Fast charging of capacitors (with a boost converter)

[T_guttata]

Hi

I'm experimenting with the charging of capacitors. In particular, I would like to charge a 1000 uF cap as fast as possible to 30V from an approx. 5V voltage source.

I have tried two boost converter circuits. I used my Peaktech P6226 power supply and measured the charging time with my Rigol Scope.

Results, see below.

I have a few questions:

1) The two boost converter circuits seem to perform similar @ Vin>10V. With Vin=5V charging time differs by a factor of 2.5. Any explanation on that?
2) Fastest charging time with the boost converter circuits is around 6ms (@ 25V input voltage). If I charge the cap without the boost converter circuit but directly with my lab power supply, it takes much longer. Can anyone explain to me the strange charging curve of my lab power supply? In theorey, a DC power supply delivering 10A should be able to charge the cap in 3ms (0.001F*30V/10A = 3ms)? Due to the large capacitors of the power supply possibly faster?
3) Any suggestions on how to reduce the charging time to <5ms with a boost converter circuit?

Thanks for your help.


MT3608:https://de.aliexpress.com/item/4001066566291.html?spm=a2g0s.9042311.0.0.27424c4d0jzQ58

Vin = 5V
Vout = 28V
t= 41ms

Vin = 10V
Vout = 28V
t= 15ms

Vin = 15V
Vout = 28V
t= 9.5ms

Vin = 20V
Vout = 28V
t= 7.8ms

Vin = 25V
Vout = 28V
t= 5.6ms

XL 6009 https://de.aliexpress.com/item/1605337399.html?spm=a2g0s.9042311.0.0.27424c4dDj0ruM

Vin =5V
Vout =30V
t= 100ms

Vin =10V
Vout =30V
t= 17ms

Vin =15V
Vout =30V
t= 9ms

Vin =20V
Vout =30V
t= 8ms

Vin =25V
Vout =30V
t= 6ms

Charged with lab power supply Peaktech P6226:

" alt="" class="bbc_img" />

[coromonadalix]

A discharged capacitor may or can act as a short for a very brief time,  some psu's will kick in current limit and switch in constant voltage  and vice versa, or get into current limit protection

It can depends of their power output in some cases

You can't instantly charge a capacitor ...  try to charge an 1 farad  loll  tried recently a 3.3 farad at 2.7 volts,  took a    while to get it charged.

As i see you time chart, the more you get to the working voltage, less time it take ...    nice experiment


just need the 30volts / vs the 30 volts of the capacitor working / usage

[amyk]

ESR will be the ultimate limit of how fast you can charge a cap.

This dissipates power in the cap itself, so beware if you are trying to do this continuously.

[Manul]

1000uF cap up to 30V in less then 5ms needs more then 6A of average current. And if the source is a boost converter working from 5V source, then it begins to be really problematic. Lets say, the output is at 25V (in the end of charge) and current is 6A, then the input current from 5V source would need to be >30A. That is way beyond the capabilities of any small converters. I mean, this is in a region of 200W peak power. Can be done of course, but not with these converters from aliexpress.

[T_guttata]

ESR will be the ultimate limit of how fast you can charge a cap.

This dissipates power in the cap itself, so beware if you are trying to do this continuously.

1000uF cap up to 30V in less then 5ms needs more then 6A of average current. And if the source is a boost converter working from 5V source, then it begins to be really problematic. Lets say, the output is at 25V (in the end of charge) and current is 6A, then the input current from 5V source would need to be >30A. That is way beyond the capabilities of any small converters. I mean, this is in a region of 200W peak power. Can be done of course, but not with these converters from aliexpress.

Yes, I'm aware that a short charging time will result in high peak current. I don't intend to do that continusly. Worst case might be a 2s period.
Therefore, I would like to optimized a boost converter circuit specifically for short high current transients.

I measured the charging time of the 1000 uF cap when using 2x4700 uF caps as a source. When the empty 1000 uF cap is connected (parallel) to the 2x4700 uF caps charged to 30V, the 1000 uF cap will be charged to 27V.

I have used various resistors to see the impact on the charging time:

100 Ohm --> 140ms
10 Ohm --> 35 ms
1 Ohm --> resistor died
0 Ohm (only bread board resistances) --> 0.4 ms

Any suggestions on better boost converters? Do you know an IC which is designed for high transient loads rather than continuous load?

[Siwastaja]

Using a voltage source and limiting current with a resistor causes two things far from optimal:
* Efficiency is limited to 50%. 50% of energy is lost in the resistors - your explicit resistor and the ESR of the capacitor.
* Charging current tapers off, you have highest current in the beginning. Last few percent takes the most time.

Build a current source. I.e., the boost converter must measure and regulate current instead of voltage. Voltage will be just a stopping (end-of-charge) limit. This way
* Efficiency theoretical maximum becomes 100%. Energy is of course still lost in the ESR of the capacitor, and in the resistances of the boost converter itself
* Charging current can be a constant value until stopped, or it can be a CC-CV curve, i.e. constant at first then taper off near the end; in any case with the same peak current, you get much shorter charging time because you can maintain this peak current.


Capacitors have specified maximum current, though. With a current source, it's trivial to ensure it's not exceeded.

Too much current causes too much heating internal to the capacitor.

Often capacitor datasheets specify ripple current for example at 120Hz or 100kHz. This means continuous, repetitive charging and discharging at that frequency, day in day out, resulting in millions of cycles quickly. Obviously if you only charge it a handful of times, the current the capacitor can take undamaged will be much higher than the ripple current rating. But sadly this value is seldom available.

Usually electrolytic capacitors are designed to be fine when connected to an unlimited-current voltage source below the maximum rated voltage of the cap, for some sane number of cycles (hunderds; not millions).

Now how to build a current source? Actually, all proper boost ICs already have some sort of current limit. See the values listed in datasheets. They are not accurate limiters, but it's better than nothing. Make sure the IC doesn't enter some fault/hickup mode with large capacitances, though. It can be quite a chore to find a DC/DC IC which specified unlimited operation against the current limit.

[T_guttata]

Thanks for the explanation. Efficiency is not that important, but I have to admit I did not know about this 50% limit.

So far i struggled to find a suitable IC. There are just two many and I don't have the specific knowledge to see the pros and cons.

When I set Vin_min = 3V, Vin_max = 10V, Vout = 30V and Iout = 10A, I get 25 results from Texas instruments:

https://www.ti.com/power-management/non-isolated-dc-dc-switching-regulators/step-up-boost/products.html#p238min=0.3;3&p238max=10;100&p634min=-70;30&p634max=30;500&p212max=90;100&p1241=Controller

That's when I decided to try the circuits from aliexpress^^

[Siwastaja]

10A puts you in the territory of DC/DC controllers + external switches. This again makes searching a bit different, because a controller may not have output current listed because that depends on the external parts. Of course the gate drive capability sets a limit of how big MOSFETs you can use, but this is quite indirect, you can't say that a controller IC X only goes up to 10A output.

10A is also in the territory of synchronous converter; or maybe not, it depends on the capacitor voltage. If you are really charging up to 30V, then a 0.6V drop in the boost diode isn't necessarily an issue at all. Charging a single cell ultracap it would be a disaster for efficiency and thermal design.

Note that in the boost converter, the input voltage "flows through" to the output non-limited through the diode (or the body diode in the synchronous converter) - i.e., controlled range starts from the input voltage -, which prevents proper current limiting in the beginning of charge. You need another MOSFET, or a different topology to actively control current from 0V output. Sometimes this isn't a problem as the resistance-limited current at 10V is obviously only one third of the resistance-limited current at 30V.