Use for the Tesla Switch

[DoubleD]

Well if you get the chance Dave I would like to see you schematic, if its possible to draw.

Cheers
Matt

Matt Here is the Latest Working Sch's that I sent to ASH that he posted on this thread earlier. One is the PWM control and the other is the actual TS circuit

ImageShack - Image Hosting :: sstsv51pwmob1.png
http://img88.imageshack.us/my.php?image=sstsv51aw8.png

This is the Complete circuit so if you want just the switching flow and control you need to look past the Floating Power supply using the Doides and such . The Diodes are NOT used for the actual Switching, they are in the control and power supply sections.

I can also send you the Sch's I used to get this working with too if you cant follow this one to find the tricky parts, the Prototype sch's are a lot more cluttered as there is a lot more components though

Dave

[DoubleD]

By using dedicated mosfet drivers as Tecstatic uses it is easy to switch the mosfets .
TC4420 or mc34151 can be used. The last one contains 2 drivers in one chip .
A good example how to use it is from Les Banki`s Pwm circuit intended for the Bob boyce cell.

Nvisser ...

Yes this is a great circuit for NORMAL fet switching in a Motor control, Inverter or similar type circuit, I use the 34151 quite a bit it is a very good driver.

But unfortunately this switching topology does not work in this Application due to the nature of how the TS really works.
This is what I was dealing with for a couple years from chip manuf. and various EE people I was working with to try to get this TS circuit to work.

You are more than welcome to build it as try though, maybe you can get it to work right on ALL the fets.

[nvisser]

Nvisser ...
You are more than welcome to build it as try though, maybe you can get it to work right on ALL the fets.

Point taken. I will study your circuit
Thank you for posting it!
This is maybe pushing it, but how about a quick circuit description
Also
How would you tune the circuit to the load and batteries

[nilrehob]

I wonder, is it really necessary with 4 batteries?

Isn't the idea just to shuffle current back and forth?
When you have two batteries in series they loose 1 amount of charge each, meaning you loose 2 amounts of charge, but only one amount of charge is arriving at the batteries in parallel where it is split between them,
So for every two amounts of charge you loose you save only one, just as if you where running it with one battery?
And even the potential is as if one battery?

So why not just use one battery and a H-bridge?
H-bridge - Wikipedia, the free encyclopedia



/Hob

[Matthew Jones]

It would seem that, that was the case if for some reason you looked at the convential in the circiut. But it doesn't end up working that way.

I know from experience that you can easily acheive 8x more work for the same amount of energy in a given charge range of the battery (IE 12.8 volts to 12.2 volts at a 4 amp discharge on a 4 amphour battery)

So if what you say was the truth you should really only get only 1.5 times more work done, not counting enviromental loss (Heat, BEMF).

I have built one with capacitors, but it lost faster.

Although the battery is treated as a capacitor in the convetional when used unconventionally they become far more.

I don't know how to make it anymore clear, other than to say build one with 4 batteries and see what you find.

Cheers
Matt

[Tecstatic]

Nvisser ...

Yes this is a great circuit for NORMAL fet switching in a Motor control, Inverter or similar type circuit, I use the 34151 quite a bit it is a very good driver.

But unfortunately this switching topology does not work in this Application due to the nature of how the TS really works.
This is what I was dealing with for a couple years from chip manuf. and various EE people I was working with to try to get this TS circuit to work.

You are more than welcome to build it as try though, maybe you can get it to work right on ALL the fets.

@DoubleD

Please enlighten me with a hint why it can not work with the Tesla switch.

Eric

[Matthew Jones]

I believe he is refering to the oscolation in the circiut. You have to have at least 2 point in which power can come out at seperate times.

Like this circiut
Astable Multivibrator.

Maybe I am wrong though.

Matt

[Tecstatic]

@DoubleD, all

What I need to know is why a circuit with the properties below cannot be used for a Tesla switch using eg. 4 x 7Ah SLA batteries :

1. The necessary number MOSFET switches are available.
2. Each MOSFET circuit is highly insulated from the other MOSFETs
3. Each MOSFET has a very low capacitive coupling to the other MOSFETs.
4. Rise and fall time in the order of 20ns.
5. The absolute timing precision of each MOSFET is better than 50ns, maybe even better than 20ns. (controller precision 8ns).
6. Each individual MOSFET can be operated fully independently from the remaining MOSFETs
7. Max voltage 600V
8. Absolute Max continuous current 60Amps, max pulsed current 230 Amps.
9. Rds on 45mOhm.
10. max dV/dt 50V/ns
11. Max drain-source leakage current during off: 10uA

Please enlighten me. Please be specific.

Thank you.

Eric

[DoubleD]

I wonder, is it really necessary with 4 batteries?

Isn't the idea just to shuffle current back and forth?
When you have two batteries in series they loose 1 amount of charge each, meaning you loose 2 amounts of charge, but only one amount of charge is arriving at the batteries in parallel where it is split between them,
So for every two amounts of charge you loose you save only one, just as if you where running it with one battery?
And even the potential is as if one battery?

So why not just use one battery and a H-bridge?
H-bridge - Wikipedia, the free encyclopedia



/Hob

Hob

Lets look at this in Normal circuit way to start with so it is not so confusing...

1st in order to "Charge " a battery you need a Higher Potential to feed it with.

With the 4 battery set up you will have the 2 supp;y batteries in series to give you a 24v Difference of Potential. Now the Load for this is actually divided between the 12v Device(Light, motor etc) AND the 2 12v batteries in parallel giving you the By the book net 0 at the Source batteries return.

So what is happening is as the Motor or what ever you are running is pulling electrons through the circuit scrubbing off 1/2 the potential the 2 parallel batteries "being technically Backwards" get the other 1/2 of the potential run through them in a Charge state.
Then when you reverse the circuit the now charged parallel batteries become the Series supply batteries and the electron charge is sent back through the system to re-charge the other set of batts AND do work on the way.

Hope this helps explain the Basics of the circuit

Dave

[DoubleD]

@DoubleD, all

What I need to know is why a circuit with the properties below cannot be used for a Tesla switch using eg. 4 x 7Ah SLA batteries :

1. The necessary number MOSFET switches are available.
2. Each MOSFET circuit is highly insulated from the other MOSFETs
3. Each MOSFET has a very low capacitive coupling to the other MOSFETs.
4. Rise and fall time in the order of 20ns.
5. The absolute timing precision of each MOSFET is better than 50ns, maybe even better than 20ns. (controller precision 8ns).
6. Each individual MOSFET can be operated fully independently from the remaining MOSFETs
7. Max voltage 600V
8. Absolute Max continuous current 60Amps, max pulsed current 230 Amps.
9. Rds on 45mOhm.
10. max dV/dt 50V/ns
11. Max drain-source leakage current during off: 10uA

Please enlighten me. Please be specific.

Thank you.

Eric

HI Eric ....

Ok I will try to explain this, It has been a while since I was in the "Figure this out stage" but I will give it a shot..

Looks like you have the basic essentials going.

What I have found is that the entire circuit needs to be Virtual as in nothing "sees" anything else directly.

As for the H-Bridge style wiring ... It is NOT designed (nor is any other Standard circuit I could find) to Block nor Switch Both directions with a DC type supply.
SCR's PNP's etc are too hard to turn off reliably here too.

Now I will just discuss one half of the circuit here, then just Mirror for the other side.

For one half there is 3 fets ..
1 for Series connection
1 for the + parallel
1 for the - parallel

Now one of these , Dang it here is where I forget , I may have to go back through t he circuit to confirm,well you can too once you know what to look for.
I THINK it was the -parallel fets ... it may be the + parallel too.
they are at either a LOW potential when turned on and the go High
or they are High and then go low..

** Edit
Ok I went back and checked the circuit .. it is the + parallel fets that have the issue and it is Switch them High and they go Low.
*

I think it is High to Low
so in order to turn them FULL ON you need a Potential that is ~18v Above what it is switching, Normal switching here so far,..............
BUT as soon as it is turned on the Potential that is being switched goes LOW and now you have 18v too much on the gate and blow the Fet !

With the, no power, transistor circuits out there you can get a way with this a bit , but this is why they have the HEAT problems with them.

Now to get around this you go VIRTUAL and the only potential the fets see is just what they are switching, there is no + or - it is just DIFFERENCE OF POTENTIAL that they see.

so for the driver and fet the - side is the fet SOURCE potential.
Now for the GATE Supply we pull the High potential from the 24v Parallel set of batteries on the other half of the circuit so now the DIFFERENCE OF POTENTIAL of the Gate is a Virtual ~12-18v above whatever the Source is.




So grab the Data sheet for the Driver chips and look over the Sch. and it should show up to you now.

Dave

[Tecstatic]

@DoubleD

Hi Dave

Thank you for the explanation.

A bit more information.

I have a "separate" switch mode power supply, which generates 0-3-15V for each separate driver circuit.

These supplies are insulated from each other, up to several thousand volts.

The controller generating the timing signals is going to be a micro controller.

The ISO721M ICs are transferring the timing pulses like an ordinary opto-coupler, it is just much faster, up to 75MHz and isolating 4000V peak, 560 V sustained working voltage.

So the gate voltage will always be in accordance with the ucontroller control timing signals, as long as the MOSFET source pin is operating within the 560 volts from uController ground sustained, or max 4000V for seldom occasions.

I got confused so I did a draw of the needed circuit as I see it.

@all

I have now attempted to be constructive, now the rest of you has to see if you can find any flaws. Please examine the attachment.

Here I have avoided all the driver circuits. They are known well functional circuits, so to make it much more readable I just consider the MOSFETs as switches.

Eric

[Tecstatic]

Having looked a little more on my new circuit, I think I get a recharge effect from the "switching" losses in the single switches if the sequence is this way, using the original attachment above.

2
1
4
3

6
5
8
7

Eric

[Tecstatic]

Yet another sequence

If we think of Bedinis statement, that a small amount of current is necessary to keep the batteries in charging mode, radient alone does not do the job, then this sequence might be better.

1. Turn on SW2
2. Turn on SW4 and SW6 simultaneously
3. Turn off SW2 (dual switch)
4. Turn off SW4 and SW6 simultaneously

5. Turn on SW5
6. Turn on SW1 and SW3 simultaneously
7. Turn off SW5 (dual switch)
8. Turn off SW1 and SW3 simultaneously


With this timing a battery "sees" this sequence:

1. Current draw when in serial
2. Radiant "spike" when the dual switch is opened and the internal MOSFET diodes of the "single switches" starts conducting.
3. Current charge in parallel, to "slam tight" the "fluffy surface charge" from the radient event.

I expect this charging effect to be strongest, if we just connect an inductor or a DC motor (and bridge).
The effect is probably weaker with a transformer, depending on the coupling factor of the coils and the magnetic circuit properties.

I have to try this in the near future.

Any comments ?

Eric

[DoubleD]

@DoubleD

Hi Dave

Thank you for the explanation.

A bit more information.

I have a "separate" switch mode power supply, which generates 0-3-15V for each separate driver circuit.

These supplies are insulated from each other, up to several thousand volts.

The controller generating the timing signals is going to be a micro controller.

The ISO721M ICs are transferring the timing pulses like an ordinary opto-coupler, it is just much faster, up to 75MHz and isolating 4000V peak, 560 V sustained working voltage.

So the gate voltage will always be in accordance with the ucontroller control timing signals, as long as the MOSFET source pin is operating within the 560 volts from uController ground sustained, or max 4000V for seldom occasions.

I got confused so I did a draw of the needed circuit as I see it.

@all

I have now attempted to be constructive, now the rest of you has to see if you can find any flaws. Please examine the attachment.

Here I have avoided all the driver circuits. They are known well functional circuits, so to make it much more readable I just consider the MOSFETs as switches.

Eric

I haven't done a full comparison of your switching but you seem to have the general idea. connect the neg between banks and run from the Pos side and you only need 6 fets if they are wired and switched right.

I tried the "Double" fet switch as it is the By the book way to do it there , but in real world it does not work , I don't remember what the deal was with it as it was about 3 years ago when I tried that method.

I have tried probably every normal combination out there in the last few years on this and the one I finally got to work is the one I posted. I don't / won't post things that I have not built, tested and must be in a replicable working form. there is too much of that "this should work" kind of crap out there already wasting every ones limited time and resources .

and as for the gate control and voltage, this is NOT the well known functional circuits that will work here. There can not be a so called Ground in this circuit nor is there an actual dedicated Supply it all has to FLOAT, the thing works on DIFFERENCE OF POTENTIAL not V+ and V- and the uP control has nothing to do with it, all it does it supply the optical trigger to set off the event.
Charge Pump high side switching scheme doesn't work here either.

Try to study the circuit I posted and understand what was done and why. It does basically appear to be a Normal circuit but it is not. When you follow what was done then maybe we can improve on it. uP control is the next step I will go with it at this point, so the control can be more precise and easily manipulated to set up and record the effects.


Later
Dave

The main thing is to make sure that the series connection is made before the parallel so that the Higher potential always flows as desired

[DoubleD]

If we think of Bedinis statement, that a small amount of current is necessary to keep the batteries in charging mode, radient alone does not do the job, then this sequence might be better.

1. Turn on SW2
2. Turn on SW4 and SW6 simultaneously
3. Turn off SW2 (dual switch)
4. Turn off SW4 and SW6 simultaneously

5. Turn on SW5
6. Turn on SW1 and SW3 simultaneously
7. Turn off SW5 (dual switch)
8. Turn off SW1 and SW3 simultaneously


With this timing a battery "sees" this sequence:

1. Current draw when in serial
2. Radiant "spike" when the dual switch is opened and the internal MOSFET diodes of the "single switches" starts conducting.
3. Current charge in parallel, to "slam tight" the "fluffy surface charge" from the radient event.

I expect this charging effect to be strongest, if we just connect an inductor or a DC motor (and bridge).
The effect is probably weaker with a transformer, depending on the coupling factor of the coils and the magnetic circuit properties.

I have to try this in the near future.

Any comments ?

Eric

Hi Eric..

OK Here Is my Take in this...

What you describe above does cover Bedinis charging stuff, but my take on this Switching circuit is WAY different, some of the same basic forces are at work here but in a different way, Kind of like with an IC engine Both the Diesel and the Gas engine work of the same type of energy (Heat transfer) they operate quite a bit different.

My theory on this (will need to get uP control to test and confirm)

The batteries are neither charged nor discharged..
when in proper tune it should do the following..
(I will treat this with all batteries described as 1 as the way the switching does this is not relevant for the theory explanation)

When a load is placed across the terminals of the battery Electron Flow does work as it travels from one post to the other, While the electrons are flowing outside the battery there is Ions flowing inside the battery, for ease of understanding the principal , we will say it is traveling from one plate to the other.
From what I have gathered from other peoples research (need to do the tests myself for proof) is that the electron flow is faster than the Ion flow.
So now the trick is ...
as soon as the electrons reach the opposite terminal the flow is stopped, now the ions in the battery start slowing down and have not yet "Reached the other plate" so to speak and now we reverse the potential and start the electron flow again, now the Ions are traveling back the other way but again we stop the flow as soon as the electrons make it to the other side and the Ions don't make it to the plate again, now reverse and repeat,etc.

Is what we should end up with is an Ionic electron Pump

Hopefully this makes enough sense to follow.

So as to prior ?'s to tuning etc, you can see now where the Load must control how hard the electrons flow in relation to what the batteries can handle and the switching controls how long the electron flow lasts and the Dead time for the Ions to slow down before reversing. If any of this is off then there will be Electron bunching, Ion collisions etc that will cause heat and drain the batteries, or overcharge and blow up the batteries.

when it works right there is energy drawn into the system as the fets will get cold and can frost up

Later
Dave