Nordic said:Link is dead
you have to goto original post, I think
and click
http://www.diyaudio.com/forums/showthread.php?postid=759700#post759700
It is possible to use an OP-amp
to drive an external discrete output stage.
The hookup is most OP-amps as well as power Op-amps, including LM3886,
have limited voltage supply max value.
This means you will need one lower voltage for OP-amp
and one higher for output transistors.
And you also need the transistors to have voltage gain.
Because Op-amp can not output more than its own supply voltage ( actually a few volts lower).
OP-amps part of the total voltage gain, should also be set
so op-amp stays stable.
Is all this possible to make?
Yes, it is.
It is not a most simple circuit design,
but not too difficult either.
to drive an external discrete output stage.
The hookup is most OP-amps as well as power Op-amps, including LM3886,
have limited voltage supply max value.
This means you will need one lower voltage for OP-amp
and one higher for output transistors.
And you also need the transistors to have voltage gain.
Because Op-amp can not output more than its own supply voltage ( actually a few volts lower).
OP-amps part of the total voltage gain, should also be set
so op-amp stays stable.
Is all this possible to make?
Yes, it is.
It is not a most simple circuit design,
but not too difficult either.
gainclone on steroids
All my gain clones are on steroids, this was my first successful prototype with lm3886 chips with 38V+- rails. By the way I use no servo's no expensive parts i.e. .1% resistors ect.... the most expensive part were the chips at 4 bucks! 214 watts at .04%THD burst point clipping repetative 20ms 298watts 1.34% THD, oh and its stable into a 3 ohm load with the addition of fan cooling.
All my gain clones are on steroids, this was my first successful prototype with lm3886 chips with 38V+- rails. By the way I use no servo's no expensive parts i.e. .1% resistors ect.... the most expensive part were the chips at 4 bucks! 214 watts at .04%THD burst point clipping repetative 20ms 298watts 1.34% THD, oh and its stable into a 3 ohm load with the addition of fan cooling.
An externally hosted image should be here but it was not working when we last tested it.
fantastic !! bridge or parallel?
gainclone on steroids
bridged parallel 2*2, I built a single lm3875t and couldnt believe my ears, Ive been searching for the perfect performing high power version this was my first success. Ive put together 5 others and have finnally found a cicuit that is sonically identical if not better then a single chip.
bridged parallel 2*2, I built a single lm3875t and couldnt believe my ears, Ive been searching for the perfect performing high power version this was my first success. Ive put together 5 others and have finnally found a cicuit that is sonically identical if not better then a single chip.
Better check the spec sheet, you dont think these chips make 56 to 76 watts on a 110V swing??? they do it on a 50V swing. they are a current driven devices. I think you are on the wrong track, you will introduce the transfer charecteristics of the transistors, notch distortion ect... with the addition of transistors. And to make a good circuit will be a little complicated.
The chips main problem with low ohm loads is dissipation your trying to remove heat from a device producing 56watts output with a foot print smaller then a single TO220AB device.
I havent tried my new amp on electrstatics, but I have a friend that has a nice pair of Martin Logans I will report back with data. I have no doubt my chips will not have an issue driving them. My home speakers are a very hard load to drive haveing line array ribbon tweeters.
The chips main problem with low ohm loads is dissipation your trying to remove heat from a device producing 56watts output with a foot print smaller then a single TO220AB device.
I havent tried my new amp on electrstatics, but I have a friend that has a nice pair of Martin Logans I will report back with data. I have no doubt my chips will not have an issue driving them. My home speakers are a very hard load to drive haveing line array ribbon tweeters.
Well, I did try Gainclone on my e-stats and it is breaking up much before it's full output power. Mine is one chip amp not bridged/parelel
but that doesn't matter it can't produce 50W.
You are right about problems regarding cooling small footprint
but there is not much we can do about it. Properly biased transistor
output will have minimal notch distortion, certainly not more than chip alone and if you make it A-class amp even better.
Adding another stage will introduce as You said transfer characteristics of the transistors but this is better than listening chip's over-current protection kicking in every time it runs out of steam. And it will.
As You probably know there is only a hand full of commercial amps capable of driving e-stats and those are not cheap.
Massive power supply, huge banks of capacitors and multiple
output devices is what is necessary to drive e-stats on the long run.
I emphasize "long run" because You might even drive e-stats
with Your amp but probably after some time it will fail because of
too much stress on the chips. Even with "big boys" I did hear
strain of the amp trying to cope with current demands. This is why I
have built high voltage direct drive tube amp, clone of Acoustat.
But if You don't intend to drive e-stats, You have nothing to worry
about. Your amp is more than enough for any other application.
All this is my experience which may differ from yours and I am looking forward to hear how Your amp did with ML. By the way, what Your friend is using to drive his MLs?
Best regards.
but that doesn't matter it can't produce 50W.
You are right about problems regarding cooling small footprint
but there is not much we can do about it. Properly biased transistor
output will have minimal notch distortion, certainly not more than chip alone and if you make it A-class amp even better.
Adding another stage will introduce as You said transfer characteristics of the transistors but this is better than listening chip's over-current protection kicking in every time it runs out of steam. And it will.
As You probably know there is only a hand full of commercial amps capable of driving e-stats and those are not cheap.
Massive power supply, huge banks of capacitors and multiple
output devices is what is necessary to drive e-stats on the long run.
I emphasize "long run" because You might even drive e-stats
with Your amp but probably after some time it will fail because of
too much stress on the chips. Even with "big boys" I did hear
strain of the amp trying to cope with current demands. This is why I
have built high voltage direct drive tube amp, clone of Acoustat.
But if You don't intend to drive e-stats, You have nothing to worry
about. Your amp is more than enough for any other application.
All this is my experience which may differ from yours and I am looking forward to hear how Your amp did with ML. By the way, what Your friend is using to drive his MLs?
Best regards.
Re: gainclone on steroids
Hi
I'm building something similar, using four LM 4780 chips. Obviously this will be lower in power than the 4 x 3886, but I guess I'll be fine.
I have one question on bridge/parallel and I'm asking here even though it's OT, bec your email button is disabled.
I am trying to get my head around the grounding of a balanced input and power supplies.
My setup will be two LM4780 per channel. Each chip's outputs are tied together for parallel, the inputs also tied together. Then each chip is fed one half of a balanced signal for ful bridge operation. Unless I'm mistaken, you're following a similar scheme?
Here's my question:
When using a balanced input, is each amp input tied to the two sides of the balanced signal, or to one side + shield/common ground? And are the two supplies sharing a common ground or can be kept independent?
Or, as in true balanced operation, it does not matter one bit and I can just drive each half through one amp without ground reference, and only the load finally sums up both the signals?
tiltedhalo said:
Hi
I'm building something similar, using four LM 4780 chips. Obviously this will be lower in power than the 4 x 3886, but I guess I'll be fine.
I have one question on bridge/parallel and I'm asking here even though it's OT, bec your email button is disabled.
I am trying to get my head around the grounding of a balanced input and power supplies.
My setup will be two LM4780 per channel. Each chip's outputs are tied together for parallel, the inputs also tied together. Then each chip is fed one half of a balanced signal for ful bridge operation. Unless I'm mistaken, you're following a similar scheme?
Here's my question:
When using a balanced input, is each amp input tied to the two sides of the balanced signal, or to one side + shield/common ground? And are the two supplies sharing a common ground or can be kept independent?
Or, as in true balanced operation, it does not matter one bit and I can just drive each half through one amp without ground reference, and only the load finally sums up both the signals?
tiltedhalo said:
An Athlon XP T-Bred B CPU has a footprint roughly 1/8th of the size of a LM4780 or 3886. And it dissipates roughly 50-60 watts at full power, and die temps with stock manufacturer-supplied heatsinks normally don't exceed 70 degrees C. There are solutions that do a much better job at lower fan speed/noise, and 50 degrees is normally a good target.
With some imagination I think some of those solutions can be modified and transplanted for use in audio, as we can still live with somewhat higher temperatures at full tilt. AFAIK an LM4780 disspates roughly 30 watts of heat with a 35 volt supply into a 4 ohm load, a 3886 half that. I know that one quiet fan run at 7V can double the cooling efficiency of a heatsink... Some food for thought, which I'm also ruminating over for my ambitous setup...
Steve Eddy said:Yeah, but when you start adding more opamps and a big discrete output stage, it's really no longer a Gainclone.
This will give you a Gainclone on steroids:
Apex PA-03
se
Is there anyway to use that as a power amplifier? Anyone have a schematic or anything? If I read the sheet properly for it, you'd be looking at 500wRMS power output from that single chip!!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.