Amplifier works from the caps at 90% of the time and in the remaining 10% is when the transformer is loading the caps and running the amp. So while the amp is using up the juice from the caps most of the time, the transformer has lot less time to it put back. So for that the transformer gets quite high peaks of loads.
I never under design anything.
I always design things for their fullest capability whether or not it gets used as such.
This also helps with reliability down the road too, especially when comes that special time when you want to crank it up and abuse the system for a while.
But again that is just me.
FWIW
jer
I always design things for their fullest capability whether or not it gets used as such.
This also helps with reliability down the road too, especially when comes that special time when you want to crank it up and abuse the system for a while.
But again that is just me.
FWIW
jer
Last edited:
Your conclusions are wrong because your data is wrong.
One example of your complete misunderstanding of the way amplifiers work is shown in Sft's following post.
Hello rol,
Please disregard my previous post.
I was thinking that it is for home use and not for walls demolished.....
I was wrong, so I as well as most supliers of power amplifiers that use undersized transformers from decades, even under rated output power of the amplifiers.
Actualy you need 2 transformers, one for each power rail, of at least 2KW each and a filtering capacity of minimum 1F per rail, only to be sure.
How about crest factor, you can see some measurements here, but again I was wrong....
@ Andrew, sfthurber, and so on: I speak about measured things and not read from the internet, but yes you have right, the efficiency of LM4780 it is not soo good...
Please disregard my previous post.
I was thinking that it is for home use and not for walls demolished.....
I was wrong, so I as well as most supliers of power amplifiers that use undersized transformers from decades, even under rated output power of the amplifiers.
Actualy you need 2 transformers, one for each power rail, of at least 2KW each and a filtering capacity of minimum 1F per rail, only to be sure.
How about crest factor, you can see some measurements here, but again I was wrong....
@ Andrew, sfthurber, and so on: I speak about measured things and not read from the internet, but yes you have right, the efficiency of LM4780 it is not soo good...
Attachments
Last edited:
I agree that the heatsink is a concern. Active cooling (a fan) is an option if acquiring a proper heatsink isn't easily doable.
I say the OP should give it a go with the transformer he has, though. It'll probably be plenty for dorm use.
And if not, save up for more iron.
He didn't say where he is living. Not in the states clearly. I really like that here in the states i can just order up toroids from AnTek, who sell 300VA trafos appropriate for chip amps for $41+ship. Unfortunately, international shipping on a 7 pound 5"x3.1" trafo is prohibitively expensive. Or at least a real kick in the pants.
I say the OP should give it a go with the transformer he has, though. It'll probably be plenty for dorm use.
And if not, save up for more iron.
He didn't say where he is living. Not in the states clearly. I really like that here in the states i can just order up toroids from AnTek, who sell 300VA trafos appropriate for chip amps for $41+ship. Unfortunately, international shipping on a 7 pound 5"x3.1" trafo is prohibitively expensive. Or at least a real kick in the pants.
With 4 ohm, 4"~5-1/2" woofers, decrease the waste load by putting 1000uF capacitor in series to the woofer. This will decrease the load on the amplifier, and, if high output, your capacitive coupled woofer will also prevent distortion at the speaker by Not madly flapping its cone for extra low pitches that it can't reproduce audibly anyway. You can put the 1000uF cap on back of the speaker, in series with the black/negative terminal. More or less capacitance can be used to fine tune the low end response of the speaker and its power handling.
Generally, to shove a 4 ohm speaker with a single chip, you'd need a TDA7293; however, LM3886's (of course LM4780's too) can do it with inverting mode t-network and more creative power decoupling--see Decibel Dungeon. Even so, a parallel amplifier, such as LM4780 Parallel (one channel per chip) or TDA7293 parallel (I've got a thread for that), is a more suitable solution for 4 ohm speakers.
Choose the LM4780 parallel if fine imaging is the main concern (lower volume listening).
Choose the TDA7293 parallel if high output with a pleasant tone is the main concern (party sound).
Generally, to shove a 4 ohm speaker with a single chip, you'd need a TDA7293; however, LM3886's (of course LM4780's too) can do it with inverting mode t-network and more creative power decoupling--see Decibel Dungeon. Even so, a parallel amplifier, such as LM4780 Parallel (one channel per chip) or TDA7293 parallel (I've got a thread for that), is a more suitable solution for 4 ohm speakers.
Choose the LM4780 parallel if fine imaging is the main concern (lower volume listening).
Choose the TDA7293 parallel if high output with a pleasant tone is the main concern (party sound).
Last edited:
I have two Audio Sector LM4780 boards soldered, and am waiting for Avel 500VA 25+25V transformer at this moment. While I was sure I knew what I was doing, because I have been running LM3886 amps for years now without problem, but suddenly I lost my confidence.
I am going to use LM4780 in parallel mode. The PS voltage should be somewhere like 35V. Couple questions:
1. How big heatsink do I need for this? Parallel, 35V, driving Magnepan MMG.
2. How critical is matching feedback resistors of two 3886 in one 4780?
Thanks,
Doug
Use the 3886 datasheet to determine the heatsink requirement for one 3886.
Double that for the 4780.
Then double again for the two 4780 boards.
And finally I recommend that you double what National suggest for the heatsink requirement.
eg.
for one 3886 with ±35Vdc while delivering maximum power driving an 8r0 test load with the heatsink in an ambient temperature of 25°C, National tell you you need a 3C/W sink.
For two 3886 you would need 1.5C/W
For 4 off 3886 you would need 0.75C/W
Then apply my doubling recommendation for cool running and you end up with 0.4C/W
If your voltage at the chips drops below 35V when delivering maximum power from all four chips then use that lower voltage to determine the National sink size. But you have to build a prototype to find out how the amp behaves in practice.
What does your previous 3886 experienece tell you about voltage at the chip when delivering full power?
Double that for the 4780.
Then double again for the two 4780 boards.
And finally I recommend that you double what National suggest for the heatsink requirement.
eg.
for one 3886 with ±35Vdc while delivering maximum power driving an 8r0 test load with the heatsink in an ambient temperature of 25°C, National tell you you need a 3C/W sink.
For two 3886 you would need 1.5C/W
For 4 off 3886 you would need 0.75C/W
Then apply my doubling recommendation for cool running and you end up with 0.4C/W
If your voltage at the chips drops below 35V when delivering maximum power from all four chips then use that lower voltage to determine the National sink size. But you have to build a prototype to find out how the amp behaves in practice.
What does your previous 3886 experienece tell you about voltage at the chip when delivering full power?
Last edited:
4 ohm speakers, parallel amplifiers--Each amplifier "sees" an 8 ohm load. Heatsink size is same as you'd use for 4 of LM3886, each driving 8 ohm speakers.
So, twice as much heatsink as your previous LM3886 stereo amplifier.
Stability plays a key factor in operating temperature and could also affect heatsink size requirements. A more stable amplifier could give you a cost savings at the heatsink.
You need to use a digital ohmmeter to precisely match the resistors. A good match will run cooler.
For speakers as nice as the Magnepan, why not use the Neurochrome Parallel86 amplifier? Parallel-86
Yeah, I should've done that. I didn't have chance to do it yet, and will report back once I do it.
Parallel86 looks very interesting. It's like opposite of minimalist approach of gainclones.
Have you had a chance to compare Neurochrome versions with minimalist versions?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.