# Series Capacitor Bank

#### JustusS

Joined Jul 23, 2019
13
Hey all, looking to find the most effecient way to charge and regulate capacitors in series.

I made a video to describe what I'm doing, hopefully it makes sense. Attached is a schematic. Thanks.

Last edited:

#### Audioguru

Joined Dec 20, 2007
11,248
Why are you bothering to charge 50V capacitors to only 3V and what will you do with the tiny amount of charge?

#### JustusS

Joined Jul 23, 2019
13
Why are you bothering to charge 50V capacitors to only 3V and what will you do with the tiny amount of charge?
The caps are just what I had, and only trying to test a concept. Disregard the cap voltage rating and charge capacity for the moment. What I'm after is effeciently bypassing series caps that have reached a certain target voltage, as well as info on why the caps at the end of the string seems to hold greater V than the caps at beginning of the string.

#### JustusS

Joined Jul 23, 2019
13
Also, how much power is dissipated in charging a cap? Is it true only half of the supplied power is stored?

#### Audioguru

Joined Dec 20, 2007
11,248
The symbol for your battery has its polarity backwards. The long bar is the (+) terminal.
It looks like the current increases as each capacitor is bypassed because you have a resistor instead of a constant current source so of course the last LED is brighter.

#### MisterBill2

Joined Jan 23, 2018
14,661
To charge each capacitor in a series string to a similar voltage the very simple way is to put similar resistors across each one. Not at all efficient, but sort of effective. It is routinely used in high voltage power supplies. Substituting a non-linear device like a zener diode can hold the charge to some value. Or use a gas-discharge tube to hold a specific voltage. Beyond that it gets fairly complex, similar to the mechanisms used to keep equal voltages in electric car battery packs. Researching those schemes will show how it is done.

#### JustusS

Joined Jul 23, 2019
13
The symbol for your battery has its polarity backwards. The long bar is the (+) terminal.
It looks like the current increases as each capacitor is bypassed because you have a resistor instead of a constant current source so of course the last LED is brighter.
The voltage source stays constant, and the resistor stays constant, so wouldn't the current through R be constant? In series, how does the current increase? Shouldn't it be the same through all series components?

#### JustusS

Joined Jul 23, 2019
13
To charge each capacitor in a series string to a similar voltage the very simple way is to put similar resistors across each one. Not at all efficient, but sort of effective. It is routinely used in high voltage power supplies. Substituting a non-linear device like a zener diode can hold the charge to some value. Or use a gas-discharge tube to hold a specific voltage. Beyond that it gets fairly complex, similar to the mechanisms used to keep equal voltages in electric car battery packs. Researching those schemes will show how it is done.
I tried zener's but they just avalanched at their rated voltage and seemed to be turned on constantly (so cap bled to 0), whereas the LEDs wouldn't turn on until their Vf was met (cap bled to 2.8V-ish fast and then slowly to 0, maybe over a day).
I'm trying to keep the charge as long as possible, so R's and Zener's would be less than ideal for my application but I don't know much about gas discharge tubes so I'll take a peek at that as well as the electric car battery pack schemes. Thanks!

#### JustusS

Joined Jul 23, 2019
13
The voltage source stays constant, and the resistor stays constant, so wouldn't the current through R be constant? In series, how does the current increase? Shouldn't it be the same through all series components?
Edit
Or is the V changing across R as the empty caps are charged? Actually decreasing current over time? Are you saying the excess that is bypassed is added up with each consecutive cap, so that the last cap gets more bypass current to deal with than the other caps had to deal with/bypass? I'm reaching here lol

I thought it might have something to do with the MOSFETs not being all the way on, so that a bit gets bypassed but also a bit more gets pushed onto the cap than anticipated due to MOSFET not completely on, and that each consecutive cap was dealing with a little bit more somehow due to somethimg like a % of a % kind of phenomenon or something.. If that makes any sense..

Last edited:

#### Sensacell

Joined Jun 19, 2012
3,180
Your parallel caps are 141uf- the current consumed by your voltmeter and circuit leakage discharges them rapidly while you measure.

Try much larger caps to prevent insanity in measurement.
The voltages you are reading vary less than 300 mV, the part-to-part variation of Vgs and the LED drops are to blame- what are you expecting to see?

More importantly, what are you REALLY trying to accomplish?

#### JustusS

Joined Jul 23, 2019
13
Your parallel caps are 141uf- the current consumed by your voltmeter and circuit leakage discharges them rapidly while you measure.

Try much larger caps to prevent insanity in measurement.
The voltages you are reading vary less than 300 mV, the part-to-part variation of Vgs and the LED drops are to blame- what are you expecting to see?

More importantly, what are you REALLY trying to accomplish?
I anticipated seeing equal V on each cap. I thought it odd the uneven distribution. I switched LEDs around and was still seeing same V distribution (increasing). I guess I could try switching MOSFETs around too to see if the distribution seems to follow a particular component or not. Perhaps it is just a variation from component to component, but do you know that for sure? I'm just trying to double check with some people to make sure that there isn't some sort of anticipated uneven V distribution in the cct design except for minute component to component variation.

I'm trying to understand this circuit so I can better understand how to design other circuits in the future. I'm curious how to regulate and hold cap charge effeciently, how to bypass effeciently, and really just how capacitors operate in different scenarios so I can increase my overall understanding.

Can you explain why when I remove D2 that C1 only holds charge to ~1.5V instead of ~3.0V?
If Vgs(th) = 1.0V(min) - 2.0V(max), and Vf of LED is ~1.5V, then technically, D1 to M1 gate to M1 source back to C1 negative side (and beyond, when button is pushed) is actually a bypass of D1 to D2 to C1 neg side [only 2.5V drop to get through, (~1.5 + 1.0V) vs (~1.5V·2)], because D1 & D2 are ~1.5Vf, actually I think closer to ~1.3Vf (but I could test with a single high Farad cap), but even then it's still a higher V drop to get through. 1.0 Vgs(th) is best case scenario according to data sheet but why then is C1 only hold ~1.5V or ~Vf of only 1 LED without the addition of the Vgs drop? Is MOSFET on a little bit even before documented min Vgs(th)? That the tiny amount of MOSFET conductance below documented Vgs(th) (possibly?) was enough to drain the tiny amount of charge in C1?

If current is being bypassed mostly through MOSFET, then first regulation LED should be brighter?? 2nd LED is in parallel with gate and gate turns on earlier than LED (1V - 2V vs 1.3Vf)?? Or total flowed current from first LED is divided when it gets to second LED + gate junction, and that's why I can anticipate uneven brightnes? Or should I anticipate uneven brightness at all?

I calculate an equivalent capacitance of 47μF, is that correct? Is each 3 parallel cap bank only holding 47/3μF, or 15.67μF?

Last edited:

#### JustusS

Joined Jul 23, 2019
13
@Sensacell, I tried using 2200μF caps and noticed a more even initial distribution (+/- 50mV). Thanks.

Thanks to everyone for all the responses.

Just one more Q to add to the list above lol..
I also noticed the first cap seems to bleed off quicker than the last 2 caps? Can I chalk this up to minute variations in component manufacturing as well?

#### Sensacell

Joined Jun 19, 2012
3,180
Voltages equal? That's like saying "pay me a lot"

Without some numbers, it's meaningless.

If you short D2, (is that what you mean? better to draw than to write) the Vf of the LED clamps the whole section to 1.5V.

#### JustusS

Joined Jul 23, 2019
13
I calculate an equivalent capacitance of 47μF, is that correct? Is each 3 parallel cap bank only holding 47/3μF, or 15.67μF?
I'm going to try to increase regulation to 4.5V (18V source instead of 9V), or 3 LEDs per cap, but in an attempt to increase gate V of MOSFET so it can turn on harder. If I'm creating a low enough R pathway with MOSFET S to D, then I should be shorting the LEDs? Or some sort of oscillation maybe between MOSFET short and LED bypass?

Last edited:

#### bertus

Joined Apr 5, 2008
22,138

#### MisterBill2

Joined Jan 23, 2018
14,661
One more consideration is that as the voltage across each cap increases, the rising gate bias tends to decrease the effective resistance of the FET. That is functionally a negative resistance, and so it is possible that the circuit is unstable. (How this post returned after my tab crashed yesterday is unclear.)

#### JustusS

Joined Jul 23, 2019
13
One more consideration is that as the voltage across each cap increases, the rising gate bias tends to decrease the effective resistance of the FET. That is functionally a negative resistance, and so it is possible that the circuit is unstable. (How this post returned after my tab crashed yesterday is unclear.)
I wonder if I had my Zener's flipped. I'm going to try them again. Also looking into transistors.

Last edited:

#### JustusS

Joined Jul 23, 2019
13
Why can't you PWM a full bridge with a power MOSFET? Is phase angle control more effecient? Zero point switching only has to turn components on twice per period, while PWM-ing a full bridge output would require a higher freq of switching compared to diac+triac, which in turn draws more power?

Which is generally more effecient?

Phase angle control + full bridge
or
Full bridge + PWM MOSFET

Last edited:

#### MisterBill2

Joined Jan 23, 2018
14,661
Why can't you PWM a full bridge with a power MOSFET? Is phase angle control more effecient? Zero point switching only has to turn components on twice per period, while PWM-ing a full bridge output would require a higher freq of switching compared to diac+triac, which in turn draws more power?

Which is generally more effecient?

Phase angle control + full bridge
or
Full bridge + PWM MOSFET
There is a quite different set of harmonics produced between the two kinds of control. In addition the control circuits are quite different, quite often phase control is simpler. So the amount and type of filtering is very different between the two kinds of control. In addition there is the consideration of response time, since phase control can only change once per cycle, while PWM can be adjusted every pulse, which is much faster with normal PWM systems.