diyAudio (
-   Class D (
-   -   switching power supply failure (

sdoom 2nd March 2004 11:42 PM

switching power supply failure
Hi @ all

first of all, please excuse my English.

I have a pretty big car amp, itīs an older 2channel Amp called Altai B0005V. Itīs the same as Caliber CA500 and is capable of handling 500W RMS in bridged mode. Well I got it from a friend of mine when it wasnīt working anymore.

So I checked the amp and was kind of depressed when I saw the mess in there. The switching power supply is divided into two parts. It uses a TL494, 8MOSFET (IRFZ44) per transformer. There are 2 transformers totalling the MOSFET to 16. Each transformer has 2 seperate input coils so that makes 4 MOSFET per transformer-coil.

One of the transformers was busted real bad and broken apart. All MOSFET were dead. So I used a transformer-core that I got from another amp that has the same size and used the same type wiring for it. So the transformers are both the same again and working.

For testing I used only 4x IRFZ34 , one per transformer coil and the amp was back to live, the musik played without distortion on both channels and also in bridge mode. The switching power supply output voltage was nice running +/-52V. The capacitors on the secondary side of the transformers are all ok, so are the diodes.

So I checked on the MOSFET and couldnīt get the IRFZ44 , only IRFZ44N and that one doesnīt have the same specs (IRFZ44=150W, IRFZ44N=110W and lower current capability). So I used IRFZ48N, they come very close to the IRFZ 44. On the spec sheet the IRFZ44 has 28mOhms@ 31A and the IRFZ48N has 16mOhms@32A.

Installed 16 brand new of those and put everything back together.

Now I have the problem that the amp starts running but after 10 minutes it kills the Capacitors on the input 12V side and then the TL494 shuts down. I thought it could also be a problem with the TL494 and replaced it with a new one, but the same problem persits. The stock capacitors on the input side were 2x 1000ĩF/25V and 2x 470ĩF/25V . I replaced those with the same 1000ĩF/40V and 470ĩF/40V but they also blow after 10 minutes....

Itīs definetly not my 12V power supply, I also tried it with a car battery.

Is this the MOSFET causing the problem? I know now that the IRFZ48N has (a bit) less resistance than the IRFZ44, could it be that the transformers suck too much current when the FETs are turned on? I have never had any problems on the TL494 replacing a broken IRFZ44 with an IRFZ48N so far.

Could it help if I take one of the IRFZ48N out per transformer coil, I will still be on the safe side concerning the MOSFET Ptot if I use 3 of them per transformer coil.



Rhesus 3rd March 2004 09:26 AM

Are you using low-esr caps intended for high frequency smps?

If you don't the caps will run way to hot.


sdoom 3rd March 2004 10:24 AM


yes I used brand new caps that are ok for the use in switching power supplies. I also tried another set of caps out of another Amplifier.

mwh-eng 4th March 2004 01:29 AM may have your original Mosfets and I think they will ship to you.

You do have a strange problem. Sounds like too much ripple current for these capacitors. With the small inductors in series with the input and other capacitors you could have an oscillation problem, but I think the probability of this is low. If you have an Oscope, monitor the voltage across the capacitors that fail, to see if an oscillation occurs. Don't let them run for 10 minutes and have another failure, though.

Using Mosfets with lower RDS will increase the ripple current a bit. The only other thing I can think of that may put extra stress on these caps, is cross current conduction in the Mosfets. If something is wrong causing the PWM to run too high of a duty cycle without enough dead time, this could cause much higher ripple current on the caps. I'm pulling at straws here and guessing some. Let us know your progress with this problem.

Good luck!

sdoom 4th March 2004 11:11 AM

thx for your reply.

Well I donīt have access to a scope, thatīs definetly the next thing I will get !

I also think that the resistance of the FET is too low. I ordered the IRFZ44N and new caps now, hopefully with those it will work.
But itīs strange because I have replaced the FET in some other amps already and never had any problem. I doublechecked it on a standard 200W AMP half year ago when I had a scope and there was really no difference between the 2 MOSFET types. But I assume now with 16 of those the TL494 cant shorten the ON-time of the FET anymore but wouldnīt I have a higher idle current ?

I checked the input voltage with a digital multimeter (good one) and it always reads 13V DC . On the AC -setting I read a few millivolts, if I connect another amp to the DC-power I get the same readings.

I will try again when the new MOSFET arrive and let you know.

Thx again

Eva 4th March 2004 12:30 PM

If you have access to an oscilloscope, you should check power supply waveforms [at MOSFET gates and drains, and at output diodes]

Capacitors may be blowing due to overheating caused by a too high ripple current being applied. You should check for temperature increase on the capacitors

The power supply may be operating at the wrong frequency [some cores could break when operated at certain frequencies], or it may be suffering cross-conduction or transformer saturation. All these problems may cause very high AC ripple currents across the input capacitors

Also, don't asume two toroid cores have the same volts*second/turn characteristic just because they are the same size. This parameter determines the number of primary turns needed for a given supply voltage and a given core. When core specifications are not available, it's a good idea to measure the saturation characteristics of the core

When choosing new mosfets, the most important parameter is the Cgs capacitance since the amplifier has its gate and control circuit resistors sized for a certain capacitance value. If you use MOSFETs with higher Cgs or Cdg capacitances you will usually experience cross-conduction or parasitisctic turn-on and increased switching losses. MOSFETs with equal or lower capacitances will work fine. The other important parameter is Rds-on but it only affects conduction losses

I like IRFZ48V for repairing car amplifier power supplies. These devices have low Cgs capacitance and are the cheapest I can get in my country

If your devices have somewhat higher Cgs capacitance than expected, then you may have to reduce the value of gate or turn-off circuit resistors to speed-up turn-off process

mwh-eng 4th March 2004 06:42 PM

Eva brought up some very good points for potential problems!

The TL494 may have trouble driving the Mosfets with higher input capacitance. The cgs and cgd can really impact the rise and fall times, and turn on/off delays. You don't have much miller effect at 13 VDC with cgd, but enough to be a potential problem. The cgd miller capacitance charge causes turn on/off delays. Without an Oscope you are pretty much in the dark on some things.

Is there a current sensing resistor from the Mosfet sources to ground? If so, you can measure the DC voltage across it and knowing the resistor value, calculate the average current. You may have a current transformer considering that this is a high powered amp. It will have a current sensing resistor somewhere on the secondary side.

You could also have a stability problem. With these DC to DC converters, you can usually hear a stability problem. If you hear an un-even bad sound, you may have a stability problem and this can certainly cause problems. If you hear a steady clear high frequency, you probably don't have a stability problem. Usually these converters run at a frequency of 25 KHz or higher so that they are silent to the human.

Another potential problem could be a leaking output (secondary) filter cap (not likely). The audio amp portion of the unit could have a problem causing excessive current draw from the converter, also. You can do some good troubleshooting with just your DVM, but having an Oscope would be much better.

If you can isolate the power amp stage from the DC to DC converter this could help. If you have a high wattage power resistor for placing a load on the converter without the power amp section connected, this may help in case the power amp is causing the problem.

Hope you find and fix the problem before burning out more new parts!

Good Luck!

sdoom 4th March 2004 07:32 PM

me again :o)

I took the whole assembly apart today because I got new parts.

The audio part seems to work fine . I connected +/- 50V and let the amp play musik over night, no excessive hheat or destortion so that shouldnībe the probelm. If I had a fault in there I assume the DC-current would be much higher. Thereīs no bad sound with this DC-DC converter, I know what sound you mean , had this once on a Kenwood amp where UA7815 and UA7915 were broken and caused a short , that sounds real ugly. :o)

I will install the new MOSFET and I replaced the caps and, maybe that was the problem, there are two small capacitors parallel to the caps that burn, they should have 100nF each. I took them out and checked them with a good capacity-meter and they had 240 and 310nF . So thatīs out of the tolreance limit. I replaced those with 250V types , so they wont fail !

The caps I have now are new and the guy who sold them said they can be used in DC-DC converters w/o any problem.

The resistors connected to the FET have 150Ohms, I have never seen that resistor value in a DC-DC converter before, usually 47-100 Ohms is what I have seen.

If that value was too high the FET is not as quick as needed in turning off , right?

OK, I will install the FET/Caps I have now and check for the resistor current as you said .

thx guys, I hope this baby gets back to live soon :o)

mwh-eng 4th March 2004 11:17 PM


Originally posted by sdoom
The resistors connected to the FET have 150Ohms, I have never seen that resistor value in a DC-DC converter before, usually 47-100 Ohms is what I have seen.

If that value was too high the FET is not as quick as needed in turning off , right?

In my experience, a resistor value of 150 ohms for the gate drive is found for medium powered Mosfets or in applications other than DC to DC converters. In my designs for commercial and military avionics, the typical value for gate drive resistors was more like 10 ohms. Generally, a higher value gate drive resistor means slower switching speeds, lower efficiency, less EMI, and less power dissipated in snubbers.

The value of gate drive resistance effects both turn on and off speed. Running high duty cycle as many of these DC to DC Converters are designed, means the chance of cross current conduction (Mosfets on the push and pull side being on simultaneously) is possible, unless the dead time (set with the timing capacitor of the PWM, usually) is selected properly to prevent this condition. I've not used the TL494 so I would need to look at the data sheet to be sure how the dead time is selected. Since this is not a design problem, hopefully, then there must be a bad or marginally bad component somewhere in this converter. However, as Eva mentioned, using Mosfets with higher input capacitance can be your problem! Without an Oscope you really need to replace the Mosfets with the same part numbers as the originals. Using Mosfets with less input capacitance is better than using Mosfets with more input capacitance than orginals, I think. You may have noise related problems, however.

Good luck and let us know what happens when you fire it up!

mwh-eng 4th March 2004 11:27 PM

As Eva also mentioned, you need to make sure that transformer core you used as a replacement is the same as original. Size is not a good enough indication of the correct core. If it is a ferrite core, then the core material is very important as well. Also, the winding method (how windings are wound on the core) can be critical in some applications.

Repairing a SMPS without an Oscope is not easy, unless the problem is minor. You have done some major repair work.

All times are GMT. The time now is 09:32 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright Đ1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2