Transformer Design

[Militoy]

”(1)You called it a distributed gap I am having some difficulty finding a good definition for that. I know its not a physical gap but what exactly is the gap in reference too.”

A distributed gap works just as a physical gap in an inductor core does – by increasing reluctance in the magnetic path. MPP cores, etc. are made up of a blend of magnetic and non-magnetic materials, with the non-magnetic oxides distributed throughout their structure. Selecting a specific Al is like setting the gap length in a conventional inductor – the longer the gap, the lower the inductance – but, the greater the DC flux that can be supported.

”(2) You say the typical perm of a 50/60Hz core is around 30k-50k with sat flux density around 17.5k gauss. Are these parameters driven by peak voltage, frequency or power? And how much flexibility exists in this(with trade offs).”

(see below)

”(3) Cross sectional area of the core. You must be working off some sort of calculator or formulas to know how many windings(250000) it would take to keep the flux below saturation for the given core. I think that tool would be very useful here. Would you mind sharing how you came to that number?”

It’s very straightforward to Google an exact formula – which varies, depending on if you are working with a sine wave or other wave shape – as well as the units you are working with (I tend to think in inches). The important thing is to get the relationship of all the factors clear in your head.

Any given magnetic core material has a finite limit of maximum flux density it can support. When driven above this level, the material loses its magnetic properties – the magnetic field collapses – and the primary winding becomes essentially a resistor, as opposed to an inductor. The transformer core is said to be “in saturation”.
In order to successfully transfer energy, with a few notable exceptions, a transformer core must support magnetic flux at some level below saturation. In the equation used to calculate magnetic flux density, the core cross-sectional area; primary turns; and frequency are all in the denominator - and so affect the primary circuit in a proportional manner. Increase primary turns, and you can proportionally reduce core cross-section or frequency. Decrease frequency - you must increase core cross-section or primary turns. Primary voltage is in the numerator of the equation – and affects the system inversely.

So – this is what makes miniaturizing a 50/60 Hz transformer so difficult. With frequency and input voltage fixed, size reduction by definition means reducing core cross-section – and that forces the designer to increase the number of primary turns to compensate. More turns squeezed into a reduced core winding area forces the designer to use smaller wire – both to accommodate the reduced area, and the greater number of turns that must be jammed in. The longer total length of the winding and smaller wire cross-section doubly increases resistive power losses and voltage drop with load (regulation).

Acceptable load regulation percentage typically becomes a limiting factor in size reduction, before temperature rise becomes an issue. It’s not unusual in a smaller commercial line-frequency transformer, to see up to 60% regulation – before reaching only 40ºC temperature rise. Another limiting factor is decreased reliability resulting from the use of extremely fine magnet wire. It doesn’t take much force to break a 48 AWG wire – which is only 0.00124 inches in diameter. Since copper expands with temperature, the designer has to consider winding techniques that protect the wire during the winding operation – as well as prevent the winding from self-destructing during regular service.

Without straying too far off-subject – a better approach at low power, operating off line at mains frequencies, is to consider a simple SMPS approach, such as a TOPswitch. Using this technology – it’s very possible to get a few watts of isolated, rectified DC power, using a transformer about the size of a thumbnail.

 

[NickS]

I would like to revisit the calculation from a couple posts back. I rounded up some equations and I want to check them against yours because I am seeing a vastly different result.
E = 4BNfAc*10^8
E = primary voltage across N turns(volts)
B = peak flux density(gauss)
N = number or turns(primary)
f = frequency(Hz)
Ac = effective cross sectional area or core(cm^2)

So for the core data I threw out before you calculated a cross sectional area of 0.015in^2 I confirmed a cross sectional area of 0.0155in^2 which converts to 0.10cm^2(for the equation) And I am using E = 120Vrms(peak voltage makes the offset worse) B = 125g and f = 60Hz.
All this yields a turns calculation of 4,000,000 (16 times that of 250,000 in your calculation). So with such a large discrepancy I am wondering if you would mind showing us the equation you use for calculating the number of turns. Or perhaps you see where I have gone wrong with my calculation.

**edit

I do find that with 250,000 windings and the same cross sectional area that the flux density rises up to the range you said it would need to be in 2.7k gauss. Is it possible you used a higher flux density to calculate that turns number?

Thanks
-Nick-

 

[Militoy]

....So for the core data I threw out before you calculated a cross sectional area of 0.015in^2 I confirmed a cross sectional area of 0.0155in^2 which converts to 0.10cm^2(for the equation) And I am using E = 120Vrms(peak voltage makes the offset worse) B = 125g and f = 60Hz.....

I calculated approximate turns to get you into the range of 2Kg (0.2T) - which would be about optimum for a core of 125µ. In the quote above, you mention calculating turns for B = 125g. That would equate to a very low AC signal – the kind you might expect as a result of ripple on a DC reactor – but not consistent with most power transformer operation. I’m not sure if you selected that level from an inductor design worksheet – or maybe you mixed up the permeability of the material with maximum flux density. Permeability (µ) is not the same as flux density – it is an analog of the ratio of flux density to magnetizing force (B/H). Actually – when a material such as you selected is listed as having a permeability of 125 – that’s the initial permeability at low level. The instantaneous permeability changes with the magnetizing force – and each material behaves in a predictable manner, with characteristics that can be described or displayed by a performance curve known as a “BH Loop”.

Edit - I neglected to add that the equation you are using is for square wave excitation - using 4.44 instead of 4 as your constant will get it closer to where you want to be for a sine wave.

 

[NickS]

I did use the permeability as the peak flux density, my mistake. So based on the relationship you gave perm = B/H. I need to know the magnetizing force H to solve for peak flux density B. So I am unclear still on how you determined the value of H or B. You said solve to get your B into the range of 2Kg. I thought flux density was a constant of the core. But you make it sound like a parameter you control. If so you must have a lookup table for what the B should be for a given application. Is that correct.

If that is correct than for practice I should be able to back calculate how large of toroid I would need to satisfy the design(as ridiculous as using such a core for this app is).
Starting with the wire lets say I want to use 1000 turns of wire on the primary. And I want the flux density to be 2Kg then I can calculate the needed cross sectional area of the toroid as 22.5cm^2. But I take issue with the fact that by choosing B this calculation is completely independent of the core permeability. Surely I must need to have that info in the equation right? Which means the core must somehow define the peak flux density or magnetizing force. How do I extract that info if it is not given?

Thanks
-Nick-

**edit
Perhaps I am thinking about this all wrong. The equation does allow us to solve for B so perhaps what I am missing is the H such that we can find a core with suitable perm. So if the B is 2kgauss to be looking for a core of perm = 125 we would have to know that we want H =16.

 

[Militoy]

Initial permeability is a constant of the core material (independant of core size and shape). In a power transformer, you don't really have to worry too much about magnetizing force, after having selected an appropriate material for the application. Usually - there aren't too many optimum choices. For most 60 Hz work - I usually select 29 gauge M6 grain-oriented steel if I'm building a conventional core type, using EI laminations. For a cut-core or toroidal design, I'll also start with M6, but in 30 gauge. For specific applications, I might use Metglass or a cobalt alloy - but the higher costs of these materials would have to be considered. If I was involved with commercial work, I would use lower cost electrical steels, such as M19 or M22. Each material has a fairly narrow range of magnetic "excitation", where core losses can be balanced against copper losses in the windings. The initial decision on flux density is based on predicted core loss, and of course avoiding saturation.

 

[NickS]

Those are beautiful fish but I would like to learn how to catch them too.

So after further reading I think the disconnect I had can me fixed by examining the B-H loops for a given core. From this we can see clearly that a J material ferrite will never be able to go beyond H = 0.1 and B = 1.5kguass. Now revisiting the core if perm = 125 and our max B = 1.5kgauss(max) then H is 12 which is out of range so for the specified ferrite the max we can achieve is really closer to B = 12.5 gauss so H = .1 oersted with perm = 125.

Now revisiting the other calc E = 4.44*B*N*f*Ac*10^-8 we get an even more rediculous number of primary windings at 36Million!

 

[Militoy]

Those are beautiful fish but I would like to learn how to catch them too.

So after further reading I think the disconnect I had can me fixed by examining the B-H loops for a given core. From this we can see clearly that a J material ferrite will never be able to go beyond H = 0.1 and B = 1.5kguass. Now revisiting the core if perm = 125 and our max B = 1.5kgauss(max) then H is 12 which is out of range so for the specified ferrite the max we can achieve is really closer to B = 12.5 gauss so H = .1 oersted with perm = 125.

Now revisiting the other calc E = 4.44*B*N*f*Ac*10^-8 we get an even more rediculous number of primary windings at 36Million!

Not sure I follow how you put that last number together - but maybe it might be better to start fresh with a more "real world" problem - go through the steps together - and "build" a transformer on paper. I'll be rushing around quite a bit over the next few days - but I should be able to get to my desk (and my calculator) from time to time...

If you are interested in going through the process, pick a scenario - switching power supply; off-line or DC input; or AC mains operation - conventional linear auto/ or isolation transformer or PFC switcher; etc - and then an output voltage (or voltages) and load power. And we can make a few decisions, pick parts from a catalog and go through the process of putting together a design. Maybe you might want to think up something that could actually be useful to you - because if it's not too exotic and doesn't require expensive or weird parts - I'll probably throw one together for us to test and then dial in - because that's just how I think.

BTW - I'll do this only once on this site - for the sole purpose of illustrating the process.

 

[NickS]

Wow, that is an awesome offer. Let me think through a good application that should be interesting to many of the readers here(I don't want to rush a chance like this).

My last calculation came from a B-H loop(for that ferrite core we had been discussing) that showed a limit of 0.1 oersteds(H). So giver the relationship that permeability = B/H and the fact that the permeability of the core was 125 it was a simple calculation to see that the most B could be for that core was 12.5 gauss(which is a long ways from the 2.5kgauss range you said it would need to be in for good opperation). Then I just solved for "N" primary turns.

Thanks
-Nick-

 

[Militoy]

...My last calculation came from a B-H loop(for that ferrite core we had been discussing) that showed a limit of 0.1 oersteds(H). So giver the relationship that permeability = B/H and the fact that the permeability of the core was 125 it was a simple calculation to see that the most B could be for that core was 12.5 gauss(which is a long ways from the 2.5kgauss range you said it would need to be in for good opperation). Then I just solved for "N" primary turns.

Thanks
-Nick-

OK - Got it - I won't get into where you got off the tracks - Let's just go through a real one. Oh, BTW - it would be preferable, if you pick a conventional 60 Hz design, if you would keep it to 1500W or less - higher than that, and the iron in the line frequency ones cost some change to move around!

 

[NickS]

I'm back. I have been looking over several AC/DC SMPS architectures and I think we can agree on the fact that our industry is trending toward higher power density so to speak(more power in smaller packages). To be practical the solution should source from mains. So I am leaning toward a half bridge converter architecture for a couple reasons
(1) High Power handling which gives us some design flexibility

(2) The voltage splitting caps located pre-transformer reduce the max input voltage by half which in turn will hopefully reduce the transformer size.

(3) Good conversion efficiency

My main reservation is that if we design for 230V mains too, then the design must also include a input doubler for 120V mains which means for us lower voltage people the overall design size increases unnecessarily but it opens up the usefulness to a wider swath of our readers.

For now Lets look at some rough calculations to weigh out the impact.
Preliminary Option 1:
Vin = rectified US mains ~150VDC
Vout = 30V (Turns ratio approx = 5:1)
Iout = 4A (roughly 120W max)

Preliminary Option 2:
Vin = rectified mains ~300V
Vout = 30V (Turns ratio approx = 10:1)
Iout = 4A (roughly 120W max)

So I am still trying to decide on a controller. I am looking at TI, national Semi, and IRF. If someone knows of a good part or has a preference let me know. Once we pick a controller we will know the frequency range our transformer will need to work at.

Militoy is there an ideal frequency we want to be at for the transformer core options?

Thanks

 

[Militoy]

OK - an SMPS design is a good start - of course, building up a workable circuit around the transformer will mean getting serious with feedback, stability, thermal considerations, layout and so on - but, I'm game. (1) and (3) above both go with the bridge or half bridge topology - but let's discuss (2) a bit more, after we go over the input rail and topology.

First of all – let’s look at a few practical considerations with the power supply itself. Your 120W supply will be useful – but in real-world situations, we will be required to comply with EMI and harmonic-content requirements – no matter if we are building for the commercial, industrial or military markets. That means we will have to employ some kind of power factor correction in our off-line supply – to meet lower-order harmonic content requirements – as well as some EMI filtering, to meet higher-frequency limits. Nobody will want you to connect a supply to the mains – even a little one - if it interferes with everything else connected to that side of the distribution station! In practical terms, the parts we will need to meet these requirements won’t be significantly different in cost, number or volume, between 100 and 200W. You can set the current limit to 120W if you need to – but your hardware will be capable of producing 200W, unless you cut every corner possible.

As to the input range and rail voltage – most modern off-line SMPS supplies manufactured in the US are designed for both 120V and 240V operation. There is an advantage to designing a supply that is optimized for one or the other input – and that is the improved power-factor performance at 240V input; that a supply designed for narrow input range operation will have over wide input range. In practical terms – this means that the wide input range supply might have a power factor of 0.995 at 120V input – and maybe 0.93 or 0.96 at 240V input – while the supply set up to shut down below 200V may have a power factor of 0.99 at 240V. The wide range numbers are still good – they can just be improved with a narrow range design. It’s good to have a supply capable of both though - Pays your money – takes your choice.

Frequency –

The advantages of operating at higher frequencies are – as in all other aspects of design – a matter of tradeoffs. Higher frequency operation, in general, can result in reduction of magnetics and capacitor volumes. There are practical limits, though – as EMI, switching losses and core losses increase with frequency. In practical terms – we’ll be operating somewhere between 100-200KHz in most cases. I tend to select transformer / inductor / capacitor sizes for other attributes, then lower operating frequency as much as practical, to improve efficiency without running out of steam.

The PFC rail voltage needs to be set at some point higher than the input rail peak voltage at high line. For 240V supplies, this is usually 380-390VDC. For 120V (only) supplies, we’re talking about 190VDC or so. This is to prevent the peaks of the input sine from “blipping” above the switchmode boost input DC rail. This will be more understandable as we get into the actual circuit.

Now to (2). I would like to understand better, what you mean by “Voltage splitting caps…reducing input voltage by half”. Modern half-bridge converters are current-fed, zero-voltage / zero-current switching – usually soft-switching – typically taking advantage of power transformer parasitics. Can you direct me to a generalized schematic / description of the topology you have in mind?

We’ll get around to control chips – once we work through the more basic issues – hope this isn’t too much to digest all at once. I’ll check in again by tomorrow evening…

 

[NickS]

OK - an SMPS design is a good start - of course, building up a workable circuit around the transformer will mean getting serious with feedback, stability, thermal considerations, layout and so on - but, I'm game. (1) and (3) above both go with the bridge or half bridge topology - but let's discuss (2) a bit more, after we go over the input rail and topology.

Sure, I am no stranger to DC-DC converters. I know the stakes are raised but I am fully ready to give adequate consideration and learning time to good design practices.

First of all – let’s look at a few practical considerations with the power supply itself. Your 120W supply will be useful – but in real-world situations, we will be required to comply with EMI and harmonic-content requirements – no matter if we are building for the commercial, industrial or military markets. That means we will have to employ some kind of power factor correction in our off-line supply – to meet lower-order harmonic content requirements – as well as some EMI filtering, to meet higher-frequency limits. Nobody will want you to connect a supply to the mains – even a little one - if it interferes with everything else connected to that side of the distribution station! In practical terms, the parts we will need to meet these requirements won’t be significantly different in cost, number or volume, between 100 and 200W. You can set the current limit to 120W if you need to – but your hardware will be capable of producing 200W, unless you cut every corner possible.

I agree completely I had already been part hunting for PFC controllers and LC filtering at the time of my last post. I just didn't mention it because it seemed beyond the scope of what you offered to help with. But I would love to hear you out about the filtering and PFC too if your willing.

As to the input range and rail voltage – most modern off-line SMPS supplies manufactured in the US are designed for both 120V and 240V operation. There is an advantage to designing a supply that is optimized for one or the other input – and that is the improved power-factor performance at 240V input; that a supply designed for narrow input range operation will have over wide input range. In practical terms – this means that the wide input range supply might have a power factor of 0.995 at 120V input – and maybe 0.93 or 0.96 at 240V input – while the supply set up to shut down below 200V may have a power factor of 0.99 at 240V. The wide range numbers are still good – they can just be improved with a narrow range design. It’s good to have a supply capable of both though - Pays your money – takes your choice.

I would like to focus on US mains. When the time comes we can address what would change for a broad input support but I would like to focus on one input voltage

Frequency –
The advantages of operating at higher frequencies are – as in all other aspects of design – a matter of tradeoffs. Higher frequency operation, in general, can result in reduction of magnetics and capacitor volumes. There are practical limits, though – as EMI, switching losses and core losses increase with frequency. In practical terms – we’ll be operating somewhere between 100-200KHz in most cases. I tend to select transformer / inductor / capacitor sizes for other attributes, then lower operating frequency as much as practical, to improve efficiency without running out of steam.

All the parts I have looked at so far give some flexibility here. We will better know what we are dealing with when we settle on a controller.

The PFC rail voltage needs to be set at some point higher than the input rail peak voltage at high line. For 240V supplies, this is usually 380-390VDC. For 120V (only) supplies, we’re talking about 190VDC or so. This is to prevent the peaks of the input sine from “blipping” above the switchmode boost input DC rail. This will be more understandable as we get into the actual circuit.

I follow.

Now to (2). I would like to understand better, what you mean by “Voltage splitting caps…reducing input voltage by half”. Modern half-bridge converters are current-fed, zero-voltage / zero-current switching – usually soft-switching – typically taking advantage of power transformer parasitics. Can you direct me to a generalized schematic / description of the topology you have in mind?

Attached is a snipped picture of a very simplified example of the Half-Bridge controller as I know it. I will link the whole document a bit later. It gives a cliffs note explanation that follows my understanding of the half-bridge topology.

We’ll get around to control chips – once we work through the more basic issues – hope this isn’t too much to digest all at once. I’ll check in again by tomorrow evening…

Nope, great so far.

Thanks
-Nick-

 

[Militoy]

...I agree completely I had already been part hunting for PFC controllers and LC filtering at the time of my last post. I just didn't mention it because it seemed beyond the scope of what you offered to help with. But I would love to hear you out about the filtering and PFC too if your willing.


I would like to focus on US mains. When the time comes we can address what would change for a broad input support but I would like to focus on one input voltage



...

Attached is a snipped picture of a very simplified example of the Half-Bridge controller as I know it. I will link the whole document a bit later. It gives a cliffs note explanation that follows my understanding of the half-bridge topology.




-Nick-

To test out the transformer design, we might as well go through the whole process - as long as you're not in too much of a hurry, we should be able to put together some working hardware in a reasonable amount of time.

Let's go ahead and optimize on US input - but set up the design to work on either. The advantage to you will be in holdup time. The energy stored in the rail caps is based on CV^2 - and with the same volume cap(s) - we will be able to just about double the holdup time if we use twice the voltage at half the capacitance. think about this - and let me know if you need me to go into it more.

This schematic is the same as the one published in Abraham Pressman's power supply design book in 1978 - but with a different figure number. I'm curious where you found it - but in any case there are some advantages in eliminating those caps. I'll expand a bit more as soon as I can - but unfortunately tonight I need to throw a drawing together for a presentation on a HV supply - actually also a half-bridge, but a resonant topology. Will attempt to drop back in tonight, or tomorrow AM (Pacific time).

 

[NickS]

I am in no hurry. This is all about absorbing information to me.

When you say rail caps would that be the HV capacitors that follow the rectifier? I understand what you are saying about energy storage. But (1) what about the added bulk of the capacitors for the higher voltage. (2) Is the improved holdup time necessary for the now specified 200W output? I am fine taking this path but I do also want to address factors that bump the design size up.

For instance the input line filtering and PFC may not be mandatory to the end user of the power supply but it is mandatory to maintain responsible design practices. So the design size will increase for these aspects with good reason.

Now supporting a higher input voltage offers versatility yes but it does cost us on size I believe. So a space saving measure may be to keep the max input voltage low. But what I have not quantified is by how much.---Base on a quick digikey search for the same capacitance same series aluminum caps the one rated for 420V was a 30% increase in volume over those rated for 250V(50% increase in board area). So my reaction would be do we need this?

Next you said there is an advantage to doing away with the splitting caps. Does it out weigh the advantage of having a lower voltage at the primary of our transformer? I am under the impression that the higher voltage would cause the core size to increase for all else equal.

The schematic I posted comes from an app note by National Semiconductor. It is a great introductory read for SMPS theory/topologies. They have also been my favorite source for DC-DC converter designs but I can't seem to find a half bridge controller from them that can handle the voltages we are looking at(that is probably my fault I am sure they have it).

So aside from the questions above. Perhaps it is time to present a rough functional block diagram. When we reach an agreeable sequence I will generate a rough schematic to work from.

120/230V Mains---Rectifier(with Volt doubler caps and switch)---Balun line Filter----PFC---Half-Bridge converter---Transformer---Filter---30VDC

Thanks

 

[Mitchekj]

If I may inject a little:

You're looking for a PFC controller, for an active boost stage I'd presume, yeah? That will give you an automatic full-range input. (120/240Vac) Focusing on 120Vac alone may be a moot point if you go that route, since you'd need to rate your bulk caps at the 400V+ level anyway as the boost stage will... well, boost the DC bus to the ~340Vdc area.

There are other methods of getting good PF without active PFC, i.e. using critical conduction mode, and to some extent, discontinuous mode. If you were really set on a 120Vac only model. Maybe even a valley fill could get a decent PF, but at this power level I don't believe it would be the way to go.

 

[NickS]

Thanks Mitch,
I admit this will be my first circuit involving power factor correction. I guess I am unclear on why 120V would be automatically boosted but 240 would not. Is there something special that happens at 340V that can't be achieved at 170? Or is it just an industry standard thing that nobody makes PFC controllers to work on 120V alone? It sounds like we will be working through a 120/230V design but for future reference I would like to understand this better.

 

[Mitchekj]

Well, I should qualify that with that's the only way I've seen it done... since 120/240 operation was always a welcome feature. I don't see why you wouldn't be able to boost to 170Vdc instead, come to think of it.

 

[Militoy]

>>“When you say rail caps would that be the HV capacitors that follow the rectifier? I understand what you are saying about energy storage. But (1) what about the added bulk of the capacitors for the higher voltage. (2) Is the improved holdup time necessary for the now specified 200W output? I am fine taking this path but I do also want to address factors that bump the design size up.”<<

Yes – the “rail” is the front-end caps after the bridge rectifier and boost inductor.

Using higher voltage caps won’t affect cap volume to your disadvantage. Holdup time will be doubled –regardless of load. If you want to limit at 120W – no problem. I was just pointing out, that for virtually the same cost / volume, you will be able to squeeze out around 200W if you need it.

Holdup difference between 120W and 200W will be 200/120 – or 67% more holdup time for 67% less load. But, between 190V and 380V rail – holdup time will (approximately) double. The volume taken up by the caps is approximately proportional to capacitance x cap voltage rating (CV). So, in the same volume, we will have caps at half the capacitance value, but rated for twice the voltage. Energy stored in the caps though, is proportional to capacitance x voltage squared (CV^2). So, you can see – there’s an advantage to be gained by storing the energy in the caps at a higher voltage – and drawing lower current during discharge. Using caps of half value – but charged at twice voltage – will double energy stored.

>>“For instance the input line filtering and PFC may not be mandatory to the end user of the power supply but it is mandatory to maintain responsible design practices. So the design size will increase for these aspects with good reason.”<<

In most practical instances, the filtering and power-factor correction is mandatory. The front-end EMI filtering does take up a considerable percentage of the “volume budget”. But without it – you are asking for trouble, or forcing the end user to add external filtering to mitigate interference with other system devices. Often – adding on “black box” filtering can lead to system stability trouble – since the designer of the external filter is essentially “blind” to the internal power supply circuit dynamics.

>>“Now supporting a higher input voltage offers versatility yes but it does cost us on size I believe. So a space saving measure may be to keep the max input voltage low. But what I have not quantified is by how much.---Base on a quick digikey search for the same capacitance same series aluminum caps the one rated for 420V was a 30% increase in volume over those rated for 250V(50% increase in board area). So my reaction would be do we need this?”<<

(See above – no net increase in total volume).

>>“Next you said there is an advantage to doing away with the splitting caps. Does it out weigh the advantage of having a lower voltage at the primary of our transformer? I am under the impression that the higher voltage would cause the core size to increase for all else equal. “<<

We can stick with the half bridge if you would like – there are several advantages to that topology – as well as disadvantages. Yes – the core size (based on core cross section) theoretically increases in topologies that utilize only one half of the core BH loop – as opposed to both halves. In a high-power design, when we are struggling with core weight – this can be an important issue. One major disadvantage the half bridge has though, is that any little glitch in drive control – noise or hiccup in the feedback – false start, etc – may result in a short circuit across the rail – and two blown-up FETs (or transistors). When core volume isn’t at an absolute premium (as it typically isn’t between 100-200W) I lean towards a 2-transistor (FET) forward converter. The circuit looks a lot like a half bridge – but with the FETs tied hard to the (+) and (-) rails – and no problem if both are on at once (unless continuously shorted). In practice – the radiating surface area of the core will force it to be large enough to suit either topology in many cases. Again – pays your money – takes your choice.

>>“The schematic I posted comes from an app note by National Semiconductor. It is a great introductory read for SMPS theory/topologies. They have also been my favorite source for DC-DC converter designs but I can't seem to find a half bridge controller from them that can handle the voltages we are looking at(that is probably my fault I am sure they have it).”<<

National is pretty good – I use some of their chips, as well as General Semi, ST Micro, Fairchild, etc. I’ll try over the next few days to put together a practical list of stabile controllers for you to look over.

>>“So aside from the questions above. Perhaps it is time to present a rough functional block diagram. When we reach an agreeable sequence I will generate a rough schematic to work from.”<<

AOK – I’ll throw up a generalized cartoon, along with the list of controllers – and we’ll work on dialing in the details. I might be stuck taking a boat-ride some time over the next week or two, and my internet access may be limited for a few days at a time. Please excuse me if I seem to vanish for awhile – this is interesting for me – and I will stick with it.

(Mitchekj)

>>“You're looking for a PFC controller, for an active boost stage I'd presume, yeah? That will give you an automatic full-range input. (120/240Vac) Focusing on 120Vac alone may be a moot point if you go that route, since you'd need to rate your bulk caps at the 400V+ level anyway as the boost stage will... well, boost the DC bus to the ~340Vdc area.

There are other methods of getting good PF without active PFC, i.e. using critical conduction mode, and to some extent, discontinuous mode. If you were really set on a 120Vac only model. Maybe even a valley fill could get a decent PF, but at this power level I don't believe it would be the way to go.”<<

A significant percentage of boost converters are set up for 190V – 250V operation, for various reasons. The narrower voltage control bandwidth can be an important factor in certain situations. Agreed – in a general-purpose 200W converter, it’s usually not a big difference – but transformer design, switching losses and EMI constraints in some cases strongly favor a lower-voltage rail.

 

[NickS]

I see your point on the capacitors. But my primary goal is size reduction. If that means we cut the power back down a little so be it. I understand that a marginal increase in volume buys more output power capabilities. For now lets back off to 120W and try to keep minimum size for this rating as the driving factor in our decisions.

I have been on board with filtering and PFC from the start so I don't think we need to discuss the merits of that any further(ie your preaching to the choir on that) unless we are comparing topologies.

As for the switching topology I gathered from your last post that you prefer a forward converter to a half bridge. Again the driving force in my mind is size. I felt half bridge would buy us the best size reduction(I am aware of the shorting risk). If that is not the case and a different topology would be more compact then I am interested in exploring it.

National does not do offline controllers according to the reply from their apps engineering department so they are off my list. I am only finding one Fairchild HB controller and it is a resonant HB in the L6599 that could work. From ST micro I have found a series of controllers L6384-L6386 that could suit us. And TI has one in the UCC25600. Are there any you would like to add or omit from this list?

Enjoy your boat ride
-Nick-

 

[Militoy]

Back in the 'world' for awhile -

The L6599 is an ST Micro LLC converter chip – I wasn’t aware Fairchild was building a version of the same controller – but this is an early version of the same chip I’ve just started working out the magnetics on, for a (very) HV application – the L6599A (the L6599 isn’t recommended for new applications). I haven’t worked with this controller before – but it’s looking pretty good to me so far. BTW – I didn’t select it myself for the application – I’m doing the “physics” end of the project – while two circuit “gurus” are handling the control end of things.

The TI part doesn’t appear to have quite the same level of performance as the ST part – but I haven’t heard or experienced any specific issues with this controller. I’ll ask around, when I call the office over the next day or two.

The ST Micro L6384-L6387 series aren’t controllers at all – they are an older family of high-voltage gate drivers – suitable for high or low-side gate drive of FETs run off a high-voltage rail.

Of the above choices – I would (obviously) lean towards the L6599A. Since I already have some on my desk – familiarity shouldn’t be a problem. I will let you know if I hear any groans when I mention the UCC25600 to the circuit boffins.