Power Supply Capacitors keep going up in Smoke?

[gary350]

What are these ferrite tube shape beads for? Will they block high frequency? I wonder if they will help protect my PS capacitors?

I wonder if I should put 2 heavy duty choke coils on the + and - side of the power supply, between the power supply and the induction heater?

The 3300 uf 50v capacitor went up is smoke first about a month ago. I replaced it with a 470 uf 200 v capacitor and it has been working fine.

Yesterday I added 2 more capacitors to the LC circuit, the Power supply capacitor went up in smoke and the bridge rectifier with small heat sink burned up.

I have seen on You Tube several people have said, the 1000v 50 amp bridge rectifiers from China self distruct at 15 amps. Several people have said put several bridge rectifiers in parallel or use higher amp diodes. Now I have 5 bridge rectifiers in parallel on a very large heat sink getting ready to do the smoke test again.

I see a circuit just like mine on You Tube the guy has 8 caps .47 uf 1000 v on the LC coil and his induction heater works much better than mine. Lowering the resonance frequency from 221 KHz to 60 KHz reduced skin effect and 8 caps increases current to the LC coil so his circuit works much better than my induction heater. I am trying to copy what the other guy has done. Not sure what other changes I should make to protect the circuit?

**broken link removed**


Is there any advantage to using 2 ferrite chokes .005 mh each on all 3 power supply capacitors?

**broken link removed**

I have changed the 2mh center tap choke coil to 8mh.

I have put 2 chokes coils 8 mh each in + and - between power supply and induction heater.

4 caps on the LC coil has lower Hz to 135 with no load and 115 with a 1/4"steel rod as the load in the LC coil.

I put a .005 mh ferrite choke on each leg of all 3 power supply capacitors.

Power supply is 15.3 VDC. Power transformer is 10.8 VAC. With induction heater ON the RF voltage is 70 volts.

So far so good, it works better than ever.

8 caps on the LC coil will lower Hz pretty close to where I want to be, target is 60 KHz with a Litz wire LC coil..

 

[crutschow]

I doubt that the inductors or ferrite beads will help your problem.
The caps and rectifiers likely blew form overvoltage and/or overcurrent.
A high current rectifier will burn out if it's not on a good heat sink.

 

[moffy]

Like crutschow said:

I doubt that the inductors or ferrite beads will help your problem.
The caps and rectifiers likely blew form overvoltage and/or overcurrent.
A high current rectifier will burn out if its not on a good heat sink.

The capacitors look massively underrated for the current/frequency you are putting through them.
You could try putting some high value metallised film capacitors in parallel with the electro, to cope with the higher frequencies/currents.

 

[gary350]

The capacitors look massively underrated for the current/frequency you are putting through them.
You could try putting some high value metallised film capacitors in parallel with the electro, to cope with the higher frequencies/currents.

The 50 volt caps and one 200 volt cap smoked. The power supply is 15.3 vdc. The RF voltage is 70 volts. I think RF is getting into the power supply.

400 volt caps in the LC circuit all smoked. The 1000 volt caps are doing good.

 

[ronv]

Aw Gary, we missed your problem all along.
The current from your 15 volt supply is very high, so 3000ufd. is not nearly enough and the ripple current in it is way to high. When you measure with the meter you measure average voltage so you think there is 15 volts but in fact it could be peak of 25 and valley of 5. This is probably why your FETs kept blowing until you added the power supply for the gates. - Not enough gate voltage. The bigger the part you put in the coil the higher the current will go, so the better you make it the worse it gets for the caps and rectifier. You can add more caps in parallel to share the current and keep the voltage up, but pretty soon the rectifier will overheat due to the high peak current to charge them.

 

[gary350]

ronv, you were right about the PS not having enough capacitors.

Sense I don't have a scope I decided I can do this by adding 1 capacitor at a time then checking the voltage each time.

With 1 3300 uf capacitor voltage = 15.3 vdc

I added 2 more 3300 uf capacitors 1 at a time voltage went up each time.

I added 2 more caps 820 uf each voltage kept going up each time.

I added 1 more 680 cap voltage is up to 18.53 volts with no load.

I added another 470 cap voltage went down to 18.47 vdc so I removed that capacitor.

I tried to add several more capacitors of all sizes but can not get higher than 18.53 vdc.

3300, 3300, 3300, 820, 820, 680 = 12,220 uf

Wow, this PS is dangerious, it holds at 17.8 volts and only drops .01 volt every 3 seconds. It will take almost 4 days to completely discharge.

Now I need a bleader resistor.

It is after midnight I am off to bed.

 

[Misterbenn]

Now I need a bleader resistor.

might be worth putting an indicator LED on it as well

 

[gary350]

Now I have a new problem.

The two 8 mh choke coils between the power supply and the induction heater are getting so hot they are smoking after a 2 minute run. The ferrite toroid that is covered in plastic smokes first. I turned the power supply off before the chokes got hot enough to burn up.

I have a 4K 10 watt bleader resistor on the PS. PS voltage is 14.8 and drops to 5 volts instantly when I turn the PS off. Then it takes 78 seconds for voltage to drop to 1 volt and about 4 more minutes to drop to .01 volts. 14.8 / 4k = .0037 ma. and this = .05 watts. The 10 watts resistor does not even get warm.

The PS 22,000 uf capacitors seem to have improved the performance of the unit. A solid steel 1/4" rod diameter is red hot in 30 seconds. Before it was taking 2 minutes before the metal rod started to turn red but it never did glow very red hot.

I still worry that RF will get into the power supply and kill the capacitors. Look at the circuit drawing. Notice the KHz meter readings. What does the 3.2, .6 and .5 KHz readings mean?

**broken link removed**

 

[ronsimpson]

Notice the KHz meter readings. What does the 3.2, .6 and .5 KHz readings mean?

Probably the low readings are because the signal is small at those points and the meter can not count correctly.

You need bigger coils.

There is a third leg on the 7812. Where is the ground leg connected?

 

[gary350]

Probably the low readings are because the signal is small at those points and the meter can not count correctly.

You need bigger coils.

There is a third leg on the 7812. Where is the ground leg connected?

Sorry, I did not draw the ground wire, I did not draw the 2 capacitors on the 7812 either. That appears to be the only thing I forgot to draw. Maybe I should revise the drawing.

Yes, I know the coils need to be larger now. Each time I make improvements that causes problems and other things need to be changed. Now I am at the place where I need a large water cooled cool. I also need to change the 8 turn coil to 6 turns then add 2 more capacitors to keep the frequency at about 60 KHz. 2 less turns will increase coil current 25%. I would like to experement with Litz wire coils before I go to copper water cooled coils.

 

[ronv]

Aren't the filter capacitors on the other side of the bridge rectifier?

 

[gary350]

Aren't the filter capacitors on the other side of the bridge rectifier?

There are 2 groups of capacitors. 6 capacitors between transformer and bridge rectifier and 8 capacitors at the LC coil. You can see both cap banks in the drawing.

 

[ronv]

They should be from +15 to ground. Not across the transformer.

 

[ronv]

Here is a simulation of your circuit as it started and a link to ltspice. I know you don't like downloads but with this you will be able to see the effects of your changes. Much cheaper.
Spice won't keep you from frying your components, so you still need to read the data sheets.
https://www.linear.com/designtools/software/

 

[ChrisP58]

**broken link removed**

Why do you have capacitors between the transformer and the bridge rectifier? The voltage there is AC, so most any capacitance there will smoke. The reason your 200V parts aren't dying is probably because they have enough ESR to keep the heat from the fault current below the failure point.

Your line filter capacitors should be right at the output of the of the rectifier, just before the 8uH inductors.

 

[gary350]

Now we are on the same page. I don't know what I was thinking when I drew that. I changed the drawing.

Once I get all the bugs worked out I want to build this into a nice portable metal case like a portable welder.

**broken link removed**

**broken link removed**

 

[ronv]

I hate to nit pic, but I will just in case.
I think the cap on the input of the 7812 wants to go to ground, not the output.
The big coil is 8mH or 8 uH.
If the 8 uH ones are the little ones you showed earlier they will be way to small as far as current goes.

 

[gary350]

Here is the circuit I have copied. I think I had make the wrong drawing on my circuit. If I number the pins on my drawing I can make it right and not change the drawing.

The choke coils are from old TVs except for the one that I hand wound with #16 enamel coated copper wire. Maybe that small #20 wire is getting hot. I used my meter to test the uh value of TV chokes that I have. I soldered a 6mh and 2mh is seriec to get 8uh for the - side. The choke in the + side is the one that is smoking when it gets hot and it has larger wire than the - side.

I think I have also been typing 8uh instead of 8mh. The chokes are all 8mh.

 

[ChrisP58]

You need to know that all components have secondary parameters that affect their operation in any given circuit beyond their basic values and ratings. And why two equivalent parts may not really be equal to each other.

This is not a complete list, just the few that might be causing you problems.

Capacitors. In addition to capacitance and voltage rating, also have ESR (Equivalent Series Resistance,) and ESL (Equivalent Series Inductance,) and Ripple Current ratings.

Inductors. Besides inductance also have resistance. And two different current ratings. Current that causes temperature rise, and current that causes the core to saturate. When an inductor saturates, it's inductance falls to zero. Also the core material can have a big impact on how an inductor of any given value will behave at different frequencies.

 

[ronsimpson]

I am looking at RonV's spice file.
Has anyone looked up the data sheets on all the capacitors? The spice models might need to be updated. ERS, ESL