sync rectification

[willib]

Syncronous rectification Fet gate control ideas for Three phase axial flux machines

i have possible solution to gate control of a sync rectifier ,a little while ago i was playing around with Hall sensors

the Hall sensor outputs a sinusoidal wave when in the presence of a alternating magnetic field , its amplitude is +/-(.6V) approximatly , depending on the strength of the magnetic field

if a hall sensor was placed at the outer edge of one coil, in the path of the magnet , on each phase , it could indicate when a phase was going positive

and a hall sensor  MAY also work attached to the outside of a coil , not in the direct flux path , because as the coil is energized by the magnets it produces its own counter magnetic field opposing the motion of the magnet across its surface

the sensor could pick up this field

i say May because i did try the hall sensor , in the flux path , but not on the outide of the coil , i shall try this tomorrow , it seems to me that this second approach may work better than the first.

unlike the scenerio in the next paragraph , the hall sensors ouput is a known small quantity whose output is independent of the phase voltage

some history

one could use the signal (phase voltage) to apply sync rectification to an alternator(alt).

but the problems associated with that method are many

the idea is to turn on a phase when it is positive

i have tried this , some time ago, on a single phase alt, using N and P channel FETs

in a sort of cris-cross pattern to drive the gates , the scheme i tried used 8 Fets driven directly from the alt, and needless to say got very complicated

this idea will not work with an alt whose phase voltage is greater than the allowable gate voltage (10-30V)

i was not prepared to even attempt an OP AMP solution to drive the gates

so thats where it ended

[Nando]

WillB:

Synchronous rectification using Hall effects is not practical, One needs to think the rectification process that varies constantly depending on the generated voltage and load.

Detecting when the coil arrives to the opposite polarity is not enough, output voltage level is needed to "see" when the voltage is going to arrive to the output load voltage.

SO, a more accurate and faster process is to place a comparator referenced to the load voltage as a reference then detecting the "0" Zero crossing to trigger the MosFet ON.

Another solution would be to use current transformers to the detect the initial low "non-synchronous" current to turn the Power MosFet to bypass the diode.

You may need floating supplies to accomplish this project if You want maximum synchronous rectification.

One with current transformers will require 6 N Power MosFests and 6 Current transformers to turn its own MosFet on when the current of diode of the MosFet is sensed BUT additional circuit to turn the MosFet OFF when the leg ( phase) voltage is going below the load voltage. --- TRICKY !!.

One solution is to detect the transformer current dropping to Zero ( before going to opposite polarity)

The six current transformers can be converted into three (3 ) if dual windings are available to drive each MosFet pair.

If supplies are used, then the negative connected supply common for the three negative rectifiers can pump supply voltage to the positive rectifier gates supply storage capacitance

You need to look this as a half N MosFet bridge where the common point is the entrance of the AC source. - ( for each phase)

The dissipation of this system for practical purposes is just the Rdson of the MosFets since the frequency is low, so the switching frequency losses are just nil.

The MosFets are operated as diodes.

The current transformers need to be designed to operate at the lowest frequency of the wind mill's generator.

How many watts do You want to "SAVE" ?.

Remember that the generator is clamped by t he battery bank, the maximum efficiency may be around 50 or so %, a MUCH better solution is to use a half bridge and do a charge controller with MPPT to attain about 85 % efficiency with additional, in this case, of harvesting around 30 % higher power, plus the efficiency of the MPPT, are You for that task ?.

Nando

[willib]

while attempting the second method it occured to me that if the sync rect was dependent on current flow to operate , it would never start

ok back to the first method ,

the first method uses the magnetic flux directly from the magnets to signal when a phase is going positive.

let me clarify my first post a bit , an OP AMP solution IS needed to amplify the hall sensor output

the center tap

i'm just throwing some ideas out here

because in sync rectification i believe needs the center tap from a star/ wye connected stator to properly operate , since that point is the only true ground or neutral point in a balanced three phase machine.

and using the center tap could simplify the electronics needed to accomplish sync rectification.

hall sensors give the position of the rotor at any point in time thats what they are used for, actually

[Flux]

You have to use the mosfets in the 3 rd quadrant. Without gate drive the bridge will work as a conventional diode bridge using the body diodes. There will be current flow and you can use this initial current to switch in the gate drive to lower the volt drop. You can sense the drop in voltage or the flow of current as a trigger. I don't think you can force commutate it by sensing the magnetic field without causing overlap problems and many other nasty things.

If you force the commutation from something other than its normal voltage sensing you have something rather similar to sparking on the brushes of a dc machine with brush shift.

If you must force commutate it I think you will need a high frequency pwm drive.

Flux

[SamoaPower]

I don't see any way for your scheme to work. Rectification needs to take place at the LINE outputs of a star or delta connected three-phase system, not at the individual PHASE outputs. You only need to look at the phase relationship between LINE and PHASE voltages to see why detecting magnet or coil flux won't work.

To emulate the ideal diode, we need to detect the voltage polarity across the diode to within a few mV of zero to switch the FETs. Actually, true zero crossing detection is not desirable because of stability issues so a bias of a few mV should be used.

The NIS6111 from ON SEMI is an interesting device in a small package. It's a hybrid device combining a high speed comparator, voltage regulator, FET driver and a power FET all in a QFN package about 0.375" square and is self powered in AC applications. It's rated at 30A but has provision for driving additional FETs for greater current. The down-side is that it's only suitable for 12V battery systems.

For those that question the value of synchronous rectification, I only need to point out that for a single standard silicon diode at 100Adc, it's disipating about 120W where a FET version is about 10W. I want most of those watts going into my battery and loads rather than unwanted heat.

[inode buddha]

Not really related, but I've wondered about using germanium diodes instead of silicon due to the lower voltage drop. I do not know that anyone makes them with sufficient power handling.

[dinges]

Schottky-diodes have a lower voltage drop than silicon diodes, but slightly more than germanium. Schottky diodes are available in high power versions. A google on 'Schottky' in this board should give you more information. The subject has been discussed before.

[Nando]

To attain 10 watts @ 100 amps; the Rdson of the MosFet has to be 0.001 Ohms which will mean that You will need a very special Vds or to place several in parallel to obtain 1 milliohm.

If You have such device, please let me know the part number and manufacturer.

Also, one does not need to be so accurate with millivolts crossing, it would best to detect the initial current and turn the proper MosFet to reduce the body diode current with the lower Rdson values.

This is low speed switching, 100 HZ is low speed, so one or two amps current detection is quite well in the range to save power as I detailed the process in an earlier message for this thread.

Nando

[Nando]

Flux:

Since he want maximum power savings, he needs to have 6 MosFets ( 3 half bridges) to be able to do the minimum dissipation procedure, just in the third quadrant, just 50 % savings.

Nando

[stephent]

Is the power saved really that much using available (at a  decent cost) fet's with their I^2 R losses vs the I x E (voltage drop of diode) diode losses?

[SamoaPower]

Yes indeed Nando, multiple FETs are needed (for high current), but practical. The NTD110N02R FET is the same die used in the NIS6111. It's rated 110A (ha!) and 4.1mOhms Rdson and is in a small DPAK package. Digikey has it for $1.20 each (10). A 'diode' module using four of these plus the NIS6111 has an equivalent Rdson of <1mOhm and will fit on a 1 x 2" board (I'm doing it). Current sharing is good, within about 10%. Six of these modules in a three phase bridge is good for better than 100A DC output with excellent efficiency. Great for 12V systems.

I disagree with your comments on switch point detection. For best efficiency, you want the body diode(s) to conduct as little (time) as possible. Switching at a few mV does this. I see no advantage to using current transformers and their necessary interface when a comparator does it all. External supplies are not necessary.

[scottsAI]

Hello willib,

Guys, guys!

Nando said it right: Remember that the generator is clamped by the battery bank, the maximum efficiency may be around 50 or so %, a MUCH better solution is to use a half bridge and do a charge controller with MPPT to attain about 85 % efficiency with additional, in this case, of harvesting around 30 % higher power, plus the efficiency of the MPPT, are You for that task ?.

Nobody answered this question.

I will elaborate further, if cut in is low speed  then at 25mph the efficiency is like 25% or less.

Normal PMA has NO use for sync rect.

Unless boosting the voltage below cut in or some other wizardry like that.

Have fun,

Scott.

[willib]

Hi Scott , although the battery does keep the output close to 12V

it doesnt really clamp the voltage to that value, but the output voltage instead is dependent on the output current , if you are pumping 50A into a single 12V battery  the voltage of the battery will be considerably higher than 12V

[Nando]

The idea of your devices for low voltage makes the problem quite complicated electronically, detection plus protection of the circuitry.

One should look for simplicity, current detection, though initiated by the body diode of the MosFet does not conduct for long, it is used to initiate the MosFet turn ON, Do not discredit the principle without the proper analysis, just immediately saying that at little time as possible, why don't You calculate the gain and losses, then give the opinion, by the way that is the topology used for such power and temperature savings for a system with about 38 amps current (Not wind mill).

For a 12 volts with such high current, it would be better to produce higher voltage and do conversion with synchronous rectification with MPPT capabilities on the side.

IPB05CN10N is another possibility, I used very low Rdson devices in quantity, and quite familiar with those you indicate.

Nando

[SamoaPower]

Scott,

I didn't understand the relevance when Nando stated it nor when you reiterate it. I agree that there are better ways to design an alternator and interface, but what has that to do with wanting to reduce losses of rectifier diodes?

Use of synchronous rectifiers has no bearing on the system design, they're simply replacing lossy diodes no matter where or how they are used.

We often read about rectifier failures and their consequences. To state that a normal PMA has no use for synchronous rectifiers is just plain wrong. Anywhere we can reduce losses is a positive step.

 

[RP]

I've seen "half wave bridge" mentioned a few times.  Can someone please explain what that is in this context?  

[willib]

Flux , what do ya think of connecting the center tap of the star/wye connection to the ground as i've redrawn it?

this is from the second picture in your diary entry





it would make the individual phases less dependent on each other ?

oh i tried a AC cap ( rather large one phisically, 6.8uF though )between the my new minigen and the diode bridge , then i used PWM to adjust the current into a smaller voltage battery ( smaller the the gen output )

With the cap in place the output current increased for the same RPM

[willib]

correction -the cap was on just one phase(parallel to the phase ) , thats all i have installed so far

[scottsAI]

Hello willib, SamoaPower,

Thanks for the replies.

Power supply industry uses sync rectifiers to recover the power losses in the rectifiers.

PMA with sync rectifiers you will NOT recover the power loss in the rectifiers.

Removing the power losses in the rectifiers will not increase the power going into the battery.

If your using voltage boost or MPPT like then there would be a reason to use sync rectifiers.

To do something as expensive as sync rectifiers I would assume your deriving some benefit from doing so other than demonstrating you can do it! Heat is a consideration, but the applications I have seen, getting a heat sink large enough is not much cost or problem. I use sysn rectifiers in Automotive power supplies to reduce heat due to the small space I have to design with.

Using the inductance in the generator coils to boost the voltage is not a good idea unless the controller is directly on the output of the generator. Electromagnetic emissions will be a Huge problem the further away the controller is..

Have fun,

Scott.

[SamoaPower]

RP,

It's not "half wave bridge", it's Half-Bridge.

Perhaps this will help:

[SamoaPower]

Scott,

"PMA with sync rectifiers you will NOT recover the power loss in the rectifiers.

Removing the power losses in the rectifiers will not increase the power going into the battery."

I find these statements rather astounding.

For a given alternator speed, we have a certain open circuit voltage, Voc. Forcing volts, Vf = Voc-Vd-Vb where Vd is the rectifier diode(s) voltage drop and Vb = battery voltage. Current into the battery, assuming it's at a state of charge that allows it to accept available current, is Ib = Vf/Ra where Ra is the alternator resistance (neglecting Flux' constant and line loss). Power into the battery is Pb = Vb x Ib.  Power disipated in the rectifier is Pb = Ib x Vd.

A numerical example: Assume Voc=20V, Vb=13v, Vd=2.4V (two diode drops as in a bridge) and Ra=1 ohm. Then, Vf=4.6V, Ib=4.6A, Pb=59.8W and Pd=11W.

If we now replace the rectifiers with synchronous types of negligible voltage drop, Vf=7V, Ib=7A, Pb=91W and Pd=0. This clearly shows we have MORE than recovered the rectifier loss and increased the power into the battery.

I admit that this simplified analysis ignores a number of second-order effects, but the message is clear.

I'd be pleased if you would justify your statements.

[RP]

Cool.  Thanks!

[scottsAI]

Hello SamoaPower,

Astounding, nope. Interesting maybe!

Considering your past posts your going to enjoy this.

You painted a very good story, for a moment I though I screwed up. Nope!

For the answer we need to take a look at a bigger picture.

Your example used 13v so assuming 10' system. 36% eff.

Vb=13v, Vd=2.4V (two diode drops as in a bridge) and Ra=1 ohm.

For low currents the diode drops well be less, keeping fixed to make it easier to simulate.

Cut in is 5mph, available wind power is 18watts, want 1 amp charge current. Setting this up to establish the voltage needed at cut in is 16.4v = 13v + 2.4v + Vra (1 * 1).

For a given alternator speed, we have a certain open circuit voltage, Voc.

All other voltages are a ratio of this voltage based on the RPM.

For a given wind speed there is a limited available power out, if loaded greater then the available power out the RPM will drop lowering the forcing voltage resulting in matching the load to available power. The power will remain based on the wind speed not the RPM, the blades are being stalled a little, so the output power will be based on the profile of the blade. Here we will assume is within range of the blade and the efficiency has not changed.

This explains why with the generator shorted it spins slowly, also why in very high speed winds it can turn enough to burn out or at least spin faster. Efficiency at this condition sucks, we are out of the power profile of the blade, second reason it spins slowly.

Example: 6mph Pwind = 31 watts, 19.68Voc, 1.58a for 31 w.

V = 13 + 2.4 + 1*1.58 = 16.98v or 2.7v mis-match causing I to exceed 1.58a. But not goto 2.7a

RPM will lower until 17.2v is reached and I = 1.8 a for a 31 watt. Pb = 23.4w

Short diodes. Vd = 0. For 5.1v mis-match. I = 2.05amps. 0.25a different. Pb = 26.65w

Had to use a quadratic solver to get the numbers above.

Equation: RI^2 + 15.4I = 31 for the first, and RI^2 + 13I = 31 for second.

Example: 15mph, Pwind = 485 watts, 49.2Voc at 9.87a

With diodes: I = 15.6amps, 31.0v Pb = 202w. Pheat = 283 w

Short diodes, I = 16.46amps, 29.46v Pb = 213w. Pheat = 272 w

Pheat is the power as heat which is the diode and stator. With the short diodes the stator heat goes up by 283 - 2.4*15.6 = 283 - 37.44 = 245.56 or 26.44 watts.

Most systems furl in the early 20's mph, let's pick 22mph for the next example:

22mph Pwind = 1531watts, 72 Vos, 21.22amps.

With diodes: I = 32.2a, 47.6Vos, Pb = 418w

Short diodes: I = 33.2, 46.2Vos, Pb = 431w, 13w gain. 0.8%

3.25watts extra for your trouble for the first 10% gain, I knew the benefit near cut-in was the best, goes down hill from there. 11 watts extra for the second 2% gain.

I did over simplify it with my statement, I don't think your doing this for 3watts gain at cut in.

Some people have put series heaters in line lets try 2 ohms. Reports were current went down a few amps, 15mph I = 10.4amps. Pb = 135w, or 68w less. Current dropped more than I was expecting.

2*10.4^2 = 216w in the heater, stator heat is 108w or half. If the heat is as usable as charging the battery then the system is much more efficient.

I'm very glad we had this talk, I have a much better understanding. This took hours to write. What do you think?

Have fun,

Scott.

[Flux]

I am not sure what effect connecting the star point will have. Probably nothing except that you need another lead and have an additional source of RF interference.

At worst it may need a more complicated gate drive. I will try and let you know.

Flux

[Flux]

Scott

It's early in the morning and I haven't looked at this in great detail, but I think you are right. If you match the prop curve by introducing losses in the line, I don't think it matters much where they come from.

The volt drop in a bridge feeding a conventional boost converter is significant and the increase in efficiency using inductors before the bridge is worth having but that is another issue.

Without some form of MPPT the power you gain from keeping the prop out of stall far outweighs anything you gain from messing with electrical efficiency. Adding line resistance at first seems stupid, but it does really work.

The thing about this is that there is no easy way to see what effects these things have unless the improvement is considerable. It is so difficult to measure small changes in performance without weeks of monitoring that if you can see an improvement then the effect is dramatic. You will never see any difference just reducing diode drop even if it is real.

As you say it makes sense to include sync rectification as part of a MPPT scheme, but I don't think it is worth considering otherwise, too costly , complicated and probably unreliable.

We all try to squeeze the last bit of power from the system but sometimes it comes at too high a price to be of real benefit, but we enjoy the fun.

Flux

[SamoaPower]

Scott,

Looks like you put some real effort into this. If it helped clarify things for you, then it was worthwhile.

I don't think I can agree that the "big picture" helps to clarify the basic issue under discussion. In fact, introducing alternator characteristic variables tends to cloud the 'little picture'.

What your example process illustrates is a poorly designed stator that's not matched very well. To see what I mean, try running your process with a cut-in threshold (zero current) of 8 mph and compare the results. They're quite different.

To examine the value of synchronous rectifiers vs. conventional rectifiers, I still think we only need the simple model.

The largest gains are toward the low end which provides a useful benefit. Even around a typical energy peak speed, the benefit is significant. A lower cut-in also results which allows us to use fewer coil turns for a given cut-in speed.

Thanks for your comments.

Apologies to willib for getting carried away.

[Flux]

Willib

I should have spotted this immediately, there is no way you can connect the star point to dc- as shown.

If you look at the bottom half of the bridge you have the 3 mosfet body diodes connected directly as a short circuited 3 phase half wave rectifier. You have an expensive eddy current brake.

Flux

[willib]

oops

those darn body diodes , they help in some cases , and hinder in others

they are not true switches , they are switches with a sometimes pesky diode

[scottsAI]

SamoaPower,

Someone of your demonstrated understanding did not agree or understand what I wrote then it's my fault for not making it clear. I think Flux got it, but he qualified it:-) I know it was early!

My example was based on your numbers. They looked reasonable considering what I have seen on this board for a 10' PMA system.

If a 10% gain at cut-in or 0.8% gain at full load is worth the effort then it's worth the effort.

Yes, some things are just fun to do.

I think we are on topic, might not be answering willib question directly, but making sure willib and others reading this post understand the benefits of doing it or not is definitely worth our time and energy? Considering the topic is not that simple! At first and second glance, removing the rectifiers loss must be improving the system efficiency, turns out the secondary effects are more of an effect than one might think.

In my judgment for the trouble, better gains can be had by boosting the voltage or MPPT. Simple single inductor voltage booster for 100w allowing the cut in to start at 10mph... but start producing at 5mph, would cost less and be of much more benefit.

From the above examples 1 mph above cut in the Pb = 26.65 w of 31 w, we may have improved from 23.4, but there are still 6 watts not used, more than we gained, goes fast down hill from there.

At 15mph, 485w, Pb = 213 w, w/sync rect. That is 43% efficient. Why worry about a 2% gain using sync rect? Unless you want to do it for fun!

If the PMA RPM remained fixed based on wind speed and not load then the simple model would be sufficient. First we must agree for a given wind speed there is a fixed limited output power from a wind generator. Based on that Idea, the RPM will have to drop to limit the power, which changes all the calculations of the benefits of using synchronous rectifiers.

The power not lost in a synchronous rectifier is not fully recoverable in power that goes into the battery of a PMA wind generator. The modest wind speed of 15mph, I showed the gain is 11 watts, yet the rectifiers had 2.4 x 15.6 = 37.44 watts, or 29% of the diode power was recovered. I was assuming a perfect zero voltage diode replacement, which won't happen, so it's even worst.

Any body that has data on their system and wonders if sync rect would yield any benefit I would be more than glade to run their numbers and show before and after. I used this quadratic solver:

http://id.mind.net/~zona/mmts/miscellaneousMath/quadraticRealSolver/quadraticRealSolver.html

To get my email is above, is an IQ test to eliminate computers.

I hope others have reviewed my work and if any errors are found Please let me know. I want to make sure I have as good as understanding of the wind generators system as possible.

Yes, I did learn a lot form doing this and was totally worth doing it!

Have fun,

Scott.

[SamoaPower]

Scott,

There's not a problem in understanding your approach to the problem, it's that, in my opinion, it's not applicable to the issue under discussion. It's not uncommon in engineering circles, that some will choose an analytical technique, valid or not,  that will support their particular position on a topic. Usually, peer review will sort out the wheat from the chaff. However, since you brought this up, let's take a closer look at it.

"My example was based on your numbers."

Really? The numbers I used were arbitrary to illustrate the point. The only alternator related number you used, of mine, was Ra=1 ohm. You chose the rotor diameter = 10', rotor efficiency = 36%, and a cut-in of 5 mph @ 1A output (not really cut-in).

These describe a machine of poor design, perhaps better suited to a 24V system. The fact that your analysis requires a quadratic solution says that the rotor is in virtual stall throughout the range. The rotor is poorly matched to the alternator. If a rotor is well stalled, it's efficiency decreases considerably, so the resultant numbers in your analysis are not valid.

Furthermore, there seems to be a problem with how you calculate the percentage gain for the sync. rect. Using your numbers (even though not valid) for 22 mph, I see 3.1% instead of your 0.8%,(431-418)/418 = 0.031 and 5.4% at 15 mph. Hummm!

If one chooses a more reasonable stator of 8 mph cut-in, the gain at 15 mph is 17.6% and at 22 mph is 8.8% with better gains at lower speeds.

These numbers are all beside the point. Using this analytical process, you are looking more at wind generator design parameters rather than at the basic issue of whether use of synchronous rectification is worthwhile. If you introduce extraneous variables, only confusion results.

"I was assuming a perfect zero voltage diode replacement, which won't happen, so it's even worst."

True. However, for casual analysis, it's generally accepted that if the lower of two values is an order of magnitude or more lower, its insignificant.

"In my judgment for the trouble, better gains can be had by boosting the voltage or MPPT. Simple single inductor voltage booster for 100w allowing the cut in to start at 10mph... but start producing at 5mph, would cost less and be of much more benefit."

Agreed. What I will do personally on my up-coming machine, is use all of the above, including synchronous rectifiers. Depending on test results, I will use a three-phase synchronous bridge plus single boost converter with sync. rect., or a triple boost/rectifier, also using sync. rect.

I'm convinced that synchronous rectification is worthwhile for 12V systems.