Hi,
I have using dcdc converter using uc3843 for more than 2years in my amplifiers.
Now my customer wants a convertor with smd mosfet devices for lower power conversion.
I am using it as boost converter to step up 12v to 24v.
I have used this with likes of z44 and irf3205. Never had any issues but now when i used FQD50N06 TO-252 mosfet it would blow mosfet on startup.
I have attached the pdf file which i have been using.
I would like any insights what might be the issue.
Freq used is around 40khz.
I have using dcdc converter using uc3843 for more than 2years in my amplifiers.
Now my customer wants a convertor with smd mosfet devices for lower power conversion.
I am using it as boost converter to step up 12v to 24v.
I have used this with likes of z44 and irf3205. Never had any issues but now when i used FQD50N06 TO-252 mosfet it would blow mosfet on startup.
I have attached the pdf file which i have been using.
I would like any insights what might be the issue.
Freq used is around 40khz.
The UC series are limited to 1Apk on its outputs, to move a larger MOSFET gate you will need an external buffer. I suggest any one with at least 3Apk as MIC44XX series or even a discrete buffer. You can try several 6A P + N MOSFET that come in SO8 package. Keep in mind the inversion of the UC output when making external buffers.
Thanks for your reply.
What i can see from datasheet is irf3205 has gate charge of 146nc and input capacitance of 3200pf and 50n06 is with 50nc gate charge and 2900pf of input capacitance.
I do not understand the fact that irf3205 should be more demanding than 50n06. But irf3205 works and 50n06 blows.
How you do explain this situation?
What i can see from datasheet is irf3205 has gate charge of 146nc and input capacitance of 3200pf and 50n06 is with 50nc gate charge and 2900pf of input capacitance.
I do not understand the fact that irf3205 should be more demanding than 50n06. But irf3205 works and 50n06 blows.
How you do explain this situation?
Try to search an old article from IR called dos and donts in driving mosfets or something like this. I have an old copy and has been actualized. This article is very juicy.
AN 937 from IR and "Design and appliction guide for high speed..." by Laszlo Balogh.
Good readings.
It happens that although the output of the UC looks pretty square under no or light cap. load, when heavily loaded by a large capacitive load (Static gate capacitance plus Miller Effect), a low pass filter is generated with driver's internal resistance. Thus the integrated square signal becomes a rippled DC voltage that biases MOSFET into class A. If the MOSFET's DC load is small, internally quickly overheats and finaly destroyed by overdisipation, helped by the fact that MOS structure has positive tempco. And cuurent is supposed to be fixed by UC's current mode operation. Then quickly the MOS channel becomes liquid.
It happens that although the output of the UC looks pretty square under no or light cap. load, when heavily loaded by a large capacitive load (Static gate capacitance plus Miller Effect), a low pass filter is generated with driver's internal resistance. Thus the integrated square signal becomes a rippled DC voltage that biases MOSFET into class A. If the MOSFET's DC load is small, internally quickly overheats and finaly destroyed by overdisipation, helped by the fact that MOS structure has positive tempco. And cuurent is supposed to be fixed by UC's current mode operation. Then quickly the MOS channel becomes liquid.
Given the remainder components are properly designed. Say: inductor doesn't saturate, lay out, proper loop stabilization and the like.
Thanks. Will update tomorrow. I have a board with space made for buffer. Will check with it tomorrow
A complimentary pair of transistors in emitter follower configuration can be a low cost buffer, but I prefer to use a gate driver IC like the very rugged IXDN602. I would change the linear regulator from 9V to something higher so that the power FET is more fully enhanced.
My prefered are complementary MOS transistors like STP3NF06 and STP3PF06. But you are free to choose other. To prevent shoot through internally between rhem, I use a ferrite bead in each drain.
Hi. I see some possible problems with this circuit. First problem is compensation, second is missing soft-start. Are you sure your 50n06 is not a fake? Test mosfet first by using a load of few amperes ,like car incadescent lamp 60w or resistor few ohms and 12v power supply. Put resistor 1k between gate and source of mosfet, drain through load to +12v , source to minus of psu. Temporarily apply +12v to gate, and watch lamp, it must light up bright , and check voktage drop on mosfet - calculate Rds based on voltage drop and current, displayed by psu ammeter. If it's higher by 20 percent , as specified in datasheet, fake probably, expect worse results if still using this .
About compensation - comp pin needs more than one capacitor for stable operation, mostly used is small cap like few nf and bigger one like 10-100nf with resistor in series. Check mosfet gate waveform with scope, using stable input voltage and resistive load - you must see stable squarewave with constant duty cycle, not jittering from low to high. Soft-start is not implemented in this ic, but can be added externally , it makes output voltage rise slower when you apply input voltage, and reduces mosfet , inductor and diode stress . Datasheet should include example of soft- start.
About compensation - comp pin needs more than one capacitor for stable operation, mostly used is small cap like few nf and bigger one like 10-100nf with resistor in series. Check mosfet gate waveform with scope, using stable input voltage and resistive load - you must see stable squarewave with constant duty cycle, not jittering from low to high. Soft-start is not implemented in this ic, but can be added externally , it makes output voltage rise slower when you apply input voltage, and reduces mosfet , inductor and diode stress . Datasheet should include example of soft- start.
Also, just looked at datasheets... There exists uc384x series without letters, series with A letter at end, with B letter, all specified to handle output of 1 ampere - there should be no problem to drive a mosfet with few nF gate capacity, just decrease gate series resistor.