Bazukaz,
Understand that I have already tried a single LM4780 in the parallel configuration and blew a chip in about 2 minutes running the ML clarity's. And I am trying to run the Ascents now!
So the question is...What is the remedy for an amplifier to be stable when driving a VERY LOW IMPEDENCE speaker??
Is it capacity??? current???What is it???
Dominick
Understand that I have already tried a single LM4780 in the parallel configuration and blew a chip in about 2 minutes running the ML clarity's. And I am trying to run the Ascents now!
So the question is...What is the remedy for an amplifier to be stable when driving a VERY LOW IMPEDENCE speaker??
Is it capacity??? current???What is it???
Dominick
tiltedhalo,
Thanks for the inp[ut, but please understand that this is my second project that spawned from my first LM4780. I am making every effort to learn the ins and outs, but I am still very much an amateur.
Please expand on "buffer". I did a search and came up with a ton of blind leads.
Thanks,
Dominick
Thanks for the inp[ut, but please understand that this is my second project that spawned from my first LM4780. I am making every effort to learn the ins and outs, but I am still very much an amateur.
Please expand on "buffer". I did a search and came up with a ton of blind leads.
Thanks,
Dominick
Stability depends on PCB design , components chosen , and chip itself.
When driving capacitive loads , amp must be somehow isolated from very low impedances at high frequencies. It may be a resistor , RL circuit, or something else.
Did you read LM3886 application notes ? Theres is a section about stability.
I have built two 3886 amps.Two inverted channels of one amp were oscillating when no input cable was connected.They heated very much , but didn't blow , even left for hours.
I guess that you made something wrong , at this is the problem.Making amp more powerful won't make it more stable - probably otherwise.
When driving capacitive loads , amp must be somehow isolated from very low impedances at high frequencies. It may be a resistor , RL circuit, or something else.
Did you read LM3886 application notes ? Theres is a section about stability.
I have built two 3886 amps.Two inverted channels of one amp were oscillating when no input cable was connected.They heated very much , but didn't blow , even left for hours.
I guess that you made something wrong , at this is the problem.Making amp more powerful won't make it more stable - probably otherwise.
LM4780
I will again say IMHO these chips are not designed to do what you want them to in a single bridge or parellel solution. There packadge disipation SAO shear inability to delivere high current to load at a respectable voltage just wont happen within their safe area of operation.
To put it in simple terms your trying to move a tractor trailer with a honda civic. 1 chip will put out around 120w parellel with a 35v/+- supply, thats a 70v swing, 120/70=1.7A, Ive seen ES's chew up amps dumping factors of ten
I would look at a dual LM4780 parellel then bridged with another dual parellel module 480W/70*2=3.4A. Done with global bootstrapped feedback some Feed forward compensation and you should get close to the qaulity of a single chip.
Back down the supply to 25+- and you should get 1/3 amp to load increase because each part of the bridge will see 4ohms, given an 8ohm load. 4ohm they will see 2 ect.... ,but at a lower voltage hence the higher current.
I will again say IMHO these chips are not designed to do what you want them to in a single bridge or parellel solution. There packadge disipation SAO shear inability to delivere high current to load at a respectable voltage just wont happen within their safe area of operation.
To put it in simple terms your trying to move a tractor trailer with a honda civic. 1 chip will put out around 120w parellel with a 35v/+- supply, thats a 70v swing, 120/70=1.7A, Ive seen ES's chew up amps dumping factors of ten
I would look at a dual LM4780 parellel then bridged with another dual parellel module 480W/70*2=3.4A. Done with global bootstrapped feedback some Feed forward compensation and you should get close to the qaulity of a single chip.
Back down the supply to 25+- and you should get 1/3 amp to load increase because each part of the bridge will see 4ohms, given an 8ohm load. 4ohm they will see 2 ect.... ,but at a lower voltage hence the higher current.
LM4780
PS I will have to correct a basic mistake the chip will drive the ES's at low frequency maybe not with the dynamics you would like, but high frequency will kill it. I took some measurments on the claritys
.1ohms at 18000 
good by to stabability hello melt down.
By the way I used two of my amps on a set of ML's they held up alright for how long I dont want to hazard, but that was at power levels high enough to cause arcking.
PS I will have to correct a basic mistake the chip will drive the ES's at low frequency maybe not with the dynamics you would like, but high frequency will kill it. I took some measurments on the claritys
.1ohms at 18000 good by to stabability hello melt down. By the way I used two of my amps on a set of ML's they held up alright for how long I dont want to hazard, but that was at power levels high enough to cause arcking.
tiltedhalo :
Where did you get that 2 chips can push 1,7 A ?
At +- 35 V rails and 4 ohm load , one chip will deliver 20/4 = 5 A avg. current.
Two parralel chips would give 10 A average current.
LM3886 have 7A internal current limiting(According to datasheet).
I have used 1 single LM 3886 chip for ESL speaker.No problems , even when trannie was saturating and actually shorting output of chip.
Where did you get that 2 chips can push 1,7 A ?
At +- 35 V rails and 4 ohm load , one chip will deliver 20/4 = 5 A avg. current.
Two parralel chips would give 10 A average current.
LM3886 have 7A internal current limiting(According to datasheet).
I have used 1 single LM 3886 chip for ESL speaker.No problems , even when trannie was saturating and actually shorting output of chip.
LM3886
Simple math and physics, V*A=W, W/V=A, If you are running 35V+ and 35V- thats a 70V R/R swing the chip is fully saturated running 67watts at .1% as per spec, that makes for 67W/70V=.95A. Yes the chip is current protected at 7 amps, but that all depends on power supply V load and the key to all that, the package dissipation of 120w.
If you actually have 70V swing at seven amps thats about 490watts thats a little over the dissipation factor of the chip.
So let back down the supply volts to 20+20- for a 40V swing at 7 amps that still 280watts.
The current limiting is for short conditions to keep the chip from going poof.
Simple math and physics, V*A=W, W/V=A, If you are running 35V+ and 35V- thats a 70V R/R swing the chip is fully saturated running 67watts at .1% as per spec, that makes for 67W/70V=.95A. Yes the chip is current protected at 7 amps, but that all depends on power supply V load and the key to all that, the package dissipation of 120w.
If you actually have 70V swing at seven amps thats about 490watts thats a little over the dissipation factor of the chip.
So let back down the supply volts to 20+20- for a 40V swing at 7 amps that still 280watts.
The current limiting is for short conditions to keep the chip from going poof.
If you're measuring output current, then 70Vp-p is not what you should be checking out.
Going on the specs that a channel is happy driving 50W into a 8ohm load (with a 35Vp-p supply):
P=(I*I)*R
I-rms = (P/R)^(1/2)
I-rms = (6.25)^(1/2)
I-rms = 2.5A
Keep in mind, this was determined from rather conservative specs and that this is also RMS power. Peak current per channel would then be 3.5A, which is 7A per chip(though I wouldn't recommend pushing it to its peak power).
If my math's outta whack, let me know, but it seems to me that these are reasonable numbers.
Also,
PDMAX = (Vcc^2) / (2*(pi^2)*RL) [as per the app sheet]
so, lets say 60Vp-p into a 0.5ohm load
PDMAX = 364W
Although this is certainly high, it is definitely attainable by someone with a mindset to make it work (even if chips may not be the best choice in this case)
Going on the specs that a channel is happy driving 50W into a 8ohm load (with a 35Vp-p supply):
P=(I*I)*R
I-rms = (P/R)^(1/2)
I-rms = (6.25)^(1/2)
I-rms = 2.5A
Keep in mind, this was determined from rather conservative specs and that this is also RMS power. Peak current per channel would then be 3.5A, which is 7A per chip(though I wouldn't recommend pushing it to its peak power).
If my math's outta whack, let me know, but it seems to me that these are reasonable numbers.
Also,
PDMAX = (Vcc^2) / (2*(pi^2)*RL) [as per the app sheet]
so, lets say 60Vp-p into a 0.5ohm load
PDMAX = 364W
Although this is certainly high, it is definitely attainable by someone with a mindset to make it work (even if chips may not be the best choice in this case)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.