keith_correa
Well-Known Member
I'm confused on why you are using different terminology [although both are correct!] for the zobel/r-c network. Isn't the r-c network that you implemented a zobel and vice-versa?
RC network - Termination for speakers / Amplifiers
- Thread starter Hari Iyer
- Start date
Yes they are the same. In fact the use of terminology is to avoid confusion. The rc network will imply to be used with amplifier termination, while the Zobel will imply to be used with woofers.
Though both circuits use a r & c the method to compute their values and it's usage is quite different.
keith_correa
Well-Known Member
For which zobel, Hari?
No for the rc network
Fantastic
Well-Known Member
Zobels and RC networks.
A Zobel network ( also called Boucherot cell) like the RC network across a driver is a resistor in series with a capacitor used across the amplifier output .
The Zobel network in the amplifier is to ensure that the amp never sees a pure inductive load which is an open circuit at HF . Some amps can oscillate at very high frequency under that condition. The resistor value is typically equal to or close to the load that the the amp is designed for. Typically 8 ohms or 10 ohms. Different capacitor values have been tried over time and typically a small range of values around 0.1uF has been found to be sufficient. This can vary from just under o.1uF to about 0.47uF depending on what the designer determines is best after doing some practical tests. Too large a value of C will increase the HF load on the amp ( lower impedance) causing additional problems at HF.This Zobel is for stability and NOT FOR SOUND !If your system sounds better with it , it could be that some tendency to oscillate has been eliminated. if you look at the oscilloscope signals you will see that some amps can have 'a small burst' of oscillation on one half of the music wave ,at times ,and is often level dependent. So while not oscillating outright it WILL affect the sound.....if present.Eliminating this will make it sound better. However no decent amplifier in the market today will have this tendency to burst into oscillation that requires the user to add RC components on them.
I might add that often such tendencies to have bursts of oscillation are solved in the lab at the design stage and does not involve using the Zobel network to cure it!
While on the subject of amplifier instability there is one more component used on some power amplifiers. This is a series inductor of very low value with a resistor across it (in parallel) between the amp output and the speaker load. This inductor isolates the amplifier output from capacitive loads which could also cause severe instability in some designs. Sometimes it could just be the capacitance of the speaker cables ! This inductor helps to prevent that from happening. Again ,decided by the designer in the lab ! The resistor is to damp the inductance . This is because the inductance and the load capacitance is an oscillatory network and will cause a dying oscillation ( ringing) on square wave tests. The resistor damps this oscillation.
RC networks 'directly across the driver' or in the crossover board are used for a different reason. This is to compensate and nullify the inductance of the speaker voice coil.Have a look at a speaker (driver) impedance curve and you will find that starting around the mid range or earlier the impedance starts to rise rapidly due to the voice coil inductance.Now crossovers are designed assuming a resistive driver , 8 ohms or what ever it is. But due to the rising impedance the crossover will not behave as expected since the load on it keeps varying with frequency. So a RC network is placed in parallel with the speaker driver which brings down this impedance to an almost level line like a resistive impedance. The crossover then behaves as designed as the load is now practically resistive. Again the resistor is approximately equal to the dc resistance of the voice coil and the capacitor value calculated to null out the rising response. The value will be specific to the driver that is used. Practically they calculate the capacitance value but select the final value by testing the value that keeps the impedance curve at the level they want. This is done by measurement and has nothing to do with the sound ! While it does affect the sound ( due to a rising HF response for an uncompensated driver) , the components are selected by measurement. You need proper instrumentation to do this effectively.
So why do some amplifiers not use a Zobel network? That's because by design the amplifier is already stable with low inductive loads. Adding a Zobel on an amp which doesn't need it for stability will not improve the sound. It's just an additional load and best avoided. However many amplifiers do use them. The user needn't add them.
So why do some speakers not use RC compensation networks ? Many good brands do. Some don't and there could be many reasons. To save on cost. Or because the electrical rising response doesn't matter as the speaker's actual acoustic response falls very rapidly compensating for it practically. Or the crossover itself can be designed without using the regular values, to take this problem into account and reduce parts count. All this depends on what the speaker designer wants to achieve.
Some of the best speakers don't have any RC compensation and even use a simple 6 dB crossover ! They can do this because the acoustic response of the drivers are taken care of in the driver design itself.The driver's own acoustic response doesn't require steeper crossovers to achieve the response they want. Such drivers tend to be quite expensive too !
So would it be good to arbitrarily add the RC network on such speakers ? It will affect the sound as it will now deviate from what it was designed for and will not be as FLAT as they made it. BUT, you might like it better with a response away from the original design. Without measuring equipment you will not know what has happened. But if you do this and like it , don't assume you made it better !
Some people like using graphic equalisers in their systems. Do they make the sound better ? They ALTER the sound to suit their taste !Some people like more salt or sweet in their food. They are altering the norm to suite their personal taste. It's nothing about 'being better or more accurate'. It's all about " I want to do this, as I like it that way !".
Making significant component changes or additions can only be done properly with sufficient instrumentation to measure and determine what you have done. Maybe the designers know technically more than the people making their own changes ? But then, you can of course experiment !
If all of this is not for you or goes over your head, it means you should not meddle with the system you have ! Just enjoy it as it is, I'm sure you will !
A Zobel network ( also called Boucherot cell) like the RC network across a driver is a resistor in series with a capacitor used across the amplifier output .
The Zobel network in the amplifier is to ensure that the amp never sees a pure inductive load which is an open circuit at HF . Some amps can oscillate at very high frequency under that condition. The resistor value is typically equal to or close to the load that the the amp is designed for. Typically 8 ohms or 10 ohms. Different capacitor values have been tried over time and typically a small range of values around 0.1uF has been found to be sufficient. This can vary from just under o.1uF to about 0.47uF depending on what the designer determines is best after doing some practical tests. Too large a value of C will increase the HF load on the amp ( lower impedance) causing additional problems at HF.This Zobel is for stability and NOT FOR SOUND !If your system sounds better with it , it could be that some tendency to oscillate has been eliminated. if you look at the oscilloscope signals you will see that some amps can have 'a small burst' of oscillation on one half of the music wave ,at times ,and is often level dependent. So while not oscillating outright it WILL affect the sound.....if present.Eliminating this will make it sound better. However no decent amplifier in the market today will have this tendency to burst into oscillation that requires the user to add RC components on them.
I might add that often such tendencies to have bursts of oscillation are solved in the lab at the design stage and does not involve using the Zobel network to cure it!
While on the subject of amplifier instability there is one more component used on some power amplifiers. This is a series inductor of very low value with a resistor across it (in parallel) between the amp output and the speaker load. This inductor isolates the amplifier output from capacitive loads which could also cause severe instability in some designs. Sometimes it could just be the capacitance of the speaker cables ! This inductor helps to prevent that from happening. Again ,decided by the designer in the lab ! The resistor is to damp the inductance . This is because the inductance and the load capacitance is an oscillatory network and will cause a dying oscillation ( ringing) on square wave tests. The resistor damps this oscillation.
RC networks 'directly across the driver' or in the crossover board are used for a different reason. This is to compensate and nullify the inductance of the speaker voice coil.Have a look at a speaker (driver) impedance curve and you will find that starting around the mid range or earlier the impedance starts to rise rapidly due to the voice coil inductance.Now crossovers are designed assuming a resistive driver , 8 ohms or what ever it is. But due to the rising impedance the crossover will not behave as expected since the load on it keeps varying with frequency. So a RC network is placed in parallel with the speaker driver which brings down this impedance to an almost level line like a resistive impedance. The crossover then behaves as designed as the load is now practically resistive. Again the resistor is approximately equal to the dc resistance of the voice coil and the capacitor value calculated to null out the rising response. The value will be specific to the driver that is used. Practically they calculate the capacitance value but select the final value by testing the value that keeps the impedance curve at the level they want. This is done by measurement and has nothing to do with the sound ! While it does affect the sound ( due to a rising HF response for an uncompensated driver) , the components are selected by measurement. You need proper instrumentation to do this effectively.
So why do some amplifiers not use a Zobel network? That's because by design the amplifier is already stable with low inductive loads. Adding a Zobel on an amp which doesn't need it for stability will not improve the sound. It's just an additional load and best avoided. However many amplifiers do use them. The user needn't add them.
So why do some speakers not use RC compensation networks ? Many good brands do. Some don't and there could be many reasons. To save on cost. Or because the electrical rising response doesn't matter as the speaker's actual acoustic response falls very rapidly compensating for it practically. Or the crossover itself can be designed without using the regular values, to take this problem into account and reduce parts count. All this depends on what the speaker designer wants to achieve.
Some of the best speakers don't have any RC compensation and even use a simple 6 dB crossover ! They can do this because the acoustic response of the drivers are taken care of in the driver design itself.The driver's own acoustic response doesn't require steeper crossovers to achieve the response they want. Such drivers tend to be quite expensive too !
So would it be good to arbitrarily add the RC network on such speakers ? It will affect the sound as it will now deviate from what it was designed for and will not be as FLAT as they made it. BUT, you might like it better with a response away from the original design. Without measuring equipment you will not know what has happened. But if you do this and like it , don't assume you made it better !
Some people like using graphic equalisers in their systems. Do they make the sound better ? They ALTER the sound to suit their taste !Some people like more salt or sweet in their food. They are altering the norm to suite their personal taste. It's nothing about 'being better or more accurate'. It's all about " I want to do this, as I like it that way !".
Making significant component changes or additions can only be done properly with sufficient instrumentation to measure and determine what you have done. Maybe the designers know technically more than the people making their own changes ? But then, you can of course experiment !
If all of this is not for you or goes over your head, it means you should not meddle with the system you have ! Just enjoy it as it is, I'm sure you will !
Last edited:
@ Fantastic
Too much gyan and nothing concrete. I dont usually read anything more than 6 to 10 lines to be honest. I am too too too lazy for that.
If you happen to be in this part of the town and have the inclination for actual discovery do ping me.
Cheers,
Too much gyan and nothing concrete. I dont usually read anything more than 6 to 10 lines to be honest. I am too too too lazy for that.
If you happen to be in this part of the town and have the inclination for actual discovery do ping me.
Cheers,
@Hari: I'm trying out 0.47 uF in series with a 8 ohms 5 watts resistor (at speaker binding post end). It's sounding rather good
Great. If you are using WW resistor then you should replace them with either non-inductive type or a Metal oxide type else there again will be noise. You can stay with 2 Watts without any issues if your amp is less than 25 Watts RMS. Mine was just 6 watts hence am using just a 1 Watt MOR with no issues.
Correction : 0.047, not 0.47 uF.
Tried the same on a friend's system and we observed similar improvements. It's worth a try.
Yes for an 8 ohms load 0.047 mfd works great and for a 4 ohms load 0.1 mfd works with similar results.
keith_correa
Well-Known Member
So, we have 3 types of zobels being discussed here:
1. Near end zobel (at the amplifier outputs)
2. Far end zobel (at the speaker binding posts - at the end of the speaker cable and before the passive x-over)
3. X-over zobel (as part of the passive x-over)
Hari has iimplemented this at # 1 for his amplifier (from what I understand). I think Joshua has implemented this at # 2 (and his speaker x-over probably has # 3 implemented too).
Am I correct in what I understand?
1. Near end zobel (at the amplifier outputs)
2. Far end zobel (at the speaker binding posts - at the end of the speaker cable and before the passive x-over)
3. X-over zobel (as part of the passive x-over)
Hari has iimplemented this at # 1 for his amplifier (from what I understand). I think Joshua has implemented this at # 2 (and his speaker x-over probably has # 3 implemented too).
Am I correct in what I understand?
keith_correa
Well-Known Member
And my other question is: If #3 was already implemented as part of the x-over, would #2 be needed? If yes, what for?
Yes, I have implemented # 1 and # 2 is part of my x-over design. I am not sure what is # 3. Any ckt diagram to look at?
keith_correa
Well-Known Member
Hari, what you understand as #2 is actually #3 - the zobel is part of the x-over.
#2 is - the zobel connected at the end of the cables at the speaker cabinet side or between the speaker binding posts. This, I think is what Joshua implemented but I'm not sure. You get commercially available cables with this built in as part of the cable at the speaker end and there are also commercially available "devices" implementing this zobel which you connect across the enclosure binding posts.
keith_correa
Well-Known Member
I thought so. Joshua, take a look at this: Loudspeaker Cable Characteristic Impedance
Maybe you can sim this with Multisim or whatever since I'm sure you know the LCR of your cables. Oh, and also - Ceramic caps and metal film resistors (as Hari said) please.
I will prefer polyester or MKT type capacitors over ceramic. Also will be available with higher voltages than the ceramic ones.
My tube amplifier output impedance is 6 ohms, my speaker cable would have a characteristic impedance of around 150 ohms ( measured and calculated). My speakers are around 3.6 ohms impedance. To match the amplifier and load i have now used a 5 ohms + 0.91uF MKP RC to terminate at the speaker binding post. Will need to tune by ear the amount of impedance damping required after 10kHz for smooth response.
The RC termination at the amplifier end will benefit a solid-state amplifier as the impedance at the amplifier end will be infinite in a SS amplifier and this could be a cause for HF oscillation at high frequency. For a tube amplifier it can be anywhere between 4 ohms to 16 ohms depending upon the tapping used. So for a tube amplifier, the termination will any not be required at the amplifier end.
When you connect a speaker cable at this amplifier end which has a source impedance of 4 ohms, the characteristics impedance of the cable could be anywhere between 100 ohms till 500 ohms depending upon the AWG size and length of the wire used. Most speakers will have an impedance anywhere between 3 ohms till 10 ohms @ around 1KHz. This speaker cable will be a mis-match between the amplifier and the load (speaker).
The idea is to use an RC series termination across the speaker binding post to match the cable impedance with the load. For this the following is required,
- calculate the self inductance of the cable based on its AWG size and length
- measure the capacitance of the cable used with an LCR meter
The resistor value is easy to determine as its same as the DCR at the speaker terminal. The capacitance needs to be calculated so that it matches the impedance of the speaker based on the formula for the Characteristic Impedance - transmission line. ie. SQRT(L/C). For my speaker and cable the value is R=3.8 ohms and C=0.22uF.
I have emailed you all the details and you can check that out.
What i have noticed with simulation is - if you are connecting your FR directly to the amplifier terminal without anything in-between them, then the speaker load to the amplifier becomes purely inductive for the entire bandwidth. This can cause some harshness at the extreme ends of the HF range. Terminating with an RC at the speaker binding post will cause the impedance phase start to become capacitive above 15KHz (in my case). IME, this removes some of the harshness in the HF as per my listening as the load is not purely inductive for the entire bandwidth.
In EI core and Double C-core type of OPTs, there is a capacitive coupling between the primary and secondary winding of the OPT due to the nature of the windings in the OPTs. In an R-core based OPTs, the capacitive coupling is absent as the primary and secondary windings are on two separate leg of the R-core. This could result in less distortion and harshness in the overall presentation of the audio signal imo. I am not an expert in this - but things to consider.
Last edited:
For excellent sound that won't break the bank, the 5 Star Award Winning Wharfedale Diamond 12.1 Bookshelf Speakers is the one to consider!
