AC Polarity swapping circuit

[m_alizd]

Hi Folks!

I'm quite excited to join here today.
I'm new to power electronics and out of interest working on a simulation on a 32Vac line driving a low power load (less than 30W). I'd like to control negative and positive sinewave cycles going to the load and chose what cycle to go next. For example I might have 10 full normal cycles going, and then at some point choose to send 2 negative cycles in a row. Ideally I'd like to have full control over pattern. Only requirements is this to happen in zero crossing and the switching time to be no more than 1ms, ideally in microseconds scale.
I looked at it as if I am multiplexing an AC source line and swapped polarity version of it onto the load with some magical AC line multiplexer.
I'm wondering:
1. Is there anything fundamentally wrong with switching polarities to any load?
2. What's the best and cheapest way to implement this?
2. Assuming use of TRIACs there any circuit which achieves the same job?

Here is the most simplified simulation in LtSpice I could come up with to have a go:


Though it's not working as I expect:


versus something similar to expected output trace:


I appreciate if anyone with more insight shed a light and point me in right direction to follow or any component that makes this happen. I've attached simulation and data if anyone wants to have a look.

Cheers,
Matt

 

[Harald Kapp]

elcome to Electronics Point

1. Is there anything fundamentally wrong with switching polarities to any load?

That depends on the load. We need more information to give a useful answer.

2. What's the best and cheapest way to implement this?

Standard answer to "what is the best...": It depends on the quality measure you apply: size, cost, efficiency, longevity, ...

2. Assuming use of TRIACs there any circuit which achieves the same job?

There will be many circuits possible to do this.

The general issue to deal with is the synchronisation of the gate triggers in such a way that by no means both U1+U4 or U2+U3 becoma active at the same time. This will short circuit the power source V1 and blow your circuit.

With respect to your simulation (where the above issue can perfectly be avoided): V(load) is not, as you may expect, the voltage across R1. It is the voltage from the node "load" to gnd. To measure the voltage across R1, you need a differential probe. Here is how to do this real quickly.

 

[m_alizd]

Thanks Harald!

That depends on the load. We need more information to give a useful answer.

The load is going to be include a switch mode LED driver/power supply with built in full-rectifier bridge.

Standard answer to "what is the best...": It depends on the quality measure you apply: size, cost, efficiency, longevity, ...

The quality measure I have is circuit which acts/switches as quick as possible (less than 250us) around zero crossing point. At this point I'm not tightening the requirements so to explore variety of solutions at least at Simulation level. If there were a variety of solutions working on paper, for prototyping purpose other requirements such as limits on spikes and amount of EMI radiations, etc would come into picture.

The general issue to deal with is the synchronisation of the gate triggers in such a way that by no means both U1+U4 or U2+U3 becoma active at the same time. This will short circuit the power source V1 and blow your circuit.

Very true. In the hand-drawn graph I made sure to have C1 and C2 gate commands mutually exclusive in every half cycle. Plus, These will be made sure to trigger slightly after voltage/current ramps up into the cycle with TRIACs surely been already off after current dropping to zero at end of previous half cycle. Hopefully this satisfy the criteria, am I right?

With respect to your simulation (where the above issue can perfectly be avoided): V(load) is not, as you may expect, the voltage across R1. It is the voltage from the node "load" to gnd. To measure the voltage across R1, you need a differential probe. Here is how to do this real quickly.

Thanks for that, V(load,N001) is now across the R1. Not sure why in first half cycle load voltage is 0. The only possibility is perhaps both polarities are connected for some reason! no clue.


A new question would be how can I characterize and measure the inductive and resistive parameters of an LED light so to model it correctly?

There will be many circuits possible to do this.

Very interested to get more clues on this to follow up.

Thanks again Harald.
Matt

 

[Harald Kapp]

The load is going to be include a switch mode LED driver/power supply with built in full-rectifier bridge.

In that ase I do not understand why you want/need that controllable rectifier in front of it.

As your triacs are switching one polarity only, consider using thyristors (scrs) which are imho better suited to this application and less likely to turn on in the "wrong" direction.

Your circuit has another issue: The control voltage for a triac is applied between gate and Anode1 (also called main terminal 1). In your circuit this is true for U2 only!
You will need level shifted gate drives for U1, U3 and U4. You may consider using photo triacs, more expressly the zero crossing variant. Or use pulse transformers for triggering the thyristors/triacs independently from the potential at the cathode/anode1.

how can I characterize and measure the inductive and resistive parameters of an LED light so to model it correctly?

You can't unless you model the controller circuit and the LEDs in total. The controller will have a rectifier, capacitors, inductors, switching elements etc. This cannot be modeled by a simple LRC model.

 

[m_alizd]

Your circuit has another issue: The control voltage for a triac is applied between gate and Anode1 (also called main terminal 1). In your circuit this is true for U2 only!

Good pick up Harald! I'll have another go at it also another version with SCRs and check the results.
Just to confirm, should I generalize this to other instances, too?

What I learned from TRIACs info sheet was Gate and MT2 should be of same polarity to conduct, both positive or both negative.

 

[WHONOES]

If I understand you requirement, may I suggest you look up *(google) a thyristor bridge.

 

[m_alizd]

Thanks Whonoes, I had a look and looks like the thyristor bridge output is DC, however, as you can see in attached image "Polarity switch.png" in my 2nd message, the output of my circuit is required to be AC. The only difference with normal AC is that positive and negative half cycles could come in a row sometimes. I reattached a clearer image of input output requirements, please see "Polarity switch output.png".

 

[m_alizd]

In fact someone did a simulation in LTspice to that generates the output power waveform that is required.
Swapping polarity is modeled by multiplying sinewave by -1 when needed.
I attached the simulation just in case.

The aim now is to come up with a circuit that generates same output as Vcd in this simulation.

 

[Harald Kapp]

looks like the thyristor bridge output is DC, ho

it usually is. But you can rearrange the scrs to create your desired waveforms.
I still do not understand why you want to create that special waveform when there is a rectifier following

 

[m_alizd]

But you can rearrange the scrs to create your desired waveforms

What control signal and waveforms can do that?

I still do not understand why you want to create that special waveform when there is a rectifier following

Sorry for confusion, the load would be away from the power supply only connected via an AC lead. And before the rectifier in the load there can be zero crossing detect to sense this.

 

[WHONOES]

What do you actually want to drive with this waveform?

 

[Harald Kapp]

Sorry for confusion, the load would be away from the power supply only connected via an AC lead. And before the rectifier in the load there can be zero crossing detect to sense this.

I'm sorry, this still does in no way explain your desire for that circuit.
What is an "AC lead"? And what is the zero crossing detector used for?

 

[Harald Kapp]

What control signal and waveforms can do that?

Basically your triac circuit but using scrs instead. Plus you need to create control signals on different voltage levels to accommodate the shifted voltage levels of the different scrs. This is where pulse transformers or optically isolated gate drivers come inro play, as mentioned earlier.

 

[m_alizd]

What is an "AC lead"? And what is the zero crossing detector used for?

Just normal two core cable carrying 12Vac.I will use Zero crossing detection to trigger another circuit that detects if positive or negative half of the cycle is happening, i.e. normal polarity or swapped polarity is coming. But for now I'm focused on the power supply end and not worried about load side at the moment.

 

[WHONOES]

In your first post, you say that you want to drive 32vac at up to 30Watts. That is well within the capability of a power amplifier. You could then use a pic or some other type of processor to construct the wave form and sequence you require to drive the amplifier.

 

[m_alizd]

In your first post, you say that you want to drive 32vac at up to 30Watts. That is well within the capability of a power amplifier. You could then use a pic or some other type of processor to construct the wave form and sequence you require to drive the amplifier.

I'm not familiar with power amplifier, any particular part you can name for me to check out?
How do they compare with SCRs pricewise?
Do they need some other circuits on the side like oscillators, etc to generate sinewave?

 

[WHONOES]

Have a look at the link below.

https://www.banggood.com/HiFi-LM388...W-8-p-1190627.html?rmmds=buy&cur_warehouse=CN
Edit: Yes you will need an oscillator to generate the required waveforms or, as suggested earlier you could use a pic or μProcessor to generate your waveform and include the logic in software to perform the half sine swapping function.

 

[Bluejets]

I think it would be a great help to know just what the outcome of all this would be used for.

normal polarity or swapped polarity is coming

Surely you mean out of phase.

 

[m_alizd]

I think it would be a great help to know just what the outcome of all this would be used for.



Surely you mean out of phase.

Yes we can say that

 

[m_alizd]

Yes you will need an oscillator to generate the required waveforms

Thanks Whonoes, as much as I like the idea, if my calculations are right, this solution seems to be much more expensive than an SCR H-Bridge which is what I'm going to figure out how to do.
As you said either an oscillator or a dedicated MCU/firmware needed to generate the waveform. The other thing I'm guessing is needed is a high power full-rectifier with ample output current capacity to feed the oscillator and power amplifier and the finally the load. This is on top of A$12.5 for LM3886. I guess it would need between $15-$20.

If I could get TRIAC/SCR H-Bridge rectifier to work, it might cost 4 pairs of SCR+Opto which can be $4-$5 altogether.

Does my calculations look reaalistic?