I've been reading here and there in this list about different people wanting to build absolute gigapower amplifiers and invariably two difficulties (will) come to light. 1/ The output devices are going to need simply enormous safe-operating-area to be able to drive a reactive speaker load given the high dc rail voltage that will be needed. 2/ That same high rail voltage and the power it will produce means that the speakers are gunna have to be wired in series-parallel or something ugly like that.
The way I see it is this: lets assume that your choice of a good solid speaker driver will take 250 watts rms and is 8 ohms impedance. Work out how many volts rms that is across 8 ohms. mulitply by 1.414 to get the peak voltage. Add ~5 to 10 volts and this gives you the total rail voltage your amp needs, half above zero and half below zero.
Now the next bit is my whole point - if you want more power, stick on another loudspeaker and another pair of output transistors (psu suitably rate of course). Ten times the power? Ten times the transistor capacity and ten times the speakers *all in parallel*. What's the point of building an amp that can swing absolutely whopping peak to peak voltages with the ridiculous demands that makes on transistors, and then find that most speakers won't handle that amount of drive voltage anyway? Just decide the maximum drive voltage that your favourite loudspeaker can handle and design your amp for this o/p voltage capability. More power = proportionally more transistors and more speakers all in parallel, the maximum output voltage always remains the same, only the current goes up.
GP.
GP.
The way I see it is this: lets assume that your choice of a good solid speaker driver will take 250 watts rms and is 8 ohms impedance. Work out how many volts rms that is across 8 ohms. mulitply by 1.414 to get the peak voltage. Add ~5 to 10 volts and this gives you the total rail voltage your amp needs, half above zero and half below zero.
Now the next bit is my whole point - if you want more power, stick on another loudspeaker and another pair of output transistors (psu suitably rate of course). Ten times the power? Ten times the transistor capacity and ten times the speakers *all in parallel*. What's the point of building an amp that can swing absolutely whopping peak to peak voltages with the ridiculous demands that makes on transistors, and then find that most speakers won't handle that amount of drive voltage anyway? Just decide the maximum drive voltage that your favourite loudspeaker can handle and design your amp for this o/p voltage capability. More power = proportionally more transistors and more speakers all in parallel, the maximum output voltage always remains the same, only the current goes up.
GP.
GP.
All the top and bottom emitter resistors go to the one line that feeds the speaker, so you take it from there the same as usual. I wasn't meaning that each extra set of transistors was independently feeding it's own speaker, rather that as you add another speaker in parallel with the rest on the rhs, you add another pair of transistors or more transistor muscle on the lhs, all feeding the same line.
GP.
GP.
More horse power.
Hello Graham, this is indeed the technique used in pro-sound amplifiers suitable for driving low impedence loads.
For given power rails, adding extra output pairs will allow low impedence load operation without exceeding ouput transistor SOAR limits.
The caveats to this is that a stronger driver stage is required, and output transistor emitter resistors to ensure current sharing.
For best performance matched output transistors are needed.
For example my Jands JC-1000 amplifier has 12 MJ-15003/4 To-3 devices per channel.
2 are drivers, and 10 are outputs.
The pre-drivers are MJE 340/350 from memory.
These are used as drivers in lesser amplifers.
This monster (50 kG +) is rated 500+500 into 4 ohms and won't blow up if driving 3 by 8 ohm loads (2.66 ohms) per channel.
Serious power and very clean, and for home purposes infinite.
Perreaux 8000 is rated to drive 2 ohms via a weeks salary worth of To-3 mosfet outputs, and do this all day, and reliably.
I agree that if bothering to build an amplifier, it must be designed for performance and reliability into low impedence loads.
Regards, Eric.
Hello Graham, this is indeed the technique used in pro-sound amplifiers suitable for driving low impedence loads.
For given power rails, adding extra output pairs will allow low impedence load operation without exceeding ouput transistor SOAR limits.
The caveats to this is that a stronger driver stage is required, and output transistor emitter resistors to ensure current sharing.
For best performance matched output transistors are needed.
For example my Jands JC-1000 amplifier has 12 MJ-15003/4 To-3 devices per channel.
2 are drivers, and 10 are outputs.
The pre-drivers are MJE 340/350 from memory.
These are used as drivers in lesser amplifers.
This monster (50 kG +) is rated 500+500 into 4 ohms and won't blow up if driving 3 by 8 ohm loads (2.66 ohms) per channel.
Serious power and very clean, and for home purposes infinite.
Perreaux 8000 is rated to drive 2 ohms via a weeks salary worth of To-3 mosfet outputs, and do this all day, and reliably.
I agree that if bothering to build an amplifier, it must be designed for performance and reliability into low impedence loads.
Regards, Eric.
Blockheads !
Hi again AudioFreak, My kids just asked me why I laughed out aloud when reading your reply.
I fear Mr slicemaster is a plain lost cause.
All his asking and no listening reminds me of a brat who thinks he knows it all and is likes to crow about it, but when pressed really does not know Jack.
Let me applaud if you can get through his thick skull !.
BTW being young is perfectly fine, I still think like I'm young too, and I knew about etiquette and respect when I was young - the above mentioned don't have a clue.
Regards, Eric.
Hi again AudioFreak, My kids just asked me why I laughed out aloud when reading your reply.
I fear Mr slicemaster is a plain lost cause.
All his asking and no listening reminds me of a brat who thinks he knows it all and is likes to crow about it, but when pressed really does not know Jack.
Let me applaud if you can get through his thick skull !.
BTW being young is perfectly fine, I still think like I'm young too, and I knew about etiquette and respect when I was young - the above mentioned don't have a clue.
Regards, Eric.
Super Low Z speakers
I've been seeing a lot of 5V 100A+ switching power supplies at flea markets and swap meets in the last couple years. I have been thinking about how to make real low Z speakers so these super cheap (usually <$20), ultra high current supplies could be used.
Ribbons come to mind- maybe something like Apogees... Or maybe rewind the voice coil in a bass driver with just a couple turns of wide, flat aluminum or copper strap.
Maybe a class D or Tripath type amp- low voltage output transistors would be no trouble at all to find...
I've been seeing a lot of 5V 100A+ switching power supplies at flea markets and swap meets in the last couple years. I have been thinking about how to make real low Z speakers so these super cheap (usually <$20), ultra high current supplies could be used.
Ribbons come to mind- maybe something like Apogees... Or maybe rewind the voice coil in a bass driver with just a couple turns of wide, flat aluminum or copper strap.
Maybe a class D or Tripath type amp- low voltage output transistors would be no trouble at all to find...
MRehorst:
The problem with higher currents is that losses in cables, PCB's etc becomes a lot higher. Also the inductances in the cables and speakers produces nasty voltage spikes at large current changes.
You need a certain rail voltage to produce a certain power at, say 8 ohms. Then you can parallel two speakers to increase the power but you can't use much less than 4 ohm loads or it will be cheaper to use two amps in redundancy than making the enormous output stage required.
There is also a problem with low Z speakers and that is that when too few turns of wire are used, the linearity suffers. That is probably why most speakers are 8 ohms. It's a good compromize.
/Marcus
The problem with higher currents is that losses in cables, PCB's etc becomes a lot higher. Also the inductances in the cables and speakers produces nasty voltage spikes at large current changes.
You need a certain rail voltage to produce a certain power at, say 8 ohms. Then you can parallel two speakers to increase the power but you can't use much less than 4 ohm loads or it will be cheaper to use two amps in redundancy than making the enormous output stage required.
There is also a problem with low Z speakers and that is that when too few turns of wire are used, the linearity suffers. That is probably why most speakers are 8 ohms. It's a good compromize.
/Marcus
I imagine that if you had a 1 turn voice coil made of flat copper strip and it was an inch wide but with only 1/4 inch inside the magnetic gap at any one time maybe the linearity would be ok?
If you had an amp that would swing +/- 5 volts at 100 amps then you could use a step-up transformer between it and the speaker. For the amp itself there are some absolutely stunning low voltage fets available nowadays e.g. IRF3703 30v 210A 0.0028 ohms(!) TO-220 case.
Seeing you can only get N-channels this good, and you can have the two 5v rails floating WRT each other, the circlotron is the only topology worth considering
GP.
If you had an amp that would swing +/- 5 volts at 100 amps then you could use a step-up transformer between it and the speaker. For the amp itself there are some absolutely stunning low voltage fets available nowadays e.g. IRF3703 30v 210A 0.0028 ohms(!) TO-220 case.
Seeing you can only get N-channels this good, and you can have the two 5v rails floating WRT each other, the circlotron is the only topology worth considering
GP.
hi power amps
Mr Feedback wrote:
"Hi again AudioFreak, My kids just asked me why I laughed out aloud when reading your reply.
I fear Mr slicemaster is a plain lost cause.
All his asking and no listening reminds me of a brat who thinks he knows it all and is likes to crow about it, but when pressed really does not know Jack.
Let me applaud if you can get through his thick skull !.
BTW being young is perfectly fine, I still think like I'm young too, and I knew about etiquette and respect when I was young - the above mentioned don't have a clue."
I have this sign on my desk:
"People who think they know it all are especially annoying to those of us who do"
Go figure
Jan Didden
Mr Feedback wrote:
"Hi again AudioFreak, My kids just asked me why I laughed out aloud when reading your reply.
I fear Mr slicemaster is a plain lost cause.
All his asking and no listening reminds me of a brat who thinks he knows it all and is likes to crow about it, but when pressed really does not know Jack.
Let me applaud if you can get through his thick skull !.
BTW being young is perfectly fine, I still think like I'm young too, and I knew about etiquette and respect when I was young - the above mentioned don't have a clue."
I have this sign on my desk:
"People who think they know it all are especially annoying to those of us who do"
Go figure
Jan Didden
I'm only going to use 4 speakers in parallel/450W amp, you know, mrfeedback. Better than my older plan to use 12 in series-parallel/900W amp isn't it. See, I'm catching on, now don't you worry.
P.S. Audiobahn makes a 5000W RMS 34" sub with a dual 1 ohm voice coil. It costs about $2900 each though, I can't imagine anyone actually buying one, but by reputation, many of you probably think I would, and use 1600W IGBT's as output devices while I'm at it.
P.S. Audiobahn makes a 5000W RMS 34" sub with a dual 1 ohm voice coil. It costs about $2900 each though, I can't imagine anyone actually buying one, but by reputation, many of you probably think I would, and use 1600W IGBT's as output devices while I'm at it.
Circlotron:
The magnetic field starts bending quite early at the ends of a regular coil. With a one turn coil the field would be almost circular around the wire, which gives really poor linearity.
I don't think your idea is feasible. First of all you don't want an output transformer. They are hard to make and expensive, especially att the currents you are talking about. Secondly, the FETs that you refer to are probably designed for switching applications and in that case they are no good for linear amps.
If you still want to go through with it, let us know how it goes.
AMPMAN:
87 dB is probably the worst efficiency in a 18" driver that I have ever seen. No wonder that they have to be able to take thousands of watts.
/Marcus
The magnetic field starts bending quite early at the ends of a regular coil. With a one turn coil the field would be almost circular around the wire, which gives really poor linearity.
I don't think your idea is feasible. First of all you don't want an output transformer. They are hard to make and expensive, especially att the currents you are talking about. Secondly, the FETs that you refer to are probably designed for switching applications and in that case they are no good for linear amps.
If you still want to go through with it, let us know how it goes.
AMPMAN:
87 dB is probably the worst efficiency in a 18" driver that I have ever seen. No wonder that they have to be able to take thousands of watts.
/Marcus
e96mlo said:
Nah, I wasn't being serious. As for the stepup transformer, I was illustrating the difficulties involved in using a + - 5v supply in a workable but probably impractical way. Better to put your money into a power tranny instead of an o/p tranny, hey?
Those fets are indeed meant primarily for switching, but is that not an open invitation for a diy'er to try then in a linear amp? I for one have had quite acceptable results using 500v n-channel-only switching fets in linear use.
Just for that, I know where to get my hands on a pair of IRF3703's. I'll put them in my bench testbed and show them who's boss!
I'll report the results.Thinking more about voicecoils, if you used copper strip I wonder how well it would work if you used multilple turns wound like a roll of tape? Not too thick though. Keep the magnet gap small.
GP.
Hard to know if you were serious or not, since I have equally crazy ideas myself from time to time.
Regarding the voice coils:
I think the problem remains as long as you don't have a voice coil that is much longer than the width if the magnet gap.
I have read though, that underhung voice coils (that is when the gap is wider than the coil is long) gives extreme linearity for the trade off of smaller Xmax and lower BL. But it still requires the coil to have enough turns to be resonably "ideal".
There is a thread around here somewhere about making your own drivers. I don't know what has been said there. It's probably worth checking out, though.
/Marcus
Regarding the voice coils:
I think the problem remains as long as you don't have a voice coil that is much longer than the width if the magnet gap.
I have read though, that underhung voice coils (that is when the gap is wider than the coil is long) gives extreme linearity for the trade off of smaller Xmax and lower BL. But it still requires the coil to have enough turns to be resonably "ideal".
There is a thread around here somewhere about making your own drivers. I don't know what has been said there. It's probably worth checking out, though.
/Marcus
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.