
Home  Forums  Rules  Articles  The diyAudio Store  Gallery  Blogs  Register  Donations  FAQ  Calendar  Search  Today's Posts  Mark Forums Read  Search 
Class D Switching Power Amplifiers and Power D/A conversion 

Please consider donating to help us continue to serve you.
Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving 

Thread Tools  Search this Thread 
5th June 2009, 11:21 AM  #1 
diyAudio Member
Join Date: Feb 2009

Zobel networks
Hi all,
just looking at designing in my zobel network to the output stage of my class D amp. A few questions before I continue. 1. Everyone seems to stress how important it is to flatten your impedance in the higher frequency range near the crossover band. What I don't get is, if there is a rise in impedance which causes some phase shift and a bit of additional impedance at the crossover band, how can it possibly result in such huge issues as people seem to say? Keep in mind I've never actually had the chance to compare the differences, so perhaps somebody can put its importance into better terms than "it messes with your crossover". 2. How does one determine the power density of your average audio signal to know how to size the resistor? 1W, 10W? I am designing a 200W/channel amp and have referenced a similar 200W design using a 1W resistor. Considering that if the impedance of the speaker itself rises to say 200300%, that means the zobel is taking most of that high frequency energy and 1W seems hardly enough. 3. Is another key reason for a zobel network in class D stability on the output filter? Many thanks in advance. James 
10th June 2009, 03:04 AM  #2 
diyAudio Member
Join Date: Feb 2009

Is this a silly question or am I perhaps barking up the wrong forum?
Thanks 
10th June 2009, 06:31 AM  #3 
diyAudio Member
Join Date: Jun 2004
Location: Warsaw

LC lowpass filters are designed for a given load resistance, so they have a desired frequency response only at a ceratain load R.
If load R is lower, they are just less steep, which is not a big problem, but if the load has too high resistance and/or series inductance (as most speakers do), they have a bump in frequency response. That's why nopostfilterfeedback class D amps without zobel boost hights, even by several dB. Zobel is there to provide low enough resistance to ground at audio hights and resonance frequency. Play a bit with simulating simple LCR ciruits and you will see. 
10th June 2009, 06:57 PM  #4 
diyAudio Member

LC filters without load result in a big resonant peak, only limited by parasitic losses in L and C.
Some amplifier control schemes require RC networks to provide a worst case maximum load resistance and limit the amount of peaking. The resistive component of the load damps the resonant characteristic of the filter. Phase response of the filter is also affected. Some postfilterfeedback schemes require a maximum load resistance for proper phase margin (to avoid the phase getting too close to 180 degree at a too low frequency).
__________________
I use to feel like the small child in The Emperor's New Clothes tale 
10th June 2009, 10:46 PM  #5 
diyAudio Member
Join Date: Feb 2009

Thanks guys,
I'll muck about in the simulator. Any idea on how one determines the power density the resistor in the zobel needs to carry? By looking at a few speaker impedance curves, it seems not uncommon to expect the real resistive impedance of the zobel to be 1/3 of the inductive impedance in the speaker as it rises. As such, the zobel network is actually taking a fair amount of the 'real power' above and beyond its designated frequency, say 1kHz. How can I make a good estimate on how to size the resistor to handle the power it must dissipate as heat? I am assuming there In one example I saw, a 1W resistor was in the zobel for a 200W amp, which implies the power density outside of the woofer range is really quite small. 1W just seems to small to me. 
11th June 2009, 01:31 AM  #6 
diyAudio Member

Note that you only need the RC network to act at and above filter resonance. In practice this means that the imopedance of the C only becomes low above 10Khz or so and the resistor only starts to dissipate substantial power above that frequency too.
__________________
I use to feel like the small child in The Emperor's New Clothes tale 
14th June 2009, 02:18 AM  #7 
diyAudio Member
Join Date: Feb 2009

That seems to disagree with some literature I have read on the subject. The literature emphasized that you try to keep the impedance flat 'after' the woofer's resonance peak. The inductive rise of the midhigh range drivers seems to always pick up very shortly after that.
Perhaps this literature was aiming to use the zobel more for the passive crossover than it was the LC low pass on the amplifier output? 
15th June 2009, 07:04 AM  #8 
diyAudio Member
Join Date: Dec 2003
Location: Nottingham UK

The Zobel network that you might place at the output of a classd amplifier serves a different purpose from the inductancecompensation Zobel that you might put at the input to a passive loudspeaker crossover network. (Which is simply to flatten the load impedance).
On a prefilter NFB classd amplifier the network reduces the Q of the filter at resonance, which reduces the peaking of the filter into a highimpedance load, thus providing a flatter frequency response. In a postfilter NFB UCD amplifier, the Zobel may still be necessary to ensure that the oscillation remains stable into a highimpedance load (such as you might get with a transformercoupled 100V output amplifer with only a few speakers connected at lowpower tappings). In this case the filter phase response without the Zobel might have a very sharp transition from 0 to 180 degrees at the filter resonance frequency. A UCD amplifer could go into a much lower frequency of selfoscillation if operated like this. The Zobel slowsdown the rateofchange of phase when operated into a high impedance. This helps the overall loop response to provide a steeper slope on the phase as it passes through the 180 degree point (the desired selfoscillation frequency) A typical Zobel here might be 10R + 220nF. 
Thread Tools  Search this Thread 


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Biamping and Zobel Networks  Won  MultiWay  38  11th October 2014 02:23 AM 
About snubbers(Zobel networks)  CECCO  Chip Amps  0  15th June 2008 09:40 AM 
How do I parallel T networks?  classd4sure  Class D  25  5th October 2006 05:48 AM 
building zobel networks  zuki  Parts  0  27th August 2006 05:52 AM 
zobel networks and imoedence reduction?  Nanook  Full Range  4  21st January 2005 08:33 PM 
New To Site?  Need Help? 