Weird boost converter waveform.

[edmond_chan85]

Hi,
I had built a boost converter with a constant 24V output voltage from an input voltage ranging from 6-23V with a power rating of 100W. The converter is operating at 20kHz. I am able to boost up the voltage until 24V. But the problem is that my converter's efficiency is quite low (around 60-70%) and I am suspecting the losses in switch. I am using IRF540N power MOSFET.

Weird thing happen when I measure the voltage across and current through IRF540N. The current and the voltage waveform are in phase!!which means that when my switch is off, current passes through. when my switch is on, no current flow. But my converter is still performing well as it is able to step up my input voltage to 24V output voltage.

Any ideas??Please help me...My target is about 90% efficiency..

The 1st attachment is my diode waveform. The 2nd attachment is my switch waveform.
p/s: sorry for my poor english and can i post figure without using attachment?

 

[blueroomelectronics]

Schematic might be helpful.

 

[edmond_chan85]

the schematic as shown

 

[smanches]

You need to set the target input/output voltages and output current that you want the efficiency at. As you deviate from these, the efficiency will go down.

What Vin and Iout should the efficiency be for?

The switch in a boost circuit does not conduct the output current, and is 180 degrees out of phase with it. Is that what you are seeing in your scope plots above?

 

[Hero999]

Where's the feedback?

With no feedback, the output voltage will depend on the load and could rise to a dangerous level causing the output capacitor and diode to blow if the load is disconnected.

 

[Ubergeek63]

Hi,
I had built a boost converter with a constant 24V output voltage from an input voltage ranging from 6-23V with a power rating of 100W. The converter is operating at 20kHz. I am able to boost up the voltage until 24V. But the problem is that my converter's efficiency is quite low (around 60-70%) and I am suspecting the losses in switch. I am using IRF540N power MOSFET.

Weird thing happen when I measure the voltage across and current through IRF540N. The current and the voltage waveform are in phase!!which means that when my switch is off, current passes through. when my switch is on, no current flow. But my converter is still performing well as it is able to step up my input voltage to 24V output voltage.

Any ideas??Please help me...My target is about 90% efficiency..

The 1st attachment is my diode waveform. The 2nd attachment is my switch waveform.
p/s: sorry for my poor english and can i post figure without using attachment?

is that 8mH? i am actually surprised you are getting that much efficiency with that FET.

you need to be able to get the inductor current up to 80A in continuous conduction mode to boost 6V to 24V at 100W

 

[edmond_chan85]

Thanks for the reply.

smanches: the efficiency is for the critical point which means that in all the Vin values, the lowest efficiency should be 90%.
Regarding the switch waveform, shouldn't it be something like this?

The current and voltage waveform should be out of phase. So, the losses will be V x I and it should be small.

 

[edmond_chan85]

To Hero999: I am sorry because I used the simple schematic here. Actually I had a feedback for the system. I knew the consequence if I didn't put the feedback. Thank you.

To Ubergeek63: is it 8mH. but according to formula that I got from books (Ned Mohan- Power Electronics), my inductor critical value to operate in CCM is just about 2.1mH. may I ask how do you said that the inductor current up to 80A??how to calculate?? Thank you.

 

[mneary]

Do you have a capacitor on your voltage source to handle the ripple current?

And pay careful attention to your diode. If you don't have a high current schottky, 5% of your power would be going there.

 

[edmond_chan85]

To mneary: Is the voltage source my input voltage?? I didn't put any capacitor there. Is it necessary? And what value should I put? A polarity capacitor or non polarity capacitor?
I am using MBR1060 schottky diode with Average Rectified Forward Current = 10A , Non-repetitive Peak Forward Surge Current = 150A. Is it ok?
Thank you.

 

[mneary]

Yes the voltage source is your input voltage. Input impedance can drain efficiency. It should be an electrolytic (polarized), low ESR capacitor. I would use a minimum of 100 microfarads (rated at 35-50V). If anyone has done the math on this cap, please chime in.

MBR1060 can have a forward voltage of 0.8V at 10A. I know you're only drawing 4A, not 10A, but even if the loss is only 0.6V that's 0.6/24 = 2.5% of your power in that one part. If you can find a lower loss part (like an active MOSFET switch) you can erase some of this loss. Study active rectification in this article **broken link removed** particularly figure 10. Sorry I don't have a practical drive circuit for you.

Unless your inductor is huge, it's probably a major source of losses. At 8 mH, it must have a lot of wire. If you can find a 3 uH inductor with maybe 1/4 the existing resistance it could help a lot. I'm just guessing here, because I know nothing about the one you're using. If you can raise the frequency and scale your inductor accordingly you have a chance to do better on those losses. Your operating frequency should be able to be a lot higher before switching losses start to be more than the inductor losses. But remember increasing the frequency alone won't do it; the savings are to be had in the inductor. You must be able to reduce the turns and use heavier wire.

Speaking of switching losses, since the MOSFET has an 'ON' resistance of only 44 milli ohms, it should be dissipating very little. If the MOSFET is hot you should look into the gate drive circuit and make sure it's fast enough.

Finally, if your converter is 70% efficient with 100W out (143W in), something is dissipating 43 watts. The worst of the problem components should not be hard to find (ouch).

 

[Ubergeek63]

To Hero999: I am sorry because I used the simple schematic here. Actually I had a feedback for the system. I knew the consequence if I didn't put the feedback. Thank you.

To Ubergeek63: is it 8mH. but according to formula that I got from books (Ned Mohan- Power Electronics), my inductor critical value to operate in CCM is just about 2.1mH. may I ask how do you said that the inductor current up to 80A??how to calculate?? Thank you.

Physics: law of conservation of energy.

6V @ 100W = 17A continuous

sorry had it backwards...

at 70% efficiency you would need 25A

if you are running close DC in the inductor you can expect 75% duty cycle on the switch:

25A x 4/3 = 35A (critical conduction mode would put you out at 70-80A)

your crappy FET is dissipating 35²x0.044=50W not including switching losses

that means you need a 1C/W heat sink and a VERY good thermal pad to keep the junction temp down to around 125C... in other words, get a better FET

Dan

 

[edmond_chan85]

Thanks very much to mneary and Ubergeek63 ideas and information.
To mneary:
I wound the inductor by myself using a round ferrite core with insulated copper size AWG 8 (3.26 mm diameter) (IF I am not mistaken)..It is really huge and bulky.
Actually my PWM is produced by microcontroller PIC16F877 and I am using MAX4420 as my gate driver and connected a Rg with 150ohm.

To Ubergeek63:
You are right, my MOSFET is very hot. What do you mean by get a better FET?? A MOSFET with smaller ON resistance or should I readjust the current??

Back to my Question. Is it possible that a MOSFET in boost converter to have such a current and voltage waveform?? And is it cause by the inductor value (size), input current, MOSFET, MOSFET driver??

 

[MrAl]

Hi there,


This design might be a bit of a stretch, therefore most likely you are going to need the best components available.

With boost converters with very high output/input voltage ratios there are limits as to what the switch resistance and input resistance combined with inductor series resistance can be. The reason being that if the combination of these is too high then the needed output voltage can not be realized. The reason for that is because the gain with duty cycle actually goes down instead of up after you reach a certain point. That means that you may not be able to get the full output voltage (24v) with low line (6v input) just because one or more of these resistances is too high.
Luckily, we can calculate the maximum value that the duty cycle D can take on for these given resistances and a good approximation is this:

Dmax=1-sqrt((Lesr+Ron+Rin)/Ro)
where
Lesr is the ESR of the inductor, and
Ron is the 'on' resistance of the switch, and
Rin is the resistance of the source (in series with the input), and
Ro is the output resistance, and
the voltage drop of the output diode is assumed in series with the output voltage.

If the value of D goes above Dmax for any input voltage (low line is worst case) the gain will actually go down meaning the output voltage will decrease instead of increase as it always does up to that point.

This relatively simple equation allows you to calculate what your inductor resistance and switch resistance need to be in order to get the output you need. Because this design is so critical, i have to wonder if you have tested it at low line yet.

One way to get a better overall converter is to parallel two mosfets. That reduces the switch 'on' resistance of course. You do have to observe proper techniques though when doing this.

As a quick example, with 0.05 ohms switch 'on' resistance and 0.05 ohms Lesr and 0.05 ohms input resistance and 6 ohms output load you will not be able to obtain the full output of 24v with only 6v input because the duty cycle required for that would exceed the calculation of Dmax above.

You can estimate efficiency by calculating the power losses in each component taking into consideration the current flow and the 'on' duty cycle of each component.

And yes, the current waveform through the MOSFET should be out of phase with the voltage across it, but make sure you have all the test equipment hooked up right.

BTW, have you mentioned the converter switching frequency yet? You might want to look at the effects of higher frequencies in the inductor using 8 gauge wire which may not be the best choice at all due to skin effects. You do have a relatively high inductance value so that should help though. You might also want to estimate core loss in the inductor.

 

[smanches]

You will not be able to get 90% over the entire range at those power requirements. Even to hit 90% at all you're going to need synchronous rectification, a much smaller inductor, higher frequency, much lower Rds(on) for the mosfets, and low ESR caps. You may then hit 90% on a single voltage point, but with 4A output, it's only a maybe.

Start with the basic circuit and then reduce losses. After you have reduced the losses as much as you can, then you know what efficiency you will have. You can't start with efficiency in mind first, as it's not 100% under your control. When you hear of SMPS with 90%+ efficiency, they are talking about very small power requirements; a few hundred ma, not many amps.

 

[Ubergeek63]

Thanks very much to mneary and Ubergeek63 ideas and information.
To mneary:
I wound the inductor by myself using a round ferrite core with insulated copper size AWG 8 (3.26 mm diameter) (IF I am not mistaken)..It is really huge and bulky.
Actually my PWM is produced by microcontroller PIC16F877 and I am using MAX4420 as my gate driver and connected a Rg with 150ohm.

To Ubergeek63:
You are right, my MOSFET is very hot. What do you mean by get a better FET?? A MOSFET with smaller ON resistance or should I readjust the current??

Rg should only be 5-10 ohms. PIC16? expensive part to be marginal... you need to be able to RELIABLY change the duty cycle EVERY cycle ... that means ALL your math has to be done in 25uS, or about 500 instructions, since it takes 25uS to get the reading from the A/D and at 20KHz is 50uS.

so, tell me, why are you using an expensive ($8 each) chip to do a crappy job (marginal at 20KHz if you are lucky) instead of a $1 switcher chip?

 

[edmond_chan85]

To MrAl : Thank you for your information. You mentioned about paralleling two mosfets. Can I use 1 gate driver to drive two mosfets? Or do I need two separate gate drivers?
My switching frequency is 20kHz.

 

[edmond_chan85]

To smanches : Thank you for your information. According to what you had said, is it normal that I got the efficiency about 60-70% with the basic circuit?
For your information, I am started with the basic circuit and now trying to reduce the losses, increasing the efficiency.

 

[edmond_chan85]

To Ubergeeks63 : I am using PIC16 because I had other thing to add on in the future work. Such as MPPT for solar system. therefore, I need PIC16 with I/O ports to measure input current, input voltage, output voltage and output current to perform MPPT.
I am wondering what is the rating for Rg. 5-10ohms with 0.25W??or higher??

 

[mneary]

Don't know what I was thinking when I suggested your existing MOSFET should run cool.

Just a sanity check: You say that this is eventually for a MPPT solar controller. If your panel max power is around 100W (23V 4.35A), at panel voltage of 6V there is no way you'll have 25A .