Best voltage regulator for batter driven applications ... DC-DC or 'DCM' ?

Go To Last Post
65 posts / 0 new

Pages

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yep! I have been doing just that and finding out exactly as you said that my load must present a resistance of 200 Ohm to the generator - very interesting stuff! Similar to what happens when using a battery.

 

Then if I have a load of say 20 Ohm and want to extract as much energy as possible I will have to use a buck. The buck converter must be set up in a way that presents 200Ohm to the motor or draw equivalent power to a 200 Ohm resistive load. This can be achieved by drawing that much power on the 20 Ohm resistor so, if the generator produces 14V RMS, the max power that can be extracted is 7^2 / 200 = 0.245W. Then the output of the buck should be set at sqrt(0.245W * 20 Ohm) = 4.9V In order to drive the 20 Ohm load at max power. I am only stating all these in order to make sure I understood them correctly.

TO THE FINDER... THE ISLE OF KOHOLINT, IS BUT AN ILLUSION... HUMAN, MONSTER, SEA, SKY... A SCENE ON THE LID OF A SLEEPER'S EYE... AWAKE THE DREAMER, AND KOHOLINT WILL VANISH MUCH LIKE A BUBBLE ON A NEEDLE... CAST-AWAY, YOU SHOULD KNOW THE TRUTH!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

-Deleted-

TO THE FINDER... THE ISLE OF KOHOLINT, IS BUT AN ILLUSION... HUMAN, MONSTER, SEA, SKY... A SCENE ON THE LID OF A SLEEPER'S EYE... AWAKE THE DREAMER, AND KOHOLINT WILL VANISH MUCH LIKE A BUBBLE ON A NEEDLE... CAST-AWAY, YOU SHOULD KNOW THE TRUTH!

Last Edited: Sat. Apr 2, 2022 - 08:53 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

I needed to use that so I can keep the PWM at a low frequency and enable fast simulation times. I

None of that is really the main thing now...you need to find out how you will handle low voltages or low/limited power input when demand is highest (empty cap), as well as how the control circuit will be powered at that moment.  Until then, there isn't much to simulate (that would make its way into the actual design). As you see the simulation itself is quite nice to have, like someone just gave you a new table saw & lathe. 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


What you said was very promising and it made me mess around the past hours. I will keep an eye out for these traps you mentioned. I would like to avoid using a microcontroller. Aiming for another generator seems ideal at this point, yes. A brushed motor could also help me avoid using rectifiers. Could the brush friction losses be smaller than rectification losses?

 

I made a small setup with the generator, two full bridge rectifiers in series, the buck converter and after that the capacitor with a Schottky diode to prevent discharging. When the buck converter's output is set much higher than the capacitor voltage, it seems to be throttling up the PWM to 100% duty cycle, essentially connecting the capacitor straight to the generator. When the capacitor closes in on the output voltage of the buck, the buck presents a big resistance to the motor (open circuit). There seems to be a sweet spot for a specific output voltage differential between the capacitor and buck's output that yields maximum efficiency. I need to find a way to adjust the buck's output voltage relative to the capacitor's charge.

 

 

The generator can supply a maximum of (Voltage genarated)/(Winding resistance) Amps to a 0Ohm load so about 70mA - which is useless because the load experiences 0 dV and gets 0 power. It can also supply a maximum power of (half of genarated Voltage)^2/(winding resistance) Watts or about 0.245mW to a load of resistance equal to the winding resistance (200 Ohm).

 

I need to regulate the buck's output in such a way that will present a 200Ohm load to the generator or, alternatively, I need the Voltage on the buck's input to be equal to half the generated voltage at all times or I can somehow keep providing close to 0.245mW to the load/capacitor.

TO THE FINDER... THE ISLE OF KOHOLINT, IS BUT AN ILLUSION... HUMAN, MONSTER, SEA, SKY... A SCENE ON THE LID OF A SLEEPER'S EYE... AWAKE THE DREAMER, AND KOHOLINT WILL VANISH MUCH LIKE A BUBBLE ON A NEEDLE... CAST-AWAY, YOU SHOULD KNOW THE TRUTH!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

I need to regulate the buck's output in such a way that will present a 200Ohm

Well, that just says Iinput avg is Vin/200 

 

If the output  was a resistor then Iout = Vout/Rout

 

Pin = Pout

 

(Vin)2/200 = (Vout)2/Rout     or simply  set  Vout = Vin*sqrt(Rout/200)  =  0.07071*Vin*sqrt(Rout)

 

however we don't have Rout for a nice fixed ratio between Vout & Iout, due to cap instead (with a cap we'd use an integrator)

 

So you'd need to form a servo (vary Vout until measured Iin equals Vin/200).  This is also simply servoing Vout to maintain a constant power Pin.

You'd have to also limit Vout within the bounds of max allowed Vcap. That make sense, since if the cap is near full (low curent) then only way to keep drawing a fixed input power is for the output voltage to go up & up &up.  

At that point the power draw on the input must come down, as it has nowhere to go.

And you'd have to measure Iin.

 

eh, might not be worth all the trouble.  Matching the generator parameters closely to the cap seems easier.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yep, I should really be looking for other motors and I am.

 

But also, I think I could implement a buck converter like that if I could set an external voltage reference for the error amplifier. Does anyone know if there is a buck converter IC that supports external voltage reference? Or that can be somehow set up to output a specific power?

TO THE FINDER... THE ISLE OF KOHOLINT, IS BUT AN ILLUSION... HUMAN, MONSTER, SEA, SKY... A SCENE ON THE LID OF A SLEEPER'S EYE... AWAKE THE DREAMER, AND KOHOLINT WILL VANISH MUCH LIKE A BUBBLE ON A NEEDLE... CAST-AWAY, YOU SHOULD KNOW THE TRUTH!

Last Edited: Sun. Apr 3, 2022 - 04:15 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Does anyone know if there is a buck converter IC that supports external voltage reference? 

No real need to, you can nudge the voltage fed back to the FB.  however, I think som of the soft start pins ramp up or affect this ref level (to make the ramp up)---you'll have to look at some parts. 

 

Never seen one to maintain a specific power.   ...this idea could be converted to a switcher  https://www.maximintegrated.com/...

 

here is some complicated fun

https://www.edn.com/circuit-deli...

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

tellSlater wrote:
two full bridge rectifiers in series
Really? Surely in parallel.

 

Ross McKenzie, Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

tellSlater wrote:

..But also, I think I could implement a buck converter like that if I could set an external voltage reference for the error amplifier.

Does anyone know if there is a buck converter IC that supports external voltage reference?

You do not need an external ref, but you may need an external amplifier.

 

 

tellSlater wrote:

Or that can be somehow set up to output a specific power?

See the Power harvesting ones I linked earlier.

 

You need two feedback loops : During CAP charge the load-line / power matching loop has control until the CAP reaches the MAX setpoint, then constant voltage control takes over.

You could do the constant voltage with a shunt regulator, maybe even a LED is simple and close enough for that, and it gives visual feedback that CAP max is reached.

  

For load line control, you compare the generator output voltage, with the generator current * 200 ( or whatever resistance you need to match)

if the current is too low, you increase the PWM loading, until you have ==.

If the current is too high, for the voltage, you reduce the PWM loading.

With your low frequency ripple plus wobble from the human operator, getting that loop stable could be a challenge.

Since you work with spice already, I'd try a light cap load post bridge, and see if the PWM control loop can simply follow the high ripple.

PWM can be 20kHz~2MHz  which is well above the ~ 10Hz ripple.

A sawtooth generator, and a couple of simple op-amps and you would be good to go.

 

An alternative would be a MCU that reads the Generator V and I and uses a lookup table to manage the PWM, and that can also limit the CAP max V.

 

 

 

Last Edited: Tue. Apr 5, 2022 - 01:24 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yeah my bad in parallel

TO THE FINDER... THE ISLE OF KOHOLINT, IS BUT AN ILLUSION... HUMAN, MONSTER, SEA, SKY... A SCENE ON THE LID OF A SLEEPER'S EYE... AWAKE THE DREAMER, AND KOHOLINT WILL VANISH MUCH LIKE A BUBBLE ON A NEEDLE... CAST-AWAY, YOU SHOULD KNOW THE TRUTH!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

tellSlater wrote:
in parallel

 

That will not be my choice.

 

Use a step-down regulator adjusted for cap voltage, and connect generators in series.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Simulations I did favored the parallel connection in terms of charging times. Could you tell me why you would prefer the series connection?

 

I did try the step-down regulator solution but it works very poorly with the capacitor as a load.

TO THE FINDER... THE ISLE OF KOHOLINT, IS BUT AN ILLUSION... HUMAN, MONSTER, SEA, SKY... A SCENE ON THE LID OF A SLEEPER'S EYE... AWAKE THE DREAMER, AND KOHOLINT WILL VANISH MUCH LIKE A BUBBLE ON A NEEDLE... CAST-AWAY, YOU SHOULD KNOW THE TRUTH!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I do not know what "poorly" means. What I know for sure is that a supercap needs protection. Do you have any?

Did you simulate cap- how the generator does see it: it can be a very low resistance. Remember our discussion about 200 ohm?

A step down should be here to protect supercap, and translate down impedance, to a reasonable value.

Seems that supercap should be initially charged to some voltage, must not be fully empty.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Due to the common center taps at the generator (12V stepper motor), the bridge rectifiers need to stay in parallel, see details below.

 

Instead of using a voltage source for the simulation, I would recommend using an equivalent current source, where only the frequency needs to be changed for different generator speeds and not the current (which is almost constant above a certain speed) as well.

 

Using a voltage source based simulation, both the frequency and the voltage would need to be changed depending on the generator speed.

 

I wonder where the given 44 mH are coming from?

Nevertheless, assuming the 44 mH are correct, a current source based simulation schematic for the given generator could look like this:
 

Be aware that instead of 14V rms voltages sources I have used 14V/200 Ohm = 70 mA rms current sources and 10 Hz generator frequency. [1]

 

To meet the generator setup, the second current source I2 has a phase delay of 25 ms, which is 1/(4 * 10Hz). Don’t simply use 90° phase shift for I2, as it would fool the simulator.

 

However, due to the two generator currents being added behind the bridge rectifiers (another advantage when using current sources), a phase shift would not affect the resulting voltage at the cap at all. And the simulation result would look like this:

 

As you can see it takes about 255 sec charging 10F to reach the wanted 2.7V.

 

Slightly different L1 and L2 values would not affect V(cap) at the given low 10 Hz generator frequency. L1/L2 could even be omitted for low generator frequencies. Then just the current source and the 200 Ohm resistor in parallel can be used for the simulations.

 

After these simulation basics I would like to recommend a slightly different approach how to finally realize the wanted LED driver at reasonable efficiency.

 

First of all I would go with a higher supply voltage by using to 2.7V supercaps in series. I’m aware that the total capacity would be 5F at twice the capacitor volume when using two 10F caps in series (or a single 5F/5.4V supercap). 

 

However, you would also get twice the energy (0.5 * C * V²) in a 5F cap at twice the voltage compared to 10F at 2.7V – almost within the same charging time (290 sec) if the currents are staying the same!

 

If space is critical, you could use two 5F/2.7V capacitors in series still having the same original energy in the charged capacitors when charging to 5.4V.

 

As the given generator got the center taps connected, you could finally use the following schematic balancing the capacitor voltages to be equal (required if two supercaps in series are not having the same capacitance). 
 

It looks more complicated, but it is only one extra wire connecting the center taps.

 

I am very confident that the higher capacitor voltage will allow to getting a higher average efficiency when driving the 3V LEDs.

For example using an LT3477 in buck-boost mode would result in almost 80 % average efficiency (which is not bad under the given conditions):
 

 

Hint: Don’t simulate the whole circuit including the generator (which would take days).
Just simulate short up/down voltage supply ramps for the LED driver separately.

 

To show the behavior of the recommended circuit, here are the ramps for the LT3477 circuit @120 mA LED current:
 

 

The simulation works down to V(cap) = 0.8V until LT3477 stops working. Not sure whether this will be the case in the real world as well (I didn’t test it). BTW: The average LED current is staying constant until the end, even though the peak-to-peak is more.

 

What still is missing in my above supercap circuits is limiting the supercap voltage.

 

Alternately to the solutions being already discussed, I would use an active dummy LED (see http://www.led-treiber.de/html/d...), which both limits the voltage and also indicates when the voltage limit is reached (to stop generating and wasting additional current into the voltage limiter). 

 

To get the right voltage threshold of 5.4V or 5.5V or whatever (depending on the supercap(s) being used) a small white LED and a yellow or green LED [2]  in series should work, but also a combination of a small 3.9V Z diode and a single red indicator LED might be OK. You need to find a good match.

 

After this detailed analysis a big question is coming up: Is it worth to spend more than 4 minutes charging the supercap by a generator and get only 150 sec (estimated) of 120 mA LED light?

 

To be honest: Instead of a supercap I would use a lithium-ion battery (and an appropriate li-ion charger IC).

Then instead of using an expensive and complex buck-boost LED driver, even a simple and low-cost linear LED driver like the MIC4802 will show a similar average efficiency of 80%, schematics see datasheet or here: http://www.led-treiber.de/html/l...

 

Regarding the average efficiency when using a 1500 mAh li-ion battery and driving a white LED at super constant 500 mA (true DC) for 2.5 hours see test results here: http://www.led-treiber.de/html/l...

 

However, charging such a battery by the given generator (for example from time to time in-between to keep the LEDs on) would be really hard work – using a USB charger instead would be much more relaxed …

 

[1] 70 mA rms ist the short circuit current and 99 mA is the peak value required for LTSpice.

[2] Both LEDs can be 2 mA ones, depending on h_fe of the bipolar transistor being used.
 

Pages