Generating DC voltage from lower PWM signal

How could I generate higher voltage with a PWM signal?

I had tried the following:

With PWM signal (93,750 kHz, 50% duty cycle, 5V) input to VPP_PUMP. If the VPP_FEEDBACK is lower, the PIC will increase the duty cycle, until the final output voltage reaches 13V. But when I turned on transistor Q3, additional load causes the output voltage to drop to about 10V. I seem not able to get the intended output 13V no matter how I try, for example experiment with different PWM frequency, various duty cycle limits.

Could anyone point me my mistake?

Thanks.

I do not see any real problem with your circuit.
The duty-cycle should be about 64% with a heavy load, but it can be much lower than that with light load. (Remember, for a continuos mode boost the duty-cycle is: DC=1-(Vin-(Vout+Vdiode)), so in your case that is about 64%). Try to limit the duty-cycle to about 80%.

The frequency does not play a big role here, it just modifies Q2's peak current.

What I suspect is happening is that the frequency may be a little too high for Q2. Try reducing it to about 50kHz. Also, try adding a resistor (10k) between the base and emitter of Q2 and reduce R9 to 1k. That is because you have no current limit and you want the transistor saturated at the peak current.

And, of course, re-check the circuit, there can always be surprises. For instance, are R14 and R22 really 10K?

Some questions:
1. Is the output 13V before you turn on the load?
2. How much current is the load drawing?
3. What's the duty cycle after you turned on the load?
4. Is everything voltage compliant? Any soft breakdown?
5. what's the ripple voltage when you turn on the load?
6. Did you try characterize your circuit with different load current?

If you found anything, do post it here to share with us your experience.

My guess is that the value of the inductor L1 is to big ..
I suggest that you try something closer to 100µH or even below ..
Also you may add 100nF capacitor next to C4 ..
Rest looks OK.

To reduce total load increase R10, R11, R14 and R22 by factor of 3.

In reply to VVV:

Actually I have a upper and lower limit for the duty cycle: 75% and 25%, respectively. After reading your post, I increase the upper limit to 80%, but it doesn't seem to improve anything.

By the way, I had recheck my circuit, seems unable to find anything wrongly.

Regarding your suggestion to reduce the frequency, actually I tried it before, half the original 93.75kHz. But it doesn't seem to improve anything as well.

Added after 18 minutes:

In reply to lastdance:

lastdance said:

Some questions:
1. Is the output 13V before you turn on the load?
2. How much current is the load drawing?
3. What's the duty cycle after you turned on the load?
4. Is everything voltage compliant? Any soft breakdown?
5. what's the ripple voltage when you turn on the load?
6. Did you try characterize your circuit with different load current?

If you found anything, do post it here to share with us your experience.

Click to expand...

1) Yes, the output was 13V, before the additional load was added.
2) Sorry, I'm not able to check this now.
3) Below is my benchmark output before turn on the load:
volt = 12.96V (measure from AN1), with multimeter I got 12.7V.
Duty Cycle: 135 -> equivalent to 26.5%

after turn,
volt = 10.50V (measure from AN1), with multimeter I got 10.2V.
the duty cycle is the max, 80%
4) Can you please explain your question again?
5) I don't have access to oscilloscope now. Perhaps I will check this next week.
6) Yes. At least 2 kind of load. With Q3 on or off. But I will try other load when I back to my post.

Thanks. Surely I will keep this thread updated.

Added after 5 minutes:

In reply to IanP:

IanP said:

My guess is that the value of the inductor L1 is to big ..

Click to expand...

The original design was 680uH. I change it to 560uH because I could not find 680uH in hand. But you said 100uH...? I think the higher the inductance, the better the result, isn't it?

Thanks.

Just comparing components used in boost converter application (based on the LM3578), I assumed that the current in your circuit will be ca. 10mA, so they used L=220µH and C (your C4) 220µF, but their frequency was only 55kHz..
If you say the ratio is in the range of 20% to 80%, maybe Q2 is not fully saturated?

Hello Ian,
Seems like I'm an alien here.

Can you please enlighten a bit how the calculation works out?

Thanks.

Here is a very good example where calculations are just calculations..

According to National Semiconductors book for boost converter the calculations of L should be as follows:
L=Vin²(Vo-Vin)/(ΔI * f * Vo²)
If Imin=2mA (through resitors R10 and R11), Imax=20mA, ΔI=18mA, Vin=5V, Vout=13V and f=100kHz then the calculated L value is 650µH.
But on the next page there is an example and the inductor is 220µH..
So here you go ..

Hello Ian,
I tried to search the 'book' that you said, no luck. Is it a hard copy or it's a PDF available online?

Thanks. I appreciate your all your contribution so far.

To other who reply, your suggestion are great as well. Although I'm not able to confirm it now (hardware change is not possible at the time being), I will do it when there is a chance.

Some theory:
https://focus.ti.com/lit/an/slva061/slva061.pdf

I would suggest you first verify the boost works with just a load resistor, instead of turning on Q3. If it works, you know to look for the problem elsewhere. So, just connect a resistor from +13V to ground. 510Ω will give you about 25mA. You will never need more than that. If the voltage stays at about 13V, then you know the problem is related to turning on Q3.

Another suggestion, if the above does not work:
Can you try to use a MOSFET instead of Q2? You can then keep the frequency at 100kHz.
You can use a very small resistor to drive it, say 100Ω.
That would ensure you do not have problems with the low speed of the bipolar.

Virus,

You gotta reduce either L1 or PWM frequency.
Be carefull (Q2 collector current will rise!)

Regards,
Regnum

Thanks very much to all.

I get the intended result now, without much change to the hardware!

The last suggestion by Regnum (in combination with earlier suggestion by other) works out perfectly, at least I get the output.

Well I would like to try all the suggestion to experiment/learn how boost power supply works out. But as I said earlier, I'm not able to do much change to the hardware now because I 'm away from my workshop. Anyway I could manage to change R9 from 10k to 4.7k, and by reduce the frequency of PWM by half, the output is 13V when loaded by turning on Q3.

Still there are some experiment to do...

Thanks again all. I appreciate all the inputs.