Buck-converter with dual-MOSFET operation for constant current output

[deen25]

Hey,

I'm a newbie trying to figure out a few crucial additions to my circuit I'd like some tips and suggestions..

Basically, I'm building a circuit that will produce a constant current to variable loads, I'm using a buck-converter model with a supply of around 10V, the inductance should be around 10-20µH with a ripple current of about 1A and a total constant output current of 50A.

Currently I'm using a LM2724A dual-MOSFET driver and an MCP1630 PWM source to control the switching of two MOSFETs. I need help with selecting an appropriate input (coupling) capacitor and the correct logic-level MOSFETs to complete the circuit.

Any suggestions for these two components? I need to look at ripple current and ripple voltage rating of the capacitor...and I'm not sure if the MOSFET current (ID) should be rated at 50A or if a lower value is acceptable?

thanks in advance!

cheers!

 

[Phrak]

Any suggestions for these two components? I need to look at ripple current and ripple voltage rating of the capacitor...and I'm not sure if the MOSFET current (ID) should be rated at 50A or if a lower value is acceptable?

A circuit drawing would be helpful, or link to the MCP1630 data sheet if it has an appropriate app schematic.

 

[deen25]

Hey,

Yeah, I should've thought of that before! thanks...

I've attached a circuit design..and the links to the datasheets of both the MCP1630 and the LM2724A...you reckon I follow the other standard capacitances as provided in the datasheet yeah?

MCP1630 - http://www.datasheetcatalog.com/datasheets_pdf/M/C/P/1/MCP1630.shtml

LM2724A - http://www.datasheetcatalog.com/datasheets_pdf/L/M/2/7/LM2724A.shtml

you could also take a look at this datasheet of MCP14628 dual-MOSFET driver which I opted for previously..it has a neat design with the MCP1630...I had to change this microchip due to unavailability.

MCP14628 - http://www.supplyframe.com/datasheet-pdf/component/microchip+technology/MCP14628EMF-datasheet&partId=1974675 cheers!

 

[Phrak]

C3 is an input filter cap, if that's what you mean. The supply may already have an output filter cap. If so, C3 can be used to take up the slack for peak current into the inductor. There are also current spikes at switching-time. For this you might want a low ESR filter capacitor (mica would be a good choice) in parallel with C3. The 10V source could be from something like an LM317 which doesn't like a lot of output capacitance. So the selection is dependent on the 10V source.

The ripple current on the input capacitor is dependent upon switching frequency over inductance.

 

[deen25]

I will not be using the LM317 in this case, most probably I will use a direct voltage supply (no output cap). The switching frequency is 100kHz. My supervisor mentioned that the this would need a coupling capacitor based on the voltage and current ripple...what value of capacitance would you suggest?

also, any suggestions for the logic-level MOSFETs that I could use here?

 

[Phrak]

I will not be using the LM317 in this case, most probably I will use a direct voltage supply (no output cap). The switching frequency is 100kHz. My supervisor mentioned that the this would need a coupling capacitor based on the voltage and current ripple...what value of capacitance would you suggest?

That's confusing. To couple what to what?

I don't see any current sense going on. The feedback looks like it's sensing voltage.

Passing through the inductor is a DC + triangle wave current. You might want to filter the AC component with an output capacitor in parallel with the load, but electrically close to the inductor.

 

[deen25]

I completely forgot!

I will be using a hall-effect transducer to measure the current through the inductor. Then this will have to be used to convert to the voltage reference for the MCP1630 as stated in the diagram I provided.

I need to know how to work out the input capacitance value...could you suggest any methods?

 

[Phrak]

I completely forgot!

I will be using a hall-effect transducer to measure the current through the inductor. Then this will have to be used to convert to the voltage reference for the MCP1630 as stated in the diagram I provided.

I need to know how to work out the input capacitance value...could you suggest any methods?

Interesting. I assume you're speaking about C3, and that it's in parallel with a high frequency cap already. The value depend on the capacity of the voltage source to respond to changes in current (current slew rate) and step changes in the set-point current of the buck converter, if this is a variable current converter. But first you need to specify the inductor inductance.

 

[berkeman]

I completely forgot!

I will be using a hall-effect transducer to measure the current through the inductor. Then this will have to be used to convert to the voltage reference for the MCP1630 as stated in the diagram I provided.

First of all, that is not what is shown in your schematic. As Phrak says, you show voltage feedback from the load, which will not give you anything like a constant current output. You need to re-draw your schematic with the Hall sensor and it's smoothing/compensation circuit, before we can be of much help. And you should be clear about how you are simulating the control loop with the Hall sensor's response Bode plot included in the loop -- what is your phase margin with the Hall sensor in the loop? What does your new compensation circuit look like?

Also, why are you using a Hall sensor on the buck inductor? Is that traditional in higher-power current-mode buck converters, as opposed to the traditional low-side sense resistor below the load? On Semiconductor makes a current mode buck IC with a lower sense voltage to increase effiiciency (it's used in their LED driver circuits).

 

[Phrak]

Redraw your schematic, and let's see what you have.

 

[deen25]

I haven't yet started working on the inductor design or the current sensing circuitry. Right now i need to figure out the initial circuit as provided in the schematic and exhibit the switching of the MOSFETs, so I'm stuck with two things the decoupling capacitor and the MOSFETs. If I complete this task my supervisor will provide information to carry on with the rest of the circuit, until then I have very limited information.

The capacitor should be low ESR suitable for current ripple of 1-1.5A. I checked out one "NTP476M10TRD(100)F" on the niccomp website but I need equivalents of this..

The MOSFET I found is IRLR2905Z...?...i need to stick to logic-level FETs

I'll see if I can work on the full schematic and attach it here soon! I'll need to bug my supervisor for more information now...thanks guys!

 

[deen25]

hey...

I've attached a simple diagram of what the entire cct will look like...all ideas and suggestions are welcome...please give your opinions about the best way to design the cct..

thanks!

 

[berkeman]

hey...

I've attached a simple diagram of what the entire cct will look like...all ideas and suggestions are welcome...please give your opinions about the best way to design the cct..

thanks!

I thought *you* were supposed to be designing the circuit... We don't do your work for you, whether in homework, coursework, or projects.

What is your training in DC-DC converter design? Why were you given this project, if you don't know how to do the design? I've let this thred stay in EE up to now, but that doesn't mean that we will do your work for you.

What are you planning for your unity gain crossover for the feedback loop? What phase margin are you shooting for? What are the characteristics of the load, and what kind of load regulation will your unity gain crossover afford?

 

[Phrak]

Everything designs around the choke. Pick one. What do the current-voltage waveforms look like at steady state, start-up, and under dynamic load?

Edit: nevermind. This has been going on so long, i'd forgotten you gave a value.

 

[deen25]

I'm working day and night on coming up with solutions to this...No..i'm not asking you to design it for me...I just thought you might have better designs from which I can adopt some ideas..

I requested for a hardware project instead of a software/simulation one and this is what I got..I don't have any previous experience with dc-dc converters..this is my first attempt! I've only done an SMPS theory module in the past...so take it a little easy on me guys...it's not like I'm not trying..

I've found some capacitors..i've attached datasheets...its just that I have trouble understanding the characteristic requirements for the circuit...which is what I'm trying to figure out..

another limitation is my components are being ordered from a particular company so I need to have specific reference for the components I want...that's another issue that's holding me back..but I'm sure I can work this out if I find the correct part.

What are you planning for your unity gain crossover for the feedback loop? What phase margin are you shooting for? What are the characteristics of the load, and what kind of load regulation will your unity gain crossover afford?

the load is assumed to be purely resistive..I haven't worked out any of the feedback parameters...like I mentioned before, I'm currently working on the design of the buck cct only..sorry for the lack of info on this part...

 

[berkeman]

Fair enough. Thanks for the additional info to put this project in context -- that definitely helps.

The Panasonic electrolytic caps look reasonable. What is the source of the "10V" input DC voltage? What sort of output impedance and capacitance does it have itself?

The 1A ripple current at 100kHz seems do-able with that series of capacitors. Do you know what range of output voltages you will need to support? That is, if you are supplying 50A as a current source, what is the maximum resistance that you need to be able to drive that 50A through.

Keep in mind that you will want to limit your maximum duty cycle for the Buck to something like 50% or 70% max. That will determine the maximum possible output voltage for a given input voltage.

Quiz Question -- Why is that?

 

[Phrak]

Panasonic defines ripple current as I_RMS. I peak-to-peak, trangular = 2 sqrt(3) I_RMS.

Why the limit on maximum duty, Berkeman? Although the PWM should include zero for open load condition, should it exist.

 

[deen25]

I'm supposed to consider variable loads but the example given to me is attached, it is a set of by-passed diodes in this case, but I shall ask my supervisor more details about the load when I see him next..they are basically devices under test (DUTs)

answer to your question I think is, that the duty cycle allows you to manipulate the output voltage...and the voltage difference across the inductor will determine the rate of change of current... ? I wanted to limit the duty cycle ratio to 0.5..

about the capacitor...in the panasonic datasheet there is a capacitor with part no. EEUFM1A681L: 680uF w/ripple current 1240mA ...would you suggest I use that or take a higher value of ripple current, something like 1500mA ?..just to be on the safe side..

I've also attached a datasheet of a MOSFET...shouldn't a logic-level MOSFET have an on-state Vgs rating of 5V or less at rtp?

 

[Phrak]

Take ripple current? I don't know, you tell me, I took an electronics class at a junior college once.

 

[berkeman]

Why the limit on maximum duty, Berkeman? Although the PWM should include zero for open load condition, should it exist.

For some topology converters, it has to do with transformer saturation:

http://www.edn.com/article/CA6479493.html?spacedesc=DesignIdeas&taxid=10501&industryid=44217

For Buck converters, I think it mainly has to do with stability of the compensation loop. 70% is a fairly typical limit for Buck converters, which defines the minimum input voltage for a given output voltage.

 

[Phrak]

For some topology converters, it has to do with transformer saturation:...

It took me some time to recall the various topologies.

In general, I find this very useful,

$$E_1 \Delta T_1 = L \Delta I_1$$

$$E_2 \Delta T_2 = L \Delta I_2$$

which is just straight from

$$e(t)=L \frac{di(t)}{dt}$$

E_1 and E_2 are the voltages applied across a winding over some cycle of length T_1 + T_2. The currents are magnetization currents.

Ignoring junctions drops, for a 12V supply and 5V output, E_1 = +7V, and E_2 = -5V, for instance, for a buck converter.

Where L is fairly constant--a constant mu, well within saturation, the flux is proportional to the current.

$$\phi = k I$$

$$kE_1 \Delta T_1 = L \Delta \Phi_1$$

$$kE_2 \Delta T_2 = L \Delta \Phi_2$$

(For some transformer topologies, replace $E_1$ with $E_1/N_1$ , and $E_2$ with $E_2/N_2$ where claming is on the secondary.


Add up the two equations and you can see that the flux could start walking away until saturation. This is a fairly general (first order) equation for a variety of topologies, except where the current bottoms-out, in some designs, in which there may be a third phase where the current flow is zero.

If not, then $$\Delta \Phi_1= - \Delta \Phi_2$$.

$$E_1 \Delta T_1 + E_2 \Delta T_2 = 0$$


It also tells why you can have an H bridge driven transformer (and even an AC line transformer) and any deviation from an ideal 50-50 duty (or ideal sinewave), won't cause everything to fail. The nonlinearity in the Phi = Phi(I) keeps the current from walking away toward an infinite DC current.

 

[berkeman]

Nice post, Phrak.

 

[Phrak]

Nice post, Phrak.

Thanks, Berkeman.

But kinda wordy--where are the governing priniples concerning flux? Maybe this will do it.

$$\oint \Phi = 0$$

for steady state over multiple cycles. And maybe, in addition,

$$\Phi = I \left(\frac{A_e N \mu}{l_e}\right)$$

Something stewartcw held my feet to the fire for last week. :uhh:

 

[patsak]

Hi everybody!

I am working on this same kind of project, that's why I am writting it here..(and sorry for my bad English, I am french..)

I want to create a constant current generator too. I will generate 15A.

Actually I already designed a buck converter, but to generate a voltage. In order to do know the values of the external component, I used an excel file which I have created with this webpage: http://powerelectronics.com/mag/606PET25.pdf

But now, I don't know how to calculate the values of external component. I changed a little my excel file but it's not really good.. I send you a picture of my actual schematic, I generate 15A, but I put maybe too high value for the output capacitor and the inductor How could I calculate it?

Actually I will use the same input voltage, and at the output I will have to keep this constant current with a load which could be between 0.1 and 0.5Ohms.

I hope I was enough clear... I really want to know how to calculate the values of the external components by a good way..

Thanks for you help!