[EVDL List]

some modest progress to report

a basic core prototype circuit has been fleshed out:
www.zev.dk/diagram3.gif
the 3 components in the power section are shown as one each but in
practice several would be in parallel. I'm planning to try 6 IGBTs to
begin with. same with the diodes and maybe 10 of the caps. if the number
gets big the power supply and driver may need to be increased as well.

it's still missing a current sensor component I have bought as well as a
thermistor input. they should be trivial to add.
some tachometer will probably be needed too but that can perhaps be
omitted for early versions.

some of you with EE experience can comment if you see something terribly
wrong. maybe even Otmar if he isn't too scared

some future considerations are inductors and shielding to protect the
low voltage section from the shocks of the HV section.
also maybe an extra connector for B+ with a resistor for the intial
charging when connecting the battery wires to avoid sparking. hold the
positive lead on that for a few seconds before connecting to B+

Dan

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

[No message]

 

[EVDL List]

Morgan LaMoore wrote:

>A few things:
>
>Why IGBT instead of MOSFET? I'm just curious of your reasoning; for
>lower voltages (below 200-300V) I'd use MOSFETs for easier paralleling
>and low voltage drop from low Rds-on.
>
>
cost vs power. 600V 80A for 2$ each at 500. I so far intend it to work
from 100-400V. IGBT seems like the choice but if you know otherwise do tell.

>I'd recommend a snubber cap of a few nF directly across the transistor
>to stop the inductive kickback of the motor from going above the
>transistor's breakdown voltage. Low inductance, low resistance, high
>peak current snubber capacitors would be best.
>
>
yeah that's a consideration. do you know any that might fit the bill?

>Also, I couldn't find any info on the low voltage power supply you
>used. And I don't know if that's necessary - just about everyone has a
>12/13.8V either from a DC-DC or accessory battery. Why not just take a
>12V input and use a linear regulator on that?
>
>
I had thought the part number would make it easy to find but google
doesn't know it seems.
here it is though: http://www.pwrx.com/Result.aspx?g=73&m=78

the reason I want that is because I want it to be selfsufficient,
especielly when adding on the DCDC. having a battery to help an electric
car is absurd. this little device seems to offer freedom from that.
costs 7$ at rell.com

>Also, if you do that, then it should be possible to make the circuit
>automatically charge the capacitors through a resistor when you enable
>power. Have the controller control the main relay. The controller is
>connected to battery power twice: once directly, and once through the
>main relay. So when the user wants to start (determined by the
>ignition switch), the controller can connect the always-on battery
>connection to the power rails through a resistor and a switch
>(MOSFET?), charging up the capacitor. Then, once the capacitor is
>charged enough, the main relay turns on, and the controller is ready.
>That way the user doesn't need to worry about pre-charging or
>anything.
>
>
as I see it, the problem with automatic precharging seems to be that you
have to switch on a connection that will be used to carry all the
current in normal operation
touching a third rail for a few seconds before first connection seems
simple enough

Dan



_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

[No message]

 

[EVDL List]

Dan,

1.
in all the circuits diagrams I have seen the line from the MCU PWM pin
to the IGBT driver has a pullup resistor to the 5v rail, an inline
resistor and a couple of caps to ground. (I am no expert so check this
with someone who is)


2.
is there any reason your not using a single high amp IGBT with something
like Cornell Dubilier SCD capacitors?

Looks easier to fabricate to me.

Dan Frederiksen wrote:
> some modest progress to report
>
> a basic core prototype circuit has been fleshed out:
> www.zev.dk/diagram3.gif
> the 3 components in the power section are shown as one each but in
> practice several would be in parallel. I'm planning to try 6 IGBTs to
> begin with. same with the diodes and maybe 10 of the caps. if the number
> gets big the power supply and driver may need to be increased as well.
>
> it's still missing a current sensor component I have bought as well as a
> thermistor input. they should be trivial to add.
> some tachometer will probably be needed too but that can perhaps be
> omitted for early versions.
>
> some of you with EE experience can comment if you see something terribly
> wrong. maybe even Otmar if he isn't too scared
>
> some future considerations are inductors and shielding to protect the
> low voltage section from the shocks of the HV section.
> also maybe an extra connector for B+ with a resistor for the intial
> charging when connecting the battery wires to avoid sparking. hold the
> positive lead on that for a few seconds before connecting to B+
>
> Dan
>
> _______________________________________________
> For subscription options, see
> http://lists.sjsu.edu/mailman/listinfo/ev
>

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

> 1.
> in all the circuits diagrams I have seen the line from the MCU PWM pin
> to the IGBT driver has a pullup resistor to the 5v rail, an inline
> resistor and a couple of caps to ground. (I am no expert so check this
> with someone who is)

Good idea - all of that may be a bit much, but a pull-up is definitely
necessary, and some filtering is probably a good idea.

> 2.
> is there any reason your not using a single high amp IGBT with something
> like Cornell Dubilier SCD capacitors?
>
> Looks easier to fabricate to me.

It also costs a lot more. A 400A IGBT can easily cost hundreds of
dollars, or you can use a bunch of small IGBTs for much cheaper.

-Morgan LaMoore

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

[No message]

 

[EVDL List]

Dan Frederiksen wrote:
> a basic core prototype circuit has been fleshed out:
> www.zev.dk/diagram3.gif

Ok, you've got a good start. I'd say about 10% of the parts you'll need
are there. It's missing a lot of practical parts you'll need to get it
to actually work reliably.

> the 3 components in the power section are shown as one each but in
> practice several would be in parallel. I'm planning to try 6 IGBTs to
> begin with. same with the diodes and maybe 10 of the caps.

IGBTs and diodes don't parallel well. There will be enough differences
between off-the-shelf parts that some will hog most of the current (and
fail), while others hardly help at all.

What voltage is this controller intended for? If under 100v or so, use
MOSFETs instead of IGBTs if you intend to parallel them. Also, under
100v you can use Schottky diodes, which have lower on-state voltage drops.

It is much easier to get big diodes so you can use as few as possible
(ideally just one). This avoids all the problems with paralleling and
matching parts.

I would strongly suggest that a first controller use a single large
module with both the transistor and diode in it. Yes, it appears to cost
more; but saves you so much work that it's worth it. You can buy them
surplus if you like. It's all designed for you, tested, debugged, and
specs are available to make it far easier to apply.

> if the number gets big the power supply and driver may need to be
> increased as well.

The gate driver already looks too small. The power supply requirements
aren't bad, though, unless you get to really huge parts.

> it's still missing a current sensor component I have bought as well as a
> thermistor input. they should be trivial to add.

Yes; but without them your first set of semiconductors will die!

> some tachometer will probably be needed too but that can perhaps be
> omitted for early versions.

That's less critical. It protects the motor, but the controller won't care.

> some future considerations are inductors and shielding to protect the
> low voltage section from the shocks of the HV section.
> also maybe an extra connector for B+ with a resistor for the intial
> charging when connecting the battery wires to avoid sparking. hold the
> positive lead on that for a few seconds before connecting to B+

As shown, your micro is "hot" to your propulsion pack. So is your
potbox, and everything else associated with this controller. This is
workable, provided that you treat every single wire to that controller
as "hot" (dangerous high voltage).

You're missing the gate pulldown resistor, to guarantee the transistor
is off even if the driver's wire falls off or the micro's output pin
floats high. You also need a series damping resistor in the gate, 10
ohms or so. You also need a zener diode to clamp the peak transient gate
voltage.

There are no provisions for a failed pot. No precharge circuitry. No way
for the micro to know if the battery voltage is applied, and within range.

The micro has woefully inadequate filtering and protection from noise.
Controllers are intensely noisy!
--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in -- Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Thomas Ward wrote:

>in all the circuits diagrams I have seen the line from the MCU PWM pin
>to the IGBT driver has a pullup resistor to the 5v rail, an inline
>resistor and a couple of caps to ground. (I am no expert so check this
>with someone who is)
>
>
I think it should be a pull down but yes I might get that. also the
resistor inline with the gate. that remains to be thought
out/experimented. I want to keep it as simple as possible so parts will
only be added if actually necessary and not just a generic design guideline

>is there any reason your not using a single high amp IGBT with something
>like Cornell Dubilier SCD capacitors?
>Looks easier to fabricate to me.
>
>
nothing is set in stone yet but as another wrote it is a much more
expensive approach. it was my initial approach as well but is only
economical if you can get surplus parts on ebay. I got a couple of 600A
module for 25$ each but as an open source design it's bad to rely on
special finds. The 600V 80A small ones are 2$ a piece in volume of 500.
or 3-4$ in singles. quite the cost difference up to the normal price of
450$ for a 600A module.
Also I think it may actually be easier to make it using small
components. everything with modules is screw on and metal bars where as
with PCB it can be more elegant. I'm actually hoping to get away with
using no buss bar at all. thick copper layer and then solder on copper
wires to the pcb to strengthen where needed. bolt on could maybe work too.

it's my impression that once the pcb is designed, a small volume fab can
produce them for a couple of dollars each

Dan

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Lee Hart wrote:

>IGBTs and diodes don't parallel well. There will be enough differences
>between off-the-shelf parts that some will hog most of the current (and
>fail), while others hardly help at all.
>
>

I'm hoping that with derating, cooling and the claimed tight parameter
distribution of the IGBT that I can just parallel them
do you know of a single 300+A diode for pcb...

>What voltage is this controller intended for?
>

it's intended for 120-380v. the low voltage power supply only works in
that range. maybe a little wider. no lower than 90v

>I would strongly suggest that a first controller use a single large
>module with both the transistor and diode in it. Yes, it appears to cost
>more;
>

it more than appears to cost more. it actually does. as I see it it's
not viable for an open source design to go with big parts.

>The gate driver already looks too small.
>

how do you figure?

>There are no provisions for a failed pot. No precharge circuitry. No way
>for the micro to know if the battery voltage is applied, and within range.
>
>

this design runs on main battery only so if the micro is on so are the
batteries. I intend manual precharge with either a built in 3rd terminal
with a resistor or an external resistor

>The micro has woefully inadequate filtering and protection from noise.
>Controllers are intensely noisy!
>
>

what would you suggest?

Dan

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Dan Frederiksen wrote:
>> IGBTs and diodes don't parallel well.

> I'm hoping that with derating, cooling and the claimed tight
> parameter distribution of the IGBT that I can just parallel them

Hope is fine in church. But engineers put their faith in measurements.
Measure the current in each of those paralleled IGBTs. You may be surprised.

> do you know of a single 300+A diode for pcb...

You don't put 300 amp diodes on PC boards. The tracks on a PC board
can't carry this kind of current.

>> I would strongly suggest that a first controller use a single large
>> module with both the transistor and diode in it.

> it more than appears to cost more. it actually does. as I see it it's
> not viable for an open source design to go with big parts.

Remember that you have to derate your semiconductors by 3:1 or so. An
IGBT advertised at 90 amps is only good for maybe 30 amps. When you go
to a module, much of this derating is already built in. They've already
included the effect of the thermal and isolation barriers in the specs.

>> The gate driver already looks too small.

> how do you figure?

Look up the "Miller effect". The apparent gate capacitance also includes
the capacitance from gate to collector (or drain) multiplied by the gain
of the transistor. With the transistor's high gain and the high Vce
voltage, it requires a substantial current to change the gate voltage
quickly.

For example, suppose you have 0.01uf of apparent gate capacitance, and
want to charge it in 150 nsec. This requires a gate drive current of I =
C dv/dt = 0.01uf x 15v / 0.15usec = 1 amp.

This drive requirement increases as you add more or bigger IGBTs, or try
to switch faster.

>> There are no provisions for a failed pot. No precharge circuitry.
>> No way for the micro to know if the battery voltage is applied, and
>> within range.

> this design runs on main battery only so if the micro is on so are
> the batteries. I intend manual precharge with either a built in 3rd
> terminal with a resistor or an external resistor

Right now, you show no precharge, protection or fail-safe circuitry at
all. You need to think about these things. They are even more important
on one-off prototypes than they are in commercial products.

>> The micro has woefully inadequate filtering and protection from
>> noise. Controllers are intensely noisy!

> what would you suggest?

Study microcontroller designs actually used in real products (not
simplified application notes). You will find several bypass capacitors
on the power supplies, extra RC filtering on inputs, no common ground
between high power and control circuitry, watchdog timers,
power-on-reset circuits external from the micro, etc.

--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in -- Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Lee, when will you put your words to practice,
not tell others how to do things, but do it yourself?
Jack

Lee Hart wrote:
> Dan Frederiksen wrote:
>
>>>IGBTs and diodes don't parallel well.
>>
>
>>I'm hoping that with derating, cooling and the claimed tight
>>parameter distribution of the IGBT that I can just parallel them
>
>
> Hope is fine in church. But engineers put their faith in measurements.
> Measure the current in each of those paralleled IGBTs. You may be surprised.
>
>
>>do you know of a single 300+A diode for pcb...
>
>
> You don't put 300 amp diodes on PC boards. The tracks on a PC board
> can't carry this kind of current.
>
>
>>>I would strongly suggest that a first controller use a single large
>>> module with both the transistor and diode in it.
>>
>
>>it more than appears to cost more. it actually does. as I see it it's
>> not viable for an open source design to go with big parts.
>
>
> Remember that you have to derate your semiconductors by 3:1 or so. An
> IGBT advertised at 90 amps is only good for maybe 30 amps. When you go
> to a module, much of this derating is already built in. They've already
> included the effect of the thermal and isolation barriers in the specs.
>
>
>>>The gate driver already looks too small.
>>
>
>>how do you figure?
>
>
> Look up the "Miller effect". The apparent gate capacitance also includes
> the capacitance from gate to collector (or drain) multiplied by the gain
> of the transistor. With the transistor's high gain and the high Vce
> voltage, it requires a substantial current to change the gate voltage
> quickly.
>
> For example, suppose you have 0.01uf of apparent gate capacitance, and
> want to charge it in 150 nsec. This requires a gate drive current of I =
> C dv/dt = 0.01uf x 15v / 0.15usec = 1 amp.
>
> This drive requirement increases as you add more or bigger IGBTs, or try
> to switch faster.
>
>
>>>There are no provisions for a failed pot. No precharge circuitry.
>>>No way for the micro to know if the battery voltage is applied, and
>>>within range.
>>
>
>>this design runs on main battery only so if the micro is on so are
>>the batteries. I intend manual precharge with either a built in 3rd
>>terminal with a resistor or an external resistor
>
>
> Right now, you show no precharge, protection or fail-safe circuitry at
> all. You need to think about these things. They are even more important
> on one-off prototypes than they are in commercial products.
>
>
>>>The micro has woefully inadequate filtering and protection from
>>>noise. Controllers are intensely noisy!
>>
>
>>what would you suggest?
>
>
> Study microcontroller designs actually used in real products (not
> simplified application notes). You will find several bypass capacitors
> on the power supplies, extra RC filtering on inputs, no common ground
> between high power and control circuitry, watchdog timers,
> power-on-reset circuits external from the micro, etc.
>


_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Jack Murray wrote:
> Lee, when will you put your words to practice,
> not tell others how to do things, but do it yourself?

I *have* built controllers! Most other parts of EVs as well (chargers,
DC/DC converters, instrumentation, battery managment systems, etc.)!

I am trying to give Dan (and others) the benefit of my experience, to
save them time and money on things that don't work. Whenever my advice
says that something won't work, I also try to provide guidance on how to
change it or what to add to fix the problem.

And, I am building EVs myself, and helping others build theirs.

--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in -- Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Jack Murray <[email protected]> wrote:
> Lee, when will you put your words to practice,
> not tell others how to do things, but do it yourself?
> Jack
>

Um, Dan is the one who's never built an EV. Lee definitely sounds like
he has more EE experience, and more experience with power electronics
in general.

Even from the little bit of "real" circuit design I've seen at my
internship, I definitely agree with almost all of Lee's points. I've
seen far more protection and filtering in a little 20W motor
controller than on Dan's schematic.

That said, I'm rooting for Dan. I'm in the same position he is; I've
never built an EV and I want to get into controller design. I know
lots of measurement and protection circuitry is necessary to keep
things from blowing up, though, and I'm grateful when Lee points out
something I hadn't thought of (or when he re-affirms an opinion I
already have).

I also think that the power electronics shouldn't go on a circuit
board. Ideally, I think they should be something like SOT-227 package
screwed onto a heatsink with busbar or thick wire connecting them.
Even the 100A solar car does this: they have the power electronics,
gate drivers, and current sense is in one box, and everything else is
on a separate circuit board.

I think it's OK for power ground and digital ground to be connected if
it's a star ground and the two are only connected at one point. Then
they aren't used as the same ground, but they're connected at one
point so no two parts of them drift too far apart.

And I disagree that the gate driver is a bit weak; that's one of the
beefiest driver IC's I've seen, capable of delivering 14A peak gate
current. I have a 600A IGBT module, and it can switch the 3000 nC gate
charge in 214ns, faster than the IGBT is capable of turning on. That
said, the driver circuit needs some good bypass caps right next to it.
(One big, high capacitance one and one with very low ESR). It could
also use some other supporting circuitry (filtering/protection on the
inputs and maybe a gate resistor to control the rate of turn-on).

-Morgan LaMoore

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Lee Hart wrote:

>I *have* built controllers! Most other parts of EVs as well (chargers,
>DC/DC converters, instrumentation, battery managment systems, etc.)!
>
>
then why haven't you made them public?

Dan

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Morgan LaMoore wrote:

>I also think that the power electronics shouldn't go on a circuit
>board. Ideally, I think they should be something like SOT-227 package
>screwed onto a heatsink with busbar or thick wire connecting them.
>
>
well the curtis use TO-220 parts and the zilla allegedly use TO-247.
both are pcb components.
I think there is some significant elegance in the pcb approach and it's
actually possible to reasonably carry a lot of power that way

not to mention... a very significant cost element

Dan

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Lee Hart wrote:
>> I *have* built controllers! Most other parts of EVs as well (chargers,
>> DC/DC converters, instrumentation, battery managment systems, etc.)!

Dan Frederiksen wrote:
> then why haven't you made them public?

I have, on the EV list, many times. But I don't sell them as products,
or publish complete "how to" details because I'm not happy with the
designs. I gave up on my own controllers because I can buy better ones
than I can build.

--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in -- Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

----- Original Message -----
From: "Dan Frederiksen" <[email protected]>
To: "Electric Vehicle Discussion List" <[email protected]>
Sent: Saturday, October 06, 2007 8:26 PM
Subject: Re: [EVDL] Open source controller design

> Lee Hart wrote:
>
>>I *have* built controllers! Most other parts of EVs as well (chargers,
>>DC/DC converters, instrumentation, battery managment systems, etc.)!
>>
>>
> then why haven't you made them public?
>
> Dan
>
> ___Gees! Louise!! I Mean Dan<g>! Lee is so damn busy working on other
> stuff. Bat Regs, other projects, to put food on his table as he is a
> family guy. Not to mention trying as time and money permits, to get the
> Sunrise into a car, kit or whatEVer to get it on the road.As you
> controller do it yurself wannabees, it isn't easy. I talked to Lee about
> it long an' dragged out driving with Lee to Fla to Tow back to MN the
> Sunrise Body ,last year. I'm NOT an EV electronic geru, but just asked
> questions as you are. The Devil is in the DETAILS. You need to make a
> failsafe control board to run the Heavy Duty electrical stuff, which you
> can get out of a catalogue. This is stock industrial electronics. YOU as
> Electrical genious, gotta suit down and THINK of what EVer could go wrong,
> failure mode. So it will"Scram" the controller; Like IF the Go pedal is
> pressed down when you key on? So the damn car doesn't take off! IF the
> Igbits or whatEVer yur using, decide, or mis -comuntate, IF that is the
> right term? lock ON and away you go, with direct drive! Embarrising in a
> parking lot! Great lawsuit stuff!I had Lee trapped in the CAR, no
> escape<g>!It's a looong way from Tampa to St Cloud MN(Sartell).They sorta
> blend together.

I used to run Generous Electric E-60 locomotives, built in the 70's,
they used to "Lock on" now and again,your guess, we engineers got pretty
darn fast at dropping the pantograph, and spinning wheels, that woulda
turned Wayland green with envey! Of course these lokies wern't in the
general public's hands. Good Thing! We "drivers" lived with this quirk, but
G.E. Shoulda FIXED that issue,early on, but like Ford with the flaming gas
tanks in the Pintos' it was cheeper to pay a few burned people off than FIX
them! The gas tanks I mean<g>!We used to be glad IF the E-60 made a trip
WITHOUT something going haywire!Giving rise to our service mark; "Getting
Halfway there is Fun" Apologies to Cunard Lines<g>!

Point I'm making, think of ALL the possable failures, and build that
board to "Think " for you, shut things down as needed. Oh it can be done,
not denighing that, just isn't easy, ask Otmar.Well, I take that back, DON'T
ask Otmar! He's pretty godamn BUSY, at this point. HE did the engineering.
Hell! Why should he just give himself away. Does Toyota publish the design
details of the Prius?How deep are yur pockets? Ya BUY the tech, that
SOMEBODY spent sleepless nights designing. Dan, what do YOU do for a living?
I'm not meaning to diss you. Are you an EE? A plumber, Carpenter, Buss or
truck driver? You wouldn't expect your employer to say" just drive the next
few runs for me for nothin', truck's out there ready, or build somebodies
deck, here's the wood, YOU put it together . "Bus is out at dock 7, gassed
up, drive it to Sandy Eggo" and not pay ya the miles that the contract with
the UTU sez you will get. IF it is a Union outfit?

OK I'm getting carried away as I often do with Transportation stuff, but
in defence of Lee. He has gone WAY beyond the call of duty, as to giving out
electrical info. You would hafta pay big bux to a electrical consulting
ferm, for his expertise. He is only one man, a very human and lovable guy,
when ya actually meet him.A shining star on the List. If I knew a FRACTION
of what he knows. Hell! I probably could FIX my #$%^ controller when it
shits out<g>!

Hammer and torch shade tree mechanic here.Slipping into my flameproof
suit.

Bob

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Lee Hart wrote:

>I have, on the EV list, many times. But I don't sell them as products,
>or publish complete "how to" details because I'm not happy with the
>designs. I gave up on my own controllers because I can buy better ones
>than I can build.
>
>
Well, I've talked about designing controllers quite a bit as you know
and as far as I can recall you haven't shown me any of it.
And I understand the responsibility of putting out a design but that's
easily fixed with caveats. simply state that it's a work in progress and
there's no guarantee it won't fail disastrously in practice. maybe
specific concerns yet to be adressed with the design. as long as you
don't vouch for it as perfect there is no problem with sharing your
work. others might spot problems or validate the design and it might end
up getting built and move the cause forward. it could literally change
the world.

and it takes no effort to put it online if you have the circuit diagrams
already. maybe a few pictures of test builds. a test drive video would
be very inspiring too if you have ever gotten that far. if you don't
have a website I will gladly host a Lee Hart section.

when we all start from scratch we don't get as far as we could. doesn't
have to be perfect to be good.

Dan


_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

Morgan LaMoore wrote:
>> I also think that the power electronics shouldn't go on a circuit
>> board. Ideally, I think they should be something like SOT-227 package
>> screwed onto a heatsink with busbar or thick wire connecting them.

Dan Frederiksen wrote:
> well the Curtis uses TO-220 parts and the Zilla allegedly uses TO-247.
> both are pcb components.

The Curtis and Zilla use dozens of small parts in parallel to carry
hundreds of amps, so the current per device is quite low. Buss bars are
stacked onto the PC board and arranged very close to the devices to
absolutely minimize the amount of PC board foil carrying these currents.
Getting this to work is tricky, because the foil has significant
resistance and heating problems, even at mere 10's of amps.

> I think there is some significant elegance in the pcb approach
> and it's actually possible to reasonably carry a lot of power that
> way not to mention... a very significant cost element

Try it, and make some measurements.

--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in -- Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev