# REGULATOR/RECTIFIER

Discussion in 'Electronic Design' started by Fernando, Oct 22, 2004.

1. ### FernandoGuest

I am needing a rectifier / regulator for motorcycles. The alternator
produces
AC three-phases. If somebody has circuit or can indicate me a Link where I
can
to find some thing, I thank.
Fernando

2. ### John PopelishGuest

Is the alternator excited by permanent magnets (always full on), or
does it have a rotating field winding (and brushes on slip rings) so
that a regulator can control it?

3. ### FernandoGuest

Hi,

It is a alternator excited by permanent magnets.

Tks
Fernando

4. ### Ken SmithGuest

This sounds like a place where an SCR phase controlled rectifier would be
the best. Can anyone thing of a reason not to go that way?

How many wires come out of the alternator and is there an internal ground
connection we have to worry about?

If we are lucky, there are three wires and no ground connections. If
there are more we will have to know what the internal connections really
are.

Basically the power side of the circuit looks like this

-------------------------------------------- Vout
! ! !
--- D1 --- D2 --- D3
TrigA- / ^ TrigB- / ^ TrigC- / ^
! ! !
A---+ B----+ C----+
! ! !
! ! !
--- D4 --- D5 --- D6
^ ^ ^
! ! !
GND GND GND

D1,2 and 3 are biggish fastish SCRs

D4,5 and 6 are large, perhaps schottky, recifiers.

5. ### Rich GriseGuest

I can't really judge, until I see the circuit you come up with to
time them.

Good Luck!
Rich

6. ### Ken SmithGuest

I just discovered a problem with the basic idea. It is not very good at
living through an overload. Unless the alternator has a largish
impedance, the current can be huge and we have to wait a long time for the
SCR to get reverse biased and shut down.

To implement the phase control switch circuit, I thing we need to insert a
current sense resistor (a)in the ground side of the bridge rectifier or
(b) the positive output of the bridge.

(a) is the better if the servo electronics runs at the ground side of
things.

(b) is better if the servo electronics floats on top of the output
voltage.

I think I would elect to go with (a). I assume that there is a battery so
that I can use its power to run the servo circuit. This way I can avoid
having to make a small supply to run the servo on.

To fire the SCR, we need to provide a positive pulse on its gate. To do
this, we need to level shift from the ground referenced servoing section
up to the SCR. Fortunately, we only need to deal with a pulse. For each
SCR, I suggest:

+Vtrig
!
!/ NPN !
----! V
! !\ e ----- D(n)
! ! R1 C1 / !
-----+ +-----/\/\/\----!!----+------- !
! ! ! !
! !/ e ---!<-----+------
----! D7
!\ PNP
!
GND

+Vtrig is a despiked version of the battery voltage. This is needed
because spikes up to 70V may be on the battery voltage wire, depending on
what else it being switched.

We also need an accurate reference voltage to compare the output to. This
could be one of the LM7805 like regulators that have improved accuracy.
Using a power regulator instead of a reference, helps for the next bit.

We need a resistor divider off the output voltage to monitor it.

We need, perhaps, an op-amp to amplify the signal from the current sense
circuit.

We need, at least one, divider connected to the windings of the alternator
to get the timing information. I will suggest 3 dividers to make the
timing easier to work out.

Now here comes the part some will love and others will hate:

Run these dividers into the ADC inputs of a PIC and use 3 digital outputs
as the trigger signals. The rest is SAMOP (Simply A Matter Of
Programming)

7. ### Rich GriseGuest

....
This got me to thinking - what a nightmare! But, what's the ripple on
unfiltered, full-wave rectified 3-phase? A lot less than you'd first
think! So, why not just rectify it and chop it? You could free-run
PWM and just filter out both components in the DC. An inductor could
do a lot of filtering at these impedance levels, too.

Good Luck!
Rich

9. ### John PopelishGuest

That type is often designed with a lot of inductance, so that the
current does not rise in proportion to speed, as long as there is a
significant load current. They also don't like to have a widely
varying load, but are matched to the loads designed into the bike. A
shunt regulator is common with this type or alternator. This is some
big transistor on a heat sink, connected across the alternator output
that sucks up all the excess current, once the voltage rises above
something like 14 or 15 volts.

I think this is an example of such a regulator:
http://www.cranecams.com/pdf/90002500.pdf
http://www.challengers101.com/RotaxChargeSys.html
http://homepage.sunrise.ch/mysunrise/joerg.hau/mot/voltreg.htm

If you try for a more efficient series chopper regulator, you have to
be prepared for wide voltage swings out of the alternator, almost
proportional to speed. So if it puts out 13 volts at 1000 rpm, it
might approach 100 volts at 10,000 rpm, under a light load. The
rectifiers may not like this.

10. ### Ken SmithGuest

Huh! Why is thinking a nightmare? Or did you mean the design
situation?

Full wave 3 phase gives 6 peaks per cycle. The peaks are 60 degrees
apart. The dips are at the 30 degree points so its cos(30)=0.866
Depends on when you consider "first". If you mean in the 1960s then you
may be right that the first time I thought about it I got it wrong. My,
normally very good, memory fails me on this subject so I can't say for
sure what I first thought the ripple would be.
(1) It invloves more power semiconductors

(2) I didn't suggest it.
"We don't need no stinking inductors"

If the PWM design is any good, the inductor doesn;t have to deal with the
supply voltage ripple.

11. ### ChrisGibboGibsonGuest

(Ken Smith) wrote:

[snip]
Have you been reading feerguy's bullshit?

Gibbo

12. ### Ken SmithGuest

"feerguy" ???? huh

I was thinking of the movieline "we don't need no stinking badges"

13. ### ChrisGibboGibsonGuest

No worries. When you're next bored do a search on feerguy or eerguy. He posts
regular bollox on here and he "don't need no stinking" anything in order to
solve all the world's energy problems (including the laws of physics
apparently).

Gibbo

14. ### Ken SmithGuest

^^^^^^

Oh him!

He isn't even funny anymore.

15. ### ChrisGibboGibsonGuest

"anymore" ? He never was, was he?

Gibbo

16. ### Ken SmithGuest

I thought it was the first time I read it. I though someone was pulling
our leg. It seemed so totally unbelievible that I took it as a joke.

17. ### ChrisGibboGibsonGuest

It's odd how different people with different personalities perceive the same
thing.

When I first read it I just assumed he was a complete cockend.

Now when I come across it again in other NGs I often wonder whether he's just
having a laugh. Which is almost the exact opposite of what you seemed to think.

Gibbo

18. ### Rich GriseGuest

Welcome to the club! ;-)
At first, you thought of the 200% ripple of the half-wave rectifier,
Not more than trying to phase-control 3-phase with SCR's!
A Man After My Own Heart!
Yes. They tell me that people go to school for many, many years
consecutively to learn the black art of switching regulator design. I've
actually been in the presence of an actual human man who actually knew how
to design ferroresonant transformers. It's like being granted an audience
with The Oracle At Delphi. But he had no clue about switchers. He and I
together wasted about three months, back in 1998 or so, trying to design
from scratch what the Chinese are now selling for thirty bucks a pop.

Thanks!
Rich

19. ### Ken SmithGuest

At first, you thought of the 200% ripple of the half-wave rectifier,

No, its more likely that I though 50% or 70.71% or even 63% thos numbes
come up a lot so it must be one of them, right.

[..]
I had to develop much of the theory for myself. When I started at it,
books on the subject were rare and very poorly written. In most cases,
the trick is to consider the switching section to have a "transport
delay". That is there is an increasing phase lag with increasing
frequency but not reduction in gain. In large signal cases, you also have
to consider the "mixer effects". This is where the load variation beats
against the switching frequency. These days, it is easier to use the
simple model to get close and then stuff it into spice for fine tuning.

Was his name Zambre? He is a man the world will certainly miss.

20. ### Rich GriseGuest

No, its more likely that I though 50% or 70.71% or even 63% thos numbes
come up a lot so it must be one of them, right.

[..]
I had to develop much of the theory for myself. When I started at it,
books on the subject were rare and very poorly written. In most cases,
the trick is to consider the switching section to have a "transport
delay". That is there is an increasing phase lag with increasing
frequency but not reduction in gain. In large signal cases, you also have
to consider the "mixer effects". This is where the load variation beats
against the switching frequency. These days, it is easier to use the
simple model to get close and then stuff it into spice for fine tuning.

Was his name Zambre? He is a man the world will certainly miss.
[/QUOTE]
Irv McMurren. He retired about 3 or 4 years ago.

Cheers!
Rich