Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

The marketing disaster that is the "Bridgeless PFC"?

Status
Not open for further replies.

cupoftea

Advanced Member level 5
Joined
Jun 13, 2021
Messages
2,506
Helped
53
Reputation
106
Reaction score
108
Trophy points
63
Activity points
13,079
Hi

Please compare the following five methods of CCM Boost PFC..(in efficiency and solution size)

1…Standard Boost PFC with mains diode bridge
2…Standard Boost PFC with Sync FET on the boost diode, and mains diode bridge
3….As (2), but with FETs across the bridge diodes, switched at 100Hz to reduce the bridge diode loss.
4..Bridgeless PFC with Synchronous FETs and 100Hz FETs.
5…Dual interleaved Boost PFC with Mains diode bridge. (UCC28070A dual boost controller)

We all know standard Boost PFC with Full Wave Mains rectifier front end.

We now have “Bridgeless PFC”. AYK, this is basically two interleaved Synchronous Boost converters, with the inductor moved upstream. -There is also another semiconductor (switched at 100Hz) , either diode or FET, downstream of the sync Boost FETs.

So yes, we reduce power loss by the amount wasted in a single mains diode, and that wasted in the Boost diode….the thing is, we could put a sync FET in parallel with the Boost diode anyway….so really, we are benefitting by the loss in a single mains rectifier only.

Now, slow silicon mains rectifiers have very low Vf anyway…and their Vf reduces with temperature…so have we really benefitted by all that much? I woudlnt mind betting that a “Bridgeless PFC” with all of its control, and synchronous FETs, 100Hz FETs, plus the extra common mode choking needed for its “common mode issue”, means that the “Bridgeless PFC” takes up more room than the standard PFC?

And why on earth have we completely skipped over the half-way house between “Bridgless PFC” and “standard Boost PFC”…that is…. “Standard Boost PFC with synchronous diode”?

I wouldnt mind betting that the smallest solution of all, is the “Standard Boost PFC with synchronous diode”….would you agree?

And also, what about “Standard Boost PFC with synchronous diode” with a mains diode bridge, with FETs switched at 100Hz across the mains diodes………………….this last one, is, would you agree, very likely to be the most efficient, and smallest solution of all of them?

Why have (2) and (3) above been completed and utterly ignored by the power electronics fraternity?

Also, "Dual interleaved Boost PFC with mains diode bridge" likely gives “Bridgeless PFC” a run for its money?....RMS conduction losses well reduced from the single stage solution.

I am concerned that “Bridgeless PFC” has become a marketing department darling, without having the advantages it purports too have. Silicon investment had to concentrate somewhere, so they chose “Bridgeless PFC”, when all told, its no great shakes?

Mains Diode Bridge:
The “Bridgeless PFC” relies for its hero_worship on villification of the losses in the mains diode bridge with silicon diodes…..but Si diodes have low Vf and the Vf reduces with temperature. Even then, Si diodes can be paralleled with controlled FETs very simply switched at 100Hz.

I mean, take apart any 1Kw SMPS that you like, and you will notice that the heatsink on the mains diode bridge is way smaller than the heatsinking of the PFC and PWM stage semiconductors….also, its likely to be so far away from the fan that no actual blown air moves over it…….and this!...this is the reason that we are going “Bridgeless PFC” ???

The last 750W offline SMPS I took apart (and had very low failure rate…never involving the diode bridge BTW) had the mains diode bridge on a very small heatsink….which actually sat on top of the mains input filter, so it actually didn’t take up any room at all..The diode bridge heatsink was at least 10x smaller than the other heatsinks....and it was as far away from the fans as could be....didnt even have fan air blowing over it at all.
_________-----______
Bridgeless PFC component count:
The most efficienct rendition of the Bridgeless PFC has Eight Semiconductors.......Two (HF) Sic FETs each with a parallel Sic Diode....Two slow Si FETs (100Hz switched), each with a parallel slow diode. Is it really worth it?

A standard Dual Boost PFC has only five semiconductors....2 FETs, 2 Sic diodes and one Mains diode bridge.

So would you agree, “Bridgeless PFC = Marketing nonsense”?
 
Last edited:

We designed a 12VDC 18/22A inline adapter for Samsung in 2003, for powering computer displays, 85 - 265Vac in,

We used our own design for the "bridgeless" PFC input stage ( variable DC us Vout ) - which is really of course a full bridge of mosfets, a slow leg and a fast leg.

As we could use large 600V mosfets, and the internal diode of the top mosfets as the boost diode we saved 2 x 0.8V in the mains BR ( 3.84W at 90Vac in - total budget for losses was 11.3W ), which really did help us achieve the overall target of 95% efficiency - which was pretty good at that time.

So yes - horses for courses.
 
Thanks, may i please ask, the "fast leg" FETs were Sic?
Also, what about solution size?....it was smaller than any other way you could have done it ?(Non Bridgeless PFC).
Also, what about EMC....this went into production?....The EMC filter was bigger than otherwise?
You give an example from 2003....so you havent thought it useful to do another Bridgeless design since?

I used to service fans on an 800V SMPS with v v low fail rate, the mains diode bridge was layed on a alu heatsink approx 3mm thick, and 3.5cm by 6cm....Not very big....and theyre saying we can save that space......whoopy-do!!!.....but then need to put a load of other stuff in there. That little heatsink layed flat over the emc filter and took no room effectively. It had pretty well zero fan air blowing on it...the bridge was never failed on any unit.

AYK, many customers are not paricularly bothered by efficiency......even with a 4kW SMPS, theyre just not bothered...and especially if its not a high volume purchase. Its cost and maybe size that may bother them.
 
Last edited:

All Si mosfets, 100kHz booster, 100 - 200kHz SRC down converter

The whole adapter was about 80mm wide x 55 high x 150mm long, encased in high conductivity black plastic with fins moulded in, there was a copper wrap ( 1.5mm ) around the bottom of the board, where the fets were, which also contacted the magnetics on the top side, via gap pad, this closely fitted to the plastic case. EMC was just within spec, CM noise generated by the boost stage was dealt with via a CM choke, I think we had one of the first working commercial synch rect on the output too, push-pull, 12V 18A continuous.
 

The following Bridgeless PFC ref design doesnt have any evidence that the product in which it got enclosed into could pass Radiated EMC....

Bridgeless PFC ref design:

..as we know, Bridgeless PFC has a common mode EMC problem, which means a likely Radiated EMC problem.

The Bridgeless PFC has the primary ground getting a huge dv/dt every 10ms......(with respect to Earth)....and AYK, primary ground can't be conveniently made small in copper area.
So no Bridgeless PFC demo board is of any use really unless it comes with some kind of Radiated EMC approval. Searching the web for Bridgeless PFC demo boards which
have a Radiated EMC pass reveals nothing.
________________--
The following expensive research paper says Common mode EMC is a "great issue" with the Bridgeless PFC...

Bridgeless boost PFC converter has high efficiency by eliminating the input diode bridge. However, Common Mode (CM) conducted EMI becomes a great issue.
If anyone can afford to view this, please may it be requested to provide here some summarisation of the findings?
 
Last edited:

If you expect a reference design done by relative newbies with PhD's to survive the rigours of mains spikes, switch on, load transient events, RFI from a nearby phone turning on, or survive thermally at full power for more than 3 minutes - then you have my pity - no reference design ever above ~ 100W will ever make the grade for production or use in the real world - this is why companies employ very experienced engineers who have a good grasp of standards and real world ( i.e. out side the lab ) conditions to turn designs into real world toloerant production units .....
--- Updated ---

From the paper author:

 
Last edited:
From the paper author:
Thankyou very much indeed Easy Peasy, that looks like the actual research paper itself. Very Much appreciated.....shows that these IEEEResearch paper things are perhaps a bit of a money-grabber scam(?)

...So i read the kindly linked paper of post #7......it just backs up the title of this whole thread.

They basically acknowledge that the Bridgeless PFC has a pretty dire common mode EMC issue.
They then put forward two techniques to combat the dire common mode EMC performance.....
1....Symmetric method
2....Balanced Method.

2....Balanced Method first...
This looks to have no improvement (over the "plain" Brigeless PFC) above 2MHz......which makes it pretty pointless, since the hardest common mode problems are above 2MHz.....the Balanced method needs two diodes and an extra small inductor to be added to the original Bridgeless PFC.......and just to note, it needs 2 Boost inductors.

1......Symetric Method.....
OK....does improve on the "plain" Bridgeless PFC........But not at all frequencies...in fact, from approx 6 MHz to 12MHz, its WORSE than the "plain" Bridgeless PFC!!!
Also, the Symetirc method requires the two Boost PFC inductors to be coupled....and this by bifilar einding them on the same core....and also needs two extra power diodes to be added.


So again......the Bridgeless PFC does not look like being able to cut it !!!!

It is extremely noticeable that in the whole of the above paper (kindly linked by Easy Peasy in #7), they do not ever dare to compare common mode EMC scans of the "Bridgeless PFC" and the "normal, traditional boost PFC " !!! ........ I WONDER WHY ???!!!

One wouldnt be thought irrational in thinking that any product in production which uses a Bridgeless PFC, probably failed Radiated Emissions but they just shipped it anyway...because who is going to take their product to a Radiated EMC chamber and check if it passes or not?...nobody!

One would not be thought terribly irrational for wondering if the The Bridgeless PFC is in fact, a dirty great jamming antenna, spewing out emissions left right and centre!
It could be thought that the Federal Communications Authority will have to make their prescence known, and acquire offtheshelf Bridgeless PFC products, and test them out to see if they really do pass Conducted and Radiated EMC.....or have just been conveniently shipped with "self-approval".
_____________________________________
Also, in the paper of #7, they fail to state that getting symetricity is also going to mean the entire PCB tracking being done symetrically.....with respect to ALL the stray L's and stray C's.....now that is not going to be a terribly easy or convenient PCB layout!...good luck to anyone who trys to achieve that without going slightly insane in the process!
______________________________________
Also, in the paper of #7, they show a heatsink being conveniently connected back to neautral......without showing that that connection from heatsink back to neutral, is going to be a nightmare to route without getting common mode emissions in the mains filter coupling straight into that neutral connection.......so their paper is a little contrived to say the least.
_________________________________________
The article of #7, also fails to say that the Poor common mode performance of the Bridgeless PFC, is even worse if the Bridgeless PFC only uses a single Boost inductor. (due to lack of symetry)
_________________________________________
The article of #7 is also very misleading , because there is no downstream converter.....they only have the PFC stage.....which is not real....there's almost always a downstream converter.....and when you have a downstream converter aswell.....your primary side PCB tracking is even more extensive, and so that strays etc are worse, and the common mode problem of the Bridgeless PFC gets even worse!
_____________________________________________
..The thing is that the GaN marketeers have got their teeth into Bridgeless PFC....so we will no doubt see engineers getting tempted into Bridgeless PFC all over the place...engineers being made to feel inferior if they dont step up to the "Bridgeless PFC chalice......when its in fact not really all that good.
_________________________________
I reckon if you want PFC then just do 'normal' Boost PFC
If you want high Power PFC, then 'normal' Dual interleaved Boost PFC.
And if you want higher efficiency, then 'normal' Dual interleaved Boost PFC, with FETs very simply switched at 100Hz, to short out the bridge diodes.
There are power supply projects all over the world, not going ahead because nobody can decide whether or not to go Bridgeless PFC or not.
___________________________________
There is not a single article on the web, showing a product comprising a Bridgeless PFC, and a full approvals pass at Conducted and Radiated EMC , as well as pictures showing how big the PCB is and how big the Mains filter is...and the BOM cost, and efficiency comparison, with "normal" PFC.... (IE, in comparison to "normal" Boost PFC.).......not one single article on the web !!!
 
Last edited:

made me smile: " So again......the Bridgeless PFC does not look like being able to cut it !!!! " from Mr CupofTea

there are squillions of Telecom grade rectifiers out there - 1-3kW, that use the so called Bridge-less approach in one form or another - all meet class B for EMC ( which includes common mode noise ) - A standard LISN measures CM+DM on the phase, and then the Neutral - so you can be sure there are clever engineering ways to overcome it. Also it is more efficient.
 
there are squillions of Telecom grade rectifiers out there - 1-3kW, that use the so called Bridge-less approach in one form or another - all meet class B for EMC
Thanks, it would be great to see the schem BOM, Layout, cooling, Heatsinks, etc etc for these, not that we would be able to. Eg, what measures did they use to pass Conducted EMC?, particularly pertaining to common mode EMC.....(and of course Radiated EMC, because poor common mode performance leads to poor Radiated EMC performance)........and did they "self-approve" for EMC?.....we know that big companies can do this....can fail EMC and still ship...without worry....big Lawyers!

One wouldlnt be considered subversive in suspecting that the Bridgeless PSUs likely fail Radiated & possibly conducted EMC , but luckily dont interfere with the comms equipment that they are supplying.

A small start-up, with no lawyers, might struggle to put out a Bridgeless PFC design.

....I mean, one would wonder if they used the bifilar wound coupled Boost inductor(s) method?....as in paper of post #7 above....and if they did, did they have that in first prototype?......a big rich company can of course afford to do a non-coupled design first, check it out, send it for pre-approvals, then re-do it if it fails badly.......a small company cant afford this kind of repitition. They have to get it right first time.

AYK, the "engineer in the street" cannot simply afford to put together a Bridgeless prototype and test if for conducted and Radiated EMC......and in truth, there is no accurate way to calculate what kind of EMC scan trace a particular power supply will produce....(ref common mode and radiated EMC) ......so without this...most poor companies wont be able to risk embarking on a Bridgeless PFC design....because Bridgeless is potentially going to fail them on conducted and Radiated.....they just cant pre-calculate this......realistically its impossible to do so...and the paper in #7 shows the far-out methods which need to be done to pass EMC with Bridgeless,,,,,extra expense.........i mean seriously, two boost PFC inductors wound bifilar on the same core....... Who is going to do a PSU which may end up needing the expense of that?....when they could do a "normal" boost PFC that only needs the one inductor.
 
Last edited:

Thanks, it would be great to see the schem BOM, Layout, cooling, Heatsinks, etc etc for these, not that we would be able to. Eg, what measures did they use to pass Conducted EMC?, particularly pertaining to common mode EMC.....(and of course Radiated EMC, because poor common mode performance leads to poor Radiated EMC performance)........and did they "self-approve" for EMC?.....we know that big companies can do this....can fail EMC and still ship...without worry....big Lawyers!

One wouldlnt be considered subversive in suspecting that the Bridgeless PSUs likely fail Radiated & possibly conducted EMC , but luckily dont interfere with the comms equipment that they are supplying.

A small start-up, with no lawyers, might struggle to put out a Bridgeless PFC design.

....I mean, one would wonder if they used the bifilar wound coupled Boost inductor(s) method?....as in paper of post #7 above....and if they did, did they have that in first prototype?......a big rich company can of course afford to do a non-coupled design first, check it out, send it for pre-approvals, then re-do it if it fails badly.......a small company cant afford this kind of repitition. They have to get it right first time.

AYK, the "engineer in the street" cannot simply afford to put together a Bridgeless prototype and test if for conducted and Radiated EMC......and in truth, there is no accurate way to calculate what kind of EMC scan trace a particular power supply will produce....(ref common mode and radiated EMC) ......so without this...most poor companies wont be able to risk embarking on a Bridgeless PFC design....because Bridgeless is potentially going to fail them on conducted and Radiated.....they just cant pre-calculate this......realistically its impossible to do so...and the paper in #7 shows the far-out methods which need to be done to pass EMC with Bridgeless,,,,,extra expense.........i mean seriously, two boost PFC inductors wound bifilar on the same core....... Who is going to do a PSU which may end up needing the expense of that?....when they could do a "normal" boost PFC that only needs the one inductor.

You can make a reasonably stab at conducted emissions (especially differential mode) through spice simulations by doing a FFT Simulation of the currents on the nodes which go to external apertures. Common mode is more difficult to estimate as the current flows are more dificult to identify but typically the CM Noise comes from the high speed switching nodes. If you know what frequencies your generating you can make a reasonable stab at a fitst pass CM filter design also. If you have a earthed box and your conducted emissions are controlled then theres little else left to radiate.
Spend time Researching Good Layout, Correct use of Screening around switching nodes and heatsinks, Creating Decent Quality Windings (control leakage etc..), Controlling Edge Speeds, Damping/Snubbing, Use of Metal Box for screening, bonding resistance in relation to EMC, Common Mode Filters, Differential Mode Filters.

Theres some great power conversion books that cover these topics in reasonable detail
 
If you know what frequencies your generating you can make a reasonable stab at a fitst pass CM filter design
Thanks, you bring very good points to the table......AYK, common mode problem spike frequencies of SMPS's are not really calculable. I used to work for an EE that did the PSUs for British military...they described picking common mode EMC components for a power supply, as like "shovelling currents into a cake mix"

If you have a earthed box and your conducted emissions are controlled then theres little else left to radiate.
.........I think you may have hit the nail on the head here.....ie...if youre going to go bridgeless PFC....maybe make sure you have a well sealed and gasketted metal enclosure, which is very well earthed.......(which it wont likely be, as most earth cables have high inductance in reference to the frequencies concerned.)

you can make a reasonable stab at a fitst pass CM filter design also.
.....AYK, with a bridgeless PFC...your "first stab" might have to involve a Bridgeless PFC with coupled Boost inductors, (as in post #7 above) wound bifilar on the same core....And your boss will be getting worried about the cost that might get racked up in producing this Bridgeless PFC based SMPS and getting it through EMC approvals.

Theres some great power conversion books that cover these topics in reasonable detail
Thanks, please do bring forward titles of those that cover the Bridgeless PFC fully including getting it through EMC.....though i have to confess, i have violated my Electronics book spending limit years ago.........few seem to be less than £100 now
 
Last edited:

"common mode problem spike frequencies of SMPS's are not really calculable. I used to work for an EE that did the PSUs for British military...they described picking common mode EMC components for a power supply, as like "shovelling currents into a cake mix"

Fundamentally noise exists because its being generated and the noise generated is typically predictable by design if you look in the right places. If you know where noise is generated you can direct it to where you want (i.e not on the cables). This can be done through a combilinatation of damping, filtering and component design (magnetics).

".........I think you may have hit the nail on the head here.....ie...if youre going to go bridgeless PFC....maybe make sure you have a well sealed and gasketted metal enclosure, which is very well earthed.......(which it wont likely be, as most earth cables have high inductance in reference to the frequencies concerned.)"

I'm not sure of bridgeless PFC specifics but a earth cable is a radiator like any other when it comes to EMC. You need to keep the current loop in the box (not on the cables) be it differential or common mode.

.....AYK, with a bridgeless PFC...your "first stab" might have to involve a Bridgeless PFC with coupled Boost inductors, (as in post #7 above) wound bifilar on the same core....And your boss will be getting worried about the cost that might get racked up in producing this Bridgeless PFC based SMPS and getting it through EMC approvals.

If its too expensive then i would consider other approaches or determine if the budget is realistic to the requirements

Good books cost £££ unfortunately but the money you spend on basic learning is far less struggling at test houses.

https://www.amazon.co.uk/Switchmode...p-0071639713/dp/0071639713/ref=dp_ob_title_bk

 
Thanks,
Interestingly, this book..

by Billings & Morey doesnt cover Bridgeless PFC at all. I have a copy, and chapter 4.xx has PFC but not bridgeless. Dont know it Pressman covers it.
 

A search revelas neither Drs Basso or Ridley seem to have material on the Bridgeless PFC (mind you they are both more into feedback loop control).
There almost certainly is not a book on the market that deals in depth with Bridgeless PFC...

.......To be worth reading, a book on Bridgeless PFC would have to show an actual mains powered product, (not in a metal earthed enclosure) being done with Bridgeless PFC, and then done with "normal" PFC.....and the two compared in terms of cost ,size, EMC scans, filter size, etc etc.

......only then would that book be worth reading.....until somebody does such a book....then Bridgeless PFC literature will be confined to the type of #7 above.

The Bridgeless PFC pretty well needs two Power inductors, in order to give it the symetry it needs to fight its inherent common mode EMC problem.....
Also, whereas in a "normal" PFC the switching node is confined to the drain of the switching FET.....IN A Bridgeless PFC, THE SWITCHING NODE IS VIRTUALLY THE ENTIRE PRIMARY SIDE COPPER TRACKING AND COMPONENTRY !!!

So whereas with your "normal" boost PFC , you can layout to minimise that limited switching node, in the Bridgeless PFC.......its spewing out almost everywhere!....and then you have to try and layout so that all your strays are symetrical, so you can minimse common mode EMC.

I first met the Bridgeless PFC in an interview with a well known Component distributor......they called me forward for interview to design a converter.....when i got there......they said they wanted a Bridgeless PFC designed, so that they could proove to their doubtful customers that it was a goer........they already had an SMPS engineer there working on it.........and he and the boss were asking me over and over...what measures you take to reduce common mode EMC.......they never got back to me and i took another job elsewhere........but this topology (Bridgeless PFC) has "problem child" credentials, its plain to see.

The following paper looks like a good comparison for Bridgeless vs "normal" but is for IEEE Payee's only...
____________________________________________
This paper by two of the authors does not show comparison
-----------------------------------------------
This one deals with zero crossing noise of the bridgeless PFC...(Yet another of its problems)
____________________________________________
Page 2 of this articale on Bridgeless PFC...

.....(LHS, bottom of second paragraph down, states that .....
{QUOTE___] These two modifications eliminate the common-mode noise problem of the bridgeless-boost PFC [----UNQUOTE].....and speaks of the addition of a second boost inductor, and 2 diodes.
....Though such a statement is in disagreement with the paper of #7 above......#7 claims that the simple addition of those two things does NOT solve the common mode EMC issue.
____________________________________________
 
Last edited:

This is certainly one area where in depth experience of boost converters and emc can aid a design project that is seeking higher efficiencies thru the use of the so called "bridge-less " approach.
 
Thanks, yes, and since its not possinle to pre-calculate what the Conducted or Radiated EMC scan trace will look like before starting a project with Bridgeless PFC, the project may well need to have more time allocated to it than if the "normal" Boost PFC were used.

Also, the burning question would be, and i am sure all of us would like to do it....is make a product in a non earthed metal enclosure, with both "normal" and "bridgeless" PFC, (plus the rest of it) and compare the size and BOM cost, particularly pertaining to the mains EMC filter.

..ie make the comparison, after Conducted and Radiated EMC had been passed for both products.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top