Maker Pro
Maker Pro

Half Bridge Converter Question

twenglish1

Feb 21, 2014
24
Joined
Feb 21, 2014
Messages
24
I am working on designing a high powered switch mode power supply, and as i have been doing research on the half bridge topology, i came across two variations:

This is the most common i have seen: http://www.electronicproducts.com/images2/F129ONSE0507a.gif

The other one i have seen has the capacitor in series with the transformer winding (see attachment)

Which is the better design? and why? the one with the capacitor in series with the transformer is the way my inverter tig welder was built
 

Attachments

  • inverter.jpg.png
    inverter.jpg.png
    7.6 KB · Views: 564

duke37

Jan 9, 2011
5,364
Joined
Jan 9, 2011
Messages
5,364
Both of the circuits have a capacitance in series with the transformer.

With two capacitors the pulse is directly coducted back to the supply or ground line.
With one capacitor the pulse is conducted to the ground line only. If this pulse is obtained from the positive supply, then the power supply has to have a capacitor to provide this.

Two capacitors would be best.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
The first circuit has a few issues but it is probably the best out of the two. What's missing.
1) suppression diodes for the coil.
2) Anti flux walk capacitor in series with the coil.

This is one of the most popular configurations for high voltage and high power SMPS. The reason is you only have to have a transistor rated at half the supply voltage which allows for faster switching devices to be used.

But for the same power output they will need to be rated at over twice the current. The DC link capacitors serve two purposes one is to supply half supply voltage to the coil and also to absorb nasty fly back spikes from the coil.

One disadvantage is these capacitors need to be large in value and have low ESR and ESL and also a high ripple current rating.

The other circuit does not look like a standard SMPS output stage, but looks like a type of tuned transformer drive. I am not familiar with this type of design. I would go for either the half bridge or full bridge types.
Adam
 

twenglish1

Feb 21, 2014
24
Joined
Feb 21, 2014
Messages
24
Could you recommend a half bridge pwm driver ic? Preferably with oscillator build in, and I'm looking to run it at about 40khz.

That second circuit came from the schematic of this harbor freight inverter welder I have, I never saw that set up before
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
The first circuit has a few issues but it is probably the best out of the two. What's missing.
1) suppression diodes for the coil.
2) Anti flux walk capacitor in series with the coil.
The two capacitors across the power supply provide DC blocking to prevent flux walking.
This is one of the most popular configurations for high voltage and high power SMPS. The reason is you only have to have a transistor rated at half the supply voltage which allows for faster switching devices to be used.
Huh? In both cases, both devices need to be rated for the full supply voltage, and in both cases, the peak-to-peak voltage seen by the primary is equal to the supply voltage.
The DC link capacitors serve two purposes one is to supply half supply voltage to the coil and also to absorb nasty fly back spikes from the coil.
Connecting two capacitors in series across a supply rail doesn't cause the centre point to "supply half supply voltage" the way resistors do. The difference between the two designs relates to the capacitor values. In the two-capacitor design, the capacitors are effectively in parallel (because there's a much bigger capacitor across the supply rails), so they only need to have half the capacitance each of the single capacitor circuit. If you go into detail, there will be other differences as well, but that is the main one.
One disadvantage is these capacitors need to be large in value and have low ESR and ESL and also a high ripple current rating.
Another advantage of the two-capacitor design. The capacitors will share the current.
The other circuit does not look like a standard SMPS output stage, but looks like a type of tuned transformer drive.
I suppose it could be a tuned circuit if the capacitance was low, but it's not; it's just used for DC blocking.
All AFAIK etc.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
Quote Kris
“The two capacitors across the power supply provide DC blocking to prevent flux walking.”

Flux walking as you know happens if the coils core hysteresis loop is not reset to its original value i.e the exact same change of voltage over time for both transistors must occur. This very rarely happens and the simple fix is to put a series capacitor on the coil. The two capacitors don’t do this; they are there to supply the energy for each switching cycle and create a half supply common point for the coil.

Quote Kris
“Huh? In both cases, both devices need to be rated for the full supply voltage, and in both cases, the peak-to-peak voltage seen by the primary is equal to the supply voltage.”

The two capacitors are called DC link capacitors and are used to provide a half supply point for one side of the coil and supply the switching current. The transistors when switching only see half supply because half exists at the centre point of the capacitors.

Quote Kris
“Connecting two capacitors in series across a supply rail doesn't cause the centre point to "supply half supply voltage" the way resistors do. “

Yes I know that but the DC is being switched in and out by the transistors which do have a resistance and with matched transistor will be approx. the same. This is then the supply divided by the two across the join point of the two transistors because at a point in time they will be on at the same time.

Quote Kris
“The difference between the two designs relates to the capacitor values. In the two-capacitor design, the capacitors are effectively in parallel (because there's a much bigger capacitor across the supply rails), so they only need to have half the capacitance each of the single capacitor circuit. If you go into detail, there will be other differences as well, but that is the main one.”

Capacitors in series add there working voltage not their value. So what are you saying for a.c I get double the working voltage and double the capacitance if I put another capacitor across them? Bear in mind this is pulsed d.c.

Quote Kris
“Another advantage of the two-capacitor design. The capacitors will share the current.”

That’s not an advantage, that’s what they are supposed to do.

A simple simulation with results attached. Voltages across the transistor and capacitor centre point.

Thanks
Adam
 

Attachments

  • SMPS.PNG
    SMPS.PNG
    11.8 KB · Views: 596
  • SMPS_SCOPE.PNG
    SMPS_SCOPE.PNG
    21.3 KB · Views: 273
Last edited by a moderator:

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
Flux walking as you know happens if the coils core hysteresis loop is not reset to its original value i.e the exact same change of voltage over time for both transistors must occur. This very rarely happens and the simple fix is to put a series capacitor on the coil. The two capacitors don’t do this; they are there to supply the energy for each switching cycle and create a half supply common point for the coil.
No, they don't "supply energy"; they block DC. They function just like a series capacitor. The voltage at the middle node will stabilise at roughly half the supply voltage, but that's not because the capacitors "create a half supply" voltage; it's because the duty cycle of the signal from the driving transistors is nominally 50%.

Whether it's a single capacitor or a pair of capacitors in parallel, the reason for the capacitor is to ensure that the time integral of the voltage across the primary averages out to zero. In other words, to block DC. Any DC offset will cause flux walking. A DC offset can be caused by various factors, such as different voltage drops across the two switching elements, asymmetry in the secondary circuit, and issues with the feedback loop causing the duty cycle to deviate from 50%, and DC blocking is the simplest way to protect the transformer from all of them.

The two capacitors are called DC link capacitors and are used to provide a half supply point for one side of the coil and supply the switching current. The transistors when switching only see half supply because half exists at the centre point of the capacitors.
No. Each transistor sees the full supply voltage. The node where the transistors connect is switching between 0V and the positive DC rail. When it's at 0V, the supply voltage is across the top device; when it's at +V, the supply voltage is across the bottom device. Maybe you're thinking of the primary. It sees only half the supply voltage in each direction; that's true with the single DC blocking capacitor as well.
Yes I know that but the DC is being switched in and out by the transistors which do have a resistance and with matched transistor will be approx. the same. This is then the supply divided by the two across the join point of the two transistors because at a point in time they will be on at the same time.
Sorry, I don't know what you mean.
Capacitors in series add there working voltage not their value. So what are you saying for a.c I get double the working voltage and double the capacitance if I put another capacitor across them? Bear in mind this is pulsed d.c.
They look like they're in series, but they're actually in parallel, as I explained, because of the large amount of capacitance across the supply rails, so their capacitances add together. I don't follow the second sentence.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
I think we will have to beg to differ on this for a momemt. :) I am talking about the first circuit posted and yes the primary of the transformer. I will dig out some of my books and have a look.
Cheers
Adam
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
I have done a bit of digging and I have changed my circuit to match the microchip drawing from the app note AN1114 I found that explains what I am trying to explain but I probably explained it all wrong. Please feel free to comment and agree or disagree. If the circuit works differently to how they explain it then I would appreciate a detailed correct explanation.


OPERATION
The switches Q1 and Q2 form one leg of the bridge, with the remaining half being formed by the capacitors C3 and C4. Therefore, it is called a half-bridge converter. The switches Q1 and Q2 create pulsating AC voltage at the transformer primary. The transformer is used to step down the pulsating primary voltage, and to provide isolation between the input voltage source VIN and the output voltage. In the steady state of operation, capacitors C3 and C4 are charged to equal voltage, which results in the junction of C3 and C4 beingcharged to half the potential of the input voltage.”

When the switch Q1 is ON for the period of TON, the dot end of the primary connects to positive VIN, and the voltage across the capacitor C4 (VC4) is applied to the transformer primary. This condition results in half of the input voltage being VIN, which is applied to the primary when the switch Q1 is ON, as shown in Figure 16 (C).

Extract from NXP document SMPS applications. The two mains bulk capacitors C1 and C2are connected in series, and an artificial input voltage mid-point is provided, shown as point A in the diagram. The two transistor switches are driven alternately, and this connects
each capacitor across the single primary winding each half cycle. Vin/2 is superimposed symmetrically across the primary in a push-pull manner. Power is transferred directly
to the output on each transistor

Indicating that the capacitors do supply power?

After the time period TS/2 when the switch Q2 turns ON,the dot end of the primary connects to the negative of VIN, and the voltage across the capacitor C3 (VC3) is applied to the transformer primary. Therefore, half of the input voltage VIN is applied to the primary when the switch Q2 is ON in the reverse direction, as shown in Figure 16 (C).

This is why I can’t see how the full supply voltage appears across the transistor as you mentioned. But please explain fully if I have misunderstood.


SHOOT THROUGH
A half-bridge converter is also prone to magnetic imbalance of the transformer core when the flux created by the switches Q1 and Q2 during the TON period is not equal. To prevent staircase saturation, the peak current mode control technique is used to decide the TON period of the switches Q1 and Q2. The maximum duty cycle of 45% with a dead-time between the two switches is used to prevent shoot-through
current from the transformer primary.”

It doesn’t mention anything about C3 and C4 having any relationship to preventing flux walking and as the capacitors effectively form the other half of the bridge circuit then they should always be there for this topology. The fact that Microchip state they are prone to flux walking. This would be misleading if C3 and C4 stopped flux walking wouldn’t it?

DC BLOCKING CAPACITOR CB
A small DC blocking capacitor is placed in series with the transformer primary, to block the DC flux in the
transformer core.”

FYI
On a side note: The other picture in the original post that I thought might be a tuned type looks like it is called a Half Bridge Resonant Converter and the main function of the capacitor is to form part of the resonant tank.


Thanks Adam
 

Attachments

  • SMPSHALFB.PNG
    SMPSHALFB.PNG
    16.6 KB · Views: 326
  • HALFCONV.PNG
    HALFCONV.PNG
    29.4 KB · Views: 408

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
Hi Adam,

Re your first schematic, there's an asymmetry between the gate drive signals for Q1 and Q2. In both cases, the MOSFET responds to the gate-source voltage. Q2 has its source at the 0V rail potential, but Q1's source jumps around between 0V and +V. You have both drive signals referenced to 0V. The two MOSFETs won't see the same gate-source drive voltage. For simulation purposes, the simple solution is to move the negative side of V2 so it connects to Q1's source.

I have done a bit of digging and I have changed my circuit to match the microchip drawing from the app note AN1114 I found that explains what I am trying to explain but I probably explained it all wrong. Please feel free to comment and agree or disagree. If the circuit works differently to how they explain it then I would appreciate a detailed correct explanation.
That's a very thorough app note and the author certainly seems to know what he's talking about.

I haven't done any work with half bridge converters, so my understanding of them is on the basic level. I'll try to point out places where I think I'm more likely to be wrong.

OPERATION
Not the clearest explanation, but it's not wrong.
Extract from NXP document SMPS applications
That's not how I would have explained it, but again, it's not wrong.
Indicating that the capacitors do supply power
Again that's not how I would explain it. The energy all comes from the input power source, via the switching devices. The capacitors just provide DC blocking, in my view.
This is why I can’t see how the full supply voltage appears across the transistor as you mentioned. But please explain fully if I have misunderstood.
The full supply voltage appears across each transistor because they're connected in series across the full supply voltage and alternate between being fully ON and fully OFF.

If you ignore the capacitors and the primary, and just look at the switching devices, you can see that the common point of the devices is swinging between V+ and 0V. When it's at V+ (top MOSFET conducting), the bottom MOSFET sees the full DC supply voltage across it, and vice versa. This is true for the switching devices in any half-bridge circuit, and any full-bridge circuit too, for that matter.
SHOOT THROUGH
"A half-bridge converter is also prone to magnetic imbalance of the transformer core when the flux created by the switches Q1 and Q2 during the TON period is not equal. To prevent staircase saturation, the peak current mode control technique is used to decide the TON period of the switches Q1 and Q2. The maximum duty cycle of 45% with a dead-time between the two switches is used to prevent shoot-through
current from the transformer primary.”

It doesn’t mention anything about C3 and C4 having any relationship to preventing flux walking and as the capacitors effectively form the other half of the bridge circuit then they should always be there for this topology. The fact that Microchip state they are prone to flux walking. This would be misleading if C3 and C4 stopped flux walking wouldn’t it?
Yes, and I think it is misleading. But I don't have any experience with half bridge converters. It seems to me that since there is no DC path from the centre point of the capacitors, by definition, they block DC, and this should ensure equal energy input in each direction, and this is what prevents flux walking.
DC BLOCKING CAPACITOR CB
"A small DC blocking capacitor is placed in series with the transformer primary, to block the DC flux in the transformer core.”
You have used some very high values for C3 and C4. If you replace C3 and C4 with half of CB's capacitance each, and delete CB, then as far as I can see, the two circuits will be roughly equivalent, and C3 and C4 will do the job of CB. This assumes there's a big capacitor across the input supply, which is true in practice.

That seems like the sensible approach to me, because it uses less (and smaller) components, but I have no experience working with bridge forward converters so I can't be sure; there could be a subtle reason to use two large capacitors and a separate DC blocking capacitor.

OTOH, I don't think that application note gives recommended values for C3 and C4. They're shown as non-polarised capacitors, which implies relatively low values, but that's not conclusive.
FYI: On a side note: The other picture in the original post that I thought might be a tuned type looks like it is called a Half Bridge Resonant Converter and the main function of the capacitor is to form part of the resonant tank.
I guess that depends on the value of the series capacitor. I kind of doubt that it is a resonant converter, but then again, I don't know much at all about resonant converters.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
Thanks Kris. It's so nice to have a good technical discussion with you. To be honest I just through in some values for the caps to keep ESR low which is what is needed. The actual value from memory is about 1uF per Watt. They normally are electrolytics.

Your right about the voltages I was getting confused about push pull being twice the voltage of half bridge. Sorry for the confusion.

The other schematic I think was just lashed together and resembles a resonant converter. I only know this because over the last two years I have been helping a KTP student who is working with a professor at the university of Plymouth on an NPC switching circuit which is effectively the same half bridge arrangement.

The drive for the top Mosfet needs to be about 25 volts. I assumed that as long as gate was say 12V higher than the source it would switch on. Is this not how the boot strapped Mosfet drivers work.
Thanks
Adam
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
What are KTP and NPC?

The best way to generate bootstrapped MOSFET drive for high-side N-channel MOSFETs uses a capacitor with one side connected to the half-bridge output point, that's charged up to a low-voltage supply rail (often 12V) through a diode when the bottom MOSFET is ON. Since it's referenced to the high-side MOSFET's source, it provides the perfect supply for the gate driver for the high-side MOSFET. The whole high-side drive circuit swings up and down as the half-bridge output swings up and down. How the drive signal is fed into that circuit can vary. An optocoupler can be used. Have a look at the high-side driver in the FAN7380: http://www.fairchildsemi.com/ds/FA/FAN7380.pdf
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
KTP is a knowledge transfer scheme for student at university. NPC is a modern way to drive high voltage motors or inductive loads, it stands for Neutral Point Clamping

Adam
 

Fish4Fun

So long, and Thanks for all the Fish!
Aug 27, 2013
481
Joined
Aug 27, 2013
Messages
481
@Kris and Adam...Great read, but I think the elephant in the room is that the OP wants to design//build a "high power SMPS" w/o a good working knowledge of existing bridge topologies/methods/designs. "High power" SMPS implies one that is line driven and likely > 500W; I think the more prudent advice might be to suggest commercially available SMPS solutions, or at least ferret out exactly what he has in mind....perhaps it is purely an intellectual pursuit in an effort to learn about SMPS design, but generally when people w/o the requisite skill sets start down the "high power. line driven" road things go badly pretty fast.

Perhaps a thread in the 'electronics chat' forum discussing the vagaries/nuances of half/full bridge forward converter design might turn into a great "sticky"....I would certainly read it ;-)

Carry On!

Fish
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
Yes Fish, I agree completely. I was assuming that the OP would read through our discussion, and if he had no idea what we were talking about, he would realise that he's bitten off more than he can chew.

Re the sticky discussing the subject, I think the Microchip app note that Adam found is a pretty good reference.
 

Arouse1973

Adam
Dec 18, 2013
5,178
Joined
Dec 18, 2013
Messages
5,178
Agree. I love stuff like this, It's the reason I stay up all night :)
 
Top