<<< Universal Electronic Switch >>>

[Matt Watts]

This project is something I determined was needed to allow us to focus on other principles without the distraction of building custom pulse drive circuits.  My goal was to develop an electronic switch that has the following characteristics:

• Digital logic control input
• High speed operation
• Fully isolated signal and power
• High voltage and current capability
• DC and AC switching similar to a true mechanical switch

What you will see in the attached images is the result of this effort, minus the heatsink.

[Matt Watts]

A couple things I have bumped into during the development cycle of these UniSwitches that I will likely add into a future revision:

• A slightly larger board with corner holes drilled and nylon standoffs screwed in place
• R3 that limits the drive current to the opto-coupler should be a 2k adjustable pot
• Test point headers need to be added for easier scope probe attachment
• Some routing work to get the switched signals as far away as possible from the control signals
• Redo hole sizes so that component removal is slightly easier

[Matt Watts]

Attached are the component spec sheets I worked from for the design.

Most everything was acquired from Digi-Key and the boards manufactured at Bay Area Circuits.  Schematic capture and layout was done using DipTrace.

Component source and pricing:

D1/2.  http://www.digikey.com/product-search/en?KeyWords=STTH6004W 2 @ $5.83
D3.  http://www.digikey.com/product-search/en?KeyWords=CP41B-WES-CK0P0154-ND 1 @ $0.34
Q1/2.  http://www.digikey.com/product-search/en?KeyWords=FGH60N60SMD-ND  2 @ $7.93
U1.  http://www.digikey.com/product-search/en?KeyWords=516-1651-5-ND 1 @ $3.92
U2.  http://www.digikey.com/product-search/en?KeyWords=CLA356-ND  1 @ $5.12
U3.  http://www.digikey.com/product-search/en?KeyWords=DG06S1212A  1 @ $20.16
J1/2/3.  http://www.digikey.com/product-search/en?KeyWords=ED1605-ND  3 @ $0.75
0.1uF Caps  http://www.digikey.com/product-search/en?KeyWords=399-4610-ND  3 @ $1.28
socket  http://www.digikey.com/product-search/en?KeyWords=ED56083-ND 1 @ $1.58
1000uF Cap  had on hand.  1
Resistors  < $0.50 each   4

PCB  ~ $25.00 each

I come up with just under $100 a pop at $92.73.  Obviously this would be cheaper to buy components in bulk and to do a little more shopping around.  Still though, if you can find a $100 hockey puck type switch that can perform this well, by all means, let me know.

[Matt Watts]

Attached is a zip file containing the DipTrace project files needed to replicate this project.

[poynt99]

Just curious, is the switch completely isolated? Just wondering as I see a cct gnd on the igbts.

[Matt Watts]

Just curious, is the switch completely isolated? Just wondering as I see a cct gnd on the igbts.

It is isolated on the signal side with the optical coupler and on the power side with the DC to DC converter.  So the GND you see is only internal and no way connected to anything external.  GND may not have been the best choice of bus type.  I suppose I could use V_EE or V_SS depending upon the final drive devices.

I did some tests already switching AC mains and found that it works as intended.  The heatsink will need to be electrically isolated with something suitable though.

[poynt99]

In terms of voltage, how low of a sine wave will it pass undistorted?

[Matt Watts]

In terms of voltage, how low of a sine wave will it pass undistorted?

Good question.  I'll have to run some tests and see.  The diodes drop around 0.8 volts, so one can expect at least that much distortion for sure.  I certainly wouldn't attempt to use these to switch audio signals, but with some tweaks an alternate version could probably be built to do much better.

I looked at a lot of those so-called hockey-pucks and none of them provided what these switches offer; the ones that came close were prohibitively expensive.

[Matt Watts]

Video intro:
http://vimeo.com/103968642

[Centraflow]

Very nice Matt

What is the shortest pulse width it can do? or did I miss that somewhere?

Really is nice and compact

regards

Mike

[Peterae]

Indeed
Nice work Matt, i would also be interested in the rise and fall times when switching a resistive load.

[Matt Watts]

Thanks guys.

I'll need to setup my pulse generator and scope with the external sync and try to find the limits this thing has.  I would suspect voltage and amperage will change the rise/fall times, so do you guys have any specific value of interest you would like me to focus on?

I think I'll start with rectified and filtered DC at about 160 volts using the same lamp as the load.  As I recall the initial amperage against an extinguished tungsten filament is pretty high.  I'll work at it until I'm certain it can do 200ns.  May take some time so please bear with me.

[Peterae]

A scope shot of a pulse of between 30-50V into a 10 Ohm resistor would show me what i need to know for my application.

I'm always on the look out for a fast driver and fet/igbt, the specs on the data sheet looks good 

Thanks
Peter

[Matt Watts]

A scope shot of a pulse of between 30-50V into a 10 Ohm resistor would show me what i need to know for my application.

Setup using 30+ volts with 8 ohm load.

http://vimeo.com/104149605

Cheers!

[TinMan]

Nice build Matt.
Would this work as an AC induction motor speed controller?.

[Matt Watts]

Nice build Matt.
Would this work as an AC induction motor speed controller?.

Thank you TinMan.

I see no reason why it would not.  If you look at my first video where I was strobing the AC mains into a light bulb, clearly the power is being chopped into little discrete pieces but the main AC 60 cycle signal is still there so the motor "should" still sync on that.

I was asked by another gentleman if the inductive kickback would destroy the UniSwitch.  To be quite honest, I don't know.  But I did propose putting two of these switches on each side of the inductor and keep their input signal synchronized, that way when they both turn off, the inductor is completely disconnected on both legs from the power source--just like a double pole, single throw mechanical switch.  In this scenario, it would seem to me electrostatic capacitance would be the only means of destroying the semi conductors.  Get that far and the next thing you could try is adding a third switch that shorts the inductor--turns on when the other two switches turn off.  Seems to me this would hold the magnetic field in the inductor during the off-time transition.  Time it right and the inductor would never get the opportunity to develop that high voltage inductive kickback.  What would actually happen...?  Someone needs to try it and found out--might lead to a reduced Lenz Effect motor controller.

[Peterae]

Hi Matt

Thanks for doing the test, good results  

That's a handy switch you have designed & built  

Cheers
Peter

[TinMan]

Thank you TinMan.

I see no reason why it would not.  If you look at my first video where I was strobing the AC mains into a light bulb, clearly the power is being chopped into little discrete pieces but the main AC 60 cycle signal is still there so the motor "should" still sync on that.

I was asked by another gentleman if the inductive kickback would destroy the UniSwitch.  To be quite honest, I don't know.  But I did propose putting two of these switches on each side of the inductor and keep their input signal synchronized, that way when they both turn off, the inductor is completely disconnected on both legs from the power source--just like a double pole, single throw mechanical switch.  In this scenario, it would seem to me electrostatic capacitance would be the only means of destroying the semi conductors.  Get that far and the next thing you could try is adding a third switch that shorts the inductor--turns on when the other two switches turn off.  Seems to me this would hold the magnetic field in the inductor during the off-time transition.  Time it right and the inductor would never get the opportunity to develop that high voltage inductive kickback.  What would actually happen...?  Someone needs to try it and found out--might lead to a reduced Lenz Effect motor controller.

Well that would be handy if it did work as a speed controller,but thinking on it a bit more,i believe that the hurtz(frequency) of the AC wave form actually has to be reduced in order to slow the speed of an AC induction motor?.I think by keeping the 60 Htz and chopping the wave,the motor would stay synced until we got rotor slip. The result would probably be lots of noise and vibration from the motor-along with a rapid loss of torque.


You could just place a 150 volt neon across the inductor to burn off most of  the inductive kickback. This should kill it enough so as it dosnt do any damage else where.

[muDped]

Excellent project Matt Watts!

High speed isolated switches such as the one you've
fabricated are very useful devices.

Switching with variable pulse width at a high frequency
such as shown in your video is an excellent way to
emulate electronically the voltage control capability of
the Variac.

Your switch could easily, as you hinted at in your video,
be adapted to make an effective desulfator/charger for
lead acid batteries.  Narrow pulses at a frequency of
One to Three KiloHerz does the job well.  With very
narrow pulses the average power delivered to the battery
would be low and quite safe.

Yes indeed, lots of creative possibilities...

[forest]

Excellent

I also concerned about flyback spike (kickback)  possibility to damage circuit. What can be done to avoid it yet still allow to experimenting with BEMF ? Now I think would be good to have a simple square wave generator. Just 50% duty cycle is eanough for me , but what I think is more important is to have a way to release just one single pulse with pressing a button

[Matt Watts]

Excellent

I also concerned about flyback spike (kickback)  possibility to damage circuit. What can be done to avoid it yet still allow to experimenting with BEMF ? Now I think would be good to have a simple square wave generator. Just 50% duty cycle is eanough for me , but what I think is more important is to have a way to release just one single pulse with pressing a button

Thank you for the compliment.  I've used these (built two of them) quite a bit lately and they seem pretty robust.  Unlike a typical MOSFET switch, these are dual and in series so they switch the current in both directions; that could be why I haven't blown them up yet.  With a flyback spike, the energy has to go somewhere and with these switches, they completely disconnect from the circuit when they turn off, so I don't think they get smacked like a single transistor would.

I've tried them on some pretty big inductors with success so far.  Can't quite understand why Mr. John B. doesn't use something similar with his SG.  From the way Mr. Lindemann explains the SG, it's real easy with their design to pop semiconductors.

Using IGBTs, they do warm up a bit if you punch 30 amp spikes through them, but you also have the flexibility to push 300 volt at a 1/2 amp instead if that better matches your output impedance.  They're about as Universal as I know how to make them.  Not super fast, not super high current; not super high voltage, just something in the middle that can be used for lots of testing.