Hi Power SMPS design question

[Tequilataz]

This is my first post on this site so let me give a little background on this project first before I get to my question(s) I built a four channel Leach Superamp for my car. For the power supply I was experimenting with modifying a few different
brands of power inverters to supply the +/- 85 volts and have had varying degrees of success. I have found that the approach different manufacturers take in their design varies greatly and that some are easier than others to modify. But this has turned into a real learning experience . While I am currently powering the amp with a modified 2000 watt inverter My rail voltages are only +/-72 volts My goal is to get the full +/- 85 so I decided to build my own from scratch. I went to all the web sites I could find (ESP,and many others) and looked at all the schematics,application notes,etc. read all the advice I could and finally decided on a design . I used the TL494 chip. I figured
I needed at least 2500 watts as the Leach Superamp at 4 ohms should ouput around 550 watts or so per channel and somewhat higher at 2 ohms depending on the value of the resistors used in the protection circuit. I read somewhere that the power supply was usually the weak link in car amps so I
I probably went overboard on the design. I decided to run a total of 30 IRF3205s(15 per side)in a push -pull design The Magnetics ZP48613TC toroid core which is their P type core material with an OD of around 3.4"ID of 2.2" and W of .55" .Although I went through their design calculations I figured 6 turns(3+3) center tapped on the primary and 40(20+20) center tapped on the secondary should be a good starting point with a frequency of around 50 KHz. I copied the drive circuit resistor set up/snubber from a known Kenwood SMPS design so I"m pretty sure that part is okay. On all the inverters I looked at If the inveter was say 750 watts it would have 2 sets of 4 gate drive transistors driving 8 mosfets on the step up portion uf the inverter. They were small signal transistors . Since I had 30 mosfets I decided to just use 2 each of the higher power MJE15030/MJE15031 transistors to drive all of them. For the sake of simplicity I wired this to run unregulated with the circuit board traces for output sensing and feedback there just in case later I changed things to run it regulated . My preliminary checks of the output waveform before connectiong the mosfets were all good .When I was testing the prototype circuit the output voltage only went up to around +/- 50 volts . So I raised the frequency to 100KHZ Since it was running unregulated my meter showed the duty cycle at 49.5%. The voltage then rose to the transformer ratio level of +/- 85 . So far so good , nothing overheating everything looked normal I then put a resistive load of 100 ohms on each half . All okay voltage not dropping no overheating. Then I figured this is going to work out okay and put a carbon pile load of 10 ohms per side . About 15 seconds went by and I was watching the soft start circuit raise the voltage.It appeared to be raising at a slower rate than with the 100 ohm load. Then I heard the mosfets popping one by one . I got to thinking what caused this. Im sure the toroid core wasn't into saturation. The transistors should have had more than enough capability to handle the load. The design books I have talk about the mosfets advantages in the ease with which they can be paralleled. My question is this Did the paralleling of all 15 mosfets per side being driven by only one set of driver transistors cause the mosfet input capacitances to sum to a larger net value and cause both halves to be in the on state at the same time when the current level reached a certain point ? I did use individual gate drive resistors for each mosfet with a value of 100 ohms paralled with a 33 ohm resistor in series with a 1N4148 reverse biased diode. And as for output filtering for this test I only used a 400 uf 400 volt cap.Any insight to this would be of great help.

[Perry Babin]

100 ohms is too high for the gate resistors if you're using IRF3205s. Try 47 ohm.

If you're using emitter follower pairs for the driver circuit, you don't need the 1N4148. If you could post a schematic of the driver circuit, it may help us to understand precisely what you have.

[luka]

[i] My question is this Did the paralleling of all 15 mosfets per side being driven by only one set of driver transistors cause the mosfet input capacitances to sum to a larger net value and cause both halves to be in the on state at the same time when the current level reached a certain point ?

No, if your driver for fets if powerful enough. But for such power, I would use more then just one core, and no need for so many fets per side!

I did use individual gate drive resistors for each mosfet with a value of 100 ohms paralled with a 33 ohm resistor in series with a 1N4148 reverse biased diode.

Too big, 22-33 is big enough for fast switching... no need for diode, if you use lower resistor for gates...

PS: didn't you check your waveforms with scope??? I mean low power will work any time, but with high power there is now way is knowing what did happen..