Understanding MOSFET datasheets

[Boardburner2]

While I favour the video approach.
There is something sad about a circuit that just gives off a whisp of smoke.
With proper design there is a big flash and bang to be had here.

Can I suggest you post your circuit here before applying power.

You may find a critique by more experienced arsonists to be helpful.

[localbroadcast]

Do the math.

Using an RC filter with PWM is not really a switch mode regulator.

Yes.. I have done the math.  And yes, a switch mode voltage regulator can be had by controlling a MOSFET with PWM and a gate driver, with the output filtered any number of ways.  I have attached a PDF document that outlines the topologies of pretty much every switch mode power supply imaginable.  It's a great document that is well worth the read.  It outlines the operation of the different switch mode regulator designs, as well as outlines the equations used for design calculations.  Also helps engineers decide which type of switch mode regulator to use for their specific needs.

In many situations, a simple lowpass RC filter is sufficient.  The specs of the resistor and capacitor can be calculated to provide the proper cutoff frequency, desired amount of output ripple, etc.  The order of filtration will determine the quality of filtration achieved.  A simple RC filter would be a 1st order passive filter.  Of course, we can make things more complex and achieve an even smoother output by increasing the filter order, using an inductor as well as capacitor and resistor, or even using active filtration with op amps instead of just passive filtration.  You will have to decide yourself what filter design is best suited for your needs.  I think a first order RLC filter will suffice.  Follow the links at the bottom for more information on PWM filter design.  These are very informative PDF files that really help wrap your head around how different filters work, and how to achieve the best efficiency for your system.

One of the more important facts that I came across when researching filtration was this: The higher the PWM frequency used, the smaller the capacitance will need to be, since capacitor reactance is inversely proportional to the frequency.
This is nice to know, because it tells us that the higher the PWM frequency we can attain, the more efficient we can make our filter, since we can minimize the capacitor and resistor sizes.  Of course, the frequency is limited by the switching speed of the MOSFET and the micro-controller being used..  So now we know why it would be a good idea to aim for a MOSFET with the best switching speed characteristics possible, as well as a low R_ds(on) value as always.

Enjoy the reading material!

Filtering PWM signals by Jim Wagner
TI - FilterPro Users Guide

Polymorph.. I hope this clues you into what you haven't quite been understanding.  Please refrain yourself from posting anything else in my thread unless you know for sure that your information is accurate.  I hate to see so much false information being spread around.. It confuses everyone.. Especially people just starting out in electronics.

[JimboZA]

my thread

That's an interesting take on things... worthy of discussion somewhere one day.

[polymorph]

I hate to see so much false information being spread around.. It confuses everyone.. Especially people just starting out in electronics.

How ironic.

[localbroadcast]

The links you gave earlier were for RC lowpass. An RC lowpass filter on PWM will give -just- as much in losses as a linear regulator.

Do the math.

Using an RC filter with PWM is not really a switch mode regulator.

I dunno.. but I doubt you can back these claims up...

[MarkT]

600V 20A IGBT 3-phase bridge with gate drivers and protection circuitry: job done.

http://uk.farnell.com/fairchild-semiconductor/fsbb20ch60f/ipm-igbt-20a-600v-spmca-027/dp/2322642

[ well I think its a more rational approach to the problem - these modules are extremely
handy ]

[polymorph]

Read a book, in between insulting me. Perhaps you can show me in one of these documents where a switch mode power supply uses a resistor as part of the smoothing network.

http://www.st.com/web/en/resource/technical/document/application_note/CD00003910.pdf

http://ww1.microchip.com/downloads/en/AppNotes/01114A.pdf

In SMPS, the series element, RS, is replaced by a
semiconductor switch, which offers very low resistance
at the ON state (minimizing conduction loss), and very
high resistance at the OFF state (blocking the
conduction). A low-pass filter using non-dissipative
passive components such as inductors and capacitors
is placed after the semiconductor switch, to provide
constant DC output voltage.

http://www.smps.us/topologies.html

http://www.onsemi.com/pub_link/Collateral/SMPSRM-D.PDF

http://www.nxp.com/documents/application_note/APPCHP2.pdf

http://www.element14.com/community/servlet/JiveServlet/previewBody/27614-102-1-76799/01207A.pdf

You are on your own.

[rockwallaby]

Localbroadcast, I think there is going to be a problem with the overall concept as I understand it which I now recall I had as well.

First to confirm, you intend to run the 3 phase induction motor as generator, IMAG, with the correct capacitors across the windings to provide the excitation flux to be able to self generate AC power.

Then to this, you intend to connect a switchmode power supply across either one, two or all three phases to then produce the 36 volts at some current.

Is this correct?

If that is correct, then there will be a problem if you place any form of rectification directly across any of the generator phases and you draw current.

The rectification of the generator AC will distort the generated waveform so badly that is will cause the self excitation to stop.

An IMAG will work reliably with a low frequency transformer based system across its phases and then into rectification at the secondary voltage.

If you place a capacitor in series with such a transformer primary winding, then you can reliably start and stop your turbine as the reactance of the capacitor will be high at low turbine speeds, thereby making the transformer out of circuit.  When the turbine gets up to speed, the capacitor reactance becomes low and then the transformer will then be in circuit. The capacitor will in theory limit the current that can flow through the transformer primary, but in reality this should not be a problem for these smaller power systems.

I only recall now that I found this to be the case when I placed a standard industrial SMPS across one of the phases of my turbine. As soon as I started to draw any significant current, I would loose all generator excitation.

Not sure how this effects your plans or how best you can work around it?

What size induction motor do you have for this, is its capacity around the 500W to 1kW mark?
Is it a 2 pole or 4 pole or 6 pole motor?
____
Paul

[localbroadcast]

Read a book, in between insulting me. Perhaps you can show me in one of these documents where a switch mode power supply uses a resistor as part of the smoothing network.

I already posted tons of links for you to peruse on the matter of filtering.  I've got enough information to work with, thanks for trying to help tho.

I plan to test the performance of a couple different filter designs.  One will be a simple straight forward RC single pole filter.  resistor in series with load, capacitor in parallel.

PWM frequency = 10kHz
R1 = 1 ohm
C1 = 10000uF

from my calculations, here's the results I'll get..

settling time = 23 ms roughly
peak-peak ripple voltage = 0.3v roughly
cut-off frequency = 16hz roughly
With a 3 amp load, I'll be losing 3 watts from the resistor.  Not too bad for a 3 amp load.

Transfer function:
G(s) = 100 / s+100

The other filter I was going to try looked more like this:

Inductor in series with load
Capacitor in parallel with load

L1 = 16 uH
C1 = 10000uF

This will give me a reactance from L1 of 1 ohm, so hopefully similar losses as the RC filter, close to 3 watts roughly.. give or take of course.


The final (and probably most successful) filter I will try will be more like this:

Inductor and resistor in series with load
Capacitor in parallel with load

L1 = 42 uH
C1 = 10000uF
R1 = 0.1 ohm

I'll be running through testing with all these different filter combinations to figure out the actual losses developed in real time, not just theoretical models.  Also want to take a look at the wave forms produced on the oscilloscope.  Should be very interesting!

[Boardburner2]

The rectification of the generator AC will distort the generated waveform so badly that is will cause the self excitation to stop.

____
Paul

At last.

Its one of the reasons that small wind turbines for charging batteries use permanent magnet generators.
The other, variable speed does not apply here.

[Boardburner2]

The rectification of the generator AC will distort the generated waveform so badly that is will cause the self excitation to stop.

____
Paul

At last.

Its one of the reasons that small wind turbines for charging batteries use permanent magnet generators.
The other, variable speed does not apply here.

With some capacitors connected to the windings, once it gets going, it self-excites and outputs a good amount of power without any issues.  I'm

When you rectify that the smoothing cap on the other side is effectively in parallel with the cap used for the excitation field for part of the cycle.

Have you considered what the effect of the pwm load will be ON THE MOTOR/generator.

IMAG do not like capacitive loads as Rockwallaby has discovered

[rockwallaby]

IMAG do not like capacitive loads as Rockwallaby has discovered

You do need the excitation caps in there and I do also have caps in series with certain loads, even my main shunt load of around 300Watts has a cap of around 45uF at 380VAC to give DC isolation when I start or stop the turbine.

Having these capacitors in series with resistive loads keeps the residual magnetism in the IMAG.

Placing more capacitors in parallel with the excitation caps will change the reactance and hence the volts you get and also unbalance the whole thing, but generally, it doesn't have a drastic effect such that rectifiers do across any phase or phases.

It is the forward conduction of the rectifier that causes the colapses the excitation flux, as the current at this point is relatively high and it is being effectively shorted out.

Most times, you also kill the residual megnetism in the rotor as well, which can be brought back by putting a good old 12volt battery across a phase for a few seconds while the turbine is not rotating.
Sorry for spelling, it's after 02h here  
____
Paul

[Boardburner2]

It is the forward conduction of the rectifier that causes the colapses the excitation flux, as the current at this point is relatively high and it is being effectively shorted out.

____
Paul

Yes the capacitor has to recharge over a very short part of the cycle.
it results In a very high Ipk.

Once charged it has no effect on operation as the motor does not see it until current is drawn.

I wondered if using an inductive filter would solve it but did not get around to experimenting.

I am thinking that what is required is a current to voltage topology.
AFAIK there are no topologies that can do that in a single stage.

[alka]

i just wanted to understand mosfet datasheets...

[Boardburner2]

i just wanted to understand mosfet datasheets...

My apologies for hijacking your thread.

I have in the past done similar with smaller similar setups.

When it comes to switch mode supplies there is a lot to be considered and it can get complex.

The buck converter mentioned works well with a voltage  source.
As already mentioned a voltage source from imag could result in loss of excitation.

A generator is often considered to be a voltage source especially the infinite grid at fixed frequency.

However a generator can be a current source also and your previous mention of experience with variable frequency drives makes me wonder if that has a bearing on what you are trying to do.

A previous argument elsewhere (in depth over many years)indicate that the self excited imag generator is a special case.
I did not understand the maths.
The previous argument involved a proposed current to voltage convertor but the proponent who successfuly did it got the huff with us and would not describe how it worked.

I just used a permanent magnet which worked fine.


Experimenting with pwm regulation could well give you some very difficult to understand results ,some of which could result in flash and bang rather than a whisp of smoke as I previously posted.

There are some here who may be able to help but you would have to post more detail.

A mosfet is a fantastic device for switching large currents with low resistance loss .
When used in switched power supplies there is more to be considered.
I do not think anyone here is being unhelpful but on its own the mosfet is not the only thing to be considered.

Often when people ask for help the assumption is they ask for help because they assume that people here know more , then argue.
It could be that in this instance you are the professor and we just do not know.


edit
I just reread that.
Style was not as intended.
Don't want to sound pompous or so.
But when considering switch mode supplies the power source has to be considered as well as the output.