Hi
I wonder if there are any experts out there that could help me with my driver output stage. I am building coil drivers for medical bug killers but I am an amateur when it comes to switch mode type output stages because the closest I came to switch modes at university was in the power supply of the main frame that use to run my Algol batch programs.
As you can see from the attached schematics I started with a single ended driver but as I pushed up the current to now 150A in to the coil the diode across the load has turned in to a 4” cube heat sink where the massive stud diode and a series 100 W resistor are mounted. So I thought why not use a second MOSFET and just get it to effectively short out the coil to dump the stored inductor energy to replace the diode contraption. At the moment the new drive circuit is switching 40A on 5V and works better than I expected. Looking at the drive wave forms and output switching speeds I decided that dead bands were not required although there is one anomaly that can be seen in the scope picture. The spike causes the upper transistor to heat more than the lower one. In fact I should say the upper transistor heats and the lower one doesn’t. I have tried a diode across the load figuring it must be the back lash from the coil but it made no difference to the spike. I would be grateful for any opinions about what the cause might be and the best way to fix it.
The circuit has been running for several weeks with no problems and the spike is actually beneficial for the application on 5V allowing the coil to be switched up to 2 MHz at 40A giving 5 times the voltage and energy at the front end of the wave. However I wish to increase the voltage to push more current and the transistors are limited to 25V so the heating of the upper transistor will be too great and the voltage of the spike would kill the transistors hence the need to stop the spike.
Thanks

 

When you turn off Q1, the voltage at the source of Q2 gets pretty high. How are you driving its gate? Is that gate drive at least 8 to 10 V greater than that source voltage? You might want to consider a P-channel top mosfet.

John

 

Try adding some capacitance across Q2's source and drain, perhaps 330pF or so to start. Single-layer ceramic caps should work OK. Multilayer caps will have too much parasitic inductance. Keep the leads as short as possible.

This will "buy time" for Q2 to turn on after Q1 has turned off.

You'll be better off having too little capacitance than too much.

 

When you turn off Q1, the voltage at the source of Q2 gets pretty high. How are you driving its gate? Is that gate drive at least 8 to 10 V greater than that source voltage? You might want to consider a P-channel top mosfet.

Hi John
The drive is from a fast MOSFET driver chip giving 12V drive signals in fact the bottom trace of the scope picture shows one of the signals and it is the same scale as the output wave on 5V per division. You can see the output switches perfectly between 0V and 5V even with more current than 40A. It is just the spike that is unwanted. If it is the energy from the load I would have thought a diode would have solved that but it made no difference that I could see so am I missing something fundamental?
Thanks

 

Try adding some capacitance across Q2's source and drain, perhaps 330pF or so to start.
This will "buy time" for Q2 to turn on after Q1 has turned off.

Hi SgtWookie
Will such a small capacitance across the output delay the MOSFET turning on? The current is very high and the impedances low. I shall have a go tomorrow.
Thanks

 

Hi SgtWookie
Will such a small capacitance across the output delay the MOSFET turning on?

The idea is not to change the turn on/turn off time of the MOSFET; it is to provide a place for the current to go until the MOSFET can turn on.

Observe what the peak voltage is without any added capacitance.

Then add a small amount of capacitance, and see how much the peak voltage decreases. That should give you an idea on how much more or less capacitance you'll need to get rid of most of the spike.

 

so the purpose of the 2nd fet is to protect the 1st fet?

is the spike wanted, like what you get from a automotive ignition coil? a flyback dump trick is using a zener and a hefty npn to dump non-wanted coil flyback energy.

can you explain "medical bug killer", is this a "zapper".

 

so the purpose of the 2nd fet is to protect the 1st fet?

Well yes and no. On the original circuit the diode dumped the stored coil energy but on the latest circuit Q2 dumps the energy better by shorting out the coil. It is now like a push pull arrangement and actively drives the current in and out.

is the spike wanted, like what you get from a automotive ignition coil? a flyback dump trick is using a zener and a hefty npn to dump non-wanted coil flyback energy.

The intention was to generate a nice square wave with as much current as possible. Spikes are dependant upon the wiring and coil characteristics but if it turns out to improve treatments I may look at reducing the coil resistance to get more current. At this time I don't know if a short spike is as good as a continuous sqaure wave. It will certainly give more penertration in to the body.

can you explain "medical bug killer", is this a "zapper".

There are many types of machines and methods of killing bugs. The common theme is that the bugs are destroyed by rattling or riping them to bits using resonance of their natural frequencies. This can be achieved by passing current through the body, placing the body in a magnetic or electric field or using ultrasonic transducers. The Zapper is a contact device that passes a current through the body but as you can imagine treatment will only occur where the current flows. I prefer electric or magnetic field devices because the field goes right through the body and requires no contact. So far I have found that the bigger the bug is the bigger the field must be but it is not quite that simple. One thing I am sure of is that my field devices kill bugs.
Unfortunately I have just dropped a screwdriver on my board and blown up the frequency generator what a pain (Can't repeat my exclamation ). I 'll have to fix that before testing if the cap will solve the spike problem.
Here is my prototype Phanotron tube device if you are interested.

 

wouldnt that natural freq of [place favorite living organism here] be about the resonance of water, ~2.4GHz band, a standard microwave for food? bugs are mostly water, right...?

i'm just not following what you are wanting to do with the flyback energy thats in the coil during the off period of the drive. you wish to dump it using a smaller solution than the diode you had? i think thats what you were wanting. either way, energy is energy and it needs to dissipate somewhere.....

how about using a bjt and zener method (attached)?

 

wouldnt that natural freq of [place favorite living organism here] be about the resonance of water, ~2.4GHz band, a standard microwave for food? bugs are mostly water, right...?

The theory is that bugs have unique gene sequences and every sequence has a unique resonant frequency enabling us to surgically destroy the bug without affecting the body. A water molecule is very small hence its resonant frequency is high but you could not use that to treat bugs because all the cells in the body would be destroyed as well.

i'm just not following what you are wanting to do with the flyback energy thats in the coil during the off period of the drive. you wish to dump it using a smaller solution than the diode you had? i think thats what you were wanting. either way, energy is energy and it needs to dissipate somewhere.....
how about using a bjt and zener method (attached)?

I am sorry but I must be missing something. My thinking is that if the coil energy can be put in using a MOSFET without heating then the same type of MOSFET should be able to dump the stored energy and it must be less energy because a lot has been lost in heating the coil. Is this thinking wrong?

 

ok, i see now.

what about complimentary FET's like the attached. V1 is 2MHz square wave drive. maybe the drive should be PWM adjustable for tuning?

 

ok, i see now.
what about complimentary FET's like the attached. V1 is 2MHz square wave drive. maybe the drive should be PWM adjustable for tuning?

Your circuit would just keep the coil on permanently wouldn't it? When one MOSFET is on the other is off and both are tied to ground.

 

Your circuit would just keep the coil on permanently wouldn't it? When one MOSFET is on the other is off and both are tied to ground.

now that i look at it again i dont think the pFET would conduct if there was no flyback voltage. Vgs would be a issue due to where the load is. question is, would the flyback get Vgs right for the channel to conduct??

 

Hi Guys
I am sort of back where I started at the start of this thread. Well I thought I was but as it turns out the position is even better there is no spike. The output pin of my PIC that is the frequency drive was blown up and the driver chip was what failed. I thought I had blown the circuit because it stopped working after I dropped the screwdriver on it but now I think it was close to failure anyway. The MOSFET driver chip shorted out the PIC pin so I had to replace the driver chip and PIC. Then I started to look at the unwanted spike with a smaller load connected but I couldn't see it. The penny dropped because when I first tested the circuit it didn't matter what load I had the spike was always there and the same size. I should have twigged it earlier that if the spike was always the same it was load independent so it was not due to the stored energy in the coil. It was due to a fault in the driver chip. I think although the waves looked correct the current drive capability was low on the side that was driving the upper MOSFET. The device could not turn on fast because there was not enough current available to empty the gate capacitance so it dissipated more heat. The spike was present because the MOSFET was not turned on hard enough to conduct the current allowing the voltage rise.
So the circuit as it stands at the moment uses two MOSFETs and a dual MOSFET driver chip that is directly driven by a PIC micro and it gives a perfect square wave in to the coil that is running at 40A but probably will end up at around 150A.
Electronics never ceases to amaze me the way something like a faulty chip can lead you up the garden path making things look really complicated but once solved ends up so simple.
Thanks for your help guys

 

Hi Guys
Things have moved on and I am now driving 600 Amps in to the coil. However after a lot of testing I believe it would be more advantageous to let the coil voltage rise to its maximum when Q1 is turned off. This way I could get a big magnetic field from the heavy current and a then a big electric field from the massive open circuit coil. My question now is if I remove Q2 and perhaps replace it with a very high value resistance how do I protect Q1 from the large voltage generated when I switch off Q1. I have looked at some ignition circuits but they tend to use high voltage transistors as Q1 then limit the voltage to their maximum but this is not suitable for my circuit because there are no high voltage transistor capable of 600 Amps, well not that I can afford. Also SCRs seem to be used but they are slow and don't have a hope of reaching the frequencies I wish to run. Any suggestions would be appreciated.
Thanks