[Your] Feedback needed on mosfet switch design

[bryce1]

Hi guys,

I want to design an opto-isolated "generic" low side switch that can PWM (400-900Hz frequency) everything 12V from solenoids to motors with about 3A max current.

TLP152 mosfet driver:
http://toshiba.semicon-storage.com/info/lookup.jsp?pid=TLP152&lang=en

Can you give me your two cents on the circuit?
I'd be especially thankful for some hints on the voltage ratings of the diodes and caps.

Thx!

[mzzj]

Hi guys,

I want to design an opto-isolated "generic" low side switch that can PWM (400-900Hz frequency) everything 12V from solenoids to motors with about 3A max current.

TLP152 mosfet driver:
http://toshiba.semicon-storage.com/info/lookup.jsp?pid=TLP152&lang=en

Can you give me your two cents on the circuit?
I'd be especially thankful for some hints on the voltage ratings of the diodes and caps.

Thx!

D6 most probably not needed.
I would beef up the C20 from 100uF to 1000uF or more
r10 10ohms sounds really low for 900Hz pwm frequency. Consider something like 20-1000ohms to reduce switching noise.
C3  probably there for nothing.

As for voltage ratings 16 volt electrolytic caps should be fine. Diodes 20 to 60 volt schotky diodes or just something general between 20 and 1000 volts. 

[PSR B1257]

I would beef up the C20 from 100uF to 1000uF or more

This is only usefull, if you know where to place it properly. Better is separate decoupling of the driver and the load terminal together with proper wiring.

The best case scenario would look like this:

[free_electron]

why do you put feed-thru capacitors on the gate ? with 100uf ( i doubt you can even find em that high , a few hundred pf , yes. even 100nf i doubt .. ) . there is not going to be much pwm on the gate anymore ...

[tautech]

why do you put feed-thru capacitors on the gate ? with 100uf ( i doubt you can even find em that high , a few hundred pf , yes. even 100nf i doubt .. ) . there is not going to be much pwm on the gate anymore ...

I'd doubt they are feed through's, it was just a convenient place to put the wire. 

[T3sl4co1l]

R16 might be a lot of sink current, check against the output rating of the opto.  It might also not be needed (does the opto have a default-low state when, e.g., supply voltage is low?).

Limited turn on/off speed, overheating and fault currents are other prime topics; if you are okay with having something that'll explode if misused, then it's fine as shown.  If you'd like a more general purpose switch, you might look at some of those "protected FET" parts out there that provide current, temperature and voltage protection, or design your own.

Tim

[bryce1]

I would beef up the C20 from 100uF to 1000uF or more

why?

[tautech]

I would beef up the C20 from 100uF to 1000uF or more

why?

When using a mosfet as a switch, you are always best to "hammer" the gate with plenty of current, increased bulk capacitance will allow that.

[T3sl4co1l]

When using a mosfet as a switch, you are always best to "hammer" the gate with plenty of current, increased bulk capacitance will allow that.

Not as a general statement.

When you can handle the large peak currents and high frequency content (with a suitably designed switching loop), and you have the EMI of that switching edge under control, then it's quite good -- less switching loss.

For slower applications where low switching loss is not important, going slower helps.  If you go slow enough, you don't even need the clamp diode!  Of course, that's load dependent, so you're still better with it.  You certainly don't want to go so slow that the transistor halfway burns out each time (>>100us??), but between "as fast as possible" (typically ~50ns for power switching grade FETs) and "a relaxed pace" (~1us), it's still quite fast for something that's switching only momentarily (>>10ms?).

For a couple amps load, I would be more than happy with 100uF local.  Don't even need that if there's another nearby in the project (say within 10cm or so?).

Tim

[mzzj]

For a couple amps load, I would be more than happy with 100uF local.  Don't even need that if there's another nearby in the project (say within 10cm or so?).

Tim

Yeah, it depends on wiring  on the input side. OP was making something "generic" so I assumed that input wiring might be 10cm or 10 meters.
100uF for that is a bit weeny with the typical 3ohm ESR. (supply voltage would drop to approx 7 volts at every turn-on and shoot up to 17v)   
Another issue would be insufficient ripple current rating.

[T3sl4co1l]

For single events, having the voltage bounded is better than unbounded; the ripple margin doesn't really matter.  Ok, the opto might not be happy with a 17V excursion.

Want to say I've seen much less than 3 ohms ESR for 100uF range capacitors.  That's still possible under low temperatures or aged conditions (dried out cap) though.  But such conditions don't really have an endpoint, so to address those failure modes, you'd have to use something other than electrolytic, entirely.

Even against 10m of wire, you'll have maybe 6uH stray inductance, which will ring with 0.1uF at 200kHz (Zo = 7.7 ohm), and with 100uF at 6.5kHz (Zo = 0.25 ohm).  Since ESR > 0.25 ohm, in series, the supply inrush/transient response will be well damped, so that's good.  With ESR < 7.7 ohm, in parallel, inrush/transient to the small bypass cap will also be well damped.  So at least there won't be ringing and squigglies due to that.

Ripple current rating won't matter to non-repetitive events, of course.  The supply becomes substantially resistive above 6kHz (where 10m of cable begins to act rather inductive), so any switching above 6kHz will be carried more and more by the cap, and the ripple rating will need to be observed.

So, somewhere around the 1kHz-50kHz range, depending on cable length and attached capacitance, will be the crossover range where the local cap matters.

Tim

[mzzj]

For single events, having the voltage bounded is better than unbounded; the ripple margin doesn't really matter.  Ok, the opto might not be happy with a 17V excursion.

Want to say I've seen much less than 3 ohms ESR for 100uF range capacitors.  That's still possible under low temperatures or aged conditions (dried out cap) though.  But such conditions don't really have an endpoint, so to address those failure modes, you'd have to use something other than electrolytic, entirely.

Looks like the "Opto" in this case is specified 10-30 volts so if there is a problem it would be the voltage droop.
No arguing that its better have damped response, my recommendation wasn't  1000uF plastic cap there 
3 Ohm esr for 100uF cap seems like some brain-fart on my side, so lets just ignore that. 0,3Ohms looks more like it and should be fine.

[bryce1]

you might look at some of those "protected FET" parts out there that provide current, temperature and voltage protection, or design your own.

Any recommendations?

[T3sl4co1l]

you might look at some of those "protected FET" parts out there that provide current, temperature and voltage protection, or design your own.

Any recommendations?

This is the sort of thing,
http://www.digikey.com/product-detail/en/VNS3NV04DPTR-E/497-13078-1-ND/3455768
Mike (@mikeselectricstuff) might have a stronger opinion on these sorts of things, he seems to use them quite regularly (and with good reason).

Tim