Placement of large input caps

Hello all,
I would really appreciate some advice on a project.
I am working on a board that is driving a relatively high current heater (~1.5A) switching @ 30V.
To stabilize voltage, I have a 5600 uF cap on the input. I also have a 1.5A fuse and reverse polarity protection via an n-channel MOSFET (4.4A).
My concern is that when I connect power, the cap will trip the fuse, as it is placed after the fuse and MOSFET and perhaps I should place it in front of the fuse and rev.pol. protection, but then the cap is not protected.
I am not very experienced with power supply design, so I would appreciate your expert advice :slight_smile:

Here’s the input circuit:

Surge currents through big capacitors are a serious design consideration.
Even capacitors as small as 100uF can have such high currents that they damage connector pins over time.

Your MOSfet won’t help much, because it’s body diode is conducting it’s current.
You can try adding a capacitor between the gate and source of the MOSfet, this may help to turn it on slowly (combined with R32).

But overall. This is not an electronics forum. A site like is much better suited for that. This forum is for helping people work with the KiCad software.

1 Like

At the risk of continuing a thread not related to KiCad, what you mean by stabilize voltage? Remove ripple or actually keep the voltage fixed? Using the formulae Q = CV and Q = It, shows that the time that the capacitor can continue to provide about 1.5 A current is of the order of a tenth of a second. Wouldn’t it be better to design an active stabilizer and incorporate soft turn on? Sorry I don’t know much beyond that.

If you expect 1.5A I would rather considered 3A fuse.
I’m not fuse expert but they are divided into fast and slow (or may be there are more classes).
I think the first question is what do you know about power source connected to your circuit.
If it is stabilized DC supply I would see nothing against replacing 5600uF with 220uF.
If it is 50 or 60 Hz transformer with Graetz bridge than may be the fuse should be before the transformer and the current stress may be will be not so strong because of transformer impedances.
If you don’t know what will be connected then you probably should not design that circuit.
But I have never designed anything with such high power consumption.

I have been designing power supplies for 40 years. I think that a full discussion is well outside/beyond the scope of this forum, but I would echo the question of why you think you need so much capacitance.

Basically I think you want both inrush current limiting and input reverse polarity protection. That can probably be done with Pch MOSFETs but I prefer Nch. To accomplish both, you ought to have two MOSFETs “back to back” with sources connected and gates connected. Then when input power is applied use a charge pump to ramp the gate voltage up gradually in order to limit the inrush current. There are “hot swap” ICs that might be able to help accomplish this if you do not want to “roll your own.” Perhaps that is the first place to look.

The Si2319 CDS may be OK for reverse polarity protection. I assume you want to save power compared to using a 60V schottky diode. But if you decide to do inrush limiting, it would probably be safest (as in avoiding self destruction) if you stick to using a “linear” MOSFET. Most MOSFETs are really intended for switching and not spending significant time in limited conduction. The “linear” FETs are more rugged in that type of operation. But if this is a hobby circuit and you do not mind replacing a blown one; learning in the process I would choose a bigger FET in a Dpak or other larger package.

Piotr I have to hand it to you again. :grinning: I have been working with power supplies but I never heard of that term. To me it is just a bridge rectifier.

A (maybe similar) “standardized” term is “high side switch”. These often have additional circuitry such as current monitoring and limiting and thermal shutdown.
Below is an example of a datasheet for such an IC. It’s not suited for you (only 5V 1A) but it is first example of the type of IC that I found.

Same here, but Piotr is right and I learnt something today. I’m surprised he didn’t call it a Pollak Bridge. :wink:

While I do not think there are any universal definitions, I think of “hot swap” as being a widely recognized term. When I type that into a search at DigiKey, I get a sub-type of power management ICs. Below is my search for that term at DigiKey. Some are listed as being able to handle considerable power. Without digging into the details, I see limits of 100V and 63 Amps. Some of these ICs provide a large amount of flexibility and control.

I tend to think of “hot swap” and “high side switch” as being somewhat different. I associate “hot swap” as protection for plugging something into a live DC circuit, while “high side switch” as more for switching a load on and off as it remains physically connected. As I say there are no universal definitions and I suspect there is a lot of function overlap.

If I type “high side switch”, DigiKey points me to “load switches”. Here is one which may be useful. I did not spend any time selecting, but I tend to view Infineon as a trustworthy manufacturer.

I think that Piotr raises another VERY important point. Your need for limiting the inrush current will depend GREATLY on what is supplying the power ahead of what you show. Specifically, its output impedance and any rise time. If your power source has a big capacitor charged up to full voltage and you will be plugging “hot” then you will definitely need some inrush current limiting. But if you have a modestly sized mains frequency transformer and bridge rectifier without much output capacitance, there may be less need. I would probably not use a blow-able fuse regardless.

If you are interested in pursuing an IC to limit inrush current, it is probably worth spending a considerable amount of time sorting through the offerings of distributors such as Digikey, Mouser, and/or others. In my experience, searching at distributors (while very flawed as in any search) offers a degree of filtering which is very helpful as compared to the baloney you will get with a general purpose search website.

Yes he has rectified the errors of our ways. :wink:

I have ever heard about Pollak bridge, but it was always for me Graetz bridge. And in Wiki they write “Today the circuit is sometimes referred to as a Graetz circuit or Graetz bridge.”

Reading about this…I was wondering what type of rectifier diode was available to these 19th century inventors? They certainly did not have silicon rectifiers, and vacuum tubes (and vt rectifier diodes) came around in the 20th century. I was unsure about selenium rectifiers but thought those probably also were invented in the 20th century. This page indicates that they were invented in 1883, so that is consistent with the invention of the bridge rectifier a few years later. The next time that you walk, bicycle, or drive across a bridge, you can be glad that the bridge is not upside down. And maybe you can thank a bridge rectifier. :grin:

My two cents:
I live in germany, and here german scientist Leo Graetz is assumed to be the inventor of the four semiconductor diode full bridge rectifier. Decades ago, at school, they introduced this to be the „Graetz bridge“ to the audience, wich I used to be part of. I think even in germany Graetz bridge sounds a bit old fashioned. But so am I.

1 Like

Wikipedia suggests that Karl Ferdinand Braun at Marburg, Germany, discovered that a point-contact semiconductor rectifies AC, in 1874.
At that time it seems to have been common that the same thing was discovered or invented by several people in different countries, independently of each other, around the same time or a few years apart.

As a side note, the swedish scientist who discovered selenium in 1817-1818 was Jöns Jacob Berzelius (usually called Jacob Berzelius).

1 Like

What is providing the source of the 28V? is it s rectifier/Pollak/Graetz ? or is it some other power source? What drove the decision for a 1.5A fuse, especially when the “current heater” is rated at ~1.5A

Something needs to be done but what depends on your unstated design drivers. At the very least it sounds like the fuse needs to uprated (2A+) since you do not want it popping during normal operational tolerances but when a FAULT has occurred and thus to limit the propagation. Likewise slow burn vs fast needs to be assessed as a slow-burn might mitigate the inrush problems

As others have noted that large cap is going to cause you problems and the main one discussed was the inrush current, but the 2nd is the possibility of an overshoot.
If this is from some 230V:12V transformer (~10:1) feeding a full bridge rectifier, then yes that capacitance is in the right order of magnitude for a 17R load (28V @ 1.5A)

its going to be bursty and the inrush will be high - NOTE: I aligned with a COSINE not a SINE because the worst-case would be attachment when there is a peak of the sine. That is like 50A+ inrush current for 1/2 cycle and quite high peaks as it tops up the capacitor.

The other issue is the stray inductance (supply, interwiring etc…) that will cause a large overshoot on the voltage.

what todo?

  1. evaluate the source of the 28V? is it some current limited output supply? if so you are in luck, if it is from a transformer+rectifier, the VA limit will help but not alot
  2. a precharge circuit, either an NTC in series so that it will drop in resistance as it heats up or a passive n-type on the return of the capacitor




your problem with the fuse is simple. You have to use a slow blow fuse here.

The current rating has to be chosen for protection against over-current or component failure (short circuit).
The slow-blow fuse will allow the high transient transient in the front capacitor.
Have a look here :

and :

Please note that the curves are different among manufacturers, there is no universal “ultra-fast”, “fast”, “slow” specs.


this problem is quite common for frequency inverterts. Often two easy solutions are chosen:

For medium to high power a charging PTC is inserted that is bridged by a relais or a semiconductor after a short time. This is expensive and needs some space.

For low power a NTC is inserted. it gets hot due to the current and becomes low impedance for operation. Very simple but likely not possible/desireable for 1.5A

Greetings, Chris

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.