Asymetric half bridge

[nix85]

I im trying to understand how AHB inverter works, seems very simple, but direction of diodes doesnt make sense. Can you please elaborate exactly what happens when transistors are ON and what exactly happens when they are OFF and cap is supposed to discharge producing another halfphase of the sinewave.



This circuit is apparently most used to run stepper motors.

Another illustration of the same circle.

https://postimg.cc/nsssS6tB

Is the circle above even correct?

This is where i heard about it, comment below the first answer.

https://www.quora.com/How-suitable-is-a-half-bridge-converter-for-DC-AC-power-conversion-within-a-certain-kW-range

Quoting the guy:

"I have used a Asymetric halfe bridge (see picture) or motor control of an SRM

I think the motor was rated to about 40kW. We used a DC voltge of 540 V and an RMS current of approcimately 270 A. Nominal speed and torque was 14300 RPM and 30 Nm. A full bridge would have caused excessive switching losses and an increased risk of switching failure effectively resulting in short circuitting

asymmetric half bridge is also capable of supplying an alternating waveform. Both transistors on results in positive voltage where as both transistor off results in negative voltage and either transistor on results in free wheeling.

http://www.mdpi.com/sensors/sensors-17-01146/article_deploy/html/images/sensors-17-01146-g002.png
"

This last link is obviously different with additional diodes in parallel with transistors. Can someone clear this up?

Another illustration of similar circle without cap in parallel.

[bob91343]

Turning on both transistors causes current to flow in the inductor.  Turning off both transistors causes the inductor current to divert through the diodes and charge the capacitor.  When the current drops sufficiently the diodes stop conducting and the circuit is ready for the next pulse.

Current always flows in the same direction through the inductor.  When the transistors are on, the magnetic field of the inductor causes the motor to move.

So you pulse the transistors to get one jump of the motor.  Wait for discharge, then pulse again.  Motor jumps again, same direction as before.

[nix85]

I just want to know can this circle be used to produce AC from DC, i don't need it to run a stepper motor of course.

Are you saying winding of the motor (what you call "inductor") comprises the integral part of the oscillating circuit? Hmm, hope to clear this up.

[nix85]

When i look at this diagram that seems to be the same only missing cap in parallel it makes more sense, im not sure why, but if we add a cap in parallel, i can see convetional current flowing

through Q1, load, Q2 to the ground, then transistors turn off, and cap discharges in reverse direction through D2, load, D1 and into itself, cycle repeats, but this is not sinewave but pulsed dc on the load. But if we just reverse how D1 and D2 are connected to the load, we do get sinewave on the load.

[bob91343]

As drawn, it doesn't oscillate.  You need a circuit to drive it.  You can't leave the transistor bases open in a practical circuit.

[nix85]

As drawn, it doesn't oscillate.  You need a circuit to drive it.  You can't leave the transistor bases open in a practical circuit.

It is assumed transistors are driven by dedicated drivers.

[MagicSmoker]

I im trying to understand how AHB inverter works, seems very simple, but direction of diodes doesnt make sense. Can you please elaborate exactly what happens when transistors are ON and what exactly happens when they are OFF and cap is supposed to discharge producing another halfphase of the sinewave.
...
This circuit is apparently most used to run stepper motors.

The asymmetric half-bridge is also used to drive switched reluctance motors and is also known as the "two-switch forward" when used as a switchmode power supply topology. It can only create unipolar pulses of current so it is not, technically, an inverter. The diodes simply provide a pathway for energy stored in the various inductances (magnetizing, leakage and stray) to return to the input supply automatically, which improves both efficiency and reliability, since the switches never experience a voltage higher than the supply (plus two diode drops, of course).

[nix85]

Since PS is on only half of the time, it's clear why full bridge is a better choice for any significant power.

I see what you mean about returning energy to PS from inductive loads, but i'm not sure if that's the purpose of these 4 diodes across transistors here.

And why the cap in parallel?

[T3sl4co1l]

I just want to know can this circle be used to produce AC from DC

Certainly, as you will measure waveforms while it is operating.

AC for what purpose?

Tim

[nix85]

I just want to know can this circle be used to produce AC from DC

Certainly, as you will measure waveforms while it is operating.

AC for what purpose?

Tim

Purpose is to produce 230V AC RMS from ~325V DC. I would prefer the answer to my last two questions, why diodes across transistors in full bridge - they are absent in some variations, and why cap in parallel.

[MagicSmoker]

Since PS is on only half of the time, it's clear why full bridge is a better choice for any significant power.

The full-bridge (and "true" half-bridge) do operate as inverters and so are not direct substitutes for the asymmetric half-bridge.

I see what you mean about returning energy to PS from inductive loads, but i'm not sure if that's the purpose of these 4 diodes across transistors here.

The anti-parallel diodes across the switches in the half/full bridge perform the same function, except they only come into conduction when all of the switches are off (ie - during the "dead time").

And why the cap in parallel?

Bridge inverters are derived from the buck converter so draw current from their supply in rectangular pulses. Without an input reservoir/decoupling capacitor present, massive spikes and ringing would be excited in the stray inductance of the supply wiring, leading to instantaneous failure of the switches.

[nix85]

Bridge inverters are derived from the buck converter so draw current from their supply in rectangular pulses. Without an input reservoir/decoupling capacitor present, massive spikes and ringing would be excited in the stray inductance of the supply wiring, leading to instantaneous failure of the switches.

So, if FETs are driven with sine, as i intend to, not square, then cap is not really needed, as there are no sudden voltage changes or high harmonics?

PS. I know square is more efficient as FETs are always either on or off, not in linear region heating up, but i dont intend to tinker with PWM, comparators and all the rest, i will provide a high quality fast drivers and big sink and hope for the best.

[T3sl4co1l]

The diodes aren't relevant then.  You are making a linear amplifier, not a switching supply.

Use big transistors.

Tim

[MagicSmoker]

So, if FETs are driven with sine, as i intend to, not square, then cap is not really needed, as there are no sudden voltage changes or high harmonics?
...

Wait, what? Setting aside the absurdity of this for anything but an audio amplifier (and even then FETs are inferior to BJTs as far as linearity is concerned), you still need supply decoupling - would you design an amplifier without reservoir capacitors on the supply lines? - and anti-parallel diodes if the load is reactive (like a transformer primary or a speaker voice coil).

[nix85]

Wait, what? Setting aside the absurdity of this for anything but an audio amplifier (and even then FETs are inferior to BJTs as far as linearity is concerned), you still need supply decoupling - would you design an amplifier without reservoir capacitors on the supply lines? - and anti-parallel diodes if the load is reactive (like a transformer primary or a speaker voice coil).

I made it clear goal is not audio amp but inverter. There's gonna be a big capacitor input filter (15000uF and half to 1H choke)  before the inverter circuit. I assure you supply is resilient to any spikes, no fear of it getting burned whatsoever, but i was anyway planning to include diodes accross FETs and reservoir caps in parallel just to be sure.

Also, i didn't know BJ behave better than FETs in linear mode, i think i will still go with FETs for efficiency.

[MagicSmoker]

...
I made it clear goal is not audio amp but inverter.

Not much difference between the two if you are operating the bridge in linear mode...

Also, i didn't know BJ behave better than FETs in linear mode, i think i will still go with FETs for efficiency.

Efficiency of the output device is all but irrelevant in linear mode. You are going to have problems with gate ringing/destructive oscillation if you stick with FETs here. Not only that, you have to use FETs intended for linear operation (typically "lateral DMOS" construction), as the far more common switching FET (vertical construction) doesn't play nice when operated with significant drain-source voltage drop. Also, the gate threshold voltage in switching FETs has a negative tempco, so the device turns on harder as it gets hotter, and to a greater extent than the positive tempco of the drain-source resistance. In other words, thermal runaway is just as much a problem for switching FETs used in linear mode as second breakdown is for BJTs.

If the goal is to make a sine wave then it really is easier to sinusoidally modulate a bridge and use an LC or LCL filter to reconstruct the sine wave for the load. That is how pretty much every "pure sine" inverter works (VFDs typically rely on the inductance of the motor windings to produce a sinusoidal current; nothing is done to integrate the voltage waveform*).



* - there are exceptions, of course; if the motor is located some distance from the VFD and/or there is a problem with bearing erosion then an LC or LCL filter will be added to the VFD output.

[nix85]

...
I made it clear goal is not audio amp but inverter.

Not much difference between the two if you are operating the bridge in linear mode...

Also, i didn't know BJ behave better than FETs in linear mode, i think i will still go with FETs for efficiency.

Efficiency of the output device is all but irrelevant in linear mode. You are going to have problems with gate ringing/destructive oscillation if you stick with FETs here. Not only that, you have to use FETs intended for linear operation (typically "lateral DMOS" construction), as the far more common switching FET (vertical construction) doesn't play nice when operated with significant drain-source voltage drop. Also, the gate threshold voltage in switching FETs has a negative tempco, so the device turns on harder as it gets hotter, and to a greater extent than the positive tempco of the drain-source resistance. In other words, thermal runaway is just as much a problem for switching FETs used in linear mode as second breakdown is for BJTs.

If the goal is to make a sine wave then it really is easier to sinusoidally modulate a bridge and use an LC or LCL filter to reconstruct the sine wave for the load. That is how pretty much every "pure sine" inverter works (VFDs typically rely on the inductance of the motor windings to produce a sinusoidal current; nothing is done to integrate the voltage waveform*).



* - there are exceptions, of course; if the motor is located some distance from the VFD and/or there is a problem with bearing erosion then an LC or LCL filter will be added to the VFD output.

Tnx for info, but "reconstruct the sine wave" is not an option here, i said PWM, comparators etc i am not getting into, at least not now and for this altho i know that is a better way. It's beyond my present knowledge and everything else for this project.

What do you think about IGBTs, do they suffer those termal issues you mention?

EDIT: Im not sure if IGBTs can even work in linear mode, i read somewhere they can altho uncommon.

[MagicSmoker]

...
What do you think about IGBTs, do they suffer those termal issues you mention?

EDIT: Im not sure if IGBTs can even work in linear mode, i read somewhere they can altho uncommon.

The only switch I can think of that would be worse for linear use than the IGBT is the thyristor. An IGBT is basically a MOSFET driving a PNP BJT with the added complication that the MOSFET only has direct control over turn-on; turn-off depends primarily on recombination time of the charge carriers (ie - it is slow).

[T3sl4co1l]

There are apparently IGBTs suitable for (if not necessarily _specified_ for) linear operation.  But MOSFETs or BJTs are better here.

BJTs are better at somewhat lower voltages, but I think you can find some with an SOA up there still.

MOSFETs and BJTs have no palpable difference in efficiency.  You're making, at best, a class B amplifier; your peak efficiency will be around 60% (I forget the exact calculation, is it up to 66%?).  Getting up to 50% is practical.

You are building an audio amplifier, whatever you may think of it -- a linear sinewave amplifier is just like any other amplifier.  The fact that you're using it at a fraction of its bandwidth, doesn't really help in any practical way.  (Even if you have more of those 1H high current chokes, and they're high enough Q to make essentially an RF amplifier but at line frequency.)

What the heck is the massive choke and cap for?  That sounds like more of a liability than anything to me.  If I were designing an inverter to run from such a supply I would at the very least add a very meaty TVS to the supply, in case that inductor lets its energy out unconstrained (say under fault conditions).  Likewise for the capacitor, a fuse at least but preferably some means of limiting fault current as well, if for no other reason than to minimize damage when something dies.

Just how much power are you looking to get out of this thing?

Tim

[nix85]

An IGBT is basically a MOSFET driving a PNP BJT with the added complication that the MOSFET only has direct control over turn-on; turn-off depends primarily on recombination time of the charge carriers (ie - it is slow).

Knew that, but beware there are IGBTs running up to 150KHz if not more.

[nix85]

There are apparently IGBTs suitable for (if not necessarily _specified_ for) linear operation.  But MOSFETs or BJTs are better here.

BJTs are better at somewhat lower voltages, but I think you can find some with an SOA up there still.

MOSFETs and BJTs have no palpable difference in efficiency.  You're making, at best, a class B amplifier; your peak efficiency will be around 60% (I forget the exact calculation, is it up to 66%?).  Getting up to 50% is practical.

You are building an audio amplifier, whatever you may think of it -- a linear sinewave amplifier is just like any other amplifier.  The fact that you're using it at a fraction of its bandwidth, doesn't really help in any practical way.  (Even if you have more of those 1H high current chokes, and they're high enough Q to make essentially an RF amplifier but at line frequency.)

What the heck is the massive choke and cap for?  That sounds like more of a liability than anything to me.  If I were designing an inverter to run from such a supply I would at the very least add a very meaty TVS to the supply, in case that inductor lets its energy out unconstrained (say under fault conditions).  Likewise for the capacitor, a fuse at least but preferably some means of limiting fault current as well, if for no other reason than to minimize damage when something dies.

Just how much power are you looking to get out of this thing?

Tim

Now that you say it's basically a class B amp with ~50% efficiency those bulky and expensive D-class inverters start to look not so bad afterall.

I see class B amp is theoretically up to 78% efficient but usually like you said about 50%. I am looking into differences between class B amp and full bridge inverter, first and obvious one is the latter uses 4 transistors instead of two. I would appreciate if you elaborate a bit on differences.

Massive choke and cap are smoothing pulsed DC from the supply into DC. Don't know the exact power right now, we are talking several kW.

[nix85]

BUMP

[MagicSmoker]

Knew that, but beware there are IGBTs running up to 150KHz if not more.

Check those datasheets a bit more carefully because they are all but certain to say that operation above 50kHz or so requires soft-switching techniques (ie - resonant or quasi-resonant operation). For hard-switched operation, 40-60kHz is usually considered the usual upper limit for state-of-the-art, 600-650V plastic-package IGBTs before switching losses exceed conduction losses. In other words, you can hard-switch a 600V/50A IGBT at 150kHz, but you might have to limit current to 5A to keep the junction temperature rise manageable.

...I am looking into differences between class B amp and full bridge inverter, first and obvious one is the latter uses 4 transistors instead of two. I would appreciate if you elaborate a bit on differences.

A totem-pole class-B amplifier is basically the same circuit as a half-bridge inverter, so a full-bridge inverter is equivalent to operating two class-B amplifiers in bridge-tied load configuration (this is popular in car audio amplifiers; not so much elsewhere).

...I humbly suggest you to start learning about overunity devices, you are already 100+ years behind on actual developments. Here's a good book to start your research. It's only 2555 pages long, so loads of fun and enlightment ahead for you.

http://www.free-energy-info.com/PJKbook.pdf

Saw this in the other thread you started about the current-fed parallel-resonant push-pull (erroneously attributed to "Mazzilli"). The management, and most of the denizens here, take a dim view on those who flaunt the laws of thermodynamics.

[Yansi]

I im trying to understand how AHB inverter works, seems very simple, but direction of diodes doesnt make sense. Can you please elaborate exactly what happens when transistors are ON and what exactly happens when they are OFF and cap is supposed to discharge producing another halfphase of the sinewave.



This circuit is apparently most used to run stepper motors.

Another illustration of the same circle.

https://postimg.cc/nsssS6tB

Is the circle above even correct?

This is where i heard about it, comment below the first answer.

https://www.quora.com/How-suitable-is-a-half-bridge-converter-for-DC-AC-power-conversion-within-a-certain-kW-range

Quoting the guy:

"I have used a Asymetric halfe bridge (see picture) or motor control of an SRM

I think the motor was rated to about 40kW. We used a DC voltge of 540 V and an RMS current of approcimately 270 A. Nominal speed and torque was 14300 RPM and 30 Nm. A full bridge would have caused excessive switching losses and an increased risk of switching failure effectively resulting in short circuitting

asymmetric half bridge is also capable of supplying an alternating waveform. Both transistors on results in positive voltage where as both transistor off results in negative voltage and either transistor on results in free wheeling.

http://www.mdpi.com/sensors/sensors-17-01146/article_deploy/html/images/sensors-17-01146-g002.png
"

This last link is obviously different with additional diodes in parallel with transistors. Can someone clear this up?

Another illustration of similar circle without cap in parallel.

That is NOT an asymmetric half bridge!  By asymmetric half-bridge, the typical terminology means really a half-bridge with asymmetric switching times for the low and high side transistor (such as a  step-down (buck) converter).

What you have is a full bridge two quadrant switch, working in quadrants I and IV.

And, this is not used to drive stepper motors, but SRM (synchronous reluctance motor), which okay, are also kind of "stepper" but not really the kind of stepper 99% of people will think of.

Also, this two quadrant switch is a pretty common in forward switching converters,  such as welders.

[MagicSmoker]

...

That is NOT an asymmetric half bridge!  By asymmetric half-bridge, the typical terminology means really a half-bridge with asymmetric switching times for the low and high side transistor (such as a  step-down (buck) converter).
...

I totally forgot about this and arguably it is the more correct definition of asymmetric half-bridge, but note that many people - including companies like STMicro - refer to the two-switch forward as the AHB. I'm not saying they are right, just that it happens. I, personally, prefer two-switch forward as it is less ambiguous about how the circuit functions. The AHB is only different from the HB by virtue of the gate timing (ie - complementary drive signals are used, instead of simultaneous).

[Yansi]

Hey! Can you refer to the exact place where they call that an ASM?  I know a bunch of the people from the IPD in person, I'll gladly let them some feedback

[MagicSmoker]

Hey! Can you refer to the exact place where they call that an ASM?  I know a bunch of the people from the IPD in person, I'll gladly let them some feedback

Here's one: https://www.st.com/en/applications/industrial-motor-control/asymmetrical-half-bridge-pwm-drive.html

Note the block labeled "power stage" - it shows a two-switch forward while the text describes it as an "asymmetrical half-bridge."