Paralleling LM4780/3886: any impact on sound quality?

[Arnop]

Hi

it seems that this forum is one of thew main contact point for chipamp freaks, maybee u can help me here.

I've just built a3886 based Amp for a friend and it worked pretty well.
Now i'd like to build one of these for my computer.
I have very big speakers, with 12inch woofers, 4Ohm, and I want to use a 2x22V Transformer.

This seems to be a little hard for a single 3886.
Now im thinking about paralleling two LM3886 or one 4780 will give more reserves.

Is this going to destroy the sound of the gainclone?
Yes, there are hundreds of threads, but I found no answer for this point...

[AndrewT]

Hi,
22Vac and 4ohm will probably kill a chipamp if driven hard.
If you're lucky the protection will cut in and just sound terrible.

Parallel operation will make the amps much more reliable and that sounds much better than a dead chip.

[Peter Daniel]

I was actually using bridged LM3875 chips with 22V AC PS and 4 ohm speakers (effective load 2 ohm per chip) for approx 6 months and I didn't experience any major problems.

When driven hard, the heatsink got pretty hot (60 deg C or so) but the protection never kicked in. On some tracks, with deep bass, clipping occured, but that's about all. The drivers (FAL) were 95 dB efficient.

Regarding sound quality when paralleling chips, it may be perceived as slight losses in immediacy, fine detail and bass control, but depending on a system and listener experience it may not be noticable as anything serious.

LM3886 has more current output than LM3875 so maybe 4ohm woofer will not be such a burden, it also depends on listening habits and how loud you like to play it.

[SimontY]

Originally posted by AndrewT
Hi,
22Vac and 4ohm will probably kill a chipamp if driven hard.
If you're lucky the protection will cut in and just sound terrible.

Parallel operation will make the amps much more reliable and that sounds much better than a dead chip.

Hi Andrew,

Clearly you're not familiar with the LM3886. Mine has been happily driving my 4ohm speakers for a year or so just fine. The chip is rated at 68w RMS into a 4ohm load. Paralleling it's obviously good for nearly double the power and current at half the impedance. My transformers are 25vac - a touch high.

It's not really possible to kill a 3886. I can play as loud as I like within reason. All that happens is the sound softens up a bit at very high levels.

Simon

[ttan98]

I suspect sometimes ago what Andrew said may not 100% correct, he said the same thing in a no. of posts on this site, I did not say anything as I do not have any direct experience.

Now that P. Daniel & SimonY have come out and dispute him, who would rather believe P.D and SimonY or Andrew?

I believe accurate technical issues should always prevail. Thanks to P.D and SimonY for responding...

[AndrewT]

Hi,

I believe accurate technical issues should always prevail

agreed.
Follow the manufacturer's design process in the data sheet and 22Vac with a reactive 4ohm load and one finds that this combination is taking a chipamp too close to it's limits, particularly with regard to case temperature.

if driven hard

note, I placed a condition that seems appropriate to Arnop's proposed use for a bass only driver.

The saving is that music is often played at average levels around 10 to 30db below the peak level. That low average level keeps the chip cool.

Peter,
is your example using a regulated supply or on PSU using near +-1mF of smoothing?

When driven hard, the heatsink got pretty hot (60 deg C or so) but the protection never kicked in. On some tracks, with deep bass, clipping occured, but that's about all. The drivers (FAL) were 95 dB efficient.

you have been very fair to also quote the conditions that you found successful. But driven hard into 95db/W speakers (presumably widebandwidth) is possibly a sutble warning that may go straight over the head of a newbie.

I feel inclined to ask just how low the supply voltage at the chip pins fell when clipping was heard?

[SimontY]

Arnop,

With some additional details we can probably tell you if you'll be ok with just the single LM3886 or not.

How loud do you tend to listen? What efficiency are your speakers? Do you intend to use it for long, loud party sessions?

Just for sake of example, my speakers dip slightly under 4ohms, are about 91db efficient and I play fairly loud, sometimes very loud (to near the limit of the speakers). My voltage rails are + and - 35vdc unloaded. My transformers are 100va @ 25vac, one per side. I have never heard clipping. I have reasonable heatsinks only and run the amp in a nearly sealed plywood chassis. It gets warm, I expect it's exceeded 60 degrees C. This could be a problem for parties etc. if caution is not excercised.

I think clipping is hard to achieve because of the high voltage rails and low va transformers, I simply run out of current at high levels (I guess, not really sure what happens).

I actually have some interest in paralleling these chips myself as this amp was only built as a test to see if it could stack up against my Roksan Caspian. It did and it's been in use for many months since! I may actively bi-amp and add an extra bass section to my speakers, which could be powered by a paralleled LM3886 (would be 2ohm load).

Simon

[Peter Daniel]

Originally posted by AndrewT
Peter,
is your example using a regulated supply or on PSU using near +-1mF of smoothing? you have been very fair to also quote the conditions that you found successful. But driven hard into 95db/W speakers (presumably widebandwidth) is possibly a sutble warning that may go straight over the head of a newbie.

I feel inclined to ask just how low the supply voltage at the chip pins fell when clipping was heard?

It is unregulated supply with 100uF capacitors directly at the chip and 1000uF at the rectifiers (2ft umbilical).

I didn't had a chance to measure the supply fluctuation when clipping. It was basically occuring only on Burmester CDIII, track 10 and Patricia Barber's Companion, track 4.

[Arnop]

Hi

thanks for your answers.

I placed a condition that seems appropriate to Arnop's proposed use for a bass only driver.

No, you misunderstood. It's a 3 way speaker with 30cm Woofers. Old Canton GLE Series.

What efficiency are your speakers?

I dont know
The manual gives two pairs of variates:

SPL 86dB (3m distance) at 5,8W and
SPL 80dB (3m distance) at 1,45W

Regarding sound quality when paralleling chips, it may be perceived as slight losses in immediacy, fine detail and bass control

Losses in bass control?
Losses in fine detail and immediacy... Well i am no HIFI expert and I dont want to say anything heretic, but maybe your single chip has just a bad bass responce, an thus you experience more emphased high and middle frequencies?
I have read somewhere that a parallel chipamp had a "deeper" sound.

But why losses in bass control?


When i compare the diagramms in the national semiconductor data sheets, i can't se any difference in distortion between 3886 at 4Ohm with 28V and the parallel circuit at 35V 4Ohm.


I don't want to create any sound, thats not how i understand HIFI.
I want no sound at all, it should be neutral. Just amplify the input

[ttan98]

Andrew,

look up this site you will find the LM3886 can drive 4.7ohm up to 90% full power using square wave.

http://dogbreath.de/Chipamps/GainCar...nCardCopy.html

any more questions you should ask the author.