Oops I think I should have posted this here - http://www.diyaudio.com/forums/showthread.php?s=&threadid=80125
Thanks
Thanks
A bridge/parallel LM3886/LM4780 won't get you half that, unfortunately. My back of the envelope math says you need approximately +/-56V (though I might have done the math wrong) to get 400W into 8 ohms and +/-70 to get 600W. No chip amp that I know of can do those rails (except the LM4702, but that is only a voltage driver, not a traditional "chip" amp)
That being said, I can't really conceive a way to get to 400-600W using just chips. I think you will have to go to a more complex design to get that much power.
David
That being said, I can't really conceive a way to get to 400-600W using just chips. I think you will have to go to a more complex design to get that much power.
David
High Power Gain Clone
You can easily achieve power levels that high and higher, upwards of a thousand watts per channel are possible. Granted with a lot of work.
Not anything agaisnt dfdye my present chip amp produces well over 400 watts per channel and uses LM3875 chips.
Watts are not all volts, oh current must there be..........
PS this is my baby chip amp I am building its big brother now.
You can easily achieve power levels that high and higher, upwards of a thousand watts per channel are possible. Granted with a lot of work.
Not anything agaisnt dfdye my present chip amp produces well over 400 watts per channel and uses LM3875 chips.
Watts are not all volts, oh current must there be..........
An externally hosted image should be here but it was not working when we last tested it.
PS this is my baby chip amp I am building its big brother now.
Re: High Power Gain Clone
Yea, so ignore my first post, if for no other reason than I screwed up the math (all of the numbers should have been swings, not +/-, not that it matters now)
None taken, and upon reconsideration, I am indeed being silly. Knee jerk ohm's law = bad assumptions.tiltedhalo said:
Yea, so ignore my first post, if for no other reason than I screwed up the math (all of the numbers should have been swings, not +/-, not that it matters now)
High Power Chip Amp
If any one wants a copy of schematics or possible boards let me know via E-Mail tiltedhalodesigns@gmail.com
Due to reasons I do not want to go into I do not post full circuits. I reserve full intellectual and physical rights to developed circuits, and would expect them only to be used for non- profit DIY projects.
If any one wants a copy of schematics or possible boards let me know via E-Mail tiltedhalodesigns@gmail.com
Due to reasons I do not want to go into I do not post full circuits. I reserve full intellectual and physical rights to developed circuits, and would expect them only to be used for non- profit DIY projects.
i know sfa about amp design, only built a couple of kits, but could you explain one thing visible from that pic please?
i notice the toroidal->bridge rectifier->filter caps, then a whole bank of caps - what are they doing/what is their purpose?
i've also looked into a way of using typical kit amps in a car but somehow upping the power but it appears the only real way of doing it is a switched power supply, meaning it's probably not really worth it. it's somehow related however, how to get big power amps from low voltage with mega amounts of current available.
chatting to a mate who has been designing some amps (but not of the audio kind), he said
means sfa to me
???
i notice the toroidal->bridge rectifier->filter caps, then a whole bank of caps - what are they doing/what is their purpose?
i've also looked into a way of using typical kit amps in a car but somehow upping the power but it appears the only real way of doing it is a switched power supply, meaning it's probably not really worth it. it's somehow related however, how to get big power amps from low voltage with mega amounts of current available.
chatting to a mate who has been designing some amps (but not of the audio kind), he said
means sfa to me
???
I have ran bridged LM3886's at +-41V in a few of my amps.
Keeping them cool was very difficult at the max heat dissipation power point.
I had to use live (not insulated from the LM3886 tabs) fan-cooled heatsinks to remove the heat.
Using mica or other tab insulation gave very early heat protection clipping problems.
Cant remeber exactly what power i was getting out of them though... it was a few years ago.
Keeping them cool was very difficult at the max heat dissipation power point.
I had to use live (not insulated from the LM3886 tabs) fan-cooled heatsinks to remove the heat.
Using mica or other tab insulation gave very early heat protection clipping problems.
Cant remeber exactly what power i was getting out of them though... it was a few years ago.
Whats that?
Phreeky82, the PC board between the two tororidal transformers have the AC input filters and tranformer and fan control circuits.
The two large power supply boards with all the capacitors , utilize a current feed through design that uses a feed ahead bridge and a take-up bridge rectifier. Followed by two HV series parellel low ESR electrolic caps, bypassed by two audio quality 4.7uf non-polar metal poly caps for high frequency stiffening, followed by an odd even bank of 4700uf 105deg electrolitics (56400uf total capacitance per channel) uses many smaller faster caps tend to let the power supply recover faster.
The whole thing is coupled to the amp boards via a current feed ahead and a seperate back loop, to help stabilize current and voltage to constant levels under the heaviest loads.
Phreeky82, the PC board between the two tororidal transformers have the AC input filters and tranformer and fan control circuits.
The two large power supply boards with all the capacitors , utilize a current feed through design that uses a feed ahead bridge and a take-up bridge rectifier. Followed by two HV series parellel low ESR electrolic caps, bypassed by two audio quality 4.7uf non-polar metal poly caps for high frequency stiffening, followed by an odd even bank of 4700uf 105deg electrolitics (56400uf total capacitance per channel) uses many smaller faster caps tend to let the power supply recover faster.
The whole thing is coupled to the amp boards via a current feed ahead and a seperate back loop, to help stabilize current and voltage to constant levels under the heaviest loads.
BPA Heat heat heat
Your BPA design amp should run as cool as a single chip given proper increase in heat sinking, if not you have a PROBLEM. This tends to be a big problem Ive seen with most BPA designs heat and lower then expected power outputs even when they sound fine to the ear.
SERVOS SUCK
even well designed ones, they are great for many applications not audio, its inherent in there physics, and they wont solve all your parallel problems they will mask a couple and create a few more.
Paralleling a chip amp is not like daisy chaining outputs on a discrete amp
the chip need to be in perfect, PERFECT comunication if not they fight. For example take a BPA200 designs 2chips parallel the two bridged total four chips chip 1 -20mv DC offset, chip two +112mv DC offset, alone not bad, together not soo good, the two together equal 92mv easy offset right? wrong. This is a parallel application that offset is amplified by the overall gain off the amp, now youve got over 20VDC 
on the out put off your first parallel set. Now lets say pair two is dissipating the same, BUT MY SPEAKER CONES are not moveing and I measure no DC offset
remember this is bridged ,push pull it will mask the individual offset of each pair.
So where is all that wasted power going into the chips 40 watts off idle heat and waste, as output power increases so does the waste, your BPA 200 is now a BPA100 and you get a nice 100 watt space heater
Your BPA design amp should run as cool as a single chip given proper increase in heat sinking, if not you have a PROBLEM. This tends to be a big problem Ive seen with most BPA designs heat and lower then expected power outputs even when they sound fine to the ear.
SERVOS SUCK
even well designed ones, they are great for many applications not audio, its inherent in there physics, and they wont solve all your parallel problems they will mask a couple and create a few more.Paralleling a chip amp is not like daisy chaining outputs on a discrete amp
the chip need to be in perfect, PERFECT comunication if not they fight. For example take a BPA200 designs 2chips parallel the two bridged total four chips chip 1 -20mv DC offset, chip two +112mv DC offset, alone not bad, together not soo good, the two together equal 92mv easy offset right? wrong. This is a parallel application that offset is amplified by the overall gain off the amp, now youve got over 20VDC on the out put off your first parallel set. Now lets say pair two is dissipating the same, BUT MY SPEAKER CONES are not moveing and I measure no DC offset remember this is bridged ,push pull it will mask the individual offset of each pair.So where is all that wasted power going into the chips 40 watts off idle heat and waste, as output power increases so does the waste, your BPA 200 is now a BPA100 and you get a nice 100 watt space heater
tiltedhalo, I'm quite keen to do the same design you have. How much did it all cost in the end?
Thanks very much by the way! I'm sure if I did it, I'd use your PCB as I am a beginner.
Oh and how does it sound? As good as a Bryston maybe? Thats what I have now but will have to sell with my active PMC's.
Thanks very much by the way! I'm sure if I did it, I'd use your PCB as I am a beginner.
Oh and how does it sound? As good as a Bryston maybe? Thats what I have now but will have to sell with my active PMC's.
Re: BPA Heat heat heat
Perhaps a single LM4780 (2 LM3886 dies on same IC) would be inherently better matched and thus give less offset per pair than using 2 separate chips?
tiltedhalo said:***
Paralleling a chip amp is not like daisy chaining outputs on a discrete amp
***
Perhaps a single LM4780 (2 LM3886 dies on same IC) would be inherently better matched and thus give less offset per pair than using 2 separate chips?
why offset is multiplied
Remember in a bridge configuration your outputs are connected via your .1 to .15 ohm output resistors giveing rise to a path for a differance voltage to appear on the negative feeb back loops of both chips, +20mv chip one -112mv chip 2 differance +92mv at the negative feed back loops, the offset differance is then amplified, this effect is like adding a strange sort of regenerative feedback. also goes to explain the oscillations some BPA exhibit.
Remember in a bridge configuration your outputs are connected via your .1 to .15 ohm output resistors giveing rise to a path for a differance voltage to appear on the negative feeb back loops of both chips, +20mv chip one -112mv chip 2 differance +92mv at the negative feed back loops, the offset differance is then amplified, this effect is like adding a strange sort of regenerative feedback. also goes to explain the oscillations some BPA exhibit.
how does it sound
My two chip amps replaced my 2 bridged Carver 4.0t's and my Krell KSA 200
Strong powerfull and controlled bass , defined mids, and un holly highs right up two secound harmonics, my amps maintain a pretty straight response curve from 5HZ to 40KHZ, the sound is actually better then a single high end chip, plus I get very little dynamic compresion even at very high out put levels, 300+.
The amps power envolope seems to get wider with greater output other then tighter.
My two chip amps replaced my 2 bridged Carver 4.0t's and my Krell KSA 200
Strong powerfull and controlled bass , defined mids, and un holly highs right up two secound harmonics, my amps maintain a pretty straight response curve from 5HZ to 40KHZ, the sound is actually better then a single high end chip, plus I get very little dynamic compresion even at very high out put levels, 300+.
The amps power envolope seems to get wider with greater output other then tighter.
useing a LM4870 for better matching
It helps to a point, you would think it would be a perfect solution due to the fact the die stamps on common substrate, the transistors are stamped on the same sub, but the resistors are laser etched, guess what the major cause for miss match is , hold your breath
the resistors.
It helps to a point, you would think it would be a perfect solution due to the fact the die stamps on common substrate, the transistors are stamped on the same sub, but the resistors are laser etched, guess what the major cause for miss match is , hold your breath
the resistors.
Re: useing a LM4870 for better matching
No data, only conjecture at this point. . . .
You bring up a very good point, but while in operation, thermal tracking will definitely contribute a great deal too, would you not agree? With two cores on the same die, temperature mismatches could be minimized. Even with resistor mismatches (which I can't think of how to test for between chips in order to match chips) I would think that the two cores of an LM4780 would be closer than two 3886's. . .tiltedhalo said:guess what the major cause for miss match is the, hold your breath
No data, only conjecture at this point. . . .
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.