though. It
may be better to cut the drive shaft, and use a motor with a shaft at
both ends. Couple the ends of the drive shaft to each end of the motor
with universal joints, so the motor mounts to the frame.

> A small 72v system might be a good option for this arrangement.
> Six Exide Orbitals under the bed, a DCP 600 or Curtis, and a 72v
> charger. Or it could be charged off of the engine with a high
> output alternator. Or both.

Right. Or, you could use a shunt motor (such as an aircraft surplus
starter-generator) and just let the driveshaft spin it as a generator to
recharge the batteries.

> The motor would just be used for slow speeds and would freewheel
> when not in use.

A classic series hybrid.

> The main drawback I see with this is I'm not sure I could sufficiently
> protect the motor from the elements.

Not really hard. Put it in a sheetmetal sleeve. Use an external blower
to blow air thru it.

> Somehow mount the motor to the engine and set it up to drive
> the engine with a belt, like an alternator or AC pump in reverse.

Possible, but harder. Many engine compartments don't have the room. You
would need a pretty stout belt or chain if it was to handle significant
power.

Another problem with this arrangement is that the electric motor and ICE
are locked together. You can't let the ICE idle and run the electric
motor at high speed or high power.

> Don't know if this is possible with all the drivetrain movement but
> how about mounting a motor between the framerails forward of the
> rear diff and driving the driveshaft with a gear of some sort welded
> to the shaft or teeth that are machined into the shaft itself.

I don't think it would work for gears, but you could probably have
pulleys on the drive shaft, and a pair of belts and motors (one left,
one right) so it equalizes the side thrust. It would probably take
multiple V-belts to handle the torque and the slight misalignment as the
drive shaft moves up/down with suspension movement.