0
\$\begingroup\$

In my project I want to use a couple of old russian VFDs. To drive them, I need 28V from a 12V power supply. The plan is to use a DC-DC boost converter as calculated here and here.

I've ported the designs to Yenka, a circuit simulator, but for some reason the MOSFET and inductor keep burning up because of sudden peeks of current of hundreds of amperes.

the circuit

That surely can't be caused by the lack of Schottky diodes in the software, a small difference in frequency of switching the diode and the voltage drop can't cause such immense currents - or can it?

\$\endgroup\$
3
  • \$\begingroup\$ The diode has to be at least an ultrafast rectifier for high-frequency switching - a 1N400x won't cut it. I also doubt that the 555 can properly drive a MOSFET if the frequency is too high. \$\endgroup\$ Dec 24, 2014 at 14:19
  • 2
    \$\begingroup\$ Expecting others on the site to reverse-engineer your calculations from your circuit is a bit much - you should state your conditions in your question (i.e. what frequency and duty cycle you were expecting) otherwise how can we tell if you made a mistake somewhere? \$\endgroup\$ Dec 24, 2014 at 14:20
  • 1
    \$\begingroup\$ (1) Try de-idealising your energy store components. eg add a small resistor in series with the inductor and maybe a small capacitor across it. I have seen cct sim pkgs that took ideal inductors at face value and ... . (2) Do yourself an immense favour and try using an MC34063 - very olde tech smps IC, very cheap, very available. MUCH more useful in real world smps apps than 555. Datasheets have all the application ccts you need and need minimal parts - no more than for 555 doing same job (and 555 cannot DO all it can). \$\endgroup\$
    – Russell McMahon
    Dec 24, 2014 at 14:25

2 Answers 2

1
\$\begingroup\$

The basic topology looks OK, so the problems are most likely due to implementation details. Since you gave little detail, we can only make some guesses:

  1. The FET isn't being driven too slowly. The total transition time should be a small fraction of the pulsing period. For example, if the oscillator is running at 100 kHz, then the pulse period is 10 µs, and the switching transition time should be small compared to that. If the gate voltage takes more than 1 µs (in this example) to get from high to low or low to high, that's not good.

    FETs have significant gate capacitance, so it takes a significant pulse of current to switch the gate from one state to the other. The digital output can probably only source or sink a few 10s of mA.

  2. Your are using too high a frequency. This works together with #1. The faster the oscillator, the faster the gate needs to transition to keep the FET fully on or fully off most of the time.

  3. At this low voltage you really should be using a Schottky diode, not the ordinary silicon diode you show. There are two reasons for this. Schottky diodes have a lower forward drop and have very fast reverse recovery time. The lower forward drop helps with efficiency. The fast reverse recovery time is very important since the FET is shorting the output during the recovery time. That really beats on the FET and the diode.

  4. 2 mH seems very large. Again, we don't know your switching speed or output current requirement, but such large inductors will have significant series resistance.

  5. The duty cycle is not optimized. For a ideal switch and diode, the forward voltage on the inductor will be 12 V, and the reverse voltage 16 V. The length of the on and off phases should be inversely proportional to those, respectively. Again, let's use 100 kHz switching frequency as example. That gives you 10 µs for the whole period. You want 16 parts of that to be on and 12 parts off, which means 5.7 µs on and 4.3 µs off. Since there will be some inefficiencies and losses, in practise the on time will be a little more relative to the off time than the purely theoretical 16/12 ratio.

\$\endgroup\$
1
\$\begingroup\$

A brief and approximate analysis: -

From a cursory check of values around the 555 you are using approximately a 2kHz switching frequency. With a 50:50 duty cycle, the inductor charge time is therefore 250 us.

It's a 2mH inductor and you are applying 12 volt across it for 250us. From the inductor equation, di/dt = 6000 amps per second (12V/2mH). Therefore after the 1st 250us I would expect the current in the inductor to rise to 1.5 amps.

At this point, the energy in the inductor is going to be (half LI^2) = 2.25 milli joules. This energy is being transferred 2000 times per second and this, as a minimum, equates to a power of 4.5 watts for the load. This will rise if the inductor isn't fully discharged of its energy because the switcher will enter continuous mode (as they normally do)

Because the load is 10kohms, the voltage across the load for this minimum power is about 212 volts (power = volts^2/R). Is this approximately what you expect?

You say you need 28 volts so I'm now assuming that your duty cycle is much smaller than 50% - maybe more like 5% on 95% off - is this what you are producing from the 555 timer?

If not then you've made an error in values around the 555 and you should fix it.

\$\endgroup\$
3
  • \$\begingroup\$ Well, I re-entered the values into the boost converter calculator: 32 kHz Vin 11.9-12.1 Vout 27.5-28.5 Iout 0.020 Amp Vripple 0.3V The results were: 56%-58.25% duty cycle, min. inductor size 2.3mH, peak inductor current under 100mA However in my simulation program there's no Schottky diode and the diode can't be set to apropriate forward voltage drop. Since you seem to be concerned about the inductor size, I had to set the frequency to 32kHz; 555 calculator: at 57% duty cycle, freq 32kHz: RA=630R, RB=1,94k, C=10nF. In my simulation everything's on fire. Would it work with a Schottky? \$\endgroup\$
    – Emil J
    Dec 25, 2014 at 23:15
  • \$\begingroup\$ Also, what alternative to a MOSFET do I have? A lower frequency of switching would require a larger inductor. Also I put a 7R resistor in series with the inductor, not much of a difference. \$\endgroup\$
    – Emil J
    Dec 25, 2014 at 23:20
  • \$\begingroup\$ How can everything be on fire in your simulator? Also check this doctronics.co.uk/555.htm#astable out and still tell me it's oscilalting at 32kHz \$\endgroup\$
    – Andy aka
    Dec 26, 2014 at 19:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.