3 phase inverter with SCR

[miso156]

Hello,

I would like to build a 3 phase inverter from old SCRs to control 12V car alternator running as a motor.
My expectation is to use typical 3 phase inverter:



With waveforms:



Are SCRs able to handle frequency from 30 to 100Hz and current about 50A?
If I connect more SCR parallel, can imbalance couse some problem, or do i need to put some serial resistance for every of them?
What is the typical way how to turn SCRs off?
Is possible to use some LC tank to force current to zero and turn the SCRs off?
Do I need to maintain V/f ratio? Means when the phases are swithed with lower frequcency I need to apply lower Voltage.
Can be lower voltage obtained by shorter pulses?

Thank you.

[Ground_Loop]

How will you turn off the SCRs

[miso156]

How will you turn off the SCRs

LC resonant circuit in series or in paralell to SCR will decrease the current under holding limit and turn it off. Or by adition SCR and discharging capacitor. See comution classes of SCR turning off.

[jmelson]

How do you commutate those SCRs?  You can apply a gate pulse to turn them ON, but you can't turn them off until some combination of input waveform and output voltage/current
allows each SCRs current to pass through zero.  I can't see anything that would turn them off with a DC input.

Jon

[james_s]

Is there a reason you want to build something with 1970s technology instead of just buying one of the many cheap and readily available brushless DC motor controllers that are available?

[miso156]

Is there a reason you want to build something with 1970s technology instead of just buying one of the many cheap and readily available brushless DC motor controllers that are available?

You are of course right, but first I want something to learn, I have nothing to do during nights. And second, its easy to make something from very good components like Mosfets, but i preffer more "softer swithing" where the fining tune has to by applied. 

[schmitt trigger]

I applaud your desire to do power electronics the hard way. Meaning with SCRs.

The only three phase SCR bridges I built in the early 80s were all naturally commutated. Force commutated bridges I only built single phase, and they are tricky.

My recommendations, based on a lot of sweat and tears (and some nasty explosions):

-SCRs are commutated by having its current decrease to zero. This means that the LC tank, pilot SCR or whatever you are using to force commutate the device, has to be able to provide the full load current magnitude, for the turnoff period.
-For reliable commutation, you have to use inverter grade SCRs. Regular phase control SCRs have widely variable commutation times. Back then, I would only use SCRs from Semikron.
-The commutating capacitors will see extremely large dI/dt pulses. Regular capacitors tend to fail prematurely. Back then General Electric made some oil filled caps suitable for inverter purposes.
-Fuse each SCR individually, THIS IS A MUST, with a fuse capable of quenching large fault currents. You want to limit the damage when a SCR fails or misfires.
-The gate driver should be very robust, and provide good insulation. Back then Coilcraft had some excellent SCR gate drive transformers.
-Speaking of gate drive, use a picket fence waveform. Don't ever rely on a single pulse for the SCR to latch.

[Zero999]

Sorry this is a bad idea. SCRs are not ideal at such a low voltage, because they have a high voltage drop, compared to MOSFETs and a car alternator will give a rubbish efficiency anyway. A good old brushed motor will give a higher efficiency than this and is much easier to control.

[miso156]

Sorry this is a bad idea. SCRs are not ideal at such a low voltage, because they have a high voltage drop, compared to MOSFETs and a car alternator will give a rubbish efficiency anyway. A good old brushed motor will give a higher efficiency than this and is much easier to control.

I know.  When I look at KTxxx thyristors they say 1.4V, but BJT like BD911 are also high 1-2V and when hfe is 20 the base current is  also not neglibigle from losses point of view.

The car alternator is only temporary solution, next i will buy some motor with higher voltage rating.

Thx anyway.

[miso156]

The only three phase SCR bridges I built in the early 80s were all naturally commutated. Force commutated bridges I only built single phase, and they are tricky.

-For reliable commutation, you have to use inverter grade SCRs. Regular phase control SCRs have widely variable commutation times. Back then, I would only use SCRs from Semikron.

Naturaly commutated, do you mean using AC supply voltage where do you can expect zero cossing for sure?
Where do you see problems in forced comutation?
What happend if SCRs in bridge have variable commutation times? - If I use LC network for comutation, I can expect much higher tolerances in LC components.

Thank you.

[MosherIV]

Naturaly commutated, do you mean using AC supply voltage where do you can expect zero cossing for sure?

Yes.
SCRs can be turned on by the control gate but not turned off.
The way to turn off SCRs is for the supply voltage to reach 0V, zero switching as you refer to. Hence they are mainly used to control AC circuits.

Your suggested circuit may work with transistor or mosfets BUT you will not be able to get negative part of AC waveform.

[Zero999]

Sorry this is a bad idea. SCRs are not ideal at such a low voltage, because they have a high voltage drop, compared to MOSFETs and a car alternator will give a rubbish efficiency anyway. A good old brushed motor will give a higher efficiency than this and is much easier to control.

I know.  When I look at KTxxx thyristors they say 1.4V, but BJT like BD911 are also high 1-2V and when hfe is 20 the base current is  also not neglibigle from losses point of view.

The car alternator is only temporary solution, next i will buy some motor with higher voltage rating.

Thx anyway.

Apart from the fact that the BD911 is far too small to drive an alternator, BJTs are also unsuitable for this application, hence why I mentioned MOSFETs, which have much lower on losses.

A quick flick through the RS-Components catalogue gives the IRFS7430, which has a maximum on resistance of 0.75mΩ, so hardly any voltage loss even at hundreds of Amps.
https://docs.rs-online.com/cba3/0900766b813c79c2.pdf

What voltage are you eventually going to be running the motor at? If it's over 600V, then you can consider IGBTs, which have similar on losses to SCRs, but are much easier to switch off.

[Circlotron]

This is the kind of project that separates the men from the boys!
The fact that the car alternator is being used as a synchronous motor means that if the motor has sufficient rotor excitation it will present a leading power factor to the inverter.  That means extra commutation components are not needed. The SCRs will commutate naturally. I have never built one like this. My experience was with DC choppers some 35 years ago.

https://www.google.com.au/search?source=hp&ei=60w9XouhJrfFz7sPjou82AQ&q=scr+inverter+synchronous+motor&oq=scr+inverter+synchronouis+motor&gs_l=psy-ab.3.0.33i22i10i29i30.950.11839..13552...0.0..0.247.5682.0j26j5....2..0....1..gws-wiz.......0i131j0j0i22i30j0i333j33i160j33i21.WnFEcZpJnx8