2 parallel Boost converters on the mains. Equal currents?

[cupoftea]

Hi,
These are two parallel Constant off time boost converters……each providing 1Kw to the same 2Kw output.
They are just meant to reduce the VA drawn from the mains, not give pure unity power factor.
Assuming that the FET currents are the same (which they will be due to the single error amplifier), then is the current in bridge1 equal to current in bridge2?
It uses L6564 in COT mode.
___----___---___----__---__---____--

As you know, the converter shown has just as good power factor (over say 1 minute) as a proper Boost PFC in cases of load jumping from max to min erratically and repeatedly. The shown converter has a "fast" voltage error amplifier loop....unlike a "Proper" boost PFC.

So anyway....just as good Power factor over one minute...yes......This is because a “proper” Boost PFC does not have an infinite output capacitor. Also, the error amp output of a “Proper” one cannot change much from mains cycle to mains cycle. So it cant possibly jump to the right level if the load suddenly goes from Max to min. -It will simply stop switching due to output overvoltage…as such, it has poor power factor in such a case.

The attached COT boost, therefore has better power factor than a “proper” boost PFC when the load is erratic and going from max to min repeatedly.
I feel convinced that you agree?

EN61000-3-2 is basically too simplifying, and assumes constant loading.

 

[KlausST]

Hi,

They are just meant to reduce the VA

I don´t get this...
How? How can two 1kW converters reduce the mains current with repect to one 2kW converter?

Klaus

 

[mtwieg]

Hi,
These are two parallel Constant off time boost converters……each providing 1Kw to the same 2Kw output.
They are just meant to reduce the VA drawn from the mains, not give pure unity power factor.
Assuming that the FET currents are the same (which they will be due to the single error amplifier), then is the current in bridge1 equal to current in bridge2?
It uses L6564 in COT mode.

If two current mode controllers share their voltage loop error amplifier (current control loops must be separate), then yes they should share average current equally.

The attached COT boost, therefore has better power factor than a “proper” boost PFC when the load is erratic and going from max to min repeatedly.
I feel convinced that you agree?
EN61000-3-2 is basically too simplifying, and assumes constant loading.

As far as I'm aware, power factor is only meant to describe operation with constant line/load conditions, there aren't any standards for evaluating PF with a quickly varying line/load, so don't try and invent one.

In general a PFC with higher bandwidth on the voltage control loop will have poorer THD on the input current, and thus lower PF. Some chips (like the UCC28070) try to cheat that rule with clever tricks, though.

 

[KlausST]

Hi,

if you want to improve ... then go for a 3 phase system.
There you may combine fast reaction time with low THD and perfect power factor.

Klaus

 

[cupoftea]

How can two 1kW converters reduce the mains current with repect to one 2kW converter?

...My apologies i didnt explain well.....Its the fact that these COT boosts significantly "improve" the power factor when compared to a totally non-PFC'd offline PSU which uses a bridge rect into an enormous smoothing cap. Thats the comparison i wish to make here.

(its just incidental that i use 2 pllel converters...its in fact because , AYK, a single 2kW Boost PFC is unadviseable due to losses. Single Boosters usually only are seen up to 1.5kW. There are of course exceptions, but generally, its not advantageous, generally.

Paralleling 2 CCM Boost PFCs using a standard controller (eg UC3854) is rather involved (UCC28070A aside for now)........it would involve hacking one of them to have the same multiplier output current as the other one, which ends up with a sea of auxiliary componentry.

The above 2 pllel COT boosters is a very easy way to get decent power factor with low component count.

 

[KlausST]

Hi,

O.K. I understand this.
But why 2x 1kW boost instead of 1x 2kW boost?

Klaus

 

[cupoftea]

A single 2kW Boost is liable to be very noisy and interfere with control of itself and other smps on the board.
1.5kW is a general maximum for PFC.
You can do it higher, but you need to damp it (to reduce noise) and you get high switching loss.
Also, components for single 2kW boost are somewhat rarer (eg torroids), than those below 1.5kW, where everything is all readily available offtheshelf.

You can do >1.5kW with low f(sw), but then everythings bigger.

So a factor is the boost pfc inductor........for a 2kW PFC from down to 120VAC.....its a lot of current......a torroid for that would need to be big with very sturdy thick windings which would be difficult to wind...you can use Litz but Litz is a nuisance to terminate reliably when unfortunately done by lowly payed assemblers. Another point is the PCB with such a single heavy component on it becomes awkward to manage in production, and even PTH components can pop open due to the board bending that happens when a board is picked up with such a big heavy inductor on it in one place....best to cut down to 1.5kw max and put the pfc's in pllel.

 

[cupoftea]

Hi.....on a similar subject.....

The attached (LTspice sim and jpeg schem) is two paralleled Boost PFCs using a standard Average currnt mode PFC controller (LT1248). (AKA UC3854, L4981B, etc etc)

They are able to be in parallel because the voltage at each of their current error amplifier inputs is the same in each case…..because one copys this voltage from the other.

This is the only way that they can be paralleled…but needs total of 4 current sense transformers……also some way of transferring the voltage at ones Currnt error amp input to the others……this is the bit that needs high component count to solve it. Can you think of a lower component count way?

I was thinking to reverse the currnt sense transformer output so that the signal is positive going…then go from there….may make the error amp signal easier to copy across?

--- Updated Sep 1, 2022 ---

Aaaahh.......so yes making the current sense txfmr output +ve does the trick, as attached.
So does anyone know if its ok to put a voltage source into the MOUT pin of LT1248?....as is done here

LT1248

https://www.analog.com/media/en/technical-documentation/data-sheets/1248fd.pdf

 

[cupoftea]

Hi,
So, The attached (LTspice and jpeg schem) shows two 1kW Boost PFCs in parallel, using two standard Average current mode control chips (LT1248). [AKA close cousins, UC3854, L4981B, etc etc]
One plays Master, other is slave. The slave copys the voltage on the Master’s MOUT pin, to its own MOUT pin….thats how they work in parallel.
However, MOUT pin has a current source in it, so an opamp buffer cannot simply be fed into it. Instead, the attached labyrinth of circuitry must be used……the slave is “hacked” so that its MOUT current source is outputting at maximum current…..then an opamp pulls down on the MOUT pin through a protection diode…so as to make the MOUT pin voltage the same as the Masters.

Can you see a lower component count solution?

Do you agree an opamp buffer cannot be simply connected to the MOUT pin?

LT1248 datasheet
https://www.analog.com/media/en/technical-documentation/data-sheets/1248fd.pdf

(My apologies in advance....It was considered this application would be popular enough to warrant a post...whilst not being "rocket science" enough to be worth keeping secret. -Just general SMPS content...the type that most countries have already outsourced due to its triviality.)

 

[cupoftea]

Hi, Just realised the post #9 is the main issue here...i wondered if i could put this as a fresh post?,... as it will not get answers this far down. It is one that would be of great interest to many on the forum, since it shows how to parallel PFCs....soomething not described anywhere on the web.

 

[scopeprobe]

Hi, Just realised the post #9 is the main issue here...i wondered if i could put this as a fresh post?,... as it will not get answers this far down. It is one that would be of great interest to many on the forum, since it shows how to parallel PFCs....soomething not described anywhere on the web.

Maybe worth looking at a proper interleaving controller which will be designed to operate like your more elaborate approach https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d46278d64ffd0178f986f0b708d4 Spend the saved money on some active rectification instead to reduce the thermals.

 

[cupoftea]

Thanks, but i am a "hobbyist", and ordered three UCC28070A's....all were dead...one had two broken legs...the other two ESD damaged.
Since i cant order through a "genuine company", i cant get good parts, so parts like UCC28070A which are more ESD sensitive than eg UC3854, are out of the question. If i order the UCC28070A, they are just going to supply me out of the "dodgy and not for good customers" box. I cant spend any more time on UCC28070A.
I think many are in the same boat on this.

 

[BradtheRad]

Thanks, but i am a "hobbyist", and ordered three UCC28070A's....all were dead...one had two broken legs...the other two ESD damaged.
Since i cant order through a "genuine company", i cant get good parts...

In acknowledgement to your status as a professional being higher than mine...

Seems like there's more and more hurdles as we try to use components that are expensive, or hard to obtain, or easily ruined the first time I use them. And increasingly more over the years as we hear about counterfeit and substandard IC's.
Yes, there's certainly an appeal to being able to put together a substitute circuit.

With your experience you're qualified to improve on my simulated twin interleaved boost converter below.

The boost converter operates as supply voltage is added to whatever comes through the inductor. Current never entirely stops flowing from the supply. As a result, differences between two branches of an interleaved boost converter are less pronounced (as compared to buck or buck-boost converters). If differences are slight then you may get away with a single control circuit driving both branches.

The control circuit rotates clock pulses to one branch, then the other. To alter duty cycle requires only a voltage change to the op amp input. (Further effort is required build a feedback circuit that automatically regulates output to 390V.)

In addition the converter exhibits surge to 700 VDC on power-up.