The capacitors are warming up (become hot)

Thread Starter

Motanache

Joined Mar 2, 2015
540
It's a problem. Because many Electronics engineers do not accept the fact that the capacitor heats up.

I made a Class D amplifier. PWM Driver ....MOSFET driver IRF2110......MOSFET IRF640.
+/-50V or 0...100V.

I tested it in my room with a laptop adapter (12V and 3A). It worked perfectly.
12V......SMPS +/-50V...audio amplifier.

When I tested it in the car, I had a surprise.
The capacitor in series with the loudspeaker it was hot. Smoke out of it. That in a few seconds.
This capacitor was 470uF / 200V and came from a PC power supply.

I did not use LC filter on out for class D. I want to test it without him. I want to make a robust device. In case of filter capacitor fails, do not have serious problems.

I just do not measure temperature on the capacitor. I have not seen anything like this anywhere.
-.................................................................

Then I removed this capacitor in bserie with the loudspeaker.
Surprize ... other heating capacitors were heating up!!!!!!:eek::eek::eek::eek::eek:

..............................................
Now let's talk a little about the capacitor.
At low frequencies we have iron core transformers. At higher frequencies, their core will heat up and should be used ferrite.

Electrolytic capacitors are used for low frequency. At higher frequencies the dipole of the electrolyte can not follow the electric field variation and and warm up(Maxwell). It would have had a ceramic capacitor, but I did not have 47uF ceramic
....................................................................................
What I want?
Help me find a solution to use the amplifier in the car.
However, in case we do not find it, I want to discuss why condensers heat up. That's why we started the last discussion.
But if we find a solution quickly and I listen to the music in the car, I do not need it anymore.
 

GopherT

Joined Nov 23, 2012
8,009
It's a problem. Because many Electronics engineers do not accept the fact that the capacitor heats up.

I made a Class D amplifier. PWM Driver ....MOSFET driver IRF2110......MOSFET IRF640.
+/-50V or 0...100V.

I tested it in my room with a laptop adapter (12V and 3A). It worked perfectly.
12V......SMPS +/-50V...audio amplifier.

When I tested it in the car, I had a surprise.
The capacitor in series with the loudspeaker it was hot. Smoke out of it. That in a few seconds.
This capacitor was 470uF / 200V and came from a PC power supply.

I did not use LC filter on out for class D. I want to test it without him. I want to make a robust device. In case of filter capacitor fails, do not have serious problems.

I just do not measure temperature on the capacitor. I have not seen anything like this anywhere.
-.................................................................

Then I removed this capacitor in bserie with the loudspeaker.
Surprize ... other heating capacitors were heating up!!!!!!:eek::eek::eek::eek::eek:

..............................................
Now let's talk a little about the capacitor.
At low frequencies we have iron core transformers. At higher frequencies, their core will heat up and should be used ferrite.

Electrolytic capacitors are used for low frequency. At higher frequencies the dipole of the electrolyte can not follow the electric field variation and and warm up(Maxwell). It would have had a ceramic capacitor, but I did not have 47uF ceramic
....................................................................................
What I want?
Help me find a solution to use the amplifier in the car.
However, in case we do not find it, I want to discuss why condensers heat up. That's why we started the last discussion.
But if we find a solution quickly and I listen to the music in the car, I do not need it anymore.
I assume you have a full bridge class D amplifier. That means current is flowing in both directions in/out of your cap. Now look at the black line on your cap that shows negative leg. There is a problem because it is not always negative on a full bridge.

So, either get a non-polar electrolytic or make one by using two 100uF caps in series (50uF equivalent).

You'll have to look up the right way to connect the two polarized electrolytic caps in series to make a non-polarized cap (I don't remember if it is + to + or - to -.. I'm going to bed.
 

Picbuster

Joined Dec 2, 2013
1,047
GopherT is correct. ( use two diodes over the cap's)
An other reason for warming up good be the impedance current is flying and energy is generated I^2 * impedance = heat.
You do need a use as stated a bi-polar or non-polar capacitor with a low ESR able to handle that current.
Your effect can also be introduced by oscillation @ a high frequency (above 20Khz).
Star with the correct capacitor still hot? Connect a scope and see what happened.
Picbuster
 

#12

Joined Nov 30, 2010
18,224
This capacitor was 470uF
(I don't remember if it is + to + or - to -.. I'm going to bed.
It doesn't matter whether its + to + or - to- as long as your diodes are correct. Stripe on the diode facing + on the capacitors.
A couple of 1000 uf caps rated at 63 V to 100 V will work.

There aren't many capable of at least 6.25 amps of ripple current.
http://www.mouser.com/Passive-Compo...ttZ1yx4atu&Rl=75hqwZgjdhf7Z1yzvauuZ1y9ihz2SGT
 

Thread Starter

Motanache

Joined Mar 2, 2015
540
There aren't many capable of at least 6.25 amps of ripple current.
Verry useful remark. Then I suspend this version with capacitor in series with speaker.


The second version-
I assume you have a full bridge class D amplifier.
.
No, it is a half bridge amplifier. The schematic is the following
http://geekcircuits.com/2010/07/class-d-amp-using-tl494-dc-to-dc-converter-chip-2/

Figure1
But it has the following changes-
1. IR2110 instead of IR2010
2. Logical NAND(CD4001) instead of XOR. I hope I maintained the correct logic.
3. PWM driver is -
Figure 2

The difference is that it has not feedback(R13).
Now, the capacitors that heat up are C1(47uF) and C6. Both are polarized.
 
Last edited:

bertus

Joined Apr 5, 2008
22,290
Hello,

You can not use an 4001 in stead of the 4030/4070.
In the schematic the 4030/4070 is used to have an inverted and a non inverted signal for the half H-bridge.
The 4001 is a NOR gate, so the lower part of the H-bridge will be always low.
The 4001 at the upper part of the H-brige will go high when both inputs are low.

Bertus
 

Thread Starter

Motanache

Joined Mar 2, 2015
540
I do not know how to show you that those capacitors are getting warm.
The thermal picture(camera) I can not found.

I have 4001 and it works. I have not kept schematic from the figure. It is important to get PWM and inverted PWM. This is not difficult considering the gates have negation. I used logic gates only as simple NOT, It is on test PCB.

PWM High ..........High MOSFET ON
PWM Low ...........Low MOSFET ON (But not both ON at the same time)

I could recording video with the phone. But I need a thermometer.
 

Thread Starter

Motanache

Joined Mar 2, 2015
540
Maybe you're helping me improvise a thermometer from an easy-to-find electronic component.
I'm still looking into the box with componets if I find a thermistor.
 

bertus

Joined Apr 5, 2008
22,290
Hello,

How can you say that the cicuit is working with a 4001?
The logic tables are quit different:

Motanache_4001_pinout_logic_table.png

Motanache_4030_pinout_logic_table.png

As you can see with the 4001 the lower Lin will stay low all the time.

Bertus
 

bertus

Joined Apr 5, 2008
22,290
Hello,

In the given schematic there is no capacitor in series with the speaker.
It is not needed as the speaker is connected to half the powersupply.
It assumes the avarage output of the half-bridge to be at half the powersupply.

As said, with the 4001 the lower part will not switch.
That way you will end up with half the signal as the speaker is on half the powersupply.

Bertus
 

Thread Starter

Motanache

Joined Mar 2, 2015
540
Hello,
As said, with the 4001 the lower part will not switch.
Only if I keep the same connection to the pins and only replace the IC.

Hello,
That way you will end up with half the signal as the speaker is on half the powersupply.
If I have a symmetrical voltage supply (as in figure 1 post #5) there is no need for that capacitor.
From R1 (figure 2 post # 5) you can adjust the continuous voltage on the loudspeaker.
However for the loudspeaker protection in order not to have accidentaly the DC through speaker I want to keep that capacitor.

I also took into consideration the simple voltage version in which I used that capacitor.
The goal was to be cheap and robust.
 

Thread Starter

Motanache

Joined Mar 2, 2015
540
There aren't many capable of at least 6.25 amps of ripple current.
Can you tell me an explanation for that?(for statement above)


Yes, I think ripple is the problem.
I will test with 100nF ceramic in parallel on these capacitors.

Initially I had another MOSFET driver that had a 200mA in MOSFET gate.In that way it was tested for 3 days.
Since I put IR2110 which has 2A MOSFET gate, has not been initial tested.
Now in a few tens of seconds the capacitors burn.

I did not put R3 and R4 in figure 1 post # 5.
The output of the IR2110 is directly connected to the MOSFET gate.
R3=R4=0ohm
 

Thread Starter

Motanache

Joined Mar 2, 2015
540
From R1 (figure 2 post # 5) you can adjust the continuous voltage on the loudspeaker.
I've already burned a lower speaker for tests. I have DC through the speaker.
It is very difficult to obtain 50% duty cycle with 0% Volume.

IR2110

In HIN i have the PWM.
In LIN I have the same PWM but passing through:
-

Depending on what IC you are using.
Put the both entrances together uses only negation (except xor).
In my case for CD4001


Pin 1 and 2 are connected together at the input PWM. And pin 3 is connected to the LIN of the IR2110
 
Last edited:
Top