Use for the Tesla Switch

[blackchisel97]

Hi Bit's
Each and every one is 10'000 uF @20V (=cheap) or 50V (initial testing)
So it's
80'000uF each bank
40'000uF @40V in series (to battery)
160'000uF @20V in parallel (from PV or PSU)

AND
they fill up to 18V not higher (ever)
And they dump to 55% "SoC" of the receiving battery, not lower (about 8V) (ever)

AND
usually they swing 10...12V back and forth (yep, 2V only)


Stevan C.

Very nice piece of work Stevan
It reminds me of John's patent, the one I had persistent thoughts about
I even made some changes and replaced Mosfets with MJL's (don't like mosfets) but didn't attempt to build yet and didn't share the circuit because it is modification of John's patent. I believe it is from 2003.


Vtech

[nvisser]

Brilliant work Stevan
Parallel to series. This is way he calls it the Tesla charger
Very clever switching using a npn on the high side instead of a pnp like in the bipolar switch.
I always wondered how to do that high side switching with npn bipolar transistors. I see now how it works as the battery + at that stage is at a lower voltage than the cap + side.
Your circuit also show what JB said about using the pnp with the npn to get a faster rise time, instead of using the parallel cap and diode.

[eternalightwithin]

Hi,
JohnK, stonewater, Lero, Genessc JangYD and other respected fellow researchers.

It's been quite a while.
To avoid being dull or boring, pick some questions, watch the vids on YT:
cest73
And i will follow up.
I do have a "kind of" TS PV Amp
1. It's unofficial
2. It's officially unsupported
3. my work is GPL, but I hereby ask JB's permission before going public.
4. private copies are permissible under non disclosure conditions only (free of charge)

Schematics I used are my own brainfart, but are strongly biased with what JB thought us here.

No guarantee, not fitness for particular application, (GPL "blah, blah"...)
Use at own risk, don't dare blame me etc...

anyone still reading?

I'm reading, definitely interested. Could you upload a higher resolution picture of your schematics. The component values are blurry. Especially on the first one.
Edit: Also your video show TL494, but schematic shows SG3525. Why did you switch?
Sincerely,

David

[StevanC]

switching the switch's switcher (LoL)

I'm reading, definitely interested. Could you upload a higher resolution picture of your schematics. The component values are blurry. Especially on the first one.
Edit: Also your video show TL494, but schematic shows SG3525. Why did you switch?
Sincerely,

David

David,
I switched for simple reasons really:
1. The TL494 is declared as a 1k cycle/sec "beast", while the SG3525 is declared as a 0.1kHz one. This alone provides advantages in slowdown.
2. The SG has a "soft start" that is brain dead" easy to implement, while 494 lack this (needs few more components to get it)
3. While I did have the mHz (miliHertz) going 494 at "zero Dead Time" it became unreliable on anything less (can't do 25%/25%25%25% at all on low speed).

So SG3525 wo in the end.
Why not SG3524 (SG1524) instead?
A. The 3525 is intended fro driving a fet = it has "source in" and "gnd" while giving output by CMOS switching it to source or GND - this could lend itself for a 4 channel setup later
B. The 3525 has a "bit" sharper rising edges than 3524, but that is of lesser concern for we drive a opto here
C. Most important: the 3525 IS declared as able to run _slower_ than the 3524, so i figured I become familiar with it, the sooner the better.

Although there is a better "instead" coming in soon (an AVR computer with usb for less than $7 piece)

The resolution is limited by the forum rules

*** stay tuned ***
Stevan C.

[StevanC]

FET vs BJT

Very nice piece of work Stevan
It reminds me of John's patent, the one I had persistent thoughts about
I even made some changes and replaced Mosfets with MJL's (don't like mosfets) but didn't attempt to build yet and didn't share the circuit because it is modification of John's patent. I believe it is from 2003.


Vtech

AS You can see, i quit "thinking" and lumped together what we did know/try:
each and every "switch" got a "driver" and each "driver" got a "opto".

But troubleshooting was really a nightmare, so I put LEDs in series to each opto-driver.

Then i could literally "see" current, and it was all clear to me:
When JB says "it's in the switching"
I believe it is in the A in to A out ratio:
Get it maxed out:
1A in -> 40A out
but on 48%/2%/48%/2% schedule
(the 2% deat time, for our switches are "imperfect" (=material, "real", and limited) )

An "ideal" switch would do:
1A -> infinite A
50% 50%
and probably ope an singularity:


I'm out ^ of here

FETs would need a separate 10V DC supply or each device (=nightmare)
Stevan C.

[StevanC]

wondering...

Brilliant work Stevan
Parallel to series. This is way he calls it the Tesla charger

I really wonder _why_ he calls it the Tesla charger?

Is it for the sheer unidirectional pulse the battery receives, Tesla was fighting so hard to explore?

Kind thanks, but I really have little credit in this here. It's JB who fought this to public attention.

Very clever switching using a npn on the high side instead of a pnp like in the bipolar switch.
I always wondered how to do that high side switching with npn bipolar transistors. I see now how it works as the battery + at that stage is at a lower voltage than the cap + side.

All i did is I begun to look at "what is _more_ positive and what is _more_ negative on a link connecting "two nodes".
"which side does current flow?" (we thing "poisitve to GND" or "HOT -> GND")

Luckily the BJT (MJL) is a current driven device (this is important here), but we need remember it needs some voltage to do so:
0.7V forward bias for base having current
1V for Vce flowing forward current
Then there is reverse breakdown (~16V on MJL)
But the BJT really is a "current" animal.

Your circuit also show what JB said about using the pnp with the npn to get a faster rise time, instead of using the parallel cap and diode.

The PNP I couldn't neither confirm nor deny. I don't really know.

He notes PNP one place, and I dug PNP drive and found Sziklay pair - the "complementary Darlington" and its PRO et CONTRA vs Darlington ("Vanilla Darlington" )

I really reconsider if I need it done "Sziklay" or a "complementary Darlington" for there is a difference how the compound device acts, and how it is "fed" to do so.

The key is both the rising edge, and the FVD (Forward Voltage Drop) for the whole transition.

Sziklay provides just the most "open" on "cue" as far as I know ATM.

But then there is the SCR and the "SCR instead" with a PNP-NPN pair, that is also able to "avalanche" sheer amperes over short time periods.

Best regards,
Stevan C.

Hopefully JB comes and comments regarding this soon, so I know what I do?

[ldissing]

Hi,
JohnK, stonewater, Lero, Genessc JangYD and other respected fellow researchers.

It's been quite a while.
To avoid being dull or boring, pick some questions

Hi StevanC,

I do still read this occasionally, and you do some nice work!

Pick some questions....Ok, and maybe a comment too.

I can not pick specific parts on the diagram you drew, because there are no part numbers, so describing it may be the best way.

Disclaimer: The questions below are not intended in any way as a criticism but only intended to help facilitate my understanding and to find out why you did what you did.

You decided to totally disconnect the capacitor banks from the power supply by using two additional "switches", i.e. the top capacitor (bank) positive is not directly connected to the power supply, but through a switch (the one to the 4th opto down from the top (or first one on the "red" led), and the the "switch" leading to the 7th opto down from top (or 4th one on the "red" led) which connects the negative of the lower cap bank to the negative of the power supply. 1. I was wondering why you thought you "needed" to do this? From a solar panel perspective, if you were charging some huge capacitor, I can see that when the discharging of the series capacitors is happening, the power supply capacitor could be charging and that would be a good thing. 2. Could you not have just used the top "switch" to disconnect the top cap bank from the power supply and connected the bottom cap bank directly to the negative of the power supply, thus reducing part count? 3. Do you think the capacitor banks need to be charged to the exact same level or something? 4. Are these "switches" I am speaking of above utilized as an attempt at a totally open looped system (i.e. negatives (and positives) of power supply and output are never connected at the same time)?

Thank you for your post and I await your thoughts!

Lero

[nvisser]

Stevan
I think this coming week I will replicate what you have done. I changed the diagram a bit by removing 2 of the switches and added 4 diodes. It looks a bit like the full TS. Can you go over the diagram if you don't mind and tell me if you think it will still do the same as yours.

[nvisser]

I am trying my best to try and understand how this type of charger can give you a gain.
Looking at its operation I see that it take much longer to charge the cap bank up to the input voltage than it takes to discharge it into the battery.
It discharge very quickly onto the low impedance battery and I don’t think the reading we see on the ammeters are true as the pulse are finished before the meter can respond properly.
I connected a 12V lamp inline on my relay par in, series out cap bank to see what happens.
Now when the cap charge up the lamp came on quite bright (18V), but when it discharged at now 36V it gave a very bright and quick light pulse.
If you go work it out with E= 0.5 CV^2 joules you will see that the energy in and out are the same.
Using 2 x 10 000uf caps you get:
For parallel charge : 0.5 x 20 000uf x 18V^2 = 3.24 J
For series discharge: 0.5 x 5 000uf x 36V^2= 3.24 J
I am not sure if the time plays a big role as discharging is only a small part of the charge pulse time.
Any thoughts??

[ldissing]

I am trying my best to try and understand how this type of charger can give you a gain.
Looking at its operation I see that it take much longer to charge the cap bank up to the input voltage than it takes to discharge it into the battery.
It discharge very quickly onto the low impedance battery and I don’t think the reading we see on the ammeters are true as the pulse are finished before the meter can respond properly.
I connected a 12V lamp inline on my relay par in, series out cap bank to see what happens.
Now when the cap charge up the lamp came on quite bright (18V), but when it discharged at now 36V it gave a very bright and quick light pulse.
If you go work it out with E= 0.5 CV^2 joules you will see that the energy in and out are the same.
Using 2 x 10 000uf caps you get:
For parallel charge : 0.5 x 20 000uf x 18V^2 = 3.24 J
For series discharge: 0.5 x 5 000uf x 36V^2= 3.24 J
I am not sure if the time plays a big role as discharging is only a small part of the charge pulse time.
Any thoughts??

Vissie,

It isn't about the joules in vs joules out. I believe the idea is to get the ions moving...and keep them moving. If you hit the battery ions with a pulse, and get them moving, then you hit them again before they stop moving, then you use less "juice" to charge the battery, but the battery still charges up fully with less input. You can do this with HV with small caps (small current) or big caps and LV (bigger current). But, you must keep the ions moving and if you can keep them moving along, you don't need a lot of "juice" to charge, you just keep them moving in charge motion.

JB talks about 125ms up to 1 second when talking about these devices and he showed on the video of the solar charger 125ms on, 125ms off. 125 ms, or thereabouts may be how long it takes after that particular pulse for the ions to stop moving in that particular battery he was charging. It may take longer, like 250ms or 1 sec. for them to stop moving, but if he hits them again, before they stop, they continue moving and charging the battery...he is getting something for less than he would have if he was just applying constant current. It still isn't "free", he had to do something to get what he was after and he didn't sulphate the battery in the process.

My 2 cents,

Leroy

[nvisser]

Thanks, it makes sense
I just wonder why Jb ask in the video "What do you think the gain of this system is?" and then he shows the input of 17.7V @ 3.2A (56W) and then the output of about 14V @10A pulses. (140W). The voltage output could even be much higher but are pulled down to the current battery voltage and of course they are much shorter pulses than the input pulses.

[ldissing]

Thanks, it makes sense
I just wonder why Jb ask in the video "What do you think the gain of this system is?" and then he shows the input of 17.7V @ 3.2A (56W) and then the output of about 14V @10A pulses. (140W). The voltage output could even be much higher but are pulled down to the current battery voltage and of course they are much shorter pulses than the input pulses.

How many apples can you get out of an orange? I think you would actually have to take the area under the curve to get the power in vs. power out, the meters are measuring wasted energy...I think JB said that. Anyway, it is not a one to one comparison and the 10As the meter was showing isn't a "true" measurement either, but the charger works...that is the point of the exercise and I'm sure those chargers will charge the hell out of the battery, so it is a mute point what the power in vs. out is/was. Made for a nice show though.

Leroy

[stonewater]

tesla switch radiant amplifier from energenx

hi everyone,

just got back from the conference in Idaho. saw JB's new tesla switch at work. it is awesome! it is a true amplifier that charges even in marginal light. I had the 20 amp model hooked up to a pair of 205 watt sunelec panels. at the end of the day the panels were in total shade, hidden by trees and the entrance to the lodge. the switch was still functioning.

[StevanC]

Sandbox type

Hi StevanC,

I do still read this occasionally, and you do some nice work!

Pick some questions....Ok, and maybe a comment too.

I can not pick specific parts on the diagram you drew, because there are no part numbers, so describing it may be the best way.

Disclaimer: The questions below are not intended in any way as a criticism but only intended to help facilitate my understanding and to find out why you did what you did.

You decided to totally disconnect the capacitor banks from the power supply by using two additional "switches", i.e. the top capacitor (bank) positive is not directly connected to the power supply, but through a switch (the one to the 4th opto down from the top (or first one on the "red" led), and the the "switch" leading to the 7th opto down from top (or 4th one on the "red" led) which connects the negative of the lower cap bank to the negative of the power supply. 1. I was wondering why you thought you "needed" to do this?

It is easy to short (bypass) a BJT, but it is much harder to put one in where a short is.
So i found all "nodes" and put a "switch" on each.
Therefrom it's all "sandbox"...

From a solar panel perspective, if you were charging some huge capacitor, I can see that when the discharging of the series capacitors is happening, the power supply capacitor could be charging and that would be a good thing.

I'm not so sure
IMHO it depends, and it _might_ be a good thing, but as well a bad?

2. Could you not have just used the top "switch" to disconnect the top cap bank from the power supply and connected the bottom cap bank directly to the negative of the power supply, thus reducing part count?

I take a clip-wire and do so. Guess what i Found out?

3. Do you think the capacitor banks need to be charged to the exact same level or something?

It seems to be desired: the "middle" BJT on the "series" side seems to like *equilibrium*. We seem to picture the potential/current_flow the wrong way around in this part of the circuit :
We consider the BJT "right" while "in Que", while i seems we should consider it at the moment it's just about to "que" - then it's all "normal", and, who says the BJT know's it's upside-down ?

4. Are these "switches" I am speaking of above utilized as an attempt at a totally open looped system (i.e. negatives (and positives) of power supply and output are never connected at the same time)?

Kind of, but no,
I just wanted a Sandbox type of setup, I could shunt with Diodes, SCRs and "shorts" until I find a way advantageous to others.

Thank you for your post and I await your thoughts!

Lero

Lero,


Those are them (thoughts),
Stevan C.

[StevanC]

The math

I am trying my best to try and understand how this type of charger can give you a gain.
Looking at its operation I see that it take much longer to charge the cap bank up to the input voltage than it takes to discharge it into the battery.
It discharge very quickly onto the low impedance battery and I don’t think the reading we see on the ammeters are true as the pulse are finished before the meter can respond properly.
I connected a 12V lamp inline on my relay par in, series out cap bank to see what happens.
Now when the cap charge up the lamp came on quite bright (18V), but when it discharged at now 36V it gave a very bright and quick light pulse.
If you go work it out with E= 0.5 CV^2 joules you will see that the energy in and out are the same.
Using 2 x 10 000uf caps you get:
For parallel charge : 0.5 x 20 000uf x 18V^2 = 3.24 J
For series discharge: 0.5 x 5 000uf x 36V^2= 3.24 J
I am not sure if the time plays a big role as discharging is only a small part of the charge pulse time.
Any thoughts??

So,
now, just how _far_ each of the two *same* joules can swing a *massive* object (the needle on the A meter)?

the _same_ energy, eventually less, certainly not _more_, can swing it *further* if we deliver it in shorter time.

TS is simply _proof_ to that.

It's in the leading edge, and the sheer rate of dump.

The same "goes further" IMHO applies to IONS inside the L.A.B. too.


Stevan C.