Buck converter with Darlington pair.

Thread Starter

pd111

Joined Apr 15, 2021
5
I am building a battery powered project, I'll be using a 4S LiPo battery. My actuators include linear actuators and stepper motors, which run on 12V and 6V respectively. Hence, I need a buck converter(2 in fact, 12V and 6V ) to maintain a constant constant input voltage to the actuators, so I was thinking of using of LM2576/LM2596 IC's for this part. However my current requirements are about 10 Amps which these IC's cannot fulfill.

1. I want to use IC's for buck operation because it will reduce my efforts of designing a feedback network. I cannot be using a IC rated 10A or above as it is out of my budget.

2.I was wondering if I could just add a Darlington pair in series with the buck converter ic so as to amplify the current. I'll be connecting the collector directly to the battery. Is this a valid approach?

3. If this is not a valid approach, pls explain why? And also guide me what should I be doing, using some IC and current boosting method OR building entire buck converter(including feedback) by myself?
 

DickCappels

Joined Aug 21, 2008
10,152
1620040333877.png
Sort of like this?

Not sure this will work well. If you use a single junction NPN it might be ok, but I think a Darlington might be too slow.
 

Ian0

Joined Aug 7, 2020
9,677
1. I want to use IC's for buck operation because it will reduce my efforts of designing a feedback network. I cannot be using a IC rated 10A or above as it is out of my budget.

2.I was wondering if I could just add a Darlington pair in series with the buck converter ic so as to amplify the current. I'll be connecting the collector directly to the battery. Is this a valid approach?

3. If this is not a valid approach, pls explain why? And also guide me what should I be doing, using some IC and current boosting method OR building entire buck converter(including feedback) by myself?
It would have been forty years ago! See datasheets for things like 78S40, TL494 etc.
Today we would use a MOSFET.
You would be much better off looking for an IC which doesn't have a built-in power device, such as TPS40200 because it is designed to drive an external power device, rather than the LM2576, which isn't, and probably wouldn't be able to turn off the power device quickly enough.
ICs without an internal power device tend to be cheaper (TP40200 is US$0.65 for 1000 off, or £1.66 each in Farnell)
 

Ian0

Joined Aug 7, 2020
9,677
View attachment 237350
Sort of like this?

Not sure this will work well. If you use a single junction NPN it might be ok, but I think a Darlington might be too slow.
I'm pretty sure it won't. The LM2576 has already got a saturation voltage of 1.5V, Adding a MOSFET that might need 4V to switch it on gives a voltage drop of 5.5V - for a small step-down ratio a linear regulator would be as efficient
The other problem is switch-off. The 2576 isn't a synchronous regulator, so there will be no pull-down whatsoever from Vout to ground - that MOSFET will take weeks to switch off!
As you said a single junction NPN would probably do, and keep the dropout voltage below 2.5V, which is still rather inefficient, but you need a pull-down resistor to get it switched off.
 

Thread Starter

pd111

Joined Apr 15, 2021
5
It would have been forty years ago! See datasheets for things like 78S40, TL494 etc.
Today we would use a MOSFET.
You would be much better off looking for an IC which doesn't have a built-in power device, such as TPS40200 because it is designed to drive an external power device, rather than the LM2576, which isn't, and probably wouldn't be able to turn off the power device quickly enough.
ICs without an internal power device tend to be cheaper (TP40200 is US$0.65 for 1000 off, or £1.66 each in Farnell)
Thanks for providing with resources, I'll need some time to check them out, once I understand them I'll definitely go ahead with them.

Why not try the good one, not the thirty-year-old one?
Actually I was aware of that IC that's why I replied quickly.

View attachment 237350
Sort of like this?

Not sure this will work well. If you use a single junction NPN it might be ok, but I think a Darlington might be too slow.
Yes sir, I was trying something on similar lines. Actually I wasn't thinking much about the speed part but now I will.
 

Thread Starter

pd111

Joined Apr 15, 2021
5
Why not try the good one, not the thirty-year-old one?
Can you guide me on building a complete buck converter as well, I mean including feedback as well. I know there are two ways to achieve this:
1. Using the traditional approach, ie comparators, sawtooth generators, error amplifiers for feedback. The problem with this is that it includes a lot of discrete components which can introduce a lot of noise.
2. Using a microcontroller with resistive potential dividers for feedback. The problem with this is that I don't have a voltage source to power microcontroller(the very reason to have a buck converter). Now I can use a separate battery dedicated to this microcontroller and put it in power down mode so as to prolong the battery. By battery I mean a coin cell or something similar.

Is second approach the best one, if I have to build any power converter with whatever constraints I want? Or are there better approaches?
 

Ian0

Joined Aug 7, 2020
9,677
Continuous time domain is always better.
If you have a sampling system, its phase response has to be well known up to the operating frequency, and you probably will have to sample at twice the switching frequency, then you have to implement the feedback and error amplifier digitally, which is far from trivial.
The TPS40200 datasheet has all the information you need, also, if you can find a copy, read TI's SLUP340, which I can't currently get to download.
I very much doubt than any of the discrete components could possibly generate as much noise as the error which would be introduced by sampling. How did you think discrete components would introduce noise?
You don't need triangle-wave generator and comparator as they are all in the TPS40200.
By the way, because of its simplistic feedback mechanism, the MC33063 is very noisy.
 

Thread Starter

pd111

Joined Apr 15, 2021
5
Continuous time domain is always better.
If you have a sampling system, its phase response has to be well known up to the operating frequency, and you probably will have to sample at twice the switching frequency, then you have to implement the feedback and error amplifier digitally, which is far from trivial.
The TPS40200 datasheet has all the information you need, also, if you can find a copy, read TI's SLUP340, which I can't currently get to download.
I very much doubt than any of the discrete components could possibly generate as much noise as the error which would be introduced by sampling. How did you think discrete components would introduce noise?
You don't need triangle-wave generator and comparator as they are all in the TPS40200.
By the way, because of its simplistic feedback mechanism, the MC33063 is very noisy.
Thanks I will study the datasheet and try to implement it.
 

Ian0

Joined Aug 7, 2020
9,677
No, it doesn't. But I think these parameters (pulses) are quite appropriate in this case.
Yes. That’s what it does. I have used one before, and it can produce ripple at all sorts of frequencies unrelated to the output frequency. If you want a noisy supply, this is the device for you!
Now that I know rather more about switched-modes and stability, voltage mode control usually needs a R+C in series, in parallel with the top feedback resistor. I was just wondering if it could improve this device, of it is was beyond hope.
 

Bordodynov

Joined May 20, 2015
3,177
I added a 1 nF capacitor and it did not improve the parameters. I increased the inductance value and the ripple decreased. But such chokes are very expensive.
 
Top