Designing a Buck Regulator That Costs a Buck

What do you mean by 'driving the mosfets' the drive for the mosfets shouldn't need 500-800mA.... Do you mean the Mosfets are switching 15-18v at 500-800mA? If so, what is this driving and why have you got a 48v bus? Why not have a separate 24v bus for the mosfets?

The whole ststem need designing as a whole ...I'm sure you have done the arithmetic...but starting with 48volts is throwing away a lot of power (48-18)x0.5 =15watts which is a lot to waste from a battery system.

These are rechargeable (lead acid?) batteries?... are they charged together (or as individual cells...I know this is unlikely...but you could tap off the various cells for different voltages then.... a horrid solution I know...but if you want cheap)

Anyhow there are a few comments to get the discussion started...

Otherwise you need a switcher to get down from 48 to 15-18 and ideally another to get from there to 5v.... complicated, messy, expensive, unreliable....(compared with starting nearer the right voltages and going linear....)

Del

Here is some more details on my system --

1. The application is for electric scooter/ bike.

2. 4X12V Sealed Lead acid battery is used as power for motor and also to run the control electronics.

3. I use 15V to drive the MOSFET gates, which consumes lesser current (not the entire 500mA~800mA), but other electronics -- hall sensor, throttle, indicator LEDs, microcontrollers etc takes a total of 500mA, which flows through the first switching regulator.

I studied some existing modules of controller for electric bikes and all of them uses some or other way of getting lower voltage from 48V bus. I will give a try on using 24V or 12V. If you can suggest me a way to design some buck regulator, it would be great.!!

Thanks.

I havn't design 'em myself, but there are some good chips available ...try Maxim or NS... search web for buck regulator circuit... there should be tons of stuff out there.

Good luck..souds like a fun project.
Del

Click on this link ... type in your requirements it comes up with plenty of alternative chips...don't know what their prices are like

Or maxim ... here ...they do good reliable app' notes, development boards, and samples. Their circuits always work

Del

Debrajdeb,

your cost limit is said to be 1$, but you didn`t tell us at what quantities. But thats an important information. At volumes of 10T ... 100T a resistor costs you only about 0.001 .. 0.005$, and a gate or transistor 0.01 to 0.05$.
At these high volumes it might be very effective to build a switcher with discrete components instead of using an integrated controller with switch which mostly is more expensive than 1$.

Please tell us your whole specs and we can help you more effective.

Are there polarity and overvoltage protection necessary ?
What about a brownout detection ?
Whats your temperature range ?
Is 500 .. 800mA a pulse current or the continuously available current without derating over environmental temperature ?

Regards Uwe

This is my reply to "uweka" -->

The volume is 50K per anum and aprox 5000 pieces per month.

Protection-- NO polarity protection required and overvoltage/ undervoltage detection is taken care by the microcontroller. We use another microcontroller that controls the motor and does all other stuffs.

Brown out-- Not required.

Temperature range -- 85deg Max.. But beyond 70deg, the microcontroller will shut down entire system (protection).

800mA is the continious current.

I want the design to be in discrete (for cost reason) and only protection that is needed from the DC/DC converter is that, even if any component in it fails, it should not allow more than 15V to pass thro to the next level. It can however get open and stop the entire system.

You need a DC to DC converter running off the 48 volts, charging up a capacitor at 15 volts. There are dedicated chips that do the clever bits and a MOSFET at the end to switch the 48 volts on and off. The frequency of which determines the voltage at the cap, which of course is sensed by the chip.....

Once you have 15 volts, you can produce your other voltages from that or make further dedicated switched mode supplies for them too......

A useful link to get you started is here:-

DC-DC converter

There are dedicated boxes around, with everything inside them, but they cost more than you are willing to put out. I personally think you are being "too cheap" as a failure here may cost you a lot more that a Buck!!

My reply to "Andy Germany" -->

Here in India, electric bike is looking like a big business and the models that we get around are all manufactured in China. The pieces are assembled in India and sold with different names. The components that are imported from China are very cheap. I would not say that they are cheap in quality, but the design is usually optimized for low cost.

I opened couple of Chinese units of "motor controller" and they use some version of DC/DC converter which are switching, but uses no inductor, but uses resistors to reduce the voltage.

When I try to design a controller and present to the Indian manufacturer, they are usually happy with the features, but not with the cost. The manufacturers are usually non-tech guys, who compares everything on money. Thats the whole driving force behind this thread!!

Meanwhile, I am checking the links that I got in this forum, to find something that match my requirement.

If the ones from China are good (purely resistive does not sound good to me, but I am no expert!) and the only problem is at failure time, add a Crowbar and a fuse, preferably replaceable!!

Make sure that the switching transistor is at least twice as good as needed in all respects and look after it and protect it from bad transients!! It cannot be that difficult!

I am thinking on similar frequency. Meanwhile, I got a buck regulator -- MAX5033, that is close to my requirement. I am giving it a try to find a lead.

Thanks everybody for your support and comments.

Hi debrajdeb,

I would hope you get lucky with the MAX5033 but I have some doubts as the datasheet says you get a maximum output voltage of 13.2V which is far less than your required 15-18V.

Nevertheless - if the output voltage might be enough for you I would suggest to take the MAX5035 with the same function but with an output capability of 1A instead of 500mA for the 5033. It costs about 25 Cents more but with the smaller one you run the risk of permanent thermal stress.

Another recommendation is to distrust Maxims datasheets as they are known in circles of developers for telling all the advantages but not telling you that they are possibly mutually exclusive.

Regards Uwe

Thanks uweka.

I have noticed the lesser output voltage of 5033. But I wanted to give it a try because higher output voltages are not available. Lets see how does it perform.

Best regards.

From the application note for MAX5035

(http://pdfserv.maxim-ic.com/en/an/AN3668.pdf)

Maximum Output Current (mA) 350 (VCONTROL= 0V)

Maximum Output Voltage (V) 12 (350mA output, internally limited by MAX5035)

You might want to go for any National chips. I spent 2 month with a MAX, I gave up and switched to National LM2674&LM2676). I came with two designs, one for 500 mA and one for 3A, which work like a charm.

Having a second look, the configuration for limiting current to 350mA if for supplying LED's. I could've look for another AN for 5035, more detailed.

Thanks "Indel" for the inputs. I got -- LM5007, which matches my requirement. Thats for voltage input upto 75V, which is good for me.

HI All,

I am also working with MAX5033 to step a 50VDC to 5V, to driving some logics.
For this buck config, Trace & values for wheeling diode, output L, BST cap, VD Cap Cin/Cout, FB line are checked and the recommended values was taken.
However the O/P was weired.... It just got some v.drop respect to the Vin (rather than a fixed 5V) and it follows the Vin if it goes 7.5V or higher... and finally the chip blowed.
model: BUSA.

thanks everybody for any Idea~

Kurry