Boost Converter Problem

[geobabu]

Hi All,
I just tried to set up a small Boost converter circuit as in the picture.
View attachment 68710

From the PIC , a 20KHz , 60% duty cycle pulse to given to the gate of the MOSFETs (IRF540), through a simple totempole driver.

Actually I am planning to setup a boost converter which can deliver around 25V @ 10Amp from a 12V 7~18AH batteries. So before connecting to this, I just tried with a small bench top supply 12V@1A.
When switch on the system, the 12V Vin is reduced to around 6 volt (ammeter shows a ~640mA current). And in the output its giving exactly 30V. I just added a screen shot of the waves (yellow - Totempole driver out, Blue - Mosfet Gate)
View attachment 68711
View attachment 68712
here in these images, a thick line (i don't know what to say that ) is generating in between the ON pulses. Please tell me what is that exactly in this case.

But the real problem is , this setup works fine for around 3 mins. after that , the input & output reduces with in 30secs. After mins
Vin -> ~ 2.5V
Vout -> ~ 7V
View attachment 68713

Please help me for solving this problem .
Thanks in advance

Geo

 

[dougy83]

That 'totem-pole driver' is ineffective. Q1 will be held on always.. in fact, there's a current path through Q1E-Q1B-Q4B-Q4E all the time.

What's with the scaling of the plots? 500mV/div? Are you using x10 probes?

 

[geobabu]

Yes , its x10 probe and scaling is 500mV/div

Geo

 

[alec_t]

That 'totem-pole driver' is ineffective. Q1 will be held on always

Try swapping the positions of Q1 and Q4, so that their emitters are coupled together and drive the FET gates.

As for the voltage drops your supply is clearly being overloaded and is probably going into shut-down mode.

 

[geobabu]

Regarding the loads,
1) I used a 5K resistor as load. in this case for around 3 mins the output voltage remained constant at 30V. Then within seconds, it suddenly drops to ~7V
2) a 200R & a 300R resistor. in both cases the out volt remained same at ~ 16V & ~17.5V resp. for about 3 mins itselfs and then its gone below.

So in such case can we say that its overloaded ?

Also regarding the totempole's output wave form, it shows a clear switch out , correct ?, anyway I will try it by swapping .

What will be that thick smash line between the high pulses denoting ?

regards,
Geo

 

[geobabu]

Try swapping the positions of Q1 and Q4, so that their emitters are coupled together and drive the FET gates.

As for the voltage drops your supply is clearly being overloaded and is probably going into shut-down mode.

My mistake ,actually in the circuit assembled, the totem pole transistor's connect is exactly as mention by you. While drawn in the circuit , its swapped.

Also regarding the switching section, I just reduced duty cycle
ie freq : 20KHz
duty Cycle : 20%
Load resistor 300R

Now it seems the output is stable at 19V , (Vin is 10.6V). And this lasts for more than 15mins. I think now its stable.

"So we simply don't give more duty cycle , if an appropriate load is not connected. " .

I believe its my first lesson need to cover in the Boost convertor design.

Thank you alec_t , dougy83

Regards,
Geo

 

[geobabu]

Hi All,
Please see the modified circuit , attached below. Its a simplified boost convertor , which worked well at boosting the voltage from a 12V@1A power supply.
In this circuit the out was regulated at 22V exactly. By connecting a 300R load, it gave an efficiency of 75% .
View attachment 68740

But a problem happened
When I replaced my bench top supply with a 12V , 7.5Ah SLA battery, with in no time, the MOSFET blasted. !!!
What may be the reason ?
Please guide me in this case.

Regards,
Geo

 

[dougy83]

The MOSFET is probably ON for too long due to its driver defaulting to the active state. Put an external resistor pullup on R5 to force Q2 on when not being driven by the microcontroller (i.e. on uC reset)

EDIT: The reason that it's blowing when connected to a 12V battery and not your 12V supply is because the battery can supply quite a large current, whereas a 12V/1A supply is limited (either inherently or by-design)

 

[geobabu]

Yes , that was very helpful . Thank you very much .
Initially there was a small delay included for the PWM initialization process. When I check in the oscilloscope, the MOSFET gate was on for around 440msec, before the PWM starts.
View attachment 68748
So, when that initialization delay is removed, then also for around 60msec, its high , as you mentioned.
View attachment 68749
10K pull up resistor is added to this line and issue is cleared. Now the MOSFET is switching without any problem .
View attachment 68750.

In the previous power supply case, the out put was regulated to 22V , but now, with the same code output voltage is regulated at 28V. Duty cycle is not going below to make the Vout 22V itself. (Now its seems to 24% duty cycle). So for checking , the duty cycle is fixed to 10%, (90% from PIC, since its in inverse mode) , but the Gate signal is seen as a improper pulse, to make the switch on , keeping the Vout in the 11.5V after the drop voltage.
View attachment 68752.

So whether this happens because, of the Pullup resistor and rest of the 2 stage transistor's delay's. Any way I just ordered TC4427, Mosfet Driver to overcome the issues of the transistor based drivers. But then also , just because of the interest to make it clear.

Once again , thank you very much
Geo

 

[dougy83]

Which signals are the yellow/blue traces connected to? Blue is the MOSFET gate, yellow is the BJT base?

Considering that in 10us the 120uH inductor current will reach 1A, 440ms will allow proportionally more (theoretically).

In the previous power supply case, the out put was regulated to 22V , but now, with the same code output voltage is regulated at 28V. Duty cycle is not going below to make the Vout 22V itself. (Now its seems to 24% duty cycle). So for checking , the duty cycle is fixed to 10%, (90% from PIC, since its in inverse mode) , but the Gate signal is seen as a improper pulse, to make the switch on , keeping the Vout in the 11.5V after the drop voltage.

It's possible that the power supply was limiting the output voltage, not your regulation method.

whether this happens because, of the Pullup resistor and rest of the 2 stage transistor's delay's. Any way I just ordered TC4427, Mosfet Driver to overcome the issues of the transistor based drivers. But then also , just because of the interest to make it clear.

The pullup won't be causing any issues (the PIC output will override it). The transistors also shouldn't be adding any significant delay (<~1us). Check your circuit perhaps.

 

[geobabu]

Which signals are the yellow/blue traces connected to? Blue is the MOSFET gate, yellow is the BJT base? .

Yes Yellow is the PWM out to the BJT base. & Blue is MOSFET Gate.

Considering that in 10us the 120uH inductor current will reach 1A, 440ms will allow proportionally more (theoretically).

Yes now I understood that. I never thought about such a loop hole existed there !! . Actually the first thing came in to mind is to parallel two MOSFETs, for more current handling . I owe you there for saving my 2 MOSFETs.

It's possible that the power supply was limiting the output voltage, not your regulation method.

The pullup won't be causing any issues (the PIC output will override it). The transistors also shouldn't be adding any significant delay (<~1us). Check your circuit perhaps .

Yes I need to check more into the circuit. Because currently the PIC controller circuit , Inductor, Capacitor & diode is placed in a PCB.
The driver circuit is assembled on another dotted board(actually my initial plan was to drive the MOSFET directly from the PIC's PWM pin !, that never happened) . And MOSFET is removed from the main PCB & placed on a Heat sink (to avoid with frequent removing, desoldering, soldering...etc already 5 are gone !!) . So all these 3 parts are connected through wires . & I believe it will additionally imply the capacitance & inductance to this circuit, especially to the Gate .

Thank you very much , for helping me.

Regards,
Geo

 

[MrAl]

Hi,

I assume you changed those two transistors to put their emitters together? That's a start, but then you'll probably need a pre driver with an open collector to drive them. That's because the max input to these two will be 5v from the PIC chip and that's not enough to drive the upper transistor emitter to a high enough voltage to drive the gate of the MOSFET unless the MOSFET is a logic level input mosfet (which the schematic shows as not being this type). Check that out.

You also do not show the values of your gate resistors which is also important to show.

Once you get the driver working you'll be ready to go on.

 

[geobabu]

Hi,
In my last post , just mentioned about a improper pulse going to the MOSFET's gate, at lower duty cycle. That happened because of the delay imposed by the base region , when switching off the Transistor in the saturation region. In the higher duty cycle, it was delayed about 1.6usec.
View attachment 68795

So a schottkey diode is added in between the base & collector of the input transistor. (also a small capacitor to llel to the base resistor,doesn't make much difference). Then a significant change occurred in that delay. (anyway there is a 120nsec delay still exists )
View attachment 68796

So the final circuit became as shown as below.
View attachment 68797

This circuit worked very well for me, by giving an efficiency around 77%. at lower load.
The output is regulated to 22.5V (changeable by setting in the PWM duty cycle) . When a 300R resistor is connected to the output, the duty cycle set to 16%. I need to check with higher load.
It may be helpful for the people like me, newer to the Boost converter design. Using a MOSFET driver will solve most of the constraints, but this helped me a lot.
Dear members , Thank you for the support given .

Additional Note: From my experience, If your soldering iron is not ESD safe, then never try to solder while the CRO probes are connected . Because my MOSFETs, PIC controller's PWM pin everything got spoiled., when I tried it ..!!


Regards,
Geo

 

[MrAl]

So everything is working ok now?

 

[geobabu]

So everything is working ok now?

Yes now its working very well. So Now I started with a Buck converter design, I think its better to start a new thread for it.

Geo

 

[WTP Pepper]

Some power supplies are fussy when having a switched current load across them even though the current is below their max output. Thurbys especially.