[SOLVED] Bi-directional AC-AC CUK converter

[Johnjacob]

I am trying to build a specific AC-AC cuk converter for a project of mine. It involves the use of 4 MOSFETs configured in bi-directional switching states. I've included a picture of the schematic.

I am essentially building a solid state 'electronic' variac(variable autotransformer). I input 80v AC 50Hz and I'd like to obtain AC voltages ranging from 0v to 80v at 50Hz, feeding and restrictive inductive load whilst being able to obtain an output current of up to 1A RMS.

I found a solution for driving the MOSFETS in a patient by Simon Greenwood, and it had described a sequence as follows:

"For a period during which input AC is positive and load current is in phase with load voltage. Transistor Q2 and transistor Q3 are held on and therefore, in conjunction with diode D2 and diode D3, provide bi-directional current paths. During this period transistors Q1 and Q4 switch alternately at high frequency in response to a high frequency control signal from the controller. For a 1st time interval of this high frequency alternation, transistor Q1 is off and transistor Q4 is on. During this interval diode D4 is forward biased and the energy transfer capacitors C1a and C1b charge through the choke inductor L1. During a second time interval of the high frequency alternation, the switching states of transistors Q1 and Q4 are reversed. Once this occurs, the energy transfer capacitors C1a and C1b discharge, driving current through the output load via inductance L2, and charging the output capacitor C2b. Circuit operation is repeated When transistor O1 is turned off and transistor O4 is turned on again. For a period during which input AC is positive and load current is in OUT of phase with load voltage, Transistor Q2 and transistor Q3 are held on. During this period transistors Q1 and Q4 switch alternately at high frequency, for a 1st time interval of this high frequency alternation, transistor Q4 is off and transistor Q1 is on. During this interval diode D1 is forward biased and the energy transfer capacitors C1a and C1b charge through the choke inductor L2. During a second time interval of the high frequency alternation, the switching states of transistors Q1 and Q4 are reversed. Once this occurs, the energy transfer capacitors discharge, driving current out through the input terminals.

At negative AC and when load current is in phase With load voltage. Transistor Q1 and transistor Q4 are held on. During this period transistors Q2 and Q3 switch alternately at high frequency. For the 1st time interval of this high frequency alternation, transistor Q2 is off and transistor Q3 is on. During a second time interval, the switching states of transistors Q2 and Q3 are reversed. Charging the output capacitor C2b. Circuit operation is repeated when transistor Q2 is turned off and Q3 is turned on again. Finally, when AC is negative and load current is out of phase with load voltage. Transistor Q1 and transistor Q4 are held on this period transistors Q2 and Q3 switch alternately at high frequency. At the 1st interval, transistor Q3 is off and transistor Q2 is on. During a second time interval, the switching states of transistors Q2 and Q3 are reversed, driving current out through the input terminals via inductance L1, and charging the input capacitor C2a. Circuit operation is repeated When transistor Q3 is turned off and transistor Q2 is turned on again."

My question is, how can I implement a simple way to drive the 4 MOSFETs in the bidirectional sequence described above?

 

[BradtheRad]

You probably wish to maintain galvanic isolation. Therefore you cannot have a control circuit which is powered from one side, and use that same control circuit to drive the opposite side.

In such situations I've seen advice to use pulse transformers. This is a galvanically-isolated control method, for operating the mosfets.

From the description, it appears that gating waveforms must be coordinated on both sides. Therefore waveforms must originate from one circuit. Are you able to confirm whether you need sine PWM, or can it be simple on-off switching?

 

[Johnjacob]

Hi, thanks for the advice.
Yes, the gating wave forms are to be coordinated on both sides and I can confirm that simple on-off switching sequence would suffice. Unfortunately I am unable to determine how exactly should I implement the simple on-off switching sequence. Once again thank you for the reply and advice.

 

[BradtheRad]

Optocouplers are commonly used to convey feedback signals, when you need galvanic isolation.

I guess the question is where to put the control circuit, on the left or the right side? We cannot always be sure the control circuit will receive power on startup.

There's a chance you can manage with a blocking oscillator, which is handy when you wish to create oscillations through a transformer. It has a winding which biases transistors on and off. In turn the transistors switch current through another winding.

 

[Johnjacob]

Hi, once again thanks for the suggestions. I was pondering with an extremely simple idea of using a ad648 op-amp with a variable dc voltage on the inverting input and a fixed saw tooth waveform at say 4KHz at the non inverting input working as a comparator producing a form of PWM signals which in turn are connected to an opto coupler feeding the MOSFET gate. Would this be a suitable solution? If so, how would I apply this to provide the switching sequence that I had mentioned in the beginning? Thanks

 

[schmitt trigger]

Reading at the patent description, my hamster died.

 

[Anna Conda]

The commonest and most robust way is to generate your gate drive signals is on to a gate drive transformer (+/- 15V say) and have sufficient isolated output windings to drive your fets, making sure the phasing is correct, thus you can turn the i/p and output sides on and off alternately (not at the same time), you may not need to modulate the drive away from 50:50 if you seek a 1 to 1 conversion - but at light loads the fet drive will need to be reduced.
This is because there will be quite a bit of dead time at light loads where all fets need to be off, else there will be a lot of circulating current which you will not want.
I suggest you put zeners or TVS's across the fets at 90% of Vds to protect them,
As the sources are tied together you can link the gates as well so only two windings are needed to drive the two sets of fets.
Also the dot notation on the main Tx needs to change if you want pos in and pos out, at the moment pos in gives neg out (the tops).
When the input fets are on the output "diode" (fets in your case) should be off, hence good to get the phasing of the gate drive correct.
Ideally at light loads you need to sense the current in the fets and turn off the fets receiving energy (i.e. acting as the diode) as soon as the thru current falls to zero, else problems will arise...
ideally run at >20kHz.

 

[Orson Cart]

You can individually control the mosfets as per your description in post 1, that way you get the benefit of the diodes (one way current flow) you need to sit down and scribble control ckts until you concieve one that fits the bill for the control you desire... a zero crossing detect on the Vin for starters, and unfortunately current sensing on the sec side to detect when you are in DCM.

 

[mtwieg]

You should look into changing your circuit so that the switches are unidirectional, but add active bridges on the input/output to control the sign of the voltage on the input/output. The active bridges will only have to switch at the line voltage, so you should get better efficiency that way.

As for gate drive, gate drive transformers aren't suitable, due to the wide range of duty cycle required. Instead I would optoisolated gate drivers with truly isolated bias supplies.

 

[Anna Conda]

Using a gate drive in a very conventional way would limit you to 0-50% or so duty cycle, but there are circuits which allow 0-100% duty cycle using a pulse effect thru the gate transformer, the ON drive can be maintained for 10mS without too much trouble, IR does a very good app note with these designs explained well.