Can someone please look at the attached schematic and tell me if it would work for driving mosfets. The input pulse frequency would be around 25Khz.
I did the following changes..
Remove R2.. changed the value of R1 to 10 ohm. Replaced the 74LS09 with the 74HC08N. It seems to be working. The Mosfet is driving a simulated (Bunch of Led's and resistors wire wound) load of 2.8Amps. I have currently 95V driving the load.
It seems to be working for the last 10 minutes without getting too hot to touch!. Mind you I do have a PC fan cooling the mosfet!. (No heat sink yet!)
I will try 15V+ but at the moment all I had was a 7812 and will have to buy / make a 15V+ supply!!!
The signal to the 74HC08 is from the parallel port and a small GWBASIC prog allows me to send pulses of varrying frequency to the driver. At 18KHz (going by the prog.) the led's seem continously on!. I do not have a scope. If I want to check that the mosfet isn't skipping a beat how would I do that..! I plan to go to 25KHz and higher if possible!!!!
One more thing. is there any software out there that would allow me to simulate the circuit in theory..?????? saving on stupid mistaks like a 4.7K resistor instead of 10 ohm one.....
The voltage excursion in your mosfet gate is limited between around 4.4 volt and 1.5 volt, not enough to keep the mosfets switching times short, and out of the linear zone, hence the temperature rise.
There are simulation software packages, all of them as good as the component model you make or get, no simulation software will ever be a replacement for knowledge and experience but they help a lot during the design stage. A good simulation run never guarantees that the actual circuit will work. So, the prototype phase is still necessary.is there any software out there that would allow me to simulate the circuit in theory..??????
A quick critique of your circuit: The gate resisistor/diode serves no purpose. The LED is harmful and will never light anyway. It is harmful because it adds an additional 1.5V to the output at logic '0'. Add the transistor's 0.6V Vbe and you get a 2.1V logic '0' which is perilously close to the MOSFET's gate to source threshold voltage.
This totem-pole gate driver should work better because it gives level-shifting. The all NPN circuit should work with 3.3V to 5V logic. This type of circuit prevents cross-conduction of the totem-poled NPNs. The 470pF is a "speed-up cap". The 500nS time-constant helps sweep out the bottom NPN's minority charge carriers on turn-off by applying a negative Vbe voltage.
The NPN / MOSFET variant minimizes component count. Use any small-signal n-channel MOSFET for the bottom transistor. It's input should be 5V logic or higher.
I agree with Kreutz. There is nothing like actually breadboarding a design and trouble-shooting your creation with a scope. It is an iterative process that repeatedly loops from schematic to breadboard to scope readings, converging to a workable circuit.
After all it's Nature your circuit has to please, not a simulator. I use Spice for a rough first cut occasionally and to establish limits of operability after design works on a breadboard. A simulator is not a primary tool for me.
Or, you can use a single stage fet driver and eliminate a bunch of IC's and real estate being taken up by discreets. Besides, the fet drivers take care of level shifting and they are also directly driveable by TTL/CMOS IC's.
We used to drive paralleled fets at 3600hz with a combination 2N4401/2N4403 totem pole and they ran pretty well. If you hit the conventional IRFZ40 style fets with 15-18vdc, they switched fast and hard with no appreciable heating. When logic levels came into being, we hit them with 10vdc using the same drivers.
Or driver was essentially the same as the original one proposed by the member except we did not have the LED and the PNP and NPN"s were switched in the totem pole - the NPN emitter was hooked direcly to ground and the PNP was hooked to V+.
Using a MOSFET driver IC takes all the pleasure out of actually designing a circuit.:-)
Needless to say we use MOSFET driver ICs for the same reasons you mentioned but it is nice to have the ability to design a discreet circuit if you have to. Another reason is it's just enjoyable to design circuits because it gives a much better understanding of the intricacy of the devices used. This knowledge cannot be gained when you use manufacturer 'cookbook' IC circuits.
The gate driver situation becomes far more interesting if you have an all n-channel MOSFET bridge. An efficient and fast discreet gate driver design is really a challange then. I have one I designed quite a while ago that works better than any monolithic gate driver but it required more parts of course.
1) Avoid using "logic-level" power MOSFETs. They are more fragile than standard threshold devices.
2) Avoid using p-channel power MOSFETs. There are no true complimentary MOSFETs like there is with bipolar transistors; a p-channel MOSFET has 4-times the Rds of an equivalent n-channel MOSFET for the same chip size and Vds rating.
3) Don't switch a MOSFET too fast or too slow. Too slow (0.1V/nS) and you get large switching losses and possible parasitic oscillation. Too fast (10V/nS) and you run the possibilty of Miller capacitance turning on the parasitic NPN transistor that forms the intrinsic diode. Tailor the gate charge/discharge currents to control switching speed.
4) Mind the MOSFET source to ground path inductance. Minimize the inductance aggressively (non-inductive sense resistors, multi-layer board with a ground plane, short trace lengths from source to ground plane, etc.).
Don't use perf-board to breadboard power MOSFETs. Use an unetched printed circuit board as a ground plane instead. Your breadboard should look like a modern-art sculpture.
5) Study the intrinsic diode reverse recovery behaviour. Avoid devices that exhibit a 'snap' recovery. Use devices that have a 'soft' reverse recovery instead. That means using a fast scope while comparing devices from different manufacturers.
6) Bypass the drain power supply bus. Use electolytics in parallel with monolithic ceramic caps. Keep all power section componet lead lengths very short; anything longer than 1/4" or 5mm is a crime. Anything needing ground gets soldered directly to the ground plane.
Thank you all for your detailed input and especially mariss for the design. I have very limited electronics background (mechanical to be exact) and I have been reading and reading and reading about stuff.....
I will remove the gate resistor and the diode. The led and the attached cap will also disappear.
I did find out that with my original design the voltage available at the gate for the mosfet is half of the voltage supplied to the 2N3904 and the pull voltage is just a tad bit less than the signal voltage.
I do agree with NC CCAMS that Mosfet Drivers are made to make life easy. But One thing people have to remember that not all are blessed with having access to such components. So it comes down to making do with what is available in the market that suits ones purpose. It also becomes a very encouraging learning exercise. Things like film resistors, Logic Level Mosfets, High speed diodes, Mosfet drivers are seldom available and if one is lucky might get some 2nd hand..?? recycled..Components that we get are usually chinese made so whatever the data sheets say about them everything is down graded by 75%.
This results in a very different way of thinking! Software simulation just cuts some of the costs of blown up components and also does the mathematics for someone like me who is learning to read the datasheets properly!
A scope is a luxury. But yesterday I did manage to pick a 2nd hand LBO-310 scope for a princely sum of 4 dollars (equivalent). It wasn't working so I gave it to a friend of mine who thinks he can get it working (he wants to use it too!). Hopefully that will be a valued addition to my soldering iron, maginifying glass and multimeter.
I have always enjoyed reading your posts and comments and have learnt a lot from them.
The list of do's and don'ts goes on the wall for me to see all the time. Once I have completed the schematic of what I am doing I no doubt will post it for further comments and learning.