RC network - Termination for speakers / Amplifiers

Hari Iyer

Well-Known Member
Joined
Feb 8, 2010
Messages
3,822
Points
113
Location
Mumbai
Recently was discussing with Bobby (Merlin Speakers) about the RC network termination suggested by him for the Merlin TSM speakers. Bobby explained in his mail in detail how the RC network when used to terminate the amplifier speaker terminal helps to reduce oscillations, ringing at HF and which in turns helps to open upon the sound stage making it more smoother and less fatigue prone.

Over the past 2 days i experimented with his theory and below are my impressions,

- tried initially with a roll-off frequency of 380KHz with a 0.1mfd polyester capacitor for a 4 ohms speaker load. And as explained by Bobby the HF did really get tamed and become more open and less strained. Listen for around 1 hr and was really very pleasing.

- tried again with a roll-off frequency of 100KHz with a 0.33mfd polyester capacitor for a 4 ohms speaker load. This did not reduce the HF as much as the first case and the effect was not that great (though present) as with the 380KHz roll-off.

With the 380KHz roll-off allowed me to increase the volume over 3 o'clock position without any sreachy or eering HF and acutally opened up details in the recording by increasing the dynamic range.

Conclusion: Simpler solutions are always better, cheaper and much easier to implement for complex problems (Theory of Constraints).
 
Last edited:
Hi hari

I am non tech guy so excuse me if my questions seems dumb

I wanted to know why not make the changes in the xover itself if there is such a pronounced Change in sound with the rc network
 
With DIY you can do it. As again explained by Bobby 97% of the amplifiers do not have this termination and this can easily go in the crossover. But having it initially outside is easy to tweak and experiment. Also gives that classy diy look to the speakers with the rc termination at the rear.
 
bakwas

if a guy really wants to make a good sounding speaker
this will be incorporated by him in his crossover

he wont leave it to chance or the amplifier maker
especially if the cost is only 10$

this is done to make money
or to hide the deficiency of the design in the first place

my personal feeling
 
Most high end amplifiers have this termination and will not require them. I have seen them in the audiophile grade Beta 22 class A amp. If you read circuit diagram you can go.to the site yourselfI and see this termination. If an high end amplifier did not have them then it falls under ultra high end where there are no oscillations in HF region and will not need them any way. The termination benefits the amplifier more than the speakers. The HF oscillations, ringing, RF, EMI noise gets filtered and grounded by this network which other wise may oscillate. The tweeter voice coil will show high impedance above 50kHz and there is no path for this HF to decay. Hence these oscillate and ring and will cause HF listening fatigue. The termination allow these spurious signals to decay. Remember all transistors and mosfet may have a bandwidth over 2MHz.

The external termination is meant for those speakers whose amplifiers are not internally terminated. In my case the tubes have an output transformer which is not terminated for sure. Hence i used this termination at the speakers terminal with benefits as explained earlier. You can find that out yourself if it benefits you too. But the value of resistor and capacitor needs to be meticulously worked out. In my case I had the benefit of using a simulator where in I can play around before implementing. If you don't use a simulator it will end up in mindless tweaking and I agree you will then find them as bakwas. Also speakers having a relatively flat impedance tend to benefit more as the load presented to the amplifier will not vary. This in turn helps to provide a relatively stable -3dB roll off for the network.
BTW I implemented them for just 50 cents.
 
Last edited:
Do the values change if RC network is used at speaker end or amplifier end?

If yes, will the values change from amplifier to amplifier if used at amps end?
 
@Hari: I tried RC series of 0.01 uF and 100 Ohms (cap to +ve speaker terminal, resistor to -ve terminal). Cut off for this is about 159 kHz. Honestly I didn't hear any improvement nor degradation. Now that you're hearing improved highs with much higher cut off, I think it's probably worth experimenting further.

One doubt: you mentioned 0.33 uF for 4 Ohm load. Does cut off freq of an RC filter depend in any way on the load impedance? For info, my speakers are 8 ohms nominal.

Also, if you use only a cap, how to calculate the fc? (my understanding is that if R is zero, denominator becomes zero).
 
@jsl

The Beta 22 amplifier shows the Resistor at the +ve end and the capacitor at the -ve end just like a zobel network. I used the simple formula C = 1/(2*pi*Fc*Z) where Z is the speaker load, Fc is the desired cutoff frequency. The resistor value can be anything you wish depending upon what load you want to present to the amplifier at the Fc. For e.g. if you want to present a load of 10 ohms at the Fc then you can add a 10 ohm resistor in series to the capacitor. At Fc and above the capacitor impedance will be negligible around a few ohms.

In your case the your nominal impedance is 8 ohms then for a cut-off of 159KHz you will need a capacitor of 0.125 microfarad using the above formula. At this freq if you want to present a load of 10 ohms to the amplifier you will need to use a 10 ohm resistor in series to the capacitor.

IMO a Fc of over 400KHz is much better than lower Fc but can be decided only with listening and experimentation. Also i believe if the speakers have a fairly constant impedance across the band-width then it will present a constant load to the amplifier and the Fc will also be relatively constant. If there are many spikes in the impedance curve the load presented to the amplifer would vary and so also the Fc.

Using only a capacitor will be very risky as it may pass all the signals above the Fc to ground and the output will get short which can blow the output stage of the amplifier. Also the resistor used here should be Metal-oxide type of around 5 W. Wire wound resistors / cement resistors will not work and may introduce noise. I have used a capacitor of MKT type, 250 Volts and a resistor of 22 Ohms MOR to begin with. I may reduce the resistor later to 10 ohms and then to 4.7 ohms after a while after listening to finalize the value.
 
Do the values change if RC network is used at speaker end or amplifier end?

If yes, will the values change from amplifier to amplifier if used at amps end?

The Fc value depends upon the speaker load and will vary depending upon what speaker impedance is connected to the amplifier. Hence having a Fc above 400KHz will benefit most speaker loads.

I would recommend you to first measure the actual speaker impedance with software like REW and then decide on the Fc, R and C values.
 
Also, AFAIK, a Zobel across the output of power amps has nothing to do with a zobel in a passive crossover. Both perform different functions.
 
Hari, the math doesn't sound right to me. If an RC network is kept at the output of an amplifier, the "R" value shouldn't decide the speaker impedance. Otherwise, it would be so easy to tailor speaker impedance - just put a resistor of suitable value and the amp will see a single flat impedance, across the entire spectrum of its operation, and that too a completely benign resistive load.

Will not the speaker impedance as seen by the amp be a function of what's inside the speaker, and not due to the value of the R in the RC filter? So further, wouldn't the RC filter values be independent of the speaker load? This was why I asked about your specifying the load in your earlier post.

Your thoughts please.
 
Zobel when used with woofers does flatten the woofer impedance, so do series resonant traps, since we also use a filter for low pass and high pass the impedance changes again.

The rd net work will be open circuit for frequency below Fc and Will come into play only for higher frequency to stabilize your amplifiers from ringing.

In Zobel network too you keep the value of the resistor equal to the woofer dc vc resistance to present this constant impedance for your crossover filter. Same for series resonant trap.
Same principles should apply here too.
 
Zobel=Boucherot Cell=Series R-C network. Application and intent of this application for amplifiers and speakers differ. What the Merlin folks are recommending this for is the application for AMPLIFIERS. Isn't that correct?
 
Recently was discussing with Bobby (Merlin Speakers) about the RC network termination suggested by him for the Merlin TSM speakers. Bobby explained in his mail in detail how the RC network when used to terminate the amplifier speaker terminal helps to reduce oscillations, ringing at HF and which in turns helps to open upon the sound stage making it more smoother and less fatigue prone.

Over the past 2 days i experimented with his theory and below are my impressions,

- tried initially with a roll-off frequency of 380KHz with a 0.1mfd polyester capacitor for a 4 ohms speaker load. And as explained by Bobby the HF did really get tamed and become more open and less strained. Listen for around 1 hr and was really very pleasing.

- tried again with a roll-off frequency of 100KHz with a 0.33mfd polyester capacitor for a 4 ohms speaker load. This did not reduce the HF as much as the first case and the effect was not that great (though present) as with the 380KHz roll-off.

With the 380KHz roll-off allowed me to increase the volume over 3 o'clock position without any sreachy or eering HF and acutally opened up details in the recording by increasing the dynamic range.

Conclusion: Simpler solutions are always better, cheaper and much easier to implement for complex problems (Theory of Constraints).
Where did you place this zobel? Before the woofer? Tweeter? Before both [before the passive x-over - assuming there was one]? At the amp output?
 
Where did you place this zobel? Before the woofer? Tweeter? Before both [before the passive x-over - assuming there was one]? At the amp output?

The rc network is at the amp output across Spkrs terminal, the Zobel before the woofer in the crossover, the series resonant trap before the tweeter in the crossover.
 
Zobel=Boucherot Cell=Series R-C network. Application and intent of this application for amplifiers and speakers differ. What the Merlin folks are recommending this for is the application for AMPLIFIERS. Isn't that correct?

Yes, the rc network benefits the amp more than the Spkrs,the Spkrs benefit is indirect.
 
Back
Top