how to design a buck converter(step down converter) with variable output voltage

[rakesh]

Am planning to design a buck converter to charge some 3000F ultracapacitor at 40A. This is the datasheet for the ultracapacitor. http://www.maxwell.com/products/ultracapacitors/docs/datasheet_k2_series_1015370.pdf . I got 30 of them which will result in a equivalent capacitance of 100F at 75V with a total esr of 8.7mohms(which is very low) . Am planning to use it in a electric car application. These ultracapacitor when charged using a battery or some other source it will act as a short circuit across its charging source. so to charge these capacitors I should develop a buck converter circuit in which its output voltage will vary from 1 to 75V. The reason for variable output voltage is that when the capacitor voltage is at 0V i would charge this capacitor at 1V.

R=V/I ; 1/40 =0.025

To have 40A of charging current the equivalent series resistance of the capacitor must be 0.025ohm . hence i will add a resistor of (0.025-0.0087=0.0163) 16.3mohm in series to the capacitor to allow only a maximum current flow of 40A.

My confusion why there are not many designs of buck converter with variable output voltage, what are the drawbacks in designing such a variable output voltage buck converter.Can it be designed??. I just wanted this charging circuit to be as efficient as possible with very little power stage during transfer cost of the circuit is not a concern since the circuit is designed mostly for very high effiiciency.

Kindly suggest me if there are other ways to charge in a more efficient way

thank you in advance.

[madires]

Since there are a lot of lab SMPSUs with variable output voltage I assume it's possible to design one  But I would be concerned about the overloading, i.e. the buck converter has to limit the maximum current being switched.

[rakesh]

Since there are a lot of lab SMPSUs with variable output voltage I assume it's possible to design one  But I would be concerned about the overloading, i.e. the buck converter has to limit the maximum current being switched.

if the smps has the capability to supply 40A on all voltages from 1 to 75V , ?I assume there wont be any overloading since the ESR of the ultracapacitor is a fixed one which is 0.025ohm.  kindly correct me if am wrong.

[madires]

if the smps has the capability to supply 40A on all voltages from 1 to 75V , ?I assume there wont be any overloading since the ESR of the ultracapacitor is a fixed one which is 0.025ohm.  kindly correct me if am wrong.

That's true for the ideal case. But you should also plan for the case that something isn't ideal or even goes wrong. 40A and ultracaps aren't toys.

[rakesh]

Am planning on adding a cutoff circuit if the current increases above 50A. could someone tell me what are the real difficulties with variable output voltage buck converters and why most buck converters are designed for  fixed output voltage.

[c4757p]

Depending on the control loop design, they can be (but don't have to be, if you can sacrifice ultra-tight regulation) hard to stabilize over a wide output voltage range.

My suggestion is to treat this as a buck-mode current source with overvoltage protection, rather than a buck-mode voltage source with overcurrent protection. It will spend most of its active life in CC mode. Amplify the current sense voltage to get the controller feedback, and resistor-OR that with a Zener diode to the output for a voltage limit. Pay attention to frequency response in the amplifier; try to make it similar to a traditional resistive feedback.

And yes, remember that 40A is *not* a toy.

[Paul Price]

Buck converters seem to be designed for constant voltage output because that is the most common use of a power supply.

By having a current sense resistor in the return wire to the switcher it is quite easy to make the constant voltage power supply a constant voltage power supply just waiting to be switched over to constant current mode.

By slowly slewing the output voltage upwards from 0 in such a constant-current supply the output current will switch into constant current mode and track the charging voltage level of the supercapacitors.

All this really takes are two comparator in the feedback loop of the switching regulator.

The first comparator sets the output voltage according to a reference voltage which is the desired output voltage by manual or computer D2A control.

A second comparator connected in the feedback pin  loop restricts the output voltage from rising above a value that causes more than the desired current to flow. The desired current to flow is  feedback from a (preferably fast slew rate op-amp) current sensing amp whose output is directly proportional to the instantaneous current. This C. Current op-amp output is fed the second comparator with its other input set to a voltage corresponding to the current you desire. The two comparator outputs  are then OR'd together (using open-collector output comparators) that feed their tied output into a third inverter (or a transistor) to feed a digital on/off signal to the feedback input of the switching regulator (which always requires a voltage greater than it's internal reference to terminate the output driving the inductor in the switcher and thus both limit current/volltage). How much higher this logic level is not so important, but always within the specs of the switcher IC to be greater than the internal reference level.

It is quite practical and simpler to use a fast MCU to do all the logic of monitoring and setting and displaying the currents and voltages, completely controlling the switching of a high power MOSFET switch  feeding the catchdiode/inductor and not use a switching regulator IC at all. The MCU can use PWM to set both output voltage and current and use fast A2D to get the instantaneous voltage and current at the output of a current sense op-amp.

[Phaedrus]

It's trivial to make a buck converter act in constant current mode. It's a little more tricky to make it efficient. Most PWM ICs for buck regulators take a feedback voltage of ~0.6-2.0V. Most I see are ~1.2V. This is a convenient voltage for a voltage regulator, and all you need is a 1-10kohm range voltage divider for your feedback loop. When you go to CC mode that voltage divider is replaced with a sense resistor, usually in the 100R range. This wastes an order of magnitude more power than the voltage divider used in the voltage regulator.

I ran into this issue when I was trying to design a driver for a 30W LED flashlight. I used TI's simulation tools to make a voltage regulator that used a TPS53319, and found a small enough inductor. As a voltage regulator it was 96% efficient at anticipated load and fit on a circular PCB 32mm across.

However when I converted it to a current regulator, the current sense resistor was dissipating almost 4W! Because the feedback voltage was 1.2V. I found a different controller with a feedback voltage of 0.6V, but it was still too high. I finally found one with a feedback voltage of 0.2V--but then I needed to relayout, and having an external mosfet and a different inductor made the whole thing untenable.



BUT. If you don't care about efficiency, making a buck regulator current source is easy.

[c4757p]

A cheap op amp with input sensitivity near the negative rail can easily amplify a small sense voltage up to the feedback voltage, though.

[Paul Price]

Phaadrus, there is no need for global warming. It is only necessary to amplify the voltage drop across a very small valued resistor (.005 ohm for instance) and then the power wasted and voltage drop are very very small.

C4757p, Although a cheap slow slew rate up amp will not work or cause the circuit to oscillate, it is important to remember that when working with supercaps, the idea to be a 1st responder to an emergency situation is paramount, use a cheap but fast op-amp to amplify the current sense and get the report of a small fire before the building burns down.

[c4757p]

You do need fast response for it to regulate well - don't go digging too deep in the bottom of the op amp barrel for anything in a 40A power supply! - but as for not burning the building down, how about a fuse? Much more reliable than god knows what errors have been accidentally introduced into a one-off circuit that hasn't undergone rigorous testing. And I'd add thermal overload protection, as well as a series diode (yes, you will incur additional losses for this) to prevent the capacitors from blowing the ass out of the MOSFET's body diode if the input is shorted.

[rakesh]

hi paul price,

thank you for your reply. could you explain more about current mode pwm control on why it uses two feedback both ouput voltage and output current while current is the parameter which we are really concerned to regulate. could you please refer me to some literature on this current mode pwm control.  I have worked with voltage mode control before but in this current mode control all the inner feedbacks and clock signals inside those pwm controller confuses me a lot. 

i would love to do it without any ic's by implementing each and every block on my own but i would like to know more about the control technique first to have a better idea of what i have to do next.

[c4757p]

"Current mode" control doesn't mean a power supply that acts as a current source. Most of them are still configured as voltage sources. The current being sensed is the switching current through the transistor. What it means is that instead of the feedback voltage directly selecting a PWM duty cycle, the feedback voltage selects a peak current threshold, and the transistor is switched such that the current peaks at that level. Instead of controlling the duty cycle, you're controlling the maximum switch current (and the duty cycle is just incidental).

[rakesh]

"Current mode" control doesn't mean a power supply that acts as a current source. Most of them are still configured as voltage sources. The current being sensed is the switching current through the transistor. What it means is that instead of the feedback voltage directly selecting a PWM duty cycle, the feedback voltage selects a peak current threshold, and the transistor is switched such that the current peaks at that level. Instead of controlling the duty cycle, you're controlling the maximum switch current (and the duty cycle is just incidental).

hi c4757p,
thanks for your reply. could you tell me why voltage is also taken as a feedback in current mode control .

[mariush]

I wonder if it's really worth designing such a power supply considering the high currents involved.

There are really cheap power supplies out there like the meanwell 24v and 36v ones on ebay, which can do 10-15A for 40-45$ a piece:

24v 15a: http://www.ebay.com/itm/Mean-Well-MW-24V-14-6A-350W-AC-DC-Switching-Power-Supply-NES-350-24-UL-Brand-New-/111120732301?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item19df4fd08d
36v - 10a http://www.ebay.com/itm/Mean-Well-MW-36V-9-7A-350W-AC-DC-Switching-Power-Supply-NES-350-36-UL-Brand-New-/111121396190?pt=LH_DefaultDomain_0&hash=item19df59f1de

These are also adjustable (for ex. the 36v is adjustable between 32-40v) .. so you could maybe split the capacitors in two banks and have tho of these 36v power supplies charge the two banks simultaneously?

Alternatively, maybe spend about 80$ on a good atx power supply that can do 12v 50-60a easily, like this one: http://www.newegg.com/Product/Product.aspx?Item=N82E16817374032 (648w on 12v, 55 amps on 4 rails)
Then use some relays to separate  the capacitors into segments of 4-5 pcs (12v/5pcs = 2.4v per cap) and charge all segments at same time with 12v
A small microcontroller should be able to monitor the charge on each set of capacitors and switch relays on and off to link together the capacitors

[c4757p]

hi c4757p,
thanks for your reply. could you tell me why voltage is also taken as a feedback in current mode control .

It's the main feedback! As I said, the feedback voltage selects the switch current (not output current) threshold and varies it to regulate the output. Typically, you want a regulated voltage output, and the switching current doesn't directly correlate with the output (at least, not in a real-life converter, though you can definitely find equations for the ideal case), so you can't just set it to a fixed value.

For a constant current supply, the output voltage will not be considered at all until it hits a limit (in the case I described above with a Zener diode, by pushing the diode into reverse breakdown and then being connected into the feedback path by the diode). Instead, you'll use the amplified output from the current sensor as the feedback signal. You still have to sense both the output quantity (voltage or current) and the switching current.

I'd recommend not using current-mode control here - it's already harder to stabilize because of the dual feedback, and CC switch-mode supplies are also harder to stabilize. Use voltage-mode control configured in constant-current output mode. If you're still confused, remember that the word "voltage" in voltage-mode control just means that the feedback is being compared against a voltage (the PWM ramp) rather than a current (the switching current).

Or, what mariush said.

Apologies for the rampant overuse of italics.

[Paul Price]

Rakesh asks, "could you explain more about current mode pwm control on why it uses two feedback both ouput voltage and output current while current is the parameter which we are really concerned to regulate. could you please refer me to some literature on this current mode pwm control.  I have worked with voltage mode control before but in this current mode control all the inner feedbacks and clock signals inside those pwm controller confuses me a lot."

The answer is simple, ohm's law. For any current to flow though a circuit, the voltage must rise high enough to cause that amount of current to flow.  For instance, if you have a .025 ohm resistor and you wish to achieve a 10-amp current to flow through that resistance, then you must apply a voltage of V=(IR) or 10 x .025, you need .25V to do this.  To get a switching or linear power supply to cause the same constant current to flow into a supercapacitor with an internal resistance of .025 ohm, the power supply output voltage must be .25V higher than the present instantaneous voltage of the supercaps.

In other words, a constant current is achieved and regulated by regulating the applied voltage to the correct value to get the load to sink the desired current, all according to ohm's law.

If a constant current power supply should be applied to the load circuit, it is best to begin with letting the voltage rise to the correct voltage, rather than fall down to it with a surge to the required level to achieve the correct current flow.. this way no surge is produced and no parts are damaged.

It is also important to have the load connected to a constant current supply before turning on the power supply.

In a switching or linear power supply, any current causes a voltage to be developed across a sense resistor that is measuring the current through the external circuit. That current is converted to a voltage by an amplifying circuit, such as an op-amp. This voltage (directly proportional to the current) is then used to control the output voltage of the supply to limit the current when the current reaches the constant-current set value.  In essence, a constant current supply switches modes at this point from constant voltage output to constant current mode output, but always doing this by voltage control.
--------------------------------------
Switch Mode Poser Supply IC's have internal clocks and oscillators and flip-flops because it is necessary find a way to allow the current forced into the inductor in a switching power supply deliver its stored energy before it is asked to store more energy.

[nctnico]

Not quite. Even with fixed timing an inductor may saturate. The current in the inductor must be allowed to reach zero but this can't be guaranteed by design. So every (modern) switched mode controller has an overcurrent protection.

[C]

As per maxwell
http://www.maxwell.com/products/ultracapacitors/docs/200904_whitepaper_designinguide.pdf

8. Series connection and balancing
Since the individual ultracapacitor cell voltage is relatively limited compared to the
majority of application requirements, it is necessary to series connect the ultracapacitors to achieve the voltage required. Because each ultracapacitor will have a slight tolerance in capacitance and resistance it is necessary to balance, or prevent, individual ultracapacitors from exceeding its rated voltage.

Both charge and discharge can make the voltage different on the stacked ultracapacitors.

Would be a good idea to read what maxwell recommends first.

C

[rakesh]

Rakesh asks, "could you explain more about current mode pwm control on why it uses two feedback both ouput voltage and output current while current is the parameter which we are really concerned to regulate. could you please refer me to some literature on this current mode pwm control.  I have worked with voltage mode control before but in this current mode control all the inner feedbacks and clock signals inside those pwm controller confuses me a lot."

The answer is simple, ohm's law. For any current to flow though a circuit, the voltage must rise high enough to cause that amount of current to flow.  For instance, if you have a .025 ohm resistor and you wish to achieve a 10-amp current to flow through that resistance, then you must apply a voltage of V=(IR) or 10 x .025, you need .25V to do this.  To get a switching or linear power supply to cause the same constant current to flow into a supercapacitor with an internal resistance of .025 ohm, the power supply output voltage must be .25V higher than the present instantaneous voltage of the supercaps.

In other words, a constant current is achieved and regulated by regulating the applied voltage to the correct value to get the load to sink the desired current, all according to ohm's law.

It is quite usual to create a constant-current power supply whose output voltage compliance will rise to the limit of the power supply's max output voltage with no load or with a load applied that is not small enough in resistance to cause the constant-current setting to be achieved..again, ohm's law.

However, it is not always desirable for a constant current power supply to rise to a voltage that must immediately cause a high current instantaneous surge into the desired load when it is connected if the actual required voltage to achieve the desired constant current must settle down to be a much smaller voltage than the maximum voltage compliance of the constant current/constant voltage power supply. After all, this upon-connecting current and voltage surge could be damaging to the circuit as could any voltage/current surge.

 As soon as the load is connected to a constant current supply without voltage regulation, the voltage must drop  to achieve the constant current level desired, again according to ohm's law, but their always some filter capacitor at the output of a constant voltage/constant current power supply that can store a lot of energy. so the current will exceed the desired set point until the voltage finally settles down to the correct value to achieve the correct current.

If a constant current power supply should be applied to the load circuit, it is best to begin with letting the voltage rise to the correct voltage, rather than fall down to it with a surge to the required level to achieve the correct current flow.. this way no surge is produced and no parts are damaged. In other words, when using a constant-current supply it is best to (regulate) control the voltage to achieve the current, always adj'ing the current by adjusting the voltage that forces the required amount to flow.

It is also important to have the load connected to a constant current supply before, perhaps even slowly increasing the current to the required setting, which is achieved by allowing the output voltage of the power supply to rise to the point of achieving the required current flow.

In a switching or linear power supply, any current causes a voltage to be developed across a sense resistor that is measuring the current through the external circuit. That current is converted to a voltage by an amplifying circuit, such as an op-amp. This voltage (directly proportional to the current) is then used to control the output voltage of the supply to limit the current when the current reaches the constant-current set value.  In essence, a constant current supply switches modes at this point from constant voltage output to constant current mode output, but always doing this by voltage control.
--------------------------------------
Switch Mode Poser Supply IC's have internal clocks and oscillators and flip-flops because it is necessary find a way to allow the current forced into the inductor in a switching power supply deliver its stored energy before it is asked to store more energy by the high-power switching transistor that attempts to force current into the inductor. Without this clocking, which usually sets a fixed frequency of operation, the current will otherwise cause the inductor magnetic field to slowly ramp up and up until the inductor saturates and at this point the inductor becomes a short circuit and soon there is smoke and people screaming in the streets. By using clocks and flip-flops, the maximum on-time/off time can be tightly controlled so that only one kick to the inductor occurs during each clock cycle and in this way the inductor is not overloaded because it has sufficient time to deliver its stored energy to the load so it is magnetically ready to handle the next voltage kick.

hi paulprice,

thank you for your reply.  It helped me a lot to clear some of my little confusion about the control strategy in a constant current mode buck converter. Could you refer me some links to understand more about the internal blocks in a current mode pwm control.

[rakesh]

I wonder if it's really worth designing such a power supply considering the high currents involved.

There are really cheap power supplies out there like the meanwell 24v and 36v ones on ebay, which can do 10-15A for 40-45$ a piece:

24v 15a: http://www.ebay.com/itm/Mean-Well-MW-24V-14-6A-350W-AC-DC-Switching-Power-Supply-NES-350-24-UL-Brand-New-/111120732301?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item19df4fd08d
36v - 10a http://www.ebay.com/itm/Mean-Well-MW-36V-9-7A-350W-AC-DC-Switching-Power-Supply-NES-350-36-UL-Brand-New-/111121396190?pt=LH_DefaultDomain_0&hash=item19df59f1de

These are also adjustable (for ex. the 36v is adjustable between 32-40v) .. so you could maybe split the capacitors in two banks and have tho of these 36v power supplies charge the two banks simultaneously?

Alternatively, maybe spend about 80$ on a good atx power supply that can do 12v 50-60a easily, like this one: http://www.newegg.com/Product/Product.aspx?Item=N82E16817374032 (648w on 12v, 55 amps on 4 rails)
Then use some relays to separate  the capacitors into segments of 4-5 pcs (12v/5pcs = 2.4v per cap) and charge all segments at same time with 12v
A small microcontroller should be able to monitor the charge on each set of capacitors and switch relays on and off to link together the capacitors

hi mariush,

I want to charge this from a 85V 150Ah battery source. I think these power supplies won't be helpful to charge these ultracapacitors in an electric car in which battery is the only available DC source.

[c4757p]

As per maxwell...

Dear god, I almost had a heart attack! I thought you were about to drag Maxwell's equations into this somehow...

[C]

Dear god, I almost had a heart attack! I thought you were about to drag Maxwell's equations into this somehow...

Sorry c4757 guess I should have added Company on that.

There are a lot of good chips built to work with ultracapacitors that would be better to start a design from.
A  ultracapacitors bank is not a simple thing unless you want Smoke, Bangs & Fire.
Could have a 777 like problem.

C
 

[mariush]

hi mariush,

I want to charge this from a 85V 150Ah battery source. I think these power supplies won't be helpful to charge these ultracapacitors in an electric car in which battery is the only available DC source.

Most modern power supplies have active PFC which means they'll accept a wide range of input voltages (usually 100-250v, you can see that on the label of that in win green atx psu) but most will actually work even with 75-80v, though they probably won't be able to output as much current.
 
So you could just pull out the bridge rectifier from the power supply pcb and put 85v DC directly there, the psu also won't mind if the battery suddenly gives more than 85v... the active pfc circuitry will then boost that to 350-380v dc the power supply uses internally which then gets downconverted to 12v.