Soft Start for >PFC< for Sine Wave AND DC

[rentner]

Hey guys. I want to build a PFC, that works with both DC and AC (sine wave and square wave AC).

The problem is, that all PFC circuits so far just work with a sine wave. Theoretically it should be possible (quite simple), to get a PFC, that works at Square waves also. I don't want to say, that this makes any sense! I just want a PFC, that WOULD work for that. It also should have a very big Vin range. 110VAC as lowest input voltage and 400V DC as highest input voltage.

Well, not even that... It should be high efficient! The problem with that, of course, is the correct FM and PWM. It should not go lower than 30khz and higher than 150khz for efficiency. One thing though: The output voltage should be atleast at 450VDC and not go lower than the max input voltage of 400V DC, that I was talking about (logic!). It is not required to be isolated in ANY way!

I was trying to do it with a 555 timer, but first, the output voltage rises as hell, 2nd, I don't know, if it is just enought, if the Frequency drops, when the Voltage is higher. So Far I don't realy understand how to realize that kind of PFC, to be honest.


What does a real PFC do? Does it modulate the frequency? The pulse width? Or does it modulate both? And also: How do I also take the output voltage in to a count? I mean, if the Voltage rises any higher than the desired (lets just say) 450VDC, how do I lower the voltage again? That seems quite complicated to me.

Hm... and what about the high turn on current? I would like, not to get 50amps +, if I turn on the Supply! Something like soft start would be great. The SG3525 supports soft start, for example. But doing this with 2 Outputs at a max of 49% pulse width? That makes it even more complicated for me.

The thing is, I realy do NOT want to use some kind of top of the range IC to get that job done. In germany it is quite hard, to get good electronics for a low budget. NE555, SG3525, TL494 or ICs like that would be a good choise for me. Does anyone have an idea?

Most important: High efficiency! (Means N-Mos with good gate and low RDSon) No high startup current!!! The circuit should be usable at 100W but with some tweaking also at 2kW. A 100W solution will work for the beginning. But I also want to go much higher.  Just to have a high efficiency bench power supply. (The stuff after the PFC will not be a problem, just the PFC is, where I need help). It must be possible, to operate it on a power inverter with square wave output!

If anyone can help, feel free to answer!

[johansen]

There is no zero crossing required.
https://www.eevblog.com/forum/projects/transition-mode-boost-converter/

one issue with square wave ac is the input capacitor. many inverters work fine with slightly capacitive loads because the inductance of the 60hz transformer reduces peak currents to reasonable levels, but the 1-10uF capacitor on the front of a pfc boost converter could easily be charged to double the supply voltage. however, the bypass diode should shunt this extra voltage to the dc bulk capacitor! so there may not be any issues.
(other than the peak currents on the leading edge of the square wave)

there are a few ways to soft start the converter, some pfc control chips do this internally and it is not adjustable.

however due to the topology limitations, the only way to soft start the inrush current is with a resistor and a relay on a time delay, or a PTC which will waste a couple watts.

Read the datasheet for the L6562 and most of your other questions will be answered.

[rentner]

Is it not possible, to soft start the PFC/Boost Converter by using a low pulse width?


Just please answer this 2 questions, because I am failing understanding it:

How do you keep the DC/DC Converters output stable? Turn off the PWM? Or modulate the switching? If so: HOW?

Also: What does the PFC realy do? Does it increase the pulse width, if the voltage drops? If I understood correctly, a boost converter can only give out a voltage with a high Vin/Vout difference, if the pulse width is close to 100%. Well, is this correct? If so, does a proper PFC increase the pulse width, if the input voltage falls down and decrease the pulse width, if the input voltage rises again? IF SO: How do you modulate the pulse width properly over the 2 Variables called Vin AND Vout, since you don't want to overshoot.

What would I have to tell a TL494 or an SG3525 at the input, to give the proper output voltage?

Let's pretend: If the pulse width is limited to lets say 50% and the input voltage does stay stable, the output voltage will climb to a certain level and than not raise any longer? If this would let it stay stable, it would only require, to modulate the input voltage. But the problem is: How does the IC even know, what pulse width is the correct one? o.O  How does this work? Magic.

I get a Buck converter done, but every Boost converter overshoots, no matter, what I do! Please help me to understand this. In things Boosting a voltage I still am a noob.

[Odysseus]

The simplest PFC scheme to understand and implement is known as "boundary", "transition", or "critical" conduction mode, and it is usually explained in most datasheets and app notes for these types of PFC controllers.  A Google search for any of those three terms will give you lots of relevant information.

Here's a brief explanation of how it works:

1) The controller turns on the mosfet switch for a certain amount of time.  During this "on-time", the inductor current linearly ramps up from zero at a rate proportional to the input voltage.
2) After this period expires, the controller turns off the mosfet switch and waits for the current in the inductor to drop to zero.  During this "off-time" the energy stored in the inductor is transferred to the output capacitor through the diode.  The duration of the "off-time" is not known ahead of time and is not fixed.  It depends on the capacitor voltage and the inductor current at the moment when the mosfet was switched off.
3) The cycle repeats.

Thus, for a given on-time, the converter presents a certain resistive load to the input.  That is, the average input current to the PFC converter will be proportional to the input voltage, and voila, we have power factor correction.

The controller does not determine a specific duty cycle.  It only decides the on-time of the mosfet switch.  If the output voltage is too low, the on-time will increase, and vice-versa.  However, the adjustment of the on-time occurs very slowly.  The time constant of the feedback loop is designed to be several times longer than the mains input period.  This is so that the "resistance" of the PFC converter remains relatively constant during each cycle of the mains input voltage.

[rentner]

Here's a brief explanation of how it works:

1) The controller turns on the mosfet switch for a certain amount of time.  During this "on-time", the inductor current linearly ramps up from zero at a rate proportional to the input voltage.
2) After this period expires, the controller turns off the mosfet switch and waits for the current in the inductor to drop to zero.  During this "off-time" the energy stored in the inductor is transferred to the output capacitor through the diode.  The duration of the "off-time" is not known ahead of time and is not fixed.  It depends on the capacitor voltage and the inductor current at the moment when the mosfet was switched off.
3) The cycle repeats.

Thanks, that is quite understandable. In other words, the IC does not realy operate at a fixed frequency, I guess? That seems quite logic!

But how does the IC know, how much current is flowing through the inductor? Is it that, why many PFC circuits use a transformer, instead of a simple inductor? Well, one problem with this: Does a transformer not just "transfer" voltage depending on the windings ratio?

Well, if I would know, how to measure the current, if should be very simple to build my own PFC Circuit - even with discrete components. That would be very, very nice. I love discrete circuits. <3





Edit: Oh, I guess, I understand it now! I would just need a current sense resistor at Ground! So there is still one question: How do I limit the current? What about a very low duty cycle oscillation? If there is a low duty cycle, the voltage should rise much slower, so this results in a lower current, when turning on. Is my theory correct that way? A very simple way for example would be:
-Load a capacitor to the peak voltage of the mains. (Just high impedance as Vref)
-Compare the output with the input.
-If the output is close to the input peak voltage, start the real PFC circuit and turn off the oscillator.

What, if I would use a very low power buck converter, to use the output voltage (which is always almost identical) and power the PFC circuit with this? This would make a very high efficiency possible, since a buck converter can reach an efficiency of almost 100% minus the diode losses.

[Odysseus]

Typically, the boost inductor has an auxiliary winding that is used to sense the voltage across the boost inductor.  After the inductor current drops to zero, the diode becomes reverse biased and the mosfet drain voltage will begins to drop.  Eventually the voltage across the boost inductor, and thus the aux winding as well, will cross zero volts, and this condition is used to tell the controller when to start another cycle.
The aux winding is used to drive a ZCD (zero current detect) pin on the controller, and the aux winding also serves to provide the supply for the controller itself.

You could absolutely build a complete boundary mode PFC controller using a 555 as the core timing block. The on-time can be adjusted by driving the rarely used control pin with an opamp feedback loop that senses the output voltage.  I have LTSpice simulations of precisely this kind of circuit.

[rentner]

If I think about it: The propably most proper way to archive a soft start on a PFC circuit is to use

1.: A Buck Converter, if the Output Voltage is low when everything is plugged in.
2.: A Boost Converter for regular Power Factor correction.


One Problem with this: How do I run the Mosfet, which is totally "floating" on the High Voltage side?


And after all: Has anyone an idea, how to make that run without any problems? o.O


I quite don't get the idea of timing. :/

[Kremmen]

I don't want to rain on your parade but let's do a quick reality check.

From your first post you make it obvious that you are not very experienced in designing power electronics. I repeat what i have said several times in the past here and elsewhere: one does not begin with off-line power circuits, one graduates into them. With the skill level your questions demonstrate, only disappointment waits for you, or worse.

Point by point response to your questions would serve little purpose - my advice is for you to spend considerable time studying the fundamentals of power electronics and especially their dynamics. The key, really the sine qua non, is to understand the mathematical representation of the power circuits and their associated control and compensation circuits and transfer functions. You cannot _design_ something you don't understand - only cookbook solutions are available.

Once again, i emphasize that i don't want to discourage you from trying but do set more realistic targets initially, until you have achieved something that works the way you wanted it to, and then proceed from there. You doubt me now, i am sure but see for yourself if you can manage the following fundamental questions, with or without supporting documentation:
- create the steady state model of the most common switcher topologies - say buck, boost, buck-boost and derive their transfer ratios as function of duty cycle. (And i mean _derive_ not just look up).
-calculate the loss model as a function of duty cycle and individual component nonidealities (you wanted a good efficiency!).
-create the canonical small-signal model for the common topologies and from those, extract the transfer functions and then calculate the poles and zeros of those transfer functions.
-then derive the loop gain and phase loci that enable you to do a stability analysis to decide if and what kind of compensation is necessary to make your switcher stable under all conditions.

After that is clear, you can start looking into the PFC. Since you didn't want to use a ready made chip, you need to calculate the power factor yourself by designing the multiplier and a modulator that controls the selected switcher to maintain optimal power factor. Here several design tradeoffs need to be balanced and i promise you a 2 kW PFC device won't happen by trial and error.

By all means try your project, but please adjust your target initally much, much lower. If not, your project is destinied to join innumerable other pies in the sky.

I recommend this book if you can find it at a reasonable price somewhere: http://www.amazon.com/Fundamentals-Power-Electronics-Robert-Erickson/dp/0792372700

I'm not your mother so it is not my place to tell you what to do. So you do as you want but please consider the above friendly advice, not a hostile putdown.

[rentner]

I licked the lines before. I know the dangers. Of course I will not just put the parts together. I start with a low voltage equivalent in LTspice and ramp the voltages up and change parameters. For safety you do not need to be afraid.


The problem is not the theory, it more is the math. I always think in math. But I don't get how a simple PWM can precisely adjust the output voltage. There must be some kind of logic, but I don't see it...

[johansen]

the biggest issue with 2KW single phase is the leakage inductance between the switch, diode, and output capacitor.
and the inductor, and high current ripple heating up the capacitor, and input current ripple generating emi.

rentner, did you read the datasheet for the L6562? it explains most of your questions.
I made a 4 phase version using adhoc phase balancing, no logic or math.

also, regarding how the output is regulated.. that's what the opamp is for, and the internal 2.5 volt referance.

the compensation network is needed to stabilize the system. typical bandwidth is on the order of 30hz, because there is often up to 15% voltage ripple on the output capacitor at 120hz.


as far as dangers go, a dedicated 120vac pfc could be designed to only output 200 volts, which is relatively safe.
but if your mosfet driver stays on for some reason, it could litterally take an eye out when the mosfet(s) explode.
or the output capacitor could explode if it can't handle the ripple current required at 2KW

[rentner]

That's why I also include the parasitics of the components in the final Versions. No perfect Caps and no Perfect Inductors.

The Ripple current will be limited to the point possible with the frequency and a very high inductance (I guess at least everything over 1mH is big).

[johansen]

That's why I also include the parasitics of the components in the final Versions. No perfect Caps and no Perfect Inductors.

The Ripple current will be limited to the point possible with the frequency and a very high inductance (I guess at least everything over 1mH is big).

Ripple current into the output capacitor is the same regardless of operating frequency, it is a function of the input to output voltage ratio.
either it can handle it or it can't.

in practice, 2kw single phase TM mode boost is out of the question.

there are two phase CCM boost smps controllers available for a few bucks.
i don't have any experience with them, only TM mode