• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

LT3799 Controller does not give Unity Power Factor...

Not open for further replies.


Well-Known Member
The LT3799 is said to be a “Flyback controller with Active PFC”.

However, when I ran the linear.com provided simulation of the LT3799 in LTspice, it is evident that the current is not really in phase with the line voltage. (as the attached waveform shows…the power factor is 0.97)

I have attached the LTspice simulation that they provided here, and have LC filtered the mains input current, so it can be easily seen. You can see that it’s not in phase with the line voltage.

Why are they saying that it is Actively Power Factor Corrected?.......when all it is doing is making the primary peak current be a scaled factor of the primary DC Bus voltage? (As you know, that provides a power factor of unity with the Boundary conduction BOOST topology…..but NOT with the Flyback topology).



Well-Known Member
Most Helpful Member
With out spending much time on this, It looks like it is working. pf=0.97
I have seen better, I have seen much worse.


Well-Known Member

The Error Amplifier inside the LT3799 is unusual. Usually, a “generic” error amplifier has an incoming signal_to_be_regulated coming into one of its pins, and then the other input pin of the error amplifier is a reference voltage.

The Error amplifier inside the LT3799 doesn’t seem to be like that. …It has a reference voltage on one pin, (or all three of them as it takes the lowest of the three CNTRL voltages) but the other input pin seems to be connected to the output of the scaled DC Bus voltage signal.

When a “generic” error amplifier is regulating, then the voltage on both of its input pins usually should be the same. But in the case of the LT3799 that would mean the scaled dc bus voltage being a constant voltage…..which it shouldn’t be…obviously.

It doesn’t make sense. Do you know what’s going on with the error amplifier inside the LT3799?

The attached LTspice simulation (and pdf schematic) shows two almost identical simulations of the LT3799. The only difference is that one has a 100n capacitor connected as feedback on its internal error amplifer…whereas the other one shorts this capacitor out…so the error amplifier behaves as a simple voltage follower…………………..the circuit waveforms for both simulation circuits, as you can see, are the same……so what on earth is the point of the Error Amplifier inside the LT3799?



Well-Known Member
Most Helpful Member
Yesterday I started to draw pictures but no time.....
1) You know how a "voltage mode" PWM works. Oscillator (saw tooth or ramp), error amp with a 1khz frequency response, comparator. The error amp decides if the output voltage is good or not. Comparator looks at saw tooth and error amp.
2) You know how a "current feed back" PWM works. The compare = error amp & current ramp. The OSC only starts a new cycle and has nothing to do with comparator.
3) The PFC-PWM that I use:
3a) Error amp has a response no faster than 10hz. (very slow) They watch the output voltage.
3b) There is a input that watches the sine wave (actually 1/2 sine wave) from the power line.
3c) Current feed back (current mode). One input of the comparator watches the current from the power line.
3d) Error amp does not directly connect to the comparator. A function must be added.
3e) The current pulled off the power line must be a sine wave!
3f) The output of the error amp goes to a (X) multiply function. (Error Amp) X (line voltage) = to comparator
4g) Comparator input is 1/2 sine wave at line frequency, height is slowly adjusted by error amp and its shape is from the power line.
This way the current pulled is shaped by the voltage of the power line.

All the above is in the LT2799 but the error amp. (error amp has been removed!)
When I build a PFC I regulate to get 400 volts. In this project you "regulate" to get 1A of LED current. The output current is not seen because you can indirectly see the LED current by watching the primary current.
Forget about temperature compensation and some other functions in what looks like the error amp error area.
Simply put, the error amp is forced to give Vref as an output. Really no error amp.
Comparator input (1) is a half sign wave with a peak of Vref.
Comparator input (2) is the current from the power line (in ramps at the switching frequency)

At low frequency (filtered) the current from the power line is sine wave with a peak of Vref.
At high frequency the current is ramps (at switching frequency) with a peak of Vref.
At any point in time the current peaks out at (power line voltage X Vref)

Current is "regulated" because current is related to Vref. The MOSFET turns off when the current reaches a set point.

There are DC to current LED drivers that do not watch the LED current directly. (not PFC) The MOSFET is turned on and off to keep the primary side current at some fixed current. Through the transformer the LED current is set. These are primary side current regulating. Because the LED is like a Zener and the voltage is almost constant, constant voltage X constant current = constant power. So from watching the primary current you can get constant power in the LEDs. (yes I know the "regulation" is not perfect!)


Active Member
Flyback - the LT3799 is an LED driver, so as Ron says, you regulate the output current by regulating the current in the primary. In your description of the error amplifier above, you are thinking of a regulated output *voltage* not output *current*.

Have a look at the LT3798 which is the equivalent part, but designed to regulate output voltage and you will see the conventional error amp with reference - and the world makes sense again...
Not open for further replies.

Latest threads

EE World Online Articles