I'm building a 450W PFC with a MC34262. It's the first time I build a PFC. I based my circuit on the reference design of page 12 of the datasheet.
I'm testing with two 100W incandescent lamps in series (the output voltage of the PFC is 400V). The problem is I'm only getting a 300V output voltage instead of 400V. Here is my schematic:
The inductor is made using two 33x19x11mm stacked yellow/white toroids (bought them here). The primary has a 1 mH inductance and the secondary has 12 turns which gives me about 20V to power the MC34262.
This is the MOSFET gate waveform while powering the PFC with 220V RMS and the two 100W incandescent lamps in series as load:
Zoomed out waveform:
This is the waveform in the drain:
In both cases, with gate and drain waveforms I have to stop the scope in order to get a clear waveform, I can't trigger the signal properly (I can take a picture if needed). I don't know if that means the output from the MC34262 is not stable and therefore I'm not getting the full output power.
Just in case here is the AC rectifier output:
As you can see the Vpk seems to be the same as the PFC output voltage.
And this is the pin 8 from the MC34262:
I'm testing with two 100W incandescent lamps in series (the output voltage of the PFC is 400V). The problem is I'm only getting a 300V output voltage instead of 400V. Here is my schematic:
The inductor is made using two 33x19x11mm stacked yellow/white toroids (bought them here). The primary has a 1 mH inductance and the secondary has 12 turns which gives me about 20V to power the MC34262.
This is the MOSFET gate waveform while powering the PFC with 220V RMS and the two 100W incandescent lamps in series as load:
Zoomed out waveform:
This is the waveform in the drain:
In both cases, with gate and drain waveforms I have to stop the scope in order to get a clear waveform, I can't trigger the signal properly (I can take a picture if needed). I don't know if that means the output from the MC34262 is not stable and therefore I'm not getting the full output power.
Just in case here is the AC rectifier output:
As you can see the Vpk seems to be the same as the PFC output voltage.
And this is the pin 8 from the MC34262:
Last edited:
I found that the voltage on the compensation pin 2 is nearly 0V and according to the datasheet the voltage should be at least 1.33V, what could be the cause?
Looks like the mosfet is cut of by the current limit not the voltage comp.
Measure over R75 and at the current sense pin 4.Perhaps there is a loss between the source and ground more then R75.
Mona
Measure over R75 and at the current sense pin 4.Perhaps there is a loss between the source and ground more then R75.
Mona
Thanks! I will check it. That resistor is kind of home made, I had to put eight 0.47 Ohm resistors in parallel to get that value (which I can't measure because my multimeter can't go that low on resistance) because I couldn't find that value in my local stores :/
Ok, I shorted R75 with a short wire and the output voltage increased to ~340V but still can't reach 400V. Also, shorting R75 makes the MC34262 really hot after a few minutes and I have to turn it off because some smoke starts to come out from it. I also replaced C35 with a 1uF electrolytic capacitor.
As Mona said: measure waveform on R75! Not short anything.
Measuring low ohms is easy: drive a known or measured current through it, while measuring voltage across the pins. R=U/I.
Layout can be critical.
what are you switching at almost 800khz???
With that inductance, that is far too high for what is needed
May inductance is "only" 120uH
C35 is not the problem. For start, use higher value Rsense resistor, one or two 0.47R will do. I think I have 0.22R and can get almost 1kw out
http://www.classdaudio.tk/index.php?page=PFC
With that inductance, that is far too high for what is needed
May inductance is "only" 120uH
C35 is not the problem. For start, use higher value Rsense resistor, one or two 0.47R will do. I think I have 0.22R and can get almost 1kw out
http://www.classdaudio.tk/index.php?page=PFC
Last edited:
Value of C37 is too high. According to datasheet it is not needed at all (if layout is correct). With not properly placed it can ruin the basic operation.
Vcc feedback value is questionable. Primary number of turns? High capacitance is neccessary to make it smooth.
The yellow-white (-26) core is very lossy, it will heat as hell.
0.05 ohm is extremely low. You don't want 20 A peak current, do you?
Vcc feedback value is questionable. Primary number of turns? High capacitance is neccessary to make it smooth.
The yellow-white (-26) core is very lossy, it will heat as hell.
0.05 ohm is extremely low. You don't want 20 A peak current, do you?
Right. If you use it, 1nF is the value, I have 1nF
You mean feedback from output? Is so, it is ok.
Yes, it does not seem to be what should be used here. Normal power material with gap will work.
No, I mean Vcc (Pin8). Been told 20V, but in what conditions? Induced voltage is variing proportionally with the difference between Vin and Vout. With 310V output it shouldn't reach 20V.
It is not intended. It is the result of the bad operation (probably due to high C37). If the circuit could work normally the frequency would drop to below 10 kHz at maximal load and peak of mains sinewave. (And this is also a problem.)
But it should work without it also. Lets see, what dragondgold will say about all that.
And didn't he say, he has 12 turns for secondary? That to me is high number, looking at what was used in datasheet.. unless he has a lot of primary turns as well, which I doubt
I calculated this inductance value using the equations in the datasheet using Po=100W as this is my minimum output power, as I understand, lower output power should give bigger inductance values, lower inductance value can make the PFC not work at low loads? Please correct me if I'm wrong.
I changed my Rsense to 0.24 Ohm (measured as suggested)
I changed C37 to 1nF as suggested now 🙂 I have 78 turns on the primary side and 12 turns on the secondary. I know this cores will give me a lot of loss but I only have them to test :/ I will be using some cores with gap I got from some TV's and printers later, it's hard to get proper cores where I live :/
As per R75, I just used the value suggested in the datasheet and also that's the value I calculated with the equations given. But I changed it to 0.22 Ohm as suggested!
This are the waveforms as requested. I'm using this to measure everything to remove undesired oscillations:
This is measured directly in parallel to R75 of 0.22 Ohm:
And this is zooming in:
I don't exactly know how this should look but there seems to be a very high frequency resonance there which I think is bad?
This is on Isense pin 4 with C37=1nF:
And this is pin 8 vcc:
The output voltage is now ~285V
Thank you all for your help! No one on my university know about this subject and it's hard reading books and not having someone to ask when something goes wrong, thank you very much to all! 🙂
Last edited:
Yes, I don't have a proper EMI filter yet but I will add it later for sure! I tried with the turn ratio given in the datasheet but I was only getting 12V for Vcc, maybe because the PFC is not working properly too.
I think I should also mention that my first tests where using 30VDC as input voltage instead of 220V RMS and shorting R73 just to test that nothing would blown up when I used 220V RMS. At no load, I got 400V at the output, but as soon as I put an small 1 KOhm load the output went down to about 25V. It's the exact same behavior I have now with 220V RMS.
I was doing some calculations, is it possible for the two toroids to be saturating? Each toroid should have a magnetic path length of 82mm (0.082m), relative permeability of 75 and a maximum flux density B of about 1.5T being optimistic. So, If I calculate the maximum allowed H it should be:
Hmax = ( 1.5 / (4*3.1415*10^(-7)*75) ) * 0.082 = 1305 A/m
And my H with a 78 primary turns a maximum inductor peak current of 5A should be:
H = (78*5)/0.082 = 4756 A/m
I have two stacked toroids, I don't know how to reflect that on the equation (permeability is reduced by half?). But that indicates a really saturated core! Could this be the problem or I'm doing something really wrong?
Hmax = ( 1.5 / (4*3.1415*10^(-7)*75) ) * 0.082 = 1305 A/m
And my H with a 78 primary turns a maximum inductor peak current of 5A should be:
H = (78*5)/0.082 = 4756 A/m
I have two stacked toroids, I don't know how to reflect that on the equation (permeability is reduced by half?). But that indicates a really saturated core! Could this be the problem or I'm doing something really wrong?
Hmax=Bmax/mu, no geometry is needed for calculation. What you calculated is excitation. And finally the result is correct, they will saturate. 2 core is no difference in this, it only makes inductance higher.
VCC=12V is enough when Vout is only 310V.
Minimum output power is meaningless. Inductance must be calculated for maximum power.
But the biggest problem I suspect is PCB layout, this is what makes current sensing incorrect.
RC filtering may compensate parasitic inductance of Rsense, but it have to be set correctly.
VCC=12V is enough when Vout is only 310V.
Minimum output power is meaningless. Inductance must be calculated for maximum power.
But the biggest problem I suspect is PCB layout, this is what makes current sensing incorrect.
RC filtering may compensate parasitic inductance of Rsense, but it have to be set correctly.
Last edited:
Ok, so I should reduce inductance and therefore number of turns to avoid the saturation of the core? I thought this white/yellow toroids where hard to saturate but the math shows it's really easy?
I will post the PCB layout, it's a 2 layer board, the path from MOSFET source to Rsense is very short and from Rsense to GND is very short too going through a via to the GND plane.
Hello!
This IC is very difficult to work with, as it is quasi resonant. Why not to try in a first instance, a fixed frequency one?
I had lots of problems with this chip in commercial SMPS's. One or two of them, I must to bypass it.
In few words, it is very hinchapelotas!
Good luck
This IC is very difficult to work with, as it is quasi resonant. Why not to try in a first instance, a fixed frequency one?
I had lots of problems with this chip in commercial SMPS's. One or two of them, I must to bypass it.
In few words, it is very hinchapelotas!
Good luck
Hi Osvaldo! Glad to see another Argentinian here! Thanks for sharing your experience, do you have any other PFC controller I can use and that is available here?
- Status
- Not open for further replies.