The open source Rectifier/Regulator

[Guest]

Well, I guess I should start a new thread for this rather than hijacking
the other thread.

As mentioned in my post there, I agreed with the person who said he was
disappointed cberkeley is charging for his design.
I didn't plan to start posting about this regulator till it was further
along, but given cberkeley's has apparently gone on hold, I figure I will put some preliminary info out.

When I first started to read that thread I had hoped that cberkeley was
going to put the design out as public domain, and that several people
would end up experimenting with it to refine and adapt the design.
Since that apparently won't happen, I am hoping that people will end up
doing so with this one.
I will put mine out as freeware and am hoping for suggestions and
feedback, to refine the design if required
Plus, when the design is coming together, I would like some help with the documentation ...

My goal is a bit different than the other regulator.
Cberkeley's design philosophy seems to be to build the best regulator he
can build, without much regard to cost.
My goal is to build the most economical regulator that is still
completely functionally adequate for all the GS series bikes.



The first thing I tried was googling to see if there are any other
designs out there that I could use.
I only found one complete design, which is at
www.takisnet.org/~abayko/vreg.pdf
Amusingly enough, it was for a Suzuki GS400E (What an unexpected
co-incidence that another Suzuki needed a regulator)
--- btw, in case anyone decides to build it, the schematic for this
one has the bottom 3 bridge diodes (D5, D6, and D7) upside-down.

Googling was kind of frustrating, because the term shunt regulator
(which is what these are) is also commonly used in other contexts in
electronics, to say nothing of all the regulators for sale in the
windpower generation market. So there may be other designs out there,
but I got tired of sorting through all the extraneous links to try to
find them.
If there is anyone out there with better googling skills than I, that can
find some more freebie regulators out there for comparison and to see
alternate ways of doing things, I would appreciate it.
keywords would include Permanent magnet alternator and shunt regulator.
You do not want any regulators that modulate current through a field
coil, they are for a different kind of alternator.



In any case, what I found was that the designs are usually a three-
phase rectifier bridge, one for each lead of the stator. Each stator
connection can also be shorted to ground via an SCR, which is how the
regulation is accomplished.

The only real variations in the designs seem to be the control circuit
that is used to tell the SCRs when to fire.
The simplest designs just have 3 zener diodes, one connected to the gate
of each SCR.
According to my Clymer manual, Suzuki's design seems to have one zener
feeding a 2 transistor amplifier which then feeds all the SCRs.
The web design mentioned above has a differential pair of transistors
which drive all the SCRs.
I have no idea of what cberkeley's control circuit is.

By the way ... throughout the course of this post I will compare my
design to the web design and what I know of cberkeley's design, this is
primarily to help any readers see both the differences and commonalities
in the designs and the reasons for the designs being the way they are.
Also, it will hopefully help readers to understand the reasons behind
the design differences and to make an informed choice for what will fit
their needs the best, or to make suggestions for changes to try.
It should not be taken to imply that any of the designs are "better" than
any of the others.

Anyway ...

At least some of the stock Suzuki regulators are somewhat different than
described above. I don't know how all are set up, but what I found for
mine is that one of the stator phases is switched through the "lights on"
switch. When the lights are off, this phase is not used at all.
When the lights are on, this phase is in the circuit.
Additionally, the switched phase is unregulated (i.e. it has no SCR)
There is conflicting info about the other 2 phases out there.
The Clymer manual says they are both regulated, but I remember seeing
somewhere that only one of them is.



Having given the background, here is where I am with my design.

My design will have all three phases regulated. Even if I leave the
"switched phase" idea in, regulating it is OK, that SCR will just be
firing with nothing to do ...

Control Circuitry:
I am dithering on the control circuitry.
The differential pair promises easy adjustability, but it has a high
quiescent current. To avoid draining the battery when the bike is off,
the voltage sense lead must be a switched lead, which I would strongly
like to avoid.
A design like Suzuki's seems good, but may be difficult to get to have
as much adjustability as I want. (yes, my regulator will be adjustable
too) However; done right, Suzuki's design should allow me not to have
to run the sense lead from a switched lead.

Since the parts for the control circuits are all dirt cheap, I will
probably build one of each just to see what I like better.



Power Circuits:
Most of the expense comes in the power circuitry.
This is where over specifying the parts costs money, and under specifying them lets the smoke out.
These are also the physically "big" parts that will determine the
enclosure size required.


The Bridge Rectifier Section:
The "right" way to build the bridge is to use a 3 phase rectifier.
These are expensive, but will allow a physically very small unit.

cberkeley uses one and a half normal (4 terminal) 40 amp bridge rectifiers.
This is an OK way to do this, but bugs me (irrationally ???) because it
does not maintain symmetry. Also, one of the rectifiers will run hotter
than the other ... probably no big deal, but it could change the timing
of when the SCRs fire, leading to unbalanced load on the stator coils
(probably wouldn't matter in actuality ... but I like Symmetry)

If space will allow, I think I will use 3 normal bridge rectifiers, and
parallel the diodes in each one. This would effectively double the
current rating, allowing the individual parts to be specified as more
available, less expensive 25 amp units. Physical size will of course
increase by one square inch ... but trilateral symmetry will be
conserved.

FWIW the author of the other regulator on the web said he used 12 amp
rectifiers successfully ... I'm not sure if he used 2 or 3 ...
... or even a single 3 phase


SCRs
I plan to use 25 amp SCRs. (at $4.47 for 3 of them)
cberkeley uses 35 amp SCRs. (at apparently $39 for all three)
The web design used 8 amp SCRs successfully.

Sizing the SCRs is perhaps the most controversial part of the design.

The rectifier bridge will see pretty much the full output current all
the time.
But the SCRs will only see the "excess" current which the alternator
makes but is not needed by the loads on the bike.
Given that the battery is always charging and the ignition circuit is
always igniting, there is at least a minimal load that will always
be there when the bike is running.
In most cases in the US, the lights will always be on as well.
(and if not, Suzuki's OEM wiring will thoughtfully disconnect one of the
phases from the alternator for you when the lights are off ...)

Further, even if there were no loads at all on the bike, each SCR will
only see a 33% duty cycle.

In one of the posts in the other thread, Nerobro said that the biggest
alternator on the GSes was a 28 amp on the GS1100s
Assuming 3-5 amps for charging and running the ignition, the 25 amp SCRs
would be (just barely) enough. All in all, given the duty cycle and the
fact that other loads will likely be on too, I am pretty confident that
it should be MORE than enough.

Of course you could also find some bigger SCRs and use them instead.
If any readers find a source of bigger SCRs for cheap, let me know and I
will uprate the design.
(note: if you wish to substitute SCRs, the ones I am using have an
insulated tab, if the ones you wish to use do not, make sure that
they can be connected to ground (assuming the heat sink is grounded))


Filter capacitor.
cberkeley is apparently providing a BEEG @SS filter capacitor in his
design to allow running the bike without a battery.
I am not. I think it's money (and a lot of space) for something that is not
needed.
Plus, I don't really trust it to filter adequately.
If you want it, you will have to add it on your own, or build the other
design


Status of the project
I have pretty much figured out all the components I need, and am
compiling a list for mail ordering them.
(complicated by the fact that I want to order the parts for 5 projects
at once:
1) the regulator
2) an LM3914 based voltage monitor for the bike
3) a Headlight Modulator for the bike
4) a brake light flasher for the bike
5) a motor controller for a radio controlled model airplane.

That's where I am with the project right now.
I'll be posting more as the project goes along.

[duaneage]

A capacitor is unnecessary. Our bikes already have a huge capacitor in the form of a battery. I like the approach to using the full wave bridges. How about a model that can be bolted to the case easily for heat transfer?

I think you can still get a LM317 in TO-3 case style for a few dollars. They can take a lot of abuse and are adjustable with a simple resistor change.

[bmac]

Very Informative

Well written and very discriptive. Reading your list of parts, you appear to have many things going on. So when do we get to hear about your test results. I realize that this is something you just don't put a sheet over and it builds itself. I give you Gentlemen alot of credit to apply your knowledge and ability to remedy a flawed design and whatever the case may be..... share with others, another avenue to resolve an issue that can be very fustrating to a bike owner.

Good luck, I will keep a eye on your thread, (progress )that is...or..... whatever


bmac

[Boondocks]

I like your approach, Martin, and feel that you're on the right track. Capacitors? We don' need no stinkin' capacitors.

The alternator output on my '80 GS1100ET is 230 watts, nominally 19 amps, so the 25 amp SCR should do just fine on a bike like mine considering that the actual current shunted will be considerably less.

In My Thread on this same subject, I mention that I had posted the link to the same vreg.pdf file back in November. Nobody paid any attention to it at the time, as the do-it-yourself bug had not yet bitten.:-D

I will be following your progress, and like your design philosophy.

[duaneage]

If you can get a self contained regulator IC for around 5 bucks and the bridges for another 10 you could do this for around 25 dollars. Wire, solder and other items are cheap enough.

Makes a great science project.

[duaneage]

I think you can still get a LM317 in TO-3 case style for a few dollars. They can take a lot of abuse and are adjustable with a simple resistor change.

Scratch the LM-317. They don't have enough capacity

[Boondocks]

At least some of the stock Suzuki regulators are somewhat different than described above. I don't know how all are set up, but what I found for mine is that one of the stator phases is switched through the "lights on" switch. When the lights are off, this phase is not used at all. When the lights are on, this phase is in the circuit. Additionally, the switched phase is unregulated (i.e. it has no SCR) There is conflicting info about the other 2 phases out there. The Clymer manual says they are both regulated, but I remember seeing somewhere that only one of them is.

Your setup is typical. For the bikes with "always on" lights and no "off" switch, one of the stator phases is routed through the headlight shell and connected there to a return wire, accomplishing nothing but added resistance and a potential problem with the extra connection. I have the Suzuki factory manual electrical schematics, and two of the three phases are regulated (with two SCR's).

Control Circuitry:
I am dithering on the control circuitry. The differential pair promises easy adjustability, but it has a high quiescent current. To avoid draining the battery when the bike is off, the voltage sense lead must be a switched lead, which I would strongly like to avoid. A design like Suzuki's seems good, but may be difficult to get to have as much adjustability as I want. (yes, my regulator will be adjustable too) However; done right, Suzuki's design should allow me not to have to run the sense lead from a switched lead.

Since the parts for the control circuits are all dirt cheap, I will probably build one of each just to see what I like better.

I vote for the differential pair method. I don't have a problem with a switched sense wire. The Suzuki method without a separate sense wire has its own pitfalls, namely the pervasive connector corrosion which acts like a resistor, causing voltage drops and false battery voltage readings. I think that a clean connection of a sense wire will give a better chance of maintaining accurate voltage readings. The ability to fine-tune the charging voltage is too good to give up.

FWIW the author of the other regulator on the web said he used 12 amp rectifiers successfully ... I'm not sure if he used 2 or 3 ... or even a single 3 phase

He used two full wave bridge rectifiers (one and one-half).

SCRs
I plan to use 25 amp SCRs. (at $4.47 for 3 of them) cberkeley uses 35 amp SCRs. (at apparently $39 for all three) The web design used 8 amp SCRs successfully.

With a 230 watt, 19 amp alternator this would work fine with my bike's actual shunt current.

Filter capacitor.
cberkeley is apparently providing a BEEG @SS filter capacitor in his design to allow running the bike without a battery. I am not. I think it's money (and a lot of space) for something that is not needed. Plus, I don't really trust it to filter adequately. If you want it, you will have to add it on your own, or build the other design

Definitely not needed, and a waste of space and money.

[bakalorz]

For the bikes with "always on" lights and no "off" switch, one of the stator phases is routed through the headlight shell and connected there to a return wire, accomplishing nothing but added resistance and a potential problem with the extra connection.

Ewww, but I guess typical of OEM design. Do people bother to reroute that to eliminate the loop ?

I vote for the differential pair method. I don't have a problem with a switched sense wire. The Suzuki method without a separate sense wire has its own pitfalls, namely the pervasive connector corrosion which acts like a resistor, causing voltage drops and false battery voltage readings. I think that a clean connection of a sense wire will give a better chance of maintaining accurate voltage readings. The ability to fine-tune the charging voltage is too good to give up.

You misunderstood. I want to try to get the leakage down low enough that the connection can be always on. That would allow either an internal to the regulator connection _or_ a seperate sense wire that doesn't have to be switched, i.e. you could run the sense directly to the battery + terminal. (probably through a fuse though ... just in case)
I agree that adjustability is a requirement

[Nerobro]

is $1.40 to much to spend to help smooth out the output? :-) I'd go for the biggest cap I could afford to put in place.

When I first was running through my mind on this, I thought of using Vregs as well. 20-30-40amps of vreg is hard to get your hands on.

You can ignore the "switched" coil on the stator. After 1979 they disabled that feature. You don't need it. Just hook the stator directly up to the r/r, and let your r/r deal with the extra current if you turn off the headlight. Or more importantly, turn off the headlight, and take the benifit of being able to charge your battery at idle!

[bakalorz]

is $1.40 to much to spend to help smooth out the output? :-) I'd go for the biggest cap I could afford to put in place.

Which cap is that and what are its specs ?

Also , I'd be ---VERY--- leery of using a cap that couldn't take the full unregulated stator output, especially if the intent is to be able to ride without the battery to keep things near 12V ...

Imagine the following:
Sam squid builds a R/R with cap, but Sam's assembly skills aren't up to snuff and it fails, running the battery dry.
No problem says Sam, I've got a cap, I can run down to the dealer and get a new battery.
On the way there is a slow truck, so Sam downshifts and twists his wrist.
Redlining it takes the stator voltage to 100 volts plus, and the failed regulator ain't stopping it ... Ka freaking BOOOOOM.

Abused Capacitors make ---really--- impressive firecrackers.

At work some body managed to significantly overvoltage a tiny little cap about the size of a pencil eraser.
It was LOUD, it sounded like a gunshot, everyone jumped a foot in the air.

In addition to overvoltage, I'd be worried about the amount of ripple current even if everything is working right, especially considering our regulating scheme.

Further, I don't get the attraction; why are you going riding without the battery ?
No matter how shot, even a toasted battery left in place is going to do a better job smoothing the alternator output than any reasonable size cap.

When I first was running through my mind on this, I thought of using Vregs as well. 20-30-40amps of vreg is hard to get your hands on.

I agree, and don't forget the voltage drop required either.

I spewed soda at the mention of an LM 317 earlier (although the thought did cross my mind too ... you CAN parallel them you know .... great minds think alike ... )

Some sort of switching regulator might be doable ... but not by me.

You can ignore the "switched" coil on the stator. After 1979 they disabled that feature. You don't need it. Just hook the stator directly up to the r/r, and let your r/r deal with the extra current if you turn off the headlight. Or more importantly, turn off the headlight, and take the benifit of being able to charge your battery at idle!

I _think_ my 81 has it, but I haven't traced the wiring yet. (clymer shows it)
The loaded alternator test showed my regulator to be toast, and thats the furthest I've taken troubleshooting the bike so far.

[Nerobro]

I am quite sure your bike does not have it. :-) My 80 550 doesn't... and I remember reading somewhere that from 80 up none of the bikes had the cutout. After 81 the bikes didn't even have the switch.

yeah, linear regulators would be fun. I do think they wouldn't like the voltage spikes coming off those rectifiers though. 100v is the low end. I wonder what voltage is really coming off the stator at 10-11krpm. Though they do claim you can run them in parallel. I think after doing the math on them I found that solution to be more expensive than electrix. :-/

Maybe it might be a good idea to put a ziener diode under the cap to drain power if the line voltage hits.. oh.. 20v? The cap I ordered for the berkley regs are 50v. The biggest benifit I see from the cap is that batteries are really pretty crappy capacitors. Even at oh.. 2000rpm, the stator has more than 15v coming from it, but a lot of that is blown away. Especially if you have a weak battery. For some reason I think that a cap would give the bike a fighting chance at lower rpms. Such as those that killed my iornbutt run. :-/

And smoother voltage regulation is better for just about any electrical asccesory.

[duaneage]

The 650s had the switch until september 81. My Feb 81 bike has a switch, it was built from 80 year parts.

If a capacitor is deployed at the output of the regulator it needs to be fuse protected. Capacitors are risky devices that can short out. This would cause a fire and be unsafe.
There are supply caps in the ignitor module but they are close to the delivery point and also are fuse protected at the fusebox.

capacitors absorb moisture, even teh condensors on the points go bad eventually.

[Planecrazy]

I will defer to Greg (Nerobro) on the exact cost estimates (he ordered for us both), but the parts for Cletus' design were NOT expensive compared to buying a used commercial unit on Ebay or building a slightly different design (as you are doing here) yourself. The thing you left out in your comparison was that Cletus has also added a really nice feature the commercial units don't have ... the ability to fine tune the voltage for the type of battery you have. I have a sealed glass mat battery, and my understanding is that they perform better and last longer if you charge them at a higher voltage than a standard lead acid "wet" cell.

Considering the "infamous" failure rate for the OEM R/R's and the fact that they often take more expensive components with them when they go, it seems like a "no-brainer" to me to spend a few extra bucks and design "the best" and most bulletproof replacement you can ... in the end it's cheap insurance.

So while I applaud your efforts at providing an "open-source" solution, my opinion is that you should still be shooting for a much superior replacement for stock, rather than simply looking for the lowest cost equivalent. My strategy for any sort of replacement, whether it be motorcycle, automotive, home appliances, etc. is to always replace with at least "one level better" than what I had, and this strategy has ultimately saved me a lot of money over time because the "better quality" replacements almost always last much longer than a cheaper alternative would have. Not always true, but generally an economical winner...

Regards,

[Boondocks]

...Considering the "infamous" failure rate for the OEM R/R's and the fact that they often take more expensive components with them when they go, it seems like a "no-brainer" to me to spend a few extra bucks and design "the best" and most bulletproof replacement you can ... in the end it's cheap insurance.

So while I applaud your efforts at providing an "open-source" solution, my opinion is that you should still be shooting for a much superior replacement for stock, rather than simply looking for the lowest cost equivalent. My strategy for any sort of replacement, whether it be motorcycle, automotive, home appliances, etc. is to always replace with at least "one level better" than what I had, and this strategy has ultimately saved me a lot of money over time because the "better quality" replacements almost always last much longer than a cheaper alternative would have. Not always true, but generally an economical winner...

Regards,

It puzzles and gives me pause to see how Cletus's design is perceived to be the holy grail of design perfection and "bulletproof" because he has specified extra tolerances on some components. His design seems to be the only one on the planet for a shunt regulator that doesn't use heat sink(s). Instead of using an inexpensive heat sink to absorb heat, he seems to think that retaining heat in more expensive components is a satisfactory solution. Not a very elegant or "better quality" solution in my opinion. I'm not moonstruck about his design, which is basically the same as the free open source vreg.pdf file without the heatsink(s) which are considered mandatory by commercial designers who have to pay for their mistakes. All I've seen so far is a box that is unwieldy and oversized for motorcycle use (with unnecessary components) whose long term practicality and viability remain to be seen. I am skeptical, and frankly the rhetoric and concept does not impress. Sorry if I offend the true believers.

[relic-rider]

Man, If there is anything worse than petty bitchin about something, it is EE's bitchin about stuff in terms i don't understand. Cut it out already. Why don't you guys get together on a design that incorporates the best of both designs and not "mine is better and cheaper than yours". Cletus was on the right track, just trying something that he thought was needed. Maybe his is overkill, but it works for his application and he was the first to design and build and make one available to everyone. Sure there was other schematics and designs before his, but where were all of you when he was BUILDING one. Sittin at your desk criticizing him. Get over it. If you can build a better one and cheaper, do it. Do it for me and all of the other non techno geeks out there, who can't or have no interest in building one ourselves. We still all need one. Our GS's RR's are CRAP. That IS the REAL problem. Sorry, My .02