Two Mosfets for AC

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
the lamp i am using for testing at the moment is 150w however i intend to go with 300+ .

Alec_t: the problem i had with the VGS voltage stemmed from my assumption that the lower fet, when turned on will conduct the peak 34V from its drain to the source - this then passes through the body diode of the top fet to the load.
I figured that when the lower fet turns on, its source would be equal to the 34V peak value at the drain - but with the flow of current inverted in relation to the opposing cycle, i assumed this to be -34V at the source of the bottom fet and therfore a large VGS value incorporating the positive capacitor value and the negative source value.


Can i pester you for a quick explanation of why this is wrong (not that i question it isnt).
I think it is likely coming from the way i am visualising the voltages and currents, making elementary errors.

Thanks for the input folks!
 

crutschow

Joined Mar 14, 2008
34,283
The problem is that the AC voltage is floating with respect to the shown circuit ground, so it's difficult to visualize what the AC voltage is relative to the MOSFETs and the MOSFET's ground.

A simulation is the best way to see what's happening. You might give the free program, LTspice from Linear Technology a try. We can send you the .asc simulation files for that if you like.
 

ronv

Joined Nov 12, 2008
3,770
You might want to reconsider Crutchow's idea to use a big triac. The current at turn on for a lamp can be as much at 10 times it's running current. Most of the circuits shown so far don't have good gate drive as they turn on making the problem worse. Not saying it can't be done, but you need to think about it -- a lot. :D
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
You might want to reconsider Crutchow's idea to use a big triac. The current at turn on for a lamp can be as much at 10 times it's running current. Most of the circuits shown so far don't have good gate drive as they turn on making the problem worse. Not saying it can't be done, but you need to think about it -- a lot. :D
I have been thinking about it, alot. i think that's why my brain gets fried haha.
I will be limiting the inrush current, i have also chosen fets able to handle this inrush.
The schematics i have been posting have an awful gate drive - i have to limit the switching frequency to 2Khz.
Though i have received both a gate drive optocoupler and power mosfet gate drive IC's which i will be using to drive the gate at higher frequencies.
This is a project for my university, so i have certain constraints to operate within.
The end goal is to effect sine wave dimming, where i can use PWM to chop the ac sinusoid, then an LC circuit to produce a variable sine wave for powering the bulb.



Did you understand Crutschow's explanation in post #18?
Here's a sim with plots of the short-circuit via FET M2 and D4:
View attachment 79494
I understood crutschows explanation in post #18 which is why i posted a new schematic in post #29.
I am being a little OCD demanding explanations of the schematics as i have fried way to many components by not accurately foreseeing errors in the design.
As far as what is confusing me at the moment - it is the voltage potential seen at the sources when conducting for positive and negative cycles.
If i figure out where my assumptions are leading me astray i will post here, because i am getting seriously muddled with the floating AC.

I am also going to try get a simulation working as suggested. i had attempted this with TINA but seemed to run into software bugs with the diodes.

Thanks again folks!
 

crutschow

Joined Mar 14, 2008
34,283
..........................................
If i figure out where my assumptions are leading me astray i will post here, because i am getting seriously muddled with the floating AC.

I am also going to try get a simulation working as suggested. i had attempted this with TINA but seemed to run into software bugs with the diodes.....................................
Yes, I have problems visualizing the various voltages also. That's why I use the simulation.
It might help if you float the circuit and ground one side of the AC (same circuit, just a different perspective).

Have you tried a different diode type?
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
Do you have to reproduce the original sine wave or can it be rectified like your previous post?
i can go either way, its a university project - so it doesnt have to be perfect.
Though i want to use ac and filter it to reduce noise and emi.
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269

I have got the simulation up and running in lt spice .
I can indeed see the gate voltage behaviour.
Being obsessive, i was wondering if anyone knew the cause of the peak during the negative portion of M2?
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
Take a look at this one.
I didn't think of that!
What are the pro's and cons of the two methods.
Two fet method obviously requires two fets
The single fet method requires a bridge rectifier or 4 adequately rated diodes - i would likely have to heat-sink the bridge.
Thanks!
 

crutschow

Joined Mar 14, 2008
34,283
I have got the simulation up and running in lt spice .
I can indeed see the gate voltage behaviour.
Being obsessive, i was wondering if anyone knew the cause of the peak during the negative portion of M2?
It's the capacitor charging current through the diode at the positive peak of the sine-wave. The path of that current is through the circuit common connection back to V1 through M2.

To find odd currents like that, just start looking at the currents in the other components. That will usually lead you to the source of the anomaly.
That's the handy thing about a simulation. You can easily look at currents and voltages anywhere in the circuit.
 

crutschow

Joined Mar 14, 2008
34,283
I didn't think of that!
What are the pro's and cons of the two methods.
Two fet method obviously requires two fets
The single fet method requires a bridge rectifier or 4 adequately rated diodes - i would likely have to heat-sink the bridge.
Thanks!
Also, the diode bridge reduces the voltage by two diode drops and is thus less efficient.
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
Also, the diode bridge reduces the voltage by two diode drops and is thus less efficient.
Indeed, though this isnt too much of an issue - my supply is 25V with my bulb rated at 24.
So i can afford to drop a little.
I cant believe that i haven't come across this before now, i have searched quite a-lot for methods to pass ac with a single fet.
Clearly not hard enough!
 

ebeowulf17

Joined Aug 12, 2014
3,307
This may be a bit off topic since your real question was about controlling AC with MOSFETs, but are you sure you want to dim a halogen bulb at all, as opposed to using an entirely different heat source?

I don't remember the exact numbers for sure, but running halogen bulbs too far below their rated wattage (less than 80%, maybe) is hard on them. IIRC, it causes the filament to gradually evaporate and deposit itself on the bulb walls, increasing resistance of the remaining filament and eventually causing it to fail.

EDIT: Here's a link to one of the more readable write ups on the subject.
 
Last edited:

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
This may be a bit off topic since your real question was about controlling AC with MOSFETs, but are you sure you want to dim a halogen bulb at all, as opposed to using an entirely different heat source?

I don't remember the exact numbers for sure, but running halogen bulbs too far below their rated wattage (less than 80%, maybe) is hard on them. IIRC, it causes the filament to gradually evaporate and deposit itself on the bulb walls, increasing resistance of the remaining filament and eventually causing it to fail.

EDIT: Here's a link to one of the more readable write ups on the subject.
Agreed, dimming halogens has traditionally been inadvisable due to the reduction in lifespan of the bulb.
To my knowledge however, dimmable halogen bulbs have been produced for a while (i believe nitrogen is used to allow dimming, but i cant find the patent on google) and PWM dimming is employed in the automotive industry.
I am actually using dimmable halogen bulbs touted to last 2000Hrs.
 

ebeowulf17

Joined Aug 12, 2014
3,307
Agreed, dimming halogens has traditionally been inadvisable due to the reduction in lifespan of the bulb.
To my knowledge however, dimmable halogen bulbs have been produced for a while (i believe nitrogen is used to allow dimming, but i cant find the patent on google) and PWM dimming is employed in the automotive industry.
I am actually using dimmable halogen bulbs touted to last 2000Hrs.
Cool. I had no idea. I learn new stuff here all the time. Thanks!
 
Top