The AVR project

[Jim Mc]

...As long as it can compensate for a change in load within a few cycles you shouldn't notice that. Make it too fast and we may have other stability issues...

I agree.  My point is that being able to compensate for load changes isn't enough.  It should also compensate for engine speed changes. (I assume this is to be used with Listeroids, not just Petteroids...)  And to be able to 'compensate' means changing the current through a large inductor (the field) in a short time (10's of milliseconds).  The only way I know to do that is with a large voltage available to force a fast current ramp through the inductance.

Also agree we need a freewheeling diode, just don't see that it needs to be a zener.

[rcavictim]

Regarding the waveform distortion seen on the Chinese ST type heads, it has been determined that the magnetic interaction of the harmonic winding when connected as a source of field current, causes this distortion.  For this reason I agree that the winding should not be used in this new AVR circuit project. 

I also like very much the suggestion of incorporating a D-cell and `flash field` manual pushbutton.

[adhall]

Jim Mc:

Thanks for all your ideas. Please note my responses below.

1.  Ditch the Z winding.  (that's the so-called harmonic winding, right?)  Instead run a power transformer and full wave bridge off the stator output for the exciter source.)

The main advantage I see to using the Z winding is that it saves having include a transformer to power the field. It also avoids having to draw the power for the field from generator output--although we are only talking about 100 W or so. Unfortunately, the Z winding is something of an unknown in that I can't say what DC voltage we will get after we filter and rectify it. We may find that the this voltage is not high enough to serve. If that turns out to be the case, then we would have to get the field power from the generator output as you propose.

2.  Plan on running that DC exciter at a fairly high DC voltage. (somewhere around twice the field requirement)  Got to, in order to change the fiield current quickly (the field is a big inductor) so that we have a hope of clearing up flicker from our slow-speed drive.

If the filter after the Z winding rectifier does what I expect it to, we will end up with close to what you are proposing here. Whether we really need to be able to change the field current that quickly will depend on the overal stability of the system. (More on that subject below.)

3.  KEEP the crowbar.  In fact make it more robust by adding a second zener to activate it sourced by the exciter DC supply.  That way, a failure in the sense transformer or its rectifier would still end up in a successful crowbar shutdown.

Good point. If we lost voltage feedback, regulator would tend to drive the field to maximum current. It makes sense to modify the crowbar circuit to test for this. By the way, there is also an overcurrent protection function built into the regulator chip. All this requires is a current sensiing resistor in the field circuit. I am inclined to go both ways here.

4.  I'd be tempted to run the LM2578A off  the sense supply instead of the exciter supply (fewer parts).  (maybe...)

This would depend on how much output the generator makes from residual magnetism alone. The complication in the existing low voltage power supply is designed to allow it to produce a regulated low voltage over a wide input range. My thinking was that the Z winding voltage would be higher than the output of the step transformer under startup conditions and that we could use that to get the low voltage power supply up and running before we are putting any current through the field.

5.  How about driving the field on the low side - N channel MOSFET? (fewer parts)

The regulator chip shown is not really suited for driving a MOSFET or an IGBT for that matter. It is capable of either sourcing or sinking current, but not both. I am looking for a regulator with a totem pole output. Once we have that, we could go either with a MOSFET or an IGBT.

6.  Of course be super careful with the grounds so the high field current doesn't affect the sense side of the regulator.

This would a major concern for any circuit board layout whether it be a printed circuit board or a hand wired board. In fact, this is something that may make building a hand wired board difficult for someone who didn't have the experience and knowledge to know how to keep this signals properly isolated from eachother.

7.  I've no idea how well this will self excite.  Might want to add provisions for a D cell and momentery 'flash' switch.

This is a big concern for me, too. I am hoping the low voltage power supply design will make flashing the rotor unneccesary.

8.  That filter cap on the sense supply shouldn't be very big for fast response to voltage fluctuations.

As I remarked above, stability is big concern as well as a big unknown. As with any other type of circuit involving feedback, there you always end balancing response time against stability. My thinking is that we start with very sluggish response and then do some testing to discover how fast we can safely make it. It may be a good idea to include an adjustment for this so that peole have the freedom to set the response time where they want it.

9.  Don't see why we need a zener across the field - regular fast rectifier should be fine (cheaper)

This was a brain fart on my part. It is actually supposed to be a Shottky diode. If the switching frequency ends up in the low kHz range, as I am hoping it will, then a standard rectifier diode should be fine here.

Best regards,
Andy Hall

[adhall]

from Andre Blanchard on January 19, 2007 at 05:57:39 AM
On my 3ph ST head I believe the AVR gets its power from one of the phases, if I remember correctly it is connected between the center tap and one of the outputs so it is getting 120AC.
It has no problem getting started except for the first time I spun it up and then there was that time there was a load on it as it spun down.

Andre:
That's a good point, it hadn't occured to me that allowing the generator to spin down with a load on would "demagnetize" the rotor. Maybe we need to disable the field current when the frequency gets too low? That wouldn't be too hard to do.

Does anyone have any thoughts about this?

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

from rcavictim on January 19, 2007 at at 08:35:15 AM
Regarding the waveform distortion seen on the Chinese ST type heads, it has been determined that the magnetic interaction of the harmonic winding when connected as a source of field current, causes this distortion.  For this reason I agree that the winding should not be used in this new AVR circuit project. 

Rcavictim:

Because this regulator is "voltage based" rather than "current based", the power for the field circuit will need to be pretty well filtered. This is needed to prevent changes in the field power supply voltage from being reflected in the rotor current--which is what causes the problems you describe. By putting the pass band of this filter far enough below the frequency response of the regulator, then any changes in the field power supply voltage will have little effect on the excitation current. The nice thing about this is that the frequency response of the field power supply filter will have no effect on the total response time of the regulator itself (unlike what would happen if you put a similar filter after the recitifier in a standard ST head).

Although, having said this, now I am rethinking "voltage based" vs. "current based". "Current based" would make the filtering requirements for the field power supply less stringent. And what we really care about is the field current after all. Well, I was talking finding about a different regulator chip anyway, so I'll see what I can find that will do this.

How about homebrewing a current based regulator using opamps? One big advantage to that idea is that opamps are general purpose devices and it isn't likely that they will all be phased out of production like a special purpose regulator chip might be. And there are a lot of opamps with the same pinout, so you can easily substitute similar devices. I'll puzzle over this and see what I can come up with.

Best regards,
Andy Hall

[Jim Mc]

Hmm.  I’m not sure you're thinking about the waveform distortion the same way as rcavictim and others have described.

The ‘prevailing wisdom’ is that distortion in the stator output sine wave is NOT caused by an unfiltered field supply.  Rather it results from drawing current from the Z winding.  Yeah, that simple.

The Z winding is mighty peculiar.  It has somehow been designed to increase its output in relation to the current being drawn from the stator.  Nice for regulation.  But an apparent side effect is that the current waveform in the Z winding is harmonic-rich, as we all know.  The twist is that this harmonic-rich current flowing in the Z winding induces a similar distortion on the stator output. (remember harmonics = distortion in sine wave systems)

So, if you reduce the Z winding current to zero, the stator output cleans up. 

That was why I (and other members) recomended ditching the Z winding altogether.

The other thing that might be troublesome is the response time of the regulator.  I agree completely that unstability in the loop is a concern.  But the main factor that slows this loop down is the field inductance.  Its huge.  Really huge.  My head (not  an ST, but similar) measures .76 H.  Making a rapid change in that highly inductive load is (a) required, if you want to regulate for engine speed fluctuations, and (b) only possible with a high field supply voltage and (I think) a shunt regulator.  (running the field in an 'H'  bridge is another possibility...but seems overly complex) 

I’ve been pondering this problem for a while and am planning a regulator.  Different from yours in that it’s linear, and as mentioned a shunt configuration.  It operates at less than 50% efficiency, so it will waste some power. 

Do you have an ST head?  Someone ought to measure the stator  output vs field drive, and the field inductance as a starting point for design.  (I'd do it, but have no ST head now)

BTW, I see you're right about the LM2578 output drive - no MOSFETS or IGBT's...

If we don’t shoot for fast response in the regulator to prevent flicker, than why bother at all? 

[rcavictim]

Hmm.  I’m not sure you're thinking about the waveform distortion the same way as rcavictim and others have described.

The ‘prevailing wisdom’ is that distortion in the stator output sine wave is NOT caused by an unfiltered field supply.  Rather it results from drawing current from the Z winding.  Yeah, that simple.

The Z winding is mighty peculiar.  It has somehow been designed to increase its output in relation to the current being drawn from the stator.  Nice for regulation.  But an apparent side effect is that the current waveform in the Z winding is harmonic-rich, as we all know.  The twist is that this harmonic-rich current flowing in the Z winding induces a similar distortion on the stator output. (remember harmonics = distortion in sine wave systems)

So, if you reduce the Z winding current to zero, the stator output cleans up. 

That was why I (and other members) recomended ditching the Z winding altogether.

Jim,

Thanx for posting this.  Excellent explanation.  You beat me to it and probably explained it better.  Andy may not have been aware that this has been discussed at some length elsewhere here recently and some actual tests made.

[adhall]

Jim Mc:

Concerning harmonic distortion:
Thanks for bringing to light the idea that the distortion problem is actually caused by current flow in the Z winding. I will search the forum to find out more about this. I will be particularly interested to learn what kinds of tests have been done and how the results lead to this conclusion.

Concerning "flicker":
Thanks also for bringing to light the rotor inductance. Having that information makes it possible to make some rough calculations to determine the minimum voltage we need in the field supply in order to respond fast enough to deal with this. By the way, Jim Mc, if you have already done these calculations, please share them with us.

 <<< My apologies to the forum -- I am having fits trying to show formulas using this editor. I am going to try to work that out off line and post the results later >>>

Best regards,
Andy Hall

[Doug]

Sorry for not being there to help....

Kind of centered on gus at this moment.....


Doug

[adhall]

Doug:

I'll pester you if something comes up in your area of expertise. In the mean time, have fun with Gus. I'm jealous. All I've got to work on right now is my old beater Dayton gas genset.

Best regards,
Andy Hall

[Doug]

I'd like to add a comment about the harmonic winding.
Who cares if the source is full of noise?
The regulator is going to be chopping up the dc made from it any way.
We are PWMing the the output and producing all kings of odd harminics any way.
We are depending on a combination of high inductance in the rotor, and some crude filtration to give us a cleaner DC source for the rotor flux than the pure harmonic winding could. The high inductance of the rotor with free wheeling diode works greatly in our favour. It can't be hard to improve on the standard Z winding

I've been thinking this over we need to get the chopping speed up say 500 hz and see what it does.
We need to run the chopper at the longest on duty cycle possible and still leave enough room to raise the rotor flux to compensate for the drop across the syncronous reactance of the stator and then still have a little more room to try and compensate for a slight lagging power factor. Its up toot the end user to try and get their loads close to unity....

Doug

One more thought:
The windings of any rotating electrical machine are a ballancing act between efficient use of copper and iron and and ideal winding ( not practical to actualy build ). By there nature they all produce odd harmonics and there's nothing we can do about it. The flicker we see on heads powered by low speed engines is more a function of changing rotor speed than harmincs that are always multiples of the fundimental.

[rcavictim]

I'd like to add a comment about the harmonic winding.
Who cares if the source is full of noise?
The regulator is going to be chopping up the dc made from it any way.

Doug,

I am not sure you understand the issue relating to the harmonic winding.  Imagine disconnecting the Z winding entirely. Now suppose you sample the output winding which goes to the load, run it through a variac and bridge rectifier to supply DC to the field.  Spin it up and set the variac so that nominal output voltage is generated against your load. The output to the load will be a clean sinewave.  With no wiring changes place a resistor across the disconnected Z winding that represents the same current that flowed when it was being used to supply the field.  Your gross output distortion will appear again.  It is the way this Z winding interacts and loads the magnetic field inside the ST head that causes the distortion in the output waveform which we are concerned about eliminating.  PWM on the field or not, by employing the Z winding as a source of power for the field supply it appears you will introduce the distortion.  The only apparent solution to me is to not use the Z winding at all.

[Doug]

I understand what you are saying and I would agree with you if not for the fact we are rectifying and chopping up the DC further just with the action of the regulator doing its job.

We could filter and clean up the output of the Z winding and still be left with harmonics generated by the switching power suply that now drives the field.

Now if we forget about that problem now and move onto actual switching power suply we are left with the questions. How fast can we switch and what will make the most practical combination of switching speed, duty cycle and filter to produce an acceptable current source.

What we don't want is realy nasty choppy power and a field flux that is varrying by as yet unknown % ( I'm just going to toss out a number and say we need to hold the line at about 2-5% riple tops ).


Here's another way to look a this a 6 phase scr drive produces a smoother dc output than 3 phase ( more pulses per second and they overlap each other more .... )
A cyclo converter will produce a less and less distorted AC wave from the lower the frequency the drive is running at even though the actual fundimental we are chopping up never changes.
These machines have nothing to do with what we are trying to acheive here but are mearly illistrate my point that once you start to chop up a current source the distortion on that source becomes less important the faster you chop and the more effective your filtering.

[rcavictim]

Doug,

You are not getting the jist of what I said.  Please read it again.

It is like the Z winding represents your big toe, and you have just stuck it into a calm pool of water.  It causes ripples in the magnetic field by being there. These ripples end up on the output of the alternator, apparently by transformer action. By leaving it completely disconnected, i.e. NOT using it as a source of power for any purpose at all, it is the same as unwinding it from the alternator frame and selling it to the scrap man.  It is not there to induce ripples in the magnetic field.

[Doug]

Now I follow you.
I never considered the Z winding having an armature reaction of its own.....
That make sence now, is that why the wave on the scope shots has what looks like a third harmonic lagging the fundimental and making more of a saw tooth than squaring up?

Hmm.
That realy sucks, I wonder if we can do something useful with the un used 50 z taps then?

Doug

[rcavictim]

Yes I think you grasp it now.  I don`t see a problem myself with taking the power needed to charge the field through the AVR circuit from the main output winding.  As for the 50 Hz taps.  Are they not just extra turns out beyond the 60 Hz taps on the same windings?  Slower speed (1500 RPM), lower induced voltage per turn, thus more turns needed to get back to 120 volts at 50 Hz.