Why Not ?

[[email protected]]

At

https://dl.dropboxusercontent.com/u/29948706/tripl.jpg

there is a screenshot of a circuit that is a bit screwed up because I don'tknow how to quite use LTSpice yet. The letters are in the way, but really it is still readable.

It is a basic voltage tripler I designed years ago to feed an audio power amp.

I get this everybody uses transformers everybody uses transformers but why not this ? You know when a capacitor is a filter is still charges and discharges just the same, the only fdifference here is like with a coupling cap,it is floating.

Is there really a material difference ? If not, this should work. There is no regulation at all, and none needed. It is simple brute force. It puts out triple what is applied. At the bottom it is fed the usual 12 volts from the battery. Then 24 volts is developed. (actually in the original design, if the input voltage went above about 16 volts the drive to the outputs would switch to in phase and it would only double the voltage)

The four capacitors are key to the voltage multiplication. I have considered making them easily replacable if this circuit is used. Actually I have built it partly, using good CGS caps of like 15,000 uF and 1N6095 diodes. I intended to use I think 2N5883 and 5885 transitors, low voltage units with alot of current capacity. Everything to minimize voltage loss. In an audio amp, every volt is important because of the watts you lose.

The amp I had oin mind for this was BTL runniing off the single ended supply, should have been maybe fifty watts into eight ohms but really, if those are really good capacitors it would have no problems with lower impedances.Also, with 18 volts on the speaker leads I came up with a pretty dandy ampprotection circuit usiing an SCR and some hefty rectifiers to clamp it alldown. In "short", if you connected a speaker wire to the car battery it would burn the wire off and not be damaged. The protection would be reset simply by turning the power switch to off, from either on or auto.

For the hell of it I think I amk going to draw out the rest of what I had. It was all hand drawn and in a notebook and got lost in a car wreck many many years ago. (I was changing the station and grabbing a beer at the same time I think, no more of that shit....)I know that since then many car amps have come out with better performance etc., but this is one of those just for the hell of it things.

I also had alot of cool shit designed for it and actually planned to put itto use but build the new features in as time passed. I would have left room in the unit for other boards. "Other" boards would include a tone controlcircuit. The bass I remember, 18 dB range, with the turnover variable from48 to 440 HZ. that's just how the numbers worked out actually and it was atotally unheard of circuit. It consisted of an actual crossover and the bass control was simply a level control, like a partial bi-amping in a way. Treble was handled similarly. Low and high mid controls were not parametric and had less range. I saw 18 dB as a bit too much for them. At the time.

Anyhoo, right now, about this power source, you can use all the transformers you want but somewhere there is still a cap charging and discharging if you want to kick up the voltage. I just eliminated the transformer. Think about the advantages in the circuit really, no freewheelers, no dampers, snubbers, any of that. All gone. Use any frequency you want, I was thinking like 6Khz. Why not ? No measures needed to switch transistors faster, drive optimization, nothing.

And then with the drive, as you can see coming in on the left of course oneinverse the other, it would damnear put out DC already without any additional filtering. Even more, who needs high frequency given this ?

I know it is not the greatest thing since sliced bread (which I don't thinkis all that great....), but it lacks certain problems involved in the design of a transformer based supply.

I originally designed it with the amps common collector, which meanss no DV/DT problems or anything of the sort, in fact the power supply part, because of that, was submersible. It's not like all the impedances were zero, butthe bases and emitters were tied together. No bias was needed of course. The thing was designed to be F_____g indestructible.

Your thoughts ?

 

[[email protected]]

There's no reason for it not to work. I fed it from another source as I didn't have the transistors in it yet and it did triple the voltage.

The OPAMP symbols were just used for convenience, they are actuall comp pairs of high current TO3 transistors, maybe 2N5885 and 5883, I think....

The drive coming in the left is two square waves, one inverted. Consider what happens with drive.

In your version (thanks for cleaning that up) when the drive goes high fromV2, while V3 is low. V3 drives U5 low which charges C1 via D6. When the drive reverses, U5 is driven high whgich dumps the charge from C1 through D5 into C2. When the drive reverses again U4 is driven high and through C2 andD4 feed the output. At this time C1 is picking up a recharge from the 12 volt line and will be ready for the next cycle.

The exact same thing is happening on the other side with U1 and 2, and C3 and 4, but in the opposite phase. This results in the output only having ripple as long as the switching time of the transistors. This of course eases output filtering requirements, the bigggest caps are in the doubler stack.

It is really just a doubler, but the output is stacked on top of the input which means the output is triple.

I know why many might think it won't work, because they expect to see one more diode in each stack and a filter. That would be easy to comprehend, butI found there is no reason to make that intermediate DC. Why ? As long as the diodes conduct when they are supposed to it works.

That means at the junction of D4 and 5, there is a 12 volt peak to peak square wave riding on 12 volts DC. This makes 24 volts total which is added tothe original 12 volts.

The duty cycle of the drive is strictly 50/50, however I did consider making it three phase for more current. In that case there would be three drivesand the duty cycle would be one third each. That would require separate invertors for each phase, and a simple ring counter to develop the waveform. What happened with that idea is that no math I know of could justify the added complexity. It would be easier to just use beefier components in the two phases.

So Spice says it won't work or do you ? It does work, perhaps Spice needs input of what the drive actually is. Can that be done ? I just put in that it is square waves ? Right now how the heck would iit know how the choppers are driven ? Of course it doesn't work without drive, which in my old set of drawings called "Xdrive". Pretty easy to figure why, inverted two phase and all, and how the thing works, I thought it was a pretty good name.

I am pretty green on Spice, I am probably going to go take in a couple of tutorials on it. I just got it last month and have been playing with it. I drew this and everything else that went with it and had every waveform figured out and almost every component value. I lost it in a car wreck that I really don't want to go into. Just say it was a very bad day.

At this ppoint I think I want even more current out of it. I want to drive a pair of amps using center tapped output chokes. If I get the 36 volts andquadruple it that way, that makes a pretty good amount of power into eightohms. I know they got car amps out there that are pretty damn good, but alot of them are rated into four or two ohms, some into one ohm. I want this into eight ohms.

I would have never told all about this if I thought there was a market for it. The market has gone the way it has gone. I want something a bit more audiophile, and the output chokes would give it a nice "sound". You know someMcIntosh amps used output chokes like that. This would be for me, and somespecail order maybe. It is not destined for Bestbuy to say the least.

Driving audio outputs into the center tapped choke presents its own set of problems but they are not all that hard to solve. One way would be to just use MOSFETs. Another way is to use the proper transformers, and that is even better. The only problem with that is it complicates the main negative feedbak path. I can deal with it.

I actually have output chokes for this. They are actually output transformers from tube amps. The main requirement is that they have 4 ansd 16 ohm taps. They do. Betweeen ground, 4 and 16 is centertapped. I have already builtamps for cars using them and their perfomance actually exceeded expectations. Something about those chokes, it's like soft clipping or whatever. The previous versions however only ran off the 12 volt supply. They'd give youisome watts with no DC convertor at all, now I want to add one.

You know another way might be to use PNP aaudio outputs. I'm not sure but Imight have four 2N5436s. These babies have 60 mil pins ! The problem is I don't believe they can take the Vce. Running off 36 volts, they will have to take at least double that. Accounting for a little headroom anything lessthan 100 Vce is no good. In fact even if I had 5885s and 5883s, they moghtbe good for the tripler, but not the output. I might look them up.

Doesn't matter, there are plenty of transistors out there. I think at the moment I'll look for something for the tripler, build the drive circuit for it. That should be a piece of cake, I'll just use a regular BTL amp chip and set it oscillate at about 6 Khz. I think that a good frequency because even slow ass transistors will work fine, and even square enough waves to keep the ripple way down will cause little or no interference. Note that without transformers in the PS, there is practically no EMI.

I thought it a decent idea. Maybe not the best, but workable.

 

[Lasse Langwadt Christensen]

Den tirsdag den 8. oktober 2013 00.07.25 UTC+2 skrev John Fields:

---

Hmmm...



Doesn't quite seem to work.

try something like 6V on the inverting input of the opamps then you might
get a square wave on the outputs

and reduce the capacitor sizes

-Lasse

 

[[email protected]]

Thanks all.

I intended on square waves all along. I think this was just not drawn righton my part after seeing the exmples. I should have just did it on paper and scanned it....

So it is a Dickerson eh ? When did that come out ? I "invented" it a damn long time ago, I mean I was half my age or less I think. I'm betting that itwas patented the day before I came up with it lol.

Either way, to get alot more current out of it I will probably need banks of caps. Originally I bought these CGS grade Mallorys or whatever for their low ESR. Getting that low ESR in a 1,000uF is not going to be easy.

Also, there is no need for any type of matching of the choppers, in fact one could be bipolar and the other MOSFET in a pair if I can conjure up a wayto simplify the drive circuit. Who knows maybe I can eliminate the inverters altogether. Maybe I can even make the thing self oscillate if I have high enough gain devices to keep them saturated.

That can all be worked on, the worst part of this is getting caps that willgive me ridiculous current for the choke type amp. This power supply is going to suffer under that load and I don't need it to lose its smoke. I might have to use one per channel.

that is why I originally considered a three phase version. I just couldn't get it com;letely figured out. It doesn't seem right to use the 33%/66% duty cycle because it doesn't do shit for efficiency. It's just another circuit. I could as easily (almost) just make one supply for each channel. It would only be 25 % more than trying to work out all the problems of the three phase deal.

How could that work ? It would be 33 % high for the top choppers which means 66 % low. During that time the top cap would have more time to charge, but the bottom cap could not be there to charge it, or could it ?

Thst would mean to push "negative" voltage into the bottom (the 12 volt) at33 % of the time. That means what, I would have to use the Q and notQs from the ring counter ? Shit, this is headache territory. For a few dollars more, I just go with dual power supplies. It also looks better on paper. It might even sound betrer. Hell, it also gives the option of building them in monoblocks.

The audiophiles would bust a nut over them.

Yup, if I go with all this, each channel will have its own tripler. It solves too many problems at once.

I'm starting to think outloud here. Time for bed, or maybe prune juice or some shit. When I think about how long ago I started this, and how far back that burner is it got stuck on, I feel old.

Mourn me, it will be valid soon lol.

I am going to build this shit, but I cannot see, I cannot pee, I cannot smell, I look like hell........

What's more, I don't even like to drive. No shit, too many assholes out there. I don't even play the radio when I drive. I'll hook up a car alernator to a bicycle and set it up like an exercise bike. Take a battery, the amp. then get some good speakers for it.

Then feed it from a PC because I don't even use disks anymore.

Almost doesn't seem worth it.

LET'S DO IT !

 

[Bill Sloman]

Thanks all.

I intended on square waves all along. I think this was just not drawn right on my part after seeing the exmples. I should have just did it on paper and scanned it....

So it is a Dickson eh ? When did that come out ? I "invented" it a damn long time ago, I mean I was half my age or less I think. I'm betting that itwas patented the day before I came up with it lol.

John Fields cites a paper from 1976. That's 37 years ago - you might be 74,but it's unlikely.

People do seem to get original ideas at much the same time. A friend of mine invented a better confocal microscope - and made quite a bit of money outof the patent, which he wouldn't have done if Tektronix hadn't dropped a very similar provisional patent dated three weeks before his patent application.

Sometimes you can work out what the common trigger was. Back in November 1986 I sent a memo to my boss suggesting that we might patent a particular way of stabilising a GaAs single-crystal puller. In March 1987 the inventor of the then standard way of pulling GaAs singel crystals showed up with his own - more or less identical - patent application, dating from rather earlier in 1986.

My insight had come from reading a paper in the journal where the inventor was not only an editor but also the action editor for that particular paper..

 

[Jasen Betts]

Thanks all.

I intended on square waves all along. I think this was just not drawn right on my part after seeing the exmples. I should have just did it on paper and scanned it....

So it is a Dickerson eh ? When did that come out ? I "invented" it a damn long time ago, I mean I was half my age or less I think. I'm betting that it was patented the day before I came up with it lol.

Either way, to get alot more current out of it I will probably need banks of caps. Originally I bought these CGS grade Mallorys or whatever for their low ESR. Getting that low ESR in a 1,000uF is not going to be easy.

Also, there is no need for any type of matching of the choppers, in fact one could be bipolar and the other MOSFET in a pair if I can conjure up a way to simplify the drive circuit. Who knows maybe I can eliminate the inverters altogether. Maybe I can even make the thing self oscillate if I have high enough gain devices to keep them saturated.

maybe use a H-bridge IC.

 

[[email protected]]

"maybe use a H-bridge IC. "

You mean like a BTL driver Same thing ?

Sure could, except that out of this i want some current. The one model out here uses surface mounts, I assure you that will not work, my current load will be way too much.

In the simplsst terms, with the output choke feeding an eight ohm load, it is going to swing +36 on one side of the speaker and as close as possible to -36 on the other side. Just one channel is effecrtively working into two ohms.

Basically 72 V peak times RMS is 50.9, times itself is 2591 andabit. Divided by eight is 323 watts. That is the power at clipping, true power is less.

But still, it pulls a shitload of current.

I know of some nice 65 amp FETs used in plasma TVs that are nice and cheap. For a stupid invertor transistor and cap, I can drive it. In facrt I can drive anything.

 

[[email protected]]

These types of circuits are mainly used to generate high voltages at low current levels. If you try to power a serious audio amp this way you will find that it is very inefficient.
Let's assume that you audio amp needs to supply 10 amps rms. Due to the voltage tripling, the input rms current will be at least 30 amps rms. But the problem is that all this current is consumed in narrow transients when the switching occurs, so you might be looking at peak currents of well over 100amps. Even if you find devices that can handle this, the ON resistance gets magnified at the output so your overall efficiency will be very poor.

Bob

 

[[email protected]]

Isn't that true of any other converter anyway ?

 

[[email protected]]

Isn't that true of any other converter anyway ?

OK, not really a flyback type, but that wouldn't be suited to this app anyway......

 

[[email protected]]

No, once you decide against storing energy in an inductor and instead only use capacitors, your efficiency when delivering large output currents drops dramatically. If there were such a thing as an ideal switch then it would be different.

Bob

 

[[email protected]]

No, once you decide against storing energy in an inductor and instead only use capacitors, your efficiency when delivering large output currents drops dramatically. If there were such a thing as an ideal switch then it would be different.

Bob

 

[[email protected]]

What is the purpose of the "boost" ?

 

[[email protected]]

That would also come in handy for driving bipolars.