Circuit to increase CEMF spikes

[Kerrowman]

I have built a circuit for which the aim is to produce high voltage back EMF spikes from the Drain of a MOSFET so I can investigate some of the properties of these voltage transients.

What I have built is shown in the attached circuit where I am using an IR2121 driver chip to encourage shorter shut off times and so produce higher voltage spikes. Going from a circuit that didn't use a driver chip to one that does has increased the voltage from about 800V to 1,040V as shown in the scope image using a 10:1 voltage divider.
 
I have read that there are ways to further reduce the FET shut off time but as electronics is not my main discipline I find them rather confusing. For example, reducing the Gate bias resistor (R5) further (to 5R?) or putting a small capacitor (1nF?) across R5 and keeping the PCB track resistance from R5 to the Gate of Q2 as short as possible.
 
I would appreciate any suggestions.
 
Thanks
 

 

[HarryA1]

They often use a diode across your R5, I gather to discharge the gate capacitor to speed it up. See this link and click on the "simulate transient" for the IPP60R099CP.

https://design.infineon.com/tinademo/designer.php?path=EXAMPLESROOT|INFINEON|Applications|Industrial|Power|&file=power_MOSFET_compare_450V.tsc&act=change.U.Infineon.IPP60R099CP_L0

I wonder if the turn off delay of the 555 propagates through the circuit? The 555 off time is 2us compared to the equivalent timer 1455 at 100ns.

 

[Kerrowman]

Thanks I will look into the links.

Am I right in thinking the diode would have its anode towards the FET?

 

[HarryA1]

Yes, as in the above link.

 

[Kerrowman]

So is the graph label ’Power Loss’ the back EMF at the drain? 
 

And do you think any capacitance across the gate resistor would help?

 

[HarryA1]

So is the graph label ’Power Loss’ the back EMF at the drain? 
 

And do you think any capacitance across the gate resistor would help?

I am not sure about their graph either. You could try  playing around with a capacitor but I am  not sure how much  it affects the voltage drop rate at the output. As you are 95% there why not use an electronic ignition circuit? There are numerous  circuits like this one on the internet. You could get up to a few thousand volts then!

 

[Kerrowman]

It's part of the investigation to use pulses derived from an inductor field collapse. I have built what you suggest but it was for another application. Thanks

 

[HarryA1]

Is the pulse width large enough for the coil current to reach maximum? I would think current should get to something like (12v - mosfet saturation voltage)/Rcoil.  Also the break down voltage of the mosfet and its diode may come in to play at high voltages?

for others:

 

[Kerrowman]

The MOSFET is rated at 600V Vds and my D1 (1000V breakdown) is there to offer some protection (I think). So far the FET runs nice and cool and seemingly unperturbed.

If you are referring to the pulse width from the 555 then they are 50% duty cycle pulses as often with a PWM setup. Each solenoid coil has a resistance of about 10Ohms from memory and there are 5 in parallel (see pic). My maximum current using the 555 trigger is about 1.75A. When using the rotor based hall sensor then about 0.6A (surprisingly, but this value is very dependent on the positioning of the Hall sensor - as with the ignition timing on an ICE).

 

[HarryA1]

Looks impressive!  Do you have an idea of the inductance of the coils? I could play with it in the simulator.

 

[Kerrowman]

Hi Harry,

 
That’s good of you to suggest doing a simulation.
 
It will be useful to you to see a diagram depicting the whole circuit and the Hall sensor option (attached) so you can see how the trigger circuit relates to the whole.
 
As mentioned I can trigger the FET (FCP260N60E in fact and not an IP160R099) to produce its 1,000V back EMF pulses at 100-5kHz using the 555 based internal trigger system or the rotor/Hall based system where it will reach a pulse repetition frequency of about 200Hz. The graphic shows the Hall sensor option but the Trigger circuit has the timer option built in at the flick of a switch.
 
This is all to investigate the behaviour of high voltage transients on batteries and is what Nicola Tesla was doing in the later stages of his life when he researched the ‘fluid-like properties’ of electricity in its longitudinal wave format. I and many others think that there are still unrecognised qualities of electrostatic fields and electricity itself that may allow us to harvest electricity from the environment. Certainly, others seem to have done so and in the inductive methodology of science, my work on this is to see if I can replicate an observable phenomenon.
 
If I get statistically significant results then I will do a paper on it and share it freely, and also how to build it for further replication, and leave it to others in the future to theorise what might be going on. Even though I’m a physicist by training, my brain isn’t young enough anymore to do the maths and algebra 

 
The next stage is to add a ‘capacitor dump’ circuit that collects the HV pulses and discharges high current pulses to the batteries and which will enhance the effect. If any of this interests you then I’m happy to share the info as I go along.
 
Anyway, all that aside, I have measured one of my coils at 380mH so let’s assume they are all the same, and there are 5 in parallel.
 
I’d be interested to see what your simulation shows and presumably any suggestions for how to tweak the back EMF up a bit further.
 
Thanks
 
Jules
 

 

[HarryA1]

Using the circuit below with  an 2.5ms wide pulse at 10v  with rise and fall times of 100ns in the simulator I get the wave forms below.



The green trace is the input at the base of the mosfet, the yellow trace is the current through one of the coils while the red trace is the voltage across the coils.  I will try increasing the pulse width until the current reaches steady state and see what is looks like and post it later.

I do not understand the current.

 

[Kerrowman]

That’s fascinating but look at the spikes - 18kV. I don’t see that in practice, more about 1kV. Would it be helpful if I showed you the actual scope trace for the gate of the mosfet?

Incidentally, I notice you do woodwork. I’m a wood turner and, if you’re interested, you can see some of my work on Instagram under the name: ‘kerrow_wood_turning’

 

[Kerrowman]

Is it relevant that each coil has a ferrite core? That certainly changes the B flux at the centre of each one.

 

[HarryA1]

It takes in the order of 200ms for the current to reach maximum in the simulator. The current through one coil is about 1.18 amperes. The voltage across the coils is about 200kv - you better wear rubber boots! Perhaps the simulator got carried away.

There is an equation: time = inductance/resistance = 63% of the charge. So t= 0.380h /10 ohms would give 38ms. That would give 0.74amps at 48ms (38ms + 10ms pulse on delay) in the second waveform. That seems about right; at the white dot.

 

[Kerrowman]

What time is 0.38 h? I’m not sure what that means.

The voltage at the main FET drain is definitely about 1,000V so I don’t know how you got 200kV.

 

[HarryA1]

It is some where in here:



So a henry = an ohm per hertz? see: https://en.wikipedia.org/wiki/Henry_(unit)

See the RL calculator here: http://learningaboutelectronics.com/Articles/RC-RL-time-constant-calculator.php#answer2

I will try it in the tina-ti simulator - we have a mutual dislike for each other! 

 

[Kerrowman]

The link with Time is that V= L.dI/dt so the Voltage produced is determined by the rate of change of the current for a given Inductance.

If 200kV is what is coming out of the simulator then something is not quite right with what’s going in to it, or an assumption somewhere.

 

[Hamza Yapici]

I am not well in English but I recommend you use the snubber circuit. A high-watt resistor, a high-voltage capacitor, and a fast diode must be included.