Paralleling LM4780/3886: any impact on sound quality?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
22Vac and 4ohm will probably kill a chipamp if driven hard.

I will have to agree with Andrew T on this one. My speakers are over 100db@1W 1M and they dip to 4 ohms at low frequency. When driven hard the LM3886 will clip and bark like a dog. I haven't had it go into protection yet because the heatsinking I use if gross overkill. I am a firm believer that once these chips do go into protection it becomes easier after that and they soon fail.
My experience with the smaller chips in a Peavy UMA or UM series commercial amplifier do prove the amplifier will not handle low impedance loads for any extended time period despite the parameters that are published.

My experience with the LM4780 is its a total waste of parts, money and time. I built a bridge version and the LM3886 will put it to shame. It will not be a total waste however because I am going to use it to drive a bipolar output stage.
 
burnedfinger's quote

"My experience with the LM4780 is its a total waste of parts, money and time. I built a bridge version and the LM3886 will put it to shame."

LM4780 application note states that the bridge version cannot handle 4 ohms...only parrallel version can. It is not a waste if you build the latter.
 
Parallel LM3886

I am using a paralleled LM3886TF in my home system now. It drives B&W 801 S2 and 15 ohm Lowthers equally well.
Reading the datasheet, the 100 dB, high impedance Lowthers was a concern. They are 12- 13 ohm dcr, and reach about 40 ohms at 20Khz.
Never heard any problems and this was listening at fairly loud levels a few times. I have driven both with a single LM3886 amp and could hear no downside of the Lowthers, which got to be a worst case scenario. The power level is 10 mw for 88dB. The possible mismatch between chips should stand out at these levels.
Needless to say, it is much better suited to drive the 86 dB, 6-8 ohm multi-driver B&W's. This combo really sings. So flat in frequency response and uncolored. Not only does it handle the bottom better, it just sounds more releaxed across the freq range than a single chip. Must be dealing with the complex crossover better.
My gut feeling is that it will take a really good discrete amp to match a paralled LM3886. and this is so much easier to build.

George
 
burnedfingers said:


My experience with the LM4780 is its a total waste of parts, money and time. I built a bridge version and the LM3886 will put it to shame. It will not be a total waste however because I am going to use it to drive a bipolar output stage.

The LM4780 is just 2 LM3886's in one package.

You seem to be the only one having such difficulty, and yet you haven't provided the forum any empirical evidence that the bass performance is deficient.
 

Attachments

  • 4780pkgpins9a-10.jpg
    4780pkgpins9a-10.jpg
    64.4 KB · Views: 499
You seem to be the only one having such difficulty, and yet you haven't provided the forum any empirical evidence that the bass performance is deficient.

Isn't that what we count on here in the land of DIY audio? There are thousands of threads with people swearing that they have just listened to the miracle interconnect cable or speaker cable or you name it and they will swear to it. This is with empirical evidence that no one questions.

My empirical evidence that is based upon my experience and or observation suddenly is without merit. Well what can I say? The amplifier will pass a low frequency sine wave without a problem but unfortunately sounds like dog do do playing music. I'm sorry! There is no need for you to take offense because I am not blaming your PC board. It may contain some flaws but nothing that will stop the amplifier from functioning. Unfortunately amplifiers do sound differently and I'm sure I can probably get at least one person to agree with that. There are some that will blame the speaker because it imposes a load of 4 ohms instead of 8 in the low frequency region. The amplifier sounds the same driving a speaker with a 8 ohm impedance in the low frequency.

I still feel its junk
 
To burnedfinger

There is a possiblity that the wiring is not complete ie the input or output wirings are too loose or number of strand of wires missing(hence too few strands of wires use leading to poor bass).

Something to consider and also parallel the LM4780 not bridged.
 
Dengbej said:


That's what I've been looking for. Thanks Peter.
Is it worth to apply discrete decoupling caps for each opamp?
Hi,
just as important, where do each of the V+ and V- supplies go to?
Which parts of the two circuits are powered by which pins?

I think it would probably benefit the performance by adding approriate decoupling to EVERY supply pin.
 
I am honestly beginning to believe that you are afraid of sending the board back to me to see what ails it -- hey, I'll pay the shipping back here to see if there is anything which needs correcting

Jack,
There is no need for me to send the board back period. There is NO problem with the quality of my work. I work in/on large commercial systems everyday and my judgement and work goes unquestioned. There is nothing that needs correcting.......

The LM3886 however does shine for what it is and unfortunately does out perform the 4780. My experience has you pointed out is largely empirical in nature and you are correct. However, I have run some brief tests however which confirm the unit does pass low frequency. Unfortunately as I pointed out the unit sounds like a dog that went to the Vet and got cut.
 
burnedfingers said:
The LM3886 however does shine for what it is and unfortunately does out perform the 4780. My experience has you pointed out is largely empirical in nature and you are correct. However, I have run some brief tests however which confirm the unit does pass low frequency. Unfortunately as I pointed out the unit sounds like a dog that went to the Vet and got cut.

I find my LM3886 great. Do you think your problems are due to the way the 4780 is designed and built? And if so would a manually configured parallel LM3886 sound better/good?

Simon
 
Idea to maybe improve bass

My chips amps have not been basic "gaincard clones" in design or layout. They have extensive filtering and large power supplies.
One small thing done that really increased the bass slam and apparent bass was local decoupling. 22 ufd Pana FM caps right on the pins.
The leads are cut very short, less than a 1/2 inch. More like a 1/4 inch. I think a small value, low impedance cap right on the power pins really will add some jump to the mix.
If anything try a smaller value than 22 ufd, 10 ufd may be even better.
The amp used now has a discrete bridge and 2 10,000 ufd TSHA caps per chip, 4 bridges and 80,000 ufd. Then smaller caps, rail to rail caps, and finally the little ones at right angles under the boards.
There are 0.033 ufd film and 0.15 ohm resistor snubbers across all 8 TSHA caps.
I do not expect a LM4780 to sound exactly like a pair of LM3886. Also would not expect to hear large differences.
These are very revealing chips. People have listened, tweaked, and listened some more. I know several have posted that there are certain brands of resistors that work best in all the positions.
It takes either a very revealing chip or a good imagination to decide which of 5 brands of resistor is best for one location in an amplifier.
Try the fast, low value caps to see if it does not improve the bass that seems lacking. If the bass seems this bad it is most likely not low end that is the problem.

George
 
Simon,

A single, parallel, or bridged 3886 will totally kick the 4780 to the curb. The 4780 has less bass responce than the 3886 does. I feel that it very well may have something to do with the design.

Panelhead,

I have also used various sized bypass caps in order to get the bass response tighter and more defined. The power supply is up to 40,000 mfd per channel on the dual mono supplies. The transformers are 850VA if my memory is correct. So the power supply should not be lacking in any way.

The LM3886 when driven with a small 160 VA transformer and a single pair of 12,000 mfd caps will put the 4780 to shame.
 
LM4780 is a dual LM3886 in a single substrate which is slightly larger than the size of a single LM3886.

It is expected the rate of power dissipation of LM3886 is much better than LM4780. This may affect the sound quality eg the bass end.

I am building the LM4780 using 1 chip as stereo and later parallel them. I would be using a large heat sink and also special thermal paste used in CPU & heatsink combination, called Artic Silver. This combination will give a better heat dissipation from the LM4780 substrate. Hence should give better sound quality.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.