Use for the Tesla Switch

[Bit's-n-Bytes]

Don't give up just yet

Hi nvisser -

Bummer - 30 diodes? Ugh!

Well, I haven't fried that many but my stack is growing.

I had to put in safety switches to isolate and help
manage battery leads - that helped. I'm using a
beefier diode than the 1n4007 (more expensive, too) but
I have the short across the diodes D9,D10 as mentioned.

You've convinced me - back to 4 trannies - I surrender!

Thanks for sharing, better days are coming -

Plazma

Team, here is a working schematic now using the IRF3205's N-FET for the diodes at D9 and D10. Please note the FET at D9 is installed reversed bias. This works and I have a test underway as I am writing this. Here is the pulse rate that I am testing;

high 6
pause 1000
low 6
pause 5
High 4
pause 800
low 4

Thanks

Bit's

[nvisser]

Dont we also need two switches on the negative sides as the same shorting are going to happen there. The only difference is the small loads in line. The bottom batteries are still shorted out by the diodes and small load.
I wish John had time to sort this problem for us.

[Bit's-n-Bytes]

Dont we also need two switches on the negative sides as the same shorting are going to happen there. The only difference is the small loads in line. The bottom batteries are still shorted out by the diodes and small load.
I wish John had time to sort this problem for us.

Vissie, the loads (i.e small light bulbs) on the negative side is what cause current flow once the Potential rises. This is the potential we are capturing (or trying to capture) before the current actually flows. You could use any number of loads, but it seems the bulbs are a little more forgiving.

Hope this helps.

Bit's

[ldissing]

Fuses

I build the latest bedini circuit that was posted .
I blew 30 diodes and 2 scr`s!! Luckily I use cheap 1n4007`s.
The first time when some of the battery leads shorted while conecting the batteries
The second time when I conected one of the bateries the wrong way round.
And the last time when I switched the flipflop on and all short out as predicted
I made sure the duty cycle was on minimum beforehand at about 1 Hz. Only a spike but it is obvious enough to short the batteries.
I will try again tommorow with transistors as the diagram stated instead of scr`s.
If it happens again, I will have to put the 4 parralel transistors into the circuit.

Perchance we should start incorporating some fuses to limit IC destruction in the future? A fuse is much easier to replace than an IC and less expensive.

Leroy

[ldissing]

Dont we also need two switches on the negative sides as the same shorting are going to happen there. The only difference is the small loads in line. The bottom batteries are still shorted out by the diodes and small load.
I wish John had time to sort this problem for us.

@NVisser:

If we do not short the grounds through a transistor as in the Brandt circuit, then the current should be limited by the load. I do not believe that in the two (2) load case as shown by JB on his board, that we need the "bottom" transistors, just the inverted ones and the ones across the positives to diodes.

In my simulations (without the bottom transistors), citfta's two diodes allows a single load to work just fine, although there is one (1) more diode drop.

@Bits:

You moved to fets, why? I will go back and search for the answer as to why, but if you get back to me before I find it, then all the better. Thanks. I think that JB does not like FETs in this setup, but to each his own.

Leroy

[Bit's-n-Bytes]

@NVisser:

If we do not short the grounds through a transistor as in the Brandt circuit, then the current should be limited by the load. I do not believe that in the two (2) load case as shown by JB on his board, that we need the "bottom" transistors, just the inverted ones and the ones across the positives to diodes.

In my simulations (without the bottom transistors), citfta's two diodes allows a single load to work just fine, although there is one (1) more diode drop.

@Bits:

You moved to fets, why? I will go back and search for the answer as to why, but if you get back to me before I find it, then all the better. Thanks. JB does not like FETs, but to each his own.

Leroy

Cause D9 was directly shorting to D1 through D3 when Q1 fires.

I may still have the problem with the FET at D9, But I am troubleshooting.

BTW, how's your build coming? are you testing yet?

Bit's

[ldissing]

Cause D9 was directly shorting to D1 through D3 when Q1 fires.

I may still have the problem with the FET at D9, But I am troubleshooting.

BTW, how's your build coming? are you testing yet?

Bit's

I will be home and creating my circuit board in 3 days time. It is all laid out and ready to go, just have to get home and get it done. I have all the parts, etc. Can't wait to get back into the swing of things again.

Leroy

P.S. This is the fifth iteration of the board layout for me...and I still haven't built one yet.

[ldissing]

Cause D9 was directly shorting to D1 through D3 when Q1 fires.

Bit's

My question was more along the lines of, "Why did you use a fet INSTEAD of a transistor?"

I'd opt for more control instead of less. You have more output pins on the PCIAXE you are using, so you could use multiple optos and then turn them on and off in specific order to get the reverse bias transistor turned off first, then the upper Q10 off, then Q9 on, then the reverse biased Q2 on, etc. The H11D1s aren't that expensive and you already have the controller, so it is a no brainer to do it. I think Dave had some things right in his controller which you could do with the PICAXE. There is no reason that I know of for each one to be turned off simultaneously (and not in some specific order), but JB might beg to differ with me, because I'm just assuming things right now...but not for long. I need to order that controller board, but was planning on just using the 3524 for now. It was working fine for me, but I'd love to be able to change things without soldering, etc. Code is easy to change, components not so much.

***EDIT
The SCR would turn off if Q10 was shut off (and OR if Q2 (SCR2) was turned on because the potential is now roughly the same, so no current can flow). So if you used a SCR, you could turn off q1 (SCR1), turn off q10, then turn on q2, then turn on q9, and bingo, no SCR latching issue, I think. JB says it will turn off, which it will if Q10 is shut off and Q2 turned on because no more current can flow once the potentials are even (minus diode drops, of course).
***EDIT

Leroy

[Bit's-n-Bytes]

My question was more along the lines of, "Why did you use a fet INSTEAD of a transistor?"

I'd opt for more control instead of less. You have more output pins on the PCIAXE you are using, so you could use multiple optos and then turn them on and off in specific order to get the reverse bias transistor turned off first, then the upper Q10 off, then Q9 on, then the reverse biased Q2 on, etc. The H11D1s aren't that expensive and you already have the controller, so it is a no brainer to do it. I think Dave had some things right in his controller which you could do with the PICAXE. There is no reason that I know of for each one to be turned off simultaneously (and not in some specific order), but JB might beg to differ with me, because I'm just assuming things right now...but not for long. I need to order that controller board, but was planning on just using the 3524 for now. It was working fine for me, but I'd love to be able to change things without soldering, etc. Code is easy to change, components not so much.

***EDIT
The SCR would turn off if Q10 was shut off (and OR if Q2 (SCR2) was turned on because the potential is now roughly the same, so no current can flow). So if you used a SCR, you could turn off q1 (SCR1), turn off q10, then turn on q2, then turn on q9, and bingo, no SCR latching issue, I think. JB says it will turn off, which it will if Q10 is shut off and Q2 turned on because no more current can flow once the potentials are even (minus diode drops, of course).
***EDIT

Leroy

First off, Cost, just trying to keep it simple. Secondly, your theory is sound, however the current flowing through D3 back to D1 cannot be stopped by Q10 so the SCR remains gated. So far the FET's have allowed proper switching (which by the way, I have installed properly biased) and am now chasing the same type of short on the negative leg. Thanks Vissie for pointing this out. There is a fundemental flaw in the way the diodes are steering at the moment. So more troubleshooting at hand. I am very excited for you to get yours built and join the ranks of testing.

Bit's

[ldissing]

First off, Cost, just trying to keep it simple. Secondly, your theory is sound, however the current flowing through D3 back to D1 cannot be stopped by Q10 so the SCR remains gated. So far the FET's have allowed proper switching (which by the way, I have installed properly biased) and am now chasing the same type of short on the negative leg. Thanks Vissie for pointing this out. There is a fundemental flaw in the way the diodes are steering at the moment. So more troubleshooting at hand. I am very excited for you to get yours built and join the ranks of testing.

Bit's

I agree that Q10 (your schematic) can not stop current from D3 back to D1, but Q9 can stop it. If Q9 is off, then no current flows from D3 to D1 or D5. Or, in other words, Q9 is the gate for current from D3 to D1 and D5. Q10 is the gate for current to D3 and D7. The transistor is the gate, so when the transistor is shut off, the SCR has no current going through it, so (if the gate signal is off), it WILL shut off.

What you have drawn is JBs circuit with FETs minus the extra optos and a PICAXE controller. When JB was showing the simplified circuit, he drew one side and that works, no doubt. When he drew the other side and connected it without the transistors, that won't work, the transistors (or fets) need to be there so you don't get the feedback. Maybe I'm still missing something, but I know that with the transistors in there and the SCRs as the reverse biased Q1 and Q2, it works. Hopefully you guys will iron all this out, so I don't blow up my board, but I do not see a problem with the circuit...and it simulates properly too. He had an extra diode in there, but it is still not enough. The D3 diode must be gated and the same for D1, but again, I should just shut up and build it and try it for myself. My previous circuit did work though, so I know it will work just fine.

If you had more signals from the PICAXE and optos, you could test this. Just turn on both of the SCRs, and leave (T9,FET9)) and (T10,Fet10) off. I do not think those SCRs will stay on which can be tested by looking at the voltage across the two batteries in series. Anyway, you are further along than me by almost a month, so I'll leave it to you.

Leroy

[Helijason]

PIC Micro?

@ Idissing

Which PIC did you decide to go with or are looking at? I'll be working with plain PICs as well and would like to be on the same page as someone else (I'm starting from scratch, have programming tools now, still looking for a scope). As I recall you were going to use a PIC with 2 PWM.

Jason

[ldissing]

Pic

@Jason:

Product Search - microchipDIRECT

Part Number : DV164121 - PICkit 2 Debug Express

This kit is ready to go, with some space for placing optos, but you have to be able to program in C. I'm an old hat at that, including interrupts and all that, so this is what I'm looking at. It also has PWM, ADC on it, and multiple IOs, so it is pretty handy. I have a friend using this on a CNC controller board.

Initially, I'm going with the PWM chip recommended by JB, but eventually, this will probably be what I use. I think the chip that Bit's is using is okay too, but it does not have PWM on it already as far a I know, but I haven't looked at it that much. Not that you necessarily need it, because you can just turn on and off outputs, but it would be nice to have already on board. Set it up and it goes....

Leroy

P.S. I'm kind of a control freak...I like to tell it what to do and I want it to do exactly what I tell it. In C, I can do most anything...

[Helijason]

@ Leroy

I'm with you on the control freak aspect of things. I love being able to modify the behavior of the circuit on the fly. I've seen in industry (manufacturing) sometimes you need a lot of bells and whistles to understand what you have, then it can be made simplier.

I agree that using the pic pwm hardware is probably the way to go about normal switching. Though I do like the idea bit's proposed (leaving the switch on one way till a condition is met). So far my concern with what bit's is doing is how consistant is the programmed pwm. I'm afraid if there is too much additional code that it will not be consistent. Have you programmed pics with c before? How deterministic was the c code? The last time I programmed pics it was in assembly and have seen the trouble that one can get into. For example, I knew a guy that coded a serial port with code. He was constantly fighting loops and extra NOPs through out the code to get the timing reliable.

Jason

[Plazma]

C on a PIC . . . PWM?

Hi Jason,

I've been running various time-critical instrumentation applications for the past 2+ years (incl PWM and PID control loops) as programmed in C. Those loops are deterministic - but I have been very careful in managing environment on the PICs I've used - that mostly means setting interrupt priorities correctly, tracking loading, and turning only certain interrupts on at certain times. On some apps where I had a lot of GUI and timed tasks, time-critical loops were put into a dedicated PIC and I had the 2 PICs microwired (a divide-to-conquer strategy). Both PICs were programmed in C.

That said, I have NOT run extremely low frequency (<= 61 Hz PWM) as the internal clock sources on my PIC for PWM/Comparator firmware are not scaled (partly by choice) to go down to the 1-2Hz or lower regions. This is not a C language/compiler problem, and not even a major firmware problem as I can easily code a timer-interrupt driven routine to accomplish what I want at the very low frequencies for TSw, and with as many IO pins and optos as I see fit to study the dynamics of the TeslaSwitch problem. And this is what I am currently working on to test, both hardware and software.

As I related in an earlier post, there are lots of good tools available - a good C compiler is just one of many - and there are quite a few viable approaches that we are sharing to deepen our collective TSw understanding.

All the Best,

Plazma

@ Leroy

I'm with you on the control freak aspect of things. I love being able to modify the behavior of the circuit on the fly. I've seen in industry (manufacturing) sometimes you need a lot of bells and whistles to understand what you have, then it can be made simplier.

I agree that using the pic pwm hardware is probably the way to go about normal switching. Though I do like the idea bit's proposed (leaving the switch on one way till a condition is met). So far my concern with what bit's is doing is how consistant is the programmed pwm. I'm afraid if there is too much additional code that it will not be consistent. Have you programmed pics with c before? How deterministic was the c code? The last time I programmed pics it was in assembly and have seen the trouble that one can get into. For example, I knew a guy that coded a serial port with code. He was constantly fighting loops and extra NOPs through out the code to get the timing reliable.

Jason

[Murlin]

Holliday Break

I see some exciting things have been happening since I have been gone. Christmas slowed down my progress.

I should resume building tonight. Finally getting a chance to solder in the diodes and hope to test the new mods on my commutator design shortly after.

regards,

Murlin