Resistor sizes in chip-amps

[lhwidget]

Hi all, hopefully this isn't too dopey...

In most designs for LM3886 chips based on the reference designs from the spec sheets, I typically see 1/4 W resistors used for the signal paths and a 2W resistor specified for the Zobel.

Is there any reason for not using 1/8 W resistors in the signal path?

Also, has anyone tried lower wattage resistors in the Zobel circuit?

[CJ900RR]

I have used SMD-resistors, 0,125W in the signal-path with no problems, and so does many commercial amps. The line-level doesn't require more.

For the zobel, it may need to handle high current sometimes, that's why you should have a higher watt-rating on them. The zobel is in the amplified area of the signal, witch can be several amps. That would destroy a low-watt resistor pretty fast...

[lhwidget]

Thanks Calle, that's what I thought about the signal path resistors.

I've seen some people using 1/2 W resistors in the Zobel, I wonder if anyone has had problems? It would seem that unless there was a constant source of oscillation, there wouldn't be much current in that path...

[Dxvideo]

I ve burned some 1W 0R1s in my LM4702 amp.. But thanks god, after that event I've never used zobel in any amps..

[peranders]

Originally posted by lhwidget
Is there any reason for not using 1/8 W resistors in the signal path?

The power rating should be chosen for your needs. It's as simple as that. You'll have to do some calculating in other words. You should consider the power dissipation in the feedback resistors since you may have high AC voltage there. This is important if you'll choose to have low ohmish values.

Originally posted by lhwidget
Also, has anyone tried lower wattage resistors in the Zobel circuit?

The resistors should cope with 20 kHz at full power. That's the minimum. Make take can be seen below, R9-R11

[digi01]

i use 0805 package resistors,no problem with 3886igc.



good work!
Zang

[kscharf]

Zobel networks are only one means of stabilizing amplifiers. Many people (Peter of AudioSector) don't like them. Another way is to put a small rf choke in series with the output to the speaker. Something like .4uh to 1uh is a typical value. Winding some #24 wire on a 1W resistor (can usually be any value above 10 ohms) makes a good choke. In a bridged amp it's usually not necessary to have a choke in both sides, just one.

[pacificblue]

The size of the Zobel resistor depends on the possible output swing. If 1/4 W burns on an LM3886, there is something severely wrong with the amplifier. Some people prefer to increase the resistor rating instead of looking for the fault and correct it.

Originally posted by Dxvideo
I ve burned some 1W 0R1s in my LM4702 amp.. But thanks god, after that event I've never used zobel in any amps..

An LM4702 cannot be compared to an LM3886 in that regard. 0R1 is also not a typical value for a Zobel network. Not really surprising that it didn't work. National shows 3R3 and 20 nF in their reference project. Shunning Zobels is the wrong conclusion from your experience.

Originally posted by peranders
The resistors should cope with 20 kHz at full power.

.. if you use the amplifier to reproduce test signals.

[peranders]

I think it's a bad idea to design parts which you easy can distroy. 20 kHz full power is a reasonable power size of the resistor but 100 kHz is better but the trade of is probably worse HF properties of the Zobel.

[jackinnj]

Originally posted by kscharf
Zobel networks are only one means of stabilizing amplifiers. Many people (Peter of AudioSector) don't like them. Another way is to put a small rf choke in series with the output to the speaker. Something like .4uh to 1uh is a typical value. Winding some #24 wire on a 1W resistor (can usually be any value above 10 ohms) makes a good choke. In a bridged amp it's usually not necessary to have a choke in both sides, just one.

The resistor lowers the "Q" of the coil -- without it you can get a sharply tuned resonant circuit consisting of the inductance of the coil, the capacitance of the speaker loads and whatever reactive elements you have on the speaker end.

I wouldn't skimp on wattage of the RC snubber -- you can get the 2W Vishay PRO2 series which are about the size of a 1/2 W carbon comp resistor.