Heat dissipation of resistors, and best practices for use in a tweeter and/or attenuation circuits

tajanes

Looking at a few Madisound and Meniscus kits’ crossover tweeter HP filter designs, I see some using 10W resistors, and in other cases 25W resistors.

So what do you all consider in your designs?  

And, in an Lpad attenuation circuit would it be best practice to double up the series resistor (where the series resistor is a lower value than in the dropping resistor)?

Thx

dynamo

I’m guessing if it’s a 25w specd it’s there for a reason. Typically that seems unnecessary on a tweeter, because there is less energy in the higher octaves, unless it’s a high power application, a single series resistor doing a lot of attenuating, or a low crossover frequency. In places like a series resistor in front of a low-crossed mid, ten watters can get hot.

I either take the simple approach of driving them hard during voicing and simply feeling the resistors, or using xsim and its power dissipation chart, which isn’t super accurate (may indicate a high power resistor when in reality it may not be necessary) for a high frequency component, but can be for mid and woofer components.

ani_101

Sometimes there are legs going across the woofers which have resistors in them. If they happen to pretty stout stout s and you are bypassing a lot of power they get hit pretty fast. I had 25w ziztors in parallel and they would toast my fingers in about 10 minutes of playtime... The project is now abandoned, but for finishing it I was considering adding heat sinks to the ziztors...

jr@mac

Ani, not heatsinking those type of resistors considerably derates them. You need to mount to a heat dissipating surface or use other types. Wire wound require elevation, so a crimp in the leads works well for that purpose. 

tajanes

just referencing HP filters

Wolf

The Zisters are likely 10W without a heatsink, 25W with, judging from the Dale equivalents.

Ani- If you are using a resistor across a woofer that is lower than 30-40 ohms without another component in series with it, it is acting as a load across the amp with only the LP coil included. It will get hot, and hotter the lower the value due to current flow.

There are 4-ways to attenuate the hump in a lowpass induced by the Fs of the woofer:

- parallel resistor to woofer, cheap and sometimes toasty. I try to stay above 40 ohms though if doing this.
- series LCR notch to take out Fs, which is expensive. PBN audio does this a lot.
- Use a smaller gauge lowpass coil to involve a higher DCR to attenuate it.
- Something like what I call a cross-pad (XPad), with a CR in series with the driver, and a lowpass coil across the CR to effectively 'slope-out' the peak. Basically 2x 6dB first order filters in parallel to dip a section of the response.

FWIW, I'm using the last option on the EMP project coming up. It had a peak in the 200Hz range due to maybe a combination of a little comb-filtering rolling off the top-end earlier, and the Fs with the lowpass. It ended up taking in sim 680uF and a 3.3mH coil to get the job done. I got a quad of 1000uF 250V polarized caps from ApexJr. to use paired in series (measured small at about 450uF) and yield a higher voltage spec of 500V compared to the 100V standard NPE. Then I parallel 2x 100uF Q4 caps to those for 650uF at the moment. We'll see what the measurements say down the road. Resistor value is simmed to work best at about 2 ohms in Xsim. If you question why I paralleled the caps like that- the 200uF spot is textbook equivalent to roughly 200Hz at 4 ohms. This should sonically be better and keep the midrange as it should be.