zobel network

[lapsteelman]

Forgive a newby if this is a painfully basic question.  I am wondering about zobel networks. 
First off, How big does the resistor have to be? I am working with the National lm1875. In the databook it looks like they just use a 1/4 watt 1 ohm resistor. In a schematic in Make magazine they use a 1 or 2 watt resistor.
The other thing I am wondering. Is there a formula for calculating the resistor and capactor value of these. For example I have no 1 ohms in my box but I have plenty of 10 ohms. Can I change the cap value and make this work?
Thanks for your time.

[J M Fahey]

I'm *sure* there is such a formula, but it's probably complex and I'm sure you'll need to know some obscure transistor parameters.
Fortunately these chipamps are designed by great brains , who know all about it, so, in a nutshell, if they ask for 1 ohm, use that.
No substitutions.

[teemuk]

There is a formula, and it actually relates to speaker parameters; I discuss that briefly in my book.

The issue is that the formula just tries to linearize the speaker's rising inductance at higher frequencies, resulting into a flatter loading. It makes sense in complex crossover -equipped systems but plain systems loaded by just a single loudspeaker don't need anything fancy and the components are often just picked more or less arbitrarily.

Typically, you want the resistor to be about the same value as the nominal load impedance but some designs go with "safer" values using even smaller resistance that causes even greater attenuation of high frequencies. Some designs use a bigger value instead, such as 10 ohms. It's often a perfectly good choice too. The capacitance depends on the sought RC filter cutoff frequency but again you often just see an arbitrarily selected value as the cutoff frequency is quite high to be of any real concern. If the resistance is small, you typically see smaller capacitance values, if the resistance is high you see higher values.

A typical all around zobel is 10R + 100nF but if a datasheet suggest something else then go with that. It doesn't really matter too much, the particular circuit is almost never audible in any way. If the designers who wrote the datasheet suggest the use of some atypical values a good chance is it's for a reason, or at least it's safer to expect so.

So.... you have plenty of 10-ohm resistors and you need a single 1-ohm resistor. Solution is simple: wire ten in parallel. If those were 1/2W resistors they essentially make up a 5W resistor in total. It's really as simple as that.

The power dissipated by the zobel under normal operating conditions is quite low, as typical output signal often contain very little high frequency content that would actually get shunted by the zobel in great extent. But if the amp oscillates the zobel gets put on a higher stress. You can estimate some ballpark values for safe dissipation ratings based on amp's output power and then decide how safe you want to go. Use of higher power dissipation there isn't really an issue but resistors with lower power dissipation ratings may go down with a puff of smoke if the amp misbehaves, is put to output very high frequencies, or if the amp just happens to be very high powered one.

[lapsteelman]

Thanks guys.  I thought about doing the 10 10ohms but it seemed a tad cumbersome.  I might try the 10 ohm + 100nf  or just wait untill I work up a Digikey order.

Thanks again

[joecool85]

Great explanation of zobel networks Teemu!