EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: electros6 on August 29, 2014, 03:26:33 am
-
hi guys,
I am doing a flyback converter using LT3803. After I designed the converter I simulate it using LTspice.Down in the attachement in LT8303-1 I simulate my design using BSS123 MOSFET which has the Qg = 1nc and the results are fine (attached). But when I simulate the same design using the IRF530 MOSFET which has the Qg=26nc the output is oscillate in a peculiar manner (attached in LT3803-2). The datasheet doesn't tell anything about the gate driving capability of the mosfet. I don't the cause of the for the oscillation . can somebody figure it out.
-
Hi
Two obvious culprits here
(a) less likely: the chip is drawing more current driving the bigger mosfet, look at the voltage on C1, it should be stable at ~ 9.5V, if not try reducing r2 to 5k
(b) most likely: It's undercompensated , and is oscillating in the control loop, compensation is achieved with a transconductance stage by connecting an impedance where c3 is. If you look at the initial turn on , the second circuit pulls up much faster (2ms instead of 5ms) , so the lower RDSon, enables a lot more current into the TFMR. What you need to do is to increase the value of C3, in 2x steps, so try 2nf, 4nf, 8nf . Put a probe on C3 , if it is too small the control voltage will be swinging up and down, if its just right you will see a slight overshoot.
(bb) If you need to use a lot of capacitance to make it stable and, as a result, it gets slugglish with a poor load step response, then you need more fiddling, put a resistor in series with c3 (like page 12 of the datsheet) try a small value like 4k7, then 10k, then 22k, then 47k . And you could possibly put a small capacitor across that resistor (like 1nF) there are app notes somewhere about this.
Other things to try
(c) put a small 100pf? (or 220,470) from FB to ground (generally more useful on real circuits with parasitics) the 200kHz voltage waveform at the sense pin should NOT have a big spike on the front edge.
(d) put about 1nF across R3
(e) put 20mR in series with C2 to get realistic ESR
(f) tweak R1 , possibly a bit bigger? 0.047ohm?
(g) make k of the TFRMR 0.99 or 0.98 , put 20mohm between transistor and primary, and put ~ 1nf across one of the isolated secondaries (these won't solve your problem , but will get more realisitic nasty waveforms on the TFMR)
-
NGATE Peak Output Current (<10?s) ......................... 1A
tRISE Gate Drive Rise Time CLOAD = 3000pF 40 ns
tFALL Gate Drive Fall Time CLOAD = 3000pF 40 ns
I think your MOSFET has just too big gate charge.
-
hi,
I tried all the ways you say but still the problem stays. So I used the 7V output to connect to Vcc and so the problem clear and it works good.
-
Ok , looks like your MOSFET is too big for this device, if you check the power dissipated by the driver it will probably be excessive , so it's dragging down the power supply, if you built this in real life, the chip would shut down on over temperature.
MOSFET selection in any converter is always a tradeoff between conduction loss, switching loss , and gate driving loss , going for lowest Ron isn't always optimum overall. In any case you've picked the wrong transistor, the IRF530 is a very old part , ore modern transistors have a better figure of merit.
This driver was designed for smaller physical size inductors, hence 200kHz operation, you need to backoff on the Rdson requirement to minimise the other losses (compared to operating at 40kHz).
You might consider adding a gate resistor 10R to 20R (to shift losses from gate driver to switching loss) with a large area MOSFET , this will slow down the switching, and probably improve the EMI signature.
The LTC3803 is designed for small DC-DC applications , so one should be looking for parts similar to the FDC2512 in the data sheet. For example the FDS86141 or FDMS86104 are around 25milliohm and 7nC and would be good upgrade choice from the FDC2512.