Tweet

Posted on 30 August 2019

High Speed, Two Quadrant DC/DC Power Supply

Free Bodo's Power Magazines!

 

  

 

The main function and benefit of synchronous rectification

To enable a converter that modulates output voltage and that generates voltage according to some predefined wave form, whilst keeping the efficiency high, is only possible using a high speed two quadrant DC/DC power supply.

By Milan Marjanovic, Texas Instruments

 

A two quadrant DC/DC power supply has bi-directional power flowforward, where the power flows from source to load - and reverse, where the power flows from load to source. This article describes how to make such a power supply using simple techniques.

One or Two Quadrant Power Supply?

As shown by the blue line in Figure 1, a single standard power supply is able to deliver only positive voltage and positive current in a forward direction. We call this kind of power supply a one quadrant power supply. If there is the possibility for reverse power flow, as indicated by the red line, then we are talking about bi-directional or, two quadrant power supplies. In this case the voltage is again only positive, but the current can be positive and negative.

One or Two Quadrant Power Supply

From this point we will talk only about Two Quadrant Power Supplies - TQPS

The basic and best example for TQPS is the synchronous buck converter. Many power engineers make the big mistake of not properly understanding synchronous rectification. They regard synchronous rectification as high efficiency. The biggest advantage, however, is pure bi-directional power flow; capability both to push the current to output and to pull the current back. We will see that this capability is directly reflected in a much better load step response - better dynamics on the output. This kind of dynamic would never be possible with a standard buck converter – buck with a diode rectifier (free wheeling diode). To summarize, Table 1 shows the three important advantages of the buck converter using synchronous rectifier technique versus a diode rectifier approach.

Three important advantages of the buck converter using synchronous rectifier technique versus a diode rectifier approach

This third capability is also very important. A system with synchronous rectification does not need the minimum load to work in continuous conduction mode (continuous current flow in the storage choke). Because of the ability to pull the current back, the duty cycle remains constant for the entire load range. While always working in continuous conduction mode, we can make a converter with a very good huge load step transient response. The worst case scenario for each power supply is load transition from no load to maximum load and from maximum load to no load; as shown in Figure 2.

Example of the huge load step (0A - 10A - 0A)

As can be seen, the system is pushing more current to the output to recover the voltage drop (marker M1), but it is pulling the same amount of energy back (marker M2) to block and reduce the voltage overshoot.

Full bridge phase shift topology and synchronous rectification

Full bridge phase shift topology - zero voltage transition - is becoming more and more popular. The main competitor is a full bridge hard switching topology. There are a few main differences between the two listed in Table 2.

Full bridge phase shift topology versus synchronous rectification

From this point of view, we can say that for low voltage applications (<100V, e.g. telecoms) there is no benefit on the power stage using phase shift topology. The main reason for using it is much simpler and more optimized gate driving. Also, there is no high frequency ringing on the switch node, which can indicate better EMI performance. Talking about high voltage design (400V range), performance is more efficient, which is a good enough reason to go straight forward with this topology. Figure 3 demystifies the full bridge phase shift principle.

The drawings demystify the full bridge phase shift principle

The gate signals for phase A and B are inverted to each other. They have the same duty cycle of 50% and the same period. Therefore the leg (A, B) is either connected to VIN or to the ground. The same holds for phase C and D as well as for leg (C, D).

The bottom left image shows the voltage across the primary, marked red, when the phase shift between leg (A, B) and (C, D) is zero degrees. In this case, both sides of the transformer are connected, either to VIN or GND. As no voltage is applied to the transformer, the output voltage of the transformer is also zero and therefore no current flows.

On the top right side the waveforms for a 90 degree phase shift are shown. The waveform for leg (A, B) remains the same, but the waveform of leg (C, D) is shifted for 90 degrees. Now, for the first quarter of the period a positive voltage is applied to the transformer. This generates a voltage on the secondary side and causes a current. During the second quarter, VIN is applied to the transformer on both taps, so the primary is now shortened and there is no voltage on the secondary side. At the third quarter, a negative voltage is applied to the transformer, which again generates a negative voltage on the secondary side. On the secondary side the voltage is rectified as shown by the hatched rectangles. At a 90 degree phase shift the total duty cycle is thus 50%.

If the phase shift is increased further, the duty cycle on the secondary side increases, and therefore output voltage rises. The maximum output voltage is achieved at a phase shift of 180 degrees and results in a duty cycle of 100% on the output, as shown in the image in the bottom right corner.

Further improvements of the loop bandwidth on galvanic isolated systems

Normally the controller sits on the primary side, generating PWM signals for the switching elements. Having already started with a half bridge topology, we need a special drive for a high side switch. Because of a floating switch node point the gate drive is floating as well. Accordingly, for low and mid power range power supplies we have the possibility of either a gate transformer or integrated half bridge driver with a boot strap diode. For the full bridge topology we already need two such symmetrical circuits. If we have a synchronous rectification on the secondary side we need to transfer additional PWM signals across the isolation barrier. This can be achieved again using a transformer or, digital isolators (ISO7920) or, an high speed optocoupler. We also need to close the feedback loop across the isolation barrier, usually again with an optocoupler. The optocoupler has a low pass behaviour, very long turn on and turn off delay times, huge CTR tolerance, aging issues and for some applications (e.g. automotive and military) is not even allowed.

To increase the frequency band width of the feedback loop, the reliability, and save the costs of the system, we can set up as shown in Figure 4.

Set up to increase the frequency band width of the feedback loop, to increase the reliability and save the costs of the system

The controller is placed on the secondary side providing the possibility of closing the feedback loop without need for an optocoupler. The signals for synchronous rectification are now on the same ground potential as referenced by the controller itself (no need for expensive digital isolators). The signals which now cross the barrier are bridge PWMs and current sensing. As stated, we can use the gate transformers for gate driving. When they are used, galvanic isolation is free. Designers have to take care with the isolation voltage needed. The same is true for the current sense transformer. This kind of setup has, unfortunately, one drawback: it needs a galvanic isolated auxiliary power supply providing voltage to the secondary side. Without this voltage the converter could never start. The synchronous rectifi- cation that we can see here makes this converter bi-directional (two quadrant power supply). In Figure 5, we can see the right and left leg of the full bridge rectifier at the secondary side. The green wave shows the storage choke current. As can be seen, the current remains continuous as does the duty cycle.

No load conditions

Applying the load at the output shifts this green line over the y axis. This can be not only in a highly positive direction, but also as stated, in a highly negative direction. The duty cycle changes only during the load transitions. In steady state condition the duty cycle remains constant for the entire load range.

Summary

We discussed a bi-directional power supply definition, synchronous buck converters, and the basics of full bridge phase shift topology. We have seen in a practical way what the main function and benefit of synchronous rectification is and how to use it in isolated systems. The layout and the position of the controller in the system play a big role as well. This all opens the door to making high speed power supplies. In the following article, “Using bidirectional power supply to modulate voltage on the output,” we talk further about power stage, sensing techniques, and choosing the right regulation methods for bidirectional power supplies.

 

 

VN:F [1.9.17_1161]
Rating: 0.0/6 (0 votes cast)

This post was written by:

- who has written 791 posts on PowerGuru - Power Electronics Information Portal.


Contact the author

Leave a Response

You must be logged in to post a comment.