Tweet
Tesla Switch
Hi guys and gals,
Here's my re-type of my musings on the Tesla Switch, after losing what I wrote last night.
Like many of us, but not all, I became interested in this project after JB became invloved. I started with the 3 battery test, as JB suggested, to learn about differential potentials and which way the potentials are flowing. I learnt that if you place two 12V batteries in series and then connect them to a single 12V battery through a load, the the single 12V battery can charge fairly quickly without the two 12V batteries in parallel losing much in voltage. If you then rotate the charged 12V battery with the other two batteries, you can get all batteries charged up. This proved what JB had posted on his web pages for years, which I often looked at but never really understood until I'd done the experiment.
Ok, great. But how does this relate to the 4 battery Tesla Switch? According to folklore it should be as easy as adding a fourth battery to the 3 battery test and just switching between battery banks. In theory, all batteries should charge up without the need to rotate batteries around. Then the batteries could be used to power a load without the batteries ever having to be charged by any external source. Sounds easy, huh?
I began by replicating the original circuit that was printed in the Eike Mueller document from 1984, where Eike had spent a couple of days with JB after the Tesla Symposium to re-design JB's Tesla Switch that was originally designed for 5V NiCd batteries. My original design was using relays and a 555 timer to perform the switching between the two battery banks. This worked OK to a point, but I was losing a lot in driving the relays - as well as blowing up 555 timers regularly, probably due to bad hookups or bad connections. Also, I could not get the timing right and eventually the batteries ran down.
I then ordered a micro-controller to do the switching and timing to replace the 555 timers. I managed to get the circuit to work, although the best results I had, no matter what timing I used, was to get all batterires to very slowly increase their voltage by 0.01V every 10 minutes or so. Even then, the batteries would get to a point where they no longer charged, or ran down. I tried various different circuits, including the ones posted on JB's web-site and some other various circuits posted by other experimenters. Still, I could not get the circuit to charge the batteries like I expected they would. I also tried different types of batteries and different sizes, along the way varying the load and the timing.
I had already seen how quickly a particular battery would respond to a load in the 3 battery test and thought "why can't I get the batteries to charge as well as the 3 battery test?" (even with 4 batteries). I started to look at the circuit. Not being an EE, I wondered why did I need so many transistors and so many diodes to make the circuit work? It became fairly clear that with so many components, the losses within the components were having a drastic effect on the results I was expecting. No matter what type of load I used or the timing used to switch between battery banks, it made little difference to the results I was seeing.
Eventually, I became frustrated and looked at a different way to run the circuit. All I needed to do was to figure out the most efficient way to place two batteries in series and two batteries in parallel to get the most charge in one side without losing too much in the other side. So, I came up with a circuit that I thought would do that, which I have attached to this post.
The attached circuit does exactly that. I've removed all of the diodes that I've seen on the original JB circuit because I can't see a need for them so far. I also borrowed JB's suggestion of isolating the transistors from the switching curcuit to protect the components and added the parallel diode and capacitors to ensure fast and sure switching of the transistors. I'm not sure at this point in time if they are needed or not, but more testing will tell.
I've used a PIC controller to perform the switching because it's simple to change the timing "on-the-fly" and I don;t have to mess with 555 timers or the like. Currently I am not switching battery banks, because I am basing my experiment on the 3 battery switch and JB's suggestion of allowing the charging battery bank to become fully charged before swapping banks.
After 8 hours of operation I've seen a gain of almost 1.5V across all battery banks. Below is a snapshot of what I have so far:
Time B1 B3 B2 B4 Total voltage
12:00 12.68 12.44 12.52 12.57 50.21
14:15 13.22 12.92 12.25 12.49 50.88
15:07 13.27 12.99 12.23 12.48 50.97
15:56 13.34 13.06 12.23 12.47 51.1
17:47 13.48 13.26 12.20 12.42 51.36
20:09 13.58 13.38 12.16 12.36 51.48
The timestamp at 12:00 was taken with all batteries resting. Further timestamps are with the switch running on just one side. Of course, when the switch is turned off I expect to see B1 & B3 drop in voltage whilst B2 & B4 will increase in volatge once the load is released.
All batteries are Trojan SCS225 marine deep cycle batteries rated at 130Ah at the C20 rate. Normally each of these batteries would take over 6 hours to charge up to these volatges on a conventional charger.
Essentially my circuit is still the 3 battery test circuit, with a fourth battery placed in parallel. The circuit places B2 & B4 in series and at the same time palces B1 & B3 in parallel. The switching is set to pulse the low side (B1 & B3) for 1 second and then pause for 0.1 seconds. This allows the charging bank to "soak up" the charge, whilst also allowing a small rest period (0.1 second) to trick the source bank to think it does not have a load on it. The load I'm using is a 12V 55W quartz halogen lamp (aka MR16) that are commonly used in 12V downlights.
Well, I think I've bored you enough with the details, but it does appear to be working how I want it to work. My plan is to continue to allow the 12V bank to continue charging until either each battery gets to 14.5V, or the 24V bank becomes too low to support the load and the 12V bank will no longer increase in charge. This will probably take overnight to happen, either way - but considering the size of the batteries I need to be patient.
Please check the circuit for errors or if you think there could be improvements, otherwise feel free to replicate yourself. I hope to post a video or pictures in the next day or so, so if there is anything you want included, feel free to let me know.
John K.
P.S. I'm glad I got to get this posted before I stuffed it up again!
Hi guys and gals,
Here's my re-type of my musings on the Tesla Switch, after losing what I wrote last night.
Like many of us, but not all, I became interested in this project after JB became invloved. I started with the 3 battery test, as JB suggested, to learn about differential potentials and which way the potentials are flowing. I learnt that if you place two 12V batteries in series and then connect them to a single 12V battery through a load, the the single 12V battery can charge fairly quickly without the two 12V batteries in parallel losing much in voltage. If you then rotate the charged 12V battery with the other two batteries, you can get all batteries charged up. This proved what JB had posted on his web pages for years, which I often looked at but never really understood until I'd done the experiment.
Ok, great. But how does this relate to the 4 battery Tesla Switch? According to folklore it should be as easy as adding a fourth battery to the 3 battery test and just switching between battery banks. In theory, all batteries should charge up without the need to rotate batteries around. Then the batteries could be used to power a load without the batteries ever having to be charged by any external source. Sounds easy, huh?
I began by replicating the original circuit that was printed in the Eike Mueller document from 1984, where Eike had spent a couple of days with JB after the Tesla Symposium to re-design JB's Tesla Switch that was originally designed for 5V NiCd batteries. My original design was using relays and a 555 timer to perform the switching between the two battery banks. This worked OK to a point, but I was losing a lot in driving the relays - as well as blowing up 555 timers regularly, probably due to bad hookups or bad connections. Also, I could not get the timing right and eventually the batteries ran down.
I then ordered a micro-controller to do the switching and timing to replace the 555 timers. I managed to get the circuit to work, although the best results I had, no matter what timing I used, was to get all batterires to very slowly increase their voltage by 0.01V every 10 minutes or so. Even then, the batteries would get to a point where they no longer charged, or ran down. I tried various different circuits, including the ones posted on JB's web-site and some other various circuits posted by other experimenters. Still, I could not get the circuit to charge the batteries like I expected they would. I also tried different types of batteries and different sizes, along the way varying the load and the timing.
I had already seen how quickly a particular battery would respond to a load in the 3 battery test and thought "why can't I get the batteries to charge as well as the 3 battery test?" (even with 4 batteries). I started to look at the circuit. Not being an EE, I wondered why did I need so many transistors and so many diodes to make the circuit work? It became fairly clear that with so many components, the losses within the components were having a drastic effect on the results I was expecting. No matter what type of load I used or the timing used to switch between battery banks, it made little difference to the results I was seeing.
Eventually, I became frustrated and looked at a different way to run the circuit. All I needed to do was to figure out the most efficient way to place two batteries in series and two batteries in parallel to get the most charge in one side without losing too much in the other side. So, I came up with a circuit that I thought would do that, which I have attached to this post.
The attached circuit does exactly that. I've removed all of the diodes that I've seen on the original JB circuit because I can't see a need for them so far. I also borrowed JB's suggestion of isolating the transistors from the switching curcuit to protect the components and added the parallel diode and capacitors to ensure fast and sure switching of the transistors. I'm not sure at this point in time if they are needed or not, but more testing will tell.
I've used a PIC controller to perform the switching because it's simple to change the timing "on-the-fly" and I don;t have to mess with 555 timers or the like. Currently I am not switching battery banks, because I am basing my experiment on the 3 battery switch and JB's suggestion of allowing the charging battery bank to become fully charged before swapping banks.
After 8 hours of operation I've seen a gain of almost 1.5V across all battery banks. Below is a snapshot of what I have so far:
Time B1 B3 B2 B4 Total voltage
12:00 12.68 12.44 12.52 12.57 50.21
14:15 13.22 12.92 12.25 12.49 50.88
15:07 13.27 12.99 12.23 12.48 50.97
15:56 13.34 13.06 12.23 12.47 51.1
17:47 13.48 13.26 12.20 12.42 51.36
20:09 13.58 13.38 12.16 12.36 51.48
The timestamp at 12:00 was taken with all batteries resting. Further timestamps are with the switch running on just one side. Of course, when the switch is turned off I expect to see B1 & B3 drop in voltage whilst B2 & B4 will increase in volatge once the load is released.
All batteries are Trojan SCS225 marine deep cycle batteries rated at 130Ah at the C20 rate. Normally each of these batteries would take over 6 hours to charge up to these volatges on a conventional charger.
Essentially my circuit is still the 3 battery test circuit, with a fourth battery placed in parallel. The circuit places B2 & B4 in series and at the same time palces B1 & B3 in parallel. The switching is set to pulse the low side (B1 & B3) for 1 second and then pause for 0.1 seconds. This allows the charging bank to "soak up" the charge, whilst also allowing a small rest period (0.1 second) to trick the source bank to think it does not have a load on it. The load I'm using is a 12V 55W quartz halogen lamp (aka MR16) that are commonly used in 12V downlights.
Well, I think I've bored you enough with the details, but it does appear to be working how I want it to work. My plan is to continue to allow the 12V bank to continue charging until either each battery gets to 14.5V, or the 24V bank becomes too low to support the load and the 12V bank will no longer increase in charge. This will probably take overnight to happen, either way - but considering the size of the batteries I need to be patient.
Please check the circuit for errors or if you think there could be improvements, otherwise feel free to replicate yourself. I hope to post a video or pictures in the next day or so, so if there is anything you want included, feel free to let me know.
John K.
P.S. I'm glad I got to get this posted before I stuffed it up again!
(I'm remembering an earlier post about perserverance . . . way to go) - looking forward to 14.5V 4x4
It's the same one that I just posted are are currently testing.
Comment