LM4780 parallel bad distorsion

[Audio Bliss]

High

I use National's schematic for parallel operation and i have very high distorsion, when i use it in bridge with National's schematic it sound clean and beautifull, Supply voltage +-34V .What ca i do? please help

Thanks

 

[zymurgn]

Can you be specific as to how you are setting up the parallel circuit as opposed to the bridge circuit? Is it two different amplifiers on different circuit boards? Is the load impedance the same for both?

 

[Audio Bliss]

Its the same pcb for parallel and same just litlle bit modified for bridge, the same chip, same power supply and same load, a subwoofer with Z 5.8 ohm , I cant use it in bridge, the load is low and the chip is running very hot but clean sound, its very weird

 

[abraxalito]

Do you have close tolerance resistors for the gain setting (Ri and Rf)? Also are your 47uF caps (connected in series with Ri) closely matched in value? If not the two amps can fight each other. Also I reckon the 0.1ohm resistors for sharing are too low, better to use 0.22ohms.

 

[tomchr]

Note that there are some resistor changes between the bridge and parallel operation.

For bridge mode, one amp is inverting and the other is non-inverting. Neglecting the phase inversion, the gains need to be equal, but as the gain of an inverting amp is R2/R1 whereas the gain of the non-inverting amp is 1+R2/R1. (R2 goes from the LM3886 out to (-) input, R1 goes to ground for the non-inverting and to the input signal for the inverting amp).

In bridge mode, the two amps are both either inverting or non-inverting. If you take the bridge amps and just connect them in parallel, they'll fight as they now have different gains. That's probably why you're having trouble.

~Tom

 

[Audio Bliss]

The resistors are 1% type, hand matched, 47uF caps are in anti-serial conectected (+ - - +) bypassed with 1uF PP but Ri aren't closed match is a little diference of 0.1ohm, Also i try 0.1 ohm to 0.3 ohm for output resistors

Tomchr I use paralell schematic values for paralell and bridge values for bridge, in bridge all values are closed match, probably the Ri value give me the trouble...i'll check it monday...

 

[Audio Bliss]

Also i have 2 lm4780 chips an i try with both, same problem in paralell

 

[AndrewT]

the PA and BA and BPA topologies are not for beginners.
There are many pitfalls that experienced designers will be aware of and design out of their implementation.

 

[abraxalito]

The resistors are 1% type, hand matched

How closely did you hand match? 0.1% is required in this circuit.

, 47uF caps are in anti-serial conectected (+ - - +) bypassed with 1uF PP

This circuit also requires these caps to be matched - something overlooked by National's apps engs. If you don't have a capacitance meter to hand match, just increase the value to 1000uF.

 

[Audio Bliss]

I'll re matche all component, the 47 caps are not very closed match indeed ... i'll post next week the results for the others beginers

Many thanks

 

[Audio Bliss]

Still the same problem with matched component, I try Ci 1000uF same problem, 4780 GND pins was grounded to sgn gnd, now i grounded them to power gnd, same problem, it sound like a broken speaker or a very hard clipping, in bridge config i listen with attention and still same distorsion but all most imperceptible, i try to supply from 2 storage batteries same distorsion, Also i have 2 lm4780 chips an i try with both, i presume one chip was damaged but the other one was new...DC offset 4mV

 

[Mark Whitney]

Could you please post photo's of amp and power supply. How does it sound with only one channel playing (single LM3886).

 

[abraxalito]

I suggest testing each of the paralleled channels individually by pulling out the 0.1R sharing resistors one at a time.

 

[Audio Bliss]

i will re-make the pcb...but i will try later as u suggest abraxalito

 

[Sangram]

One thing you can try is much higher value current sharing resistors, 0.5 ohms is a good starting point for poorly matched chip halves.

You sure you have genuine chips?

 

[Audio Bliss]

allready try with different values for output resistors, i bought them from farnell

 

[tomchr]

the PA and BA and BPA topologies are not for beginners.
There are many pitfalls that experienced designers will be aware of and design out of their implementation.

Would you shed some light on these common pitfalls?

~Tom

 

[tomchr]

The resistors are 1% type, hand matched, 47uF caps are in anti-serial conectected (+ - - +) bypassed with 1uF PP but Ri aren't closed match is a little diference of 0.1ohm, Also i try 0.1 ohm to 0.3 ohm for output resistors

That should be fine then.

Tomchr I use paralell schematic values for paralell and bridge values for bridge, in bridge all values are closed match, probably the Ri value give me the trouble...i'll check it monday...

I'm sure tons of schematics are floating around. I don't know which you consider the "parallel" and "bridge". Here is what I am assuming:

Parallel = Figure 6 in the BPA200 App Note (NSC/TI).
Bridge = Figure 3 in the BPA200 App Note (NSC/TI).

Those should work. For the parallel amp, it is critical that the two amplifiers have the same gain. That's why 1 % resistors are recommended as a minimum with 0.1 % suggested. 0.1 % resistors are actually not that expensive these days. For the bridge amp, the inverting amp should have a gain of -10 (or -20) and the non-inverting amp a gain of +10 (or +20). That's a little bit of a challenge as you need to find a pair of standard value resistors that satisfy,

1+Rf1/Ri1 = Rf2/Ri2.

That's why in the BPA200 schematic (Fig. 13), 1k, 20.5k, and 21.5k are used. You could use 1k, 10k, 11k if you wanted a gain of 10.

I see you're using a home-brew PCB. Make sure you have proper decoupling of the ICs.

~Tom

 

[peranders]

How hot is the chip? Do you have any heatsink?

 

[tomchr]

How hot is the chip? Do you have any heatsink?

+1

Another thing to be aware of with the bridged configuration is that each amp drives half the load impedance. So if you're driving a 4 ohm speaker, each amp "sees" a 2 ohm load. Make sure you select the supply voltage and heat sink accordingly. You should still be able to play at low volume without issues assuming that the IC is properly cooled so you don't engage the SPiKe protection.

~Tom