Topic: A way to maximize your speakers

[JojoD818]

Forgive me, but from what I'm seeing it is actually a low pass filter with a Zobel network right on top of it. No wonder you get a better (or "cleaner") reproduction from small speakers because that limits the low frequency components being fed to the speakers, hence, less mechanical distortion. Heck, you can even pump up the volume.

Yes but Zobel networks have a lot of uses, from power amps to power supplies to transmission. It's use in power amps is not to match impedance but to protect the output from destructive high frequency oscillations due to capacitive loads.

But yes, it can also be used for impedance matching, but is usually included in the design of speaker crossovers where it is more popularly known as an impedance equalizer.

[rascal101]

It is still an RC circuit. Yes am familiar that a C introduces a pole so by having a series R you have a non zero divisor so oscillation eliminated.

Doesn't the C (in series) form a high pass filter? Only reason I included it is so that I can place the snubber. I've already tried it with big and small speakers.

Please try it  

[rascal101]

If we look at the resulting circuit, it would be an RLC. The C in series with the L (speaker) forms a series resonant circuit. Series C and parallel L (to output) means high pass so it is a high pass circuit.

With the resulting C at 135uF, low freq response should be fairly good.

[Stagea]

Why won't it interfere with transient response when the signal peaks had been altered?

Aside from insertion losses, won't it cause reduced electrical damping when you need it most (the circuit presents a higher impedance at lower frequencies)?

[rascal101]

You compute values for your R and C such that RC time constant is low and resulting frequency is > 100KHz.

The RC snubber circuit is in parallel with a C. If you compute the total impedance, it is lower than the impedance of the C alone.

Please do take time to read on passive rc snubber circuits. Although the circuit is not primarily aimed for audio, I believe there is similarity. For those with doubts, please test the circuit at the bench and do listening test 

[JojoD818]

If we look at the resulting circuit, it would be an RLC. The C in series with the L (speaker) forms a series resonant circuit. Series C and parallel L (to output) means high pass so it is a high pass circuit.

With the resulting C at 135uF, low freq response should be fairly good.

My bad, that's what I meant, a passive high pass filter. Just the same, it removes low frequency components of the signal.

Furthermore, with a 135uf cap (//270uf) and an 8 ohm load speaker, the corner frequency cutoff should be around 150Hz. Is that fairly good to you?

I guess for small speakers or pc speakers that can easily be made "basag ang tunog", a reduction of the low frequencies would benefit them. But I certainly can't imagine doing this on my bookshelf or floorstander speakers.

[aHobbit]

I believe it does.

I didn't say it was about cleaning the signal. I said better and "cleaner" signal. I said "cleaner" because the snubber helps reduce EMI emissions (Sorry but I couldn't help that my mind still rakes with SMPS).

I really didn't study much about these Zobel Networks except I read that it's used for impedance matching. But the RC circuit I'm using here is primarily for leading edge peak voltage reduction and damping. The reason I'm saying that is because you have so much noise from source to amp and you really need a way to reduce these noise without interfering with transient response of your audio system. Further, it really looks like the way we do it on power supply circuits.

You should try it  

The zobel network in the speaker output in amplifiers do 'clean' the signal.

The problem is that the word "clean" is a descriptive language, rather than technical.

Technically, the zobel in the speaker path removes the higher frequency (probably beyond 20khz above) going the way of the speakers - so it "cleans" up. These frequencies are the result, or the effect of many switching going on in the amps (like what you said - signals crossing the zero line/switching of transistors - or high-freq EMI/RFI that found its way into the amps circuit).

The way your circuit is designed is that it tries to remove another band of frequencies in the path of your speaker (value of these frequencies are determined by the values of your R & C) by presenting to the amplifier a very big resistance values to these frequencies - thus, attenuating the unwanted signals to the speakers.

The same RC circuit presents a "shorted" circuit to all desired frequencies.

For speakers not capable of driving some frequencies (they just transform them into more noise or distortions impacting on other frequencies it can play), such move will indeed make sound "cleaner" or "better".

However, I think, if this is done by anyone who is not capable to determine the pros & cons of his speaker drivers (limitation in its frequency handling) will just do it in a manner of hit & miss, not knowing what really happened. It may also lead to a certain point, that since you may use only a limited test sounds to hear the difference, it may impact on the sound of another music (if it is a different genre).

Of course, experimentation is always a good exercise.  

[rascal101]

The Zobel network is an RC circuit in parallel with the speaker or load. It is different when you have a series RC. The series RC forms a voltage clamp to the C in parallel with it. The idea is to reduce peak voltage and do some damping so you can achieve better sound.

You can always use higher capacitor (instead of 270uF in series) so you can lower corner freq. And recompute your RC snubber.

[JojoD818]

The Zobel network is an RC circuit in parallel with the speaker or load. It is different when you have a series RC. The series RC forms a voltage clamp to the C in parallel with it. The idea is to reduce peak voltage and do some damping so you can achieve better sound.

You can always use higher capacitor (instead of 270uF in series) so you can lower corner freq.

I know, that's why I even drew this so there won't be any confusion on which circuit we are all talking about.

[rascal101]

Granted it may not work on other speaker drivers it is at least an approach worth looking at. So far all speakers I have modified worked for me so when I get to one where it doesn't work perhaps a redesign would be in order 

[Stagea]

You compute values for your R and C such that RC time constant is low and resulting frequency is > 100KHz.

The RC snubber circuit is in parallel with a C. If you compute the total impedance, it is lower than the impedance of the C alone.

Please do take time to read on passive rc snubber circuits. Although the circuit is not primarily aimed for audio, I believe there is similarity. For those with doubts, please test the circuit at the bench and do listening test 

CMIIW, but wouldn't the lower part form a first order high pass filter with an F3 of about 98-147Hz for a driver that exhibits 6 to 4 ohms of impedance at this frequency (typical "8 ohm nominal" speaks these days). If I got this right, then the lows would be rolled off and the driver damping would be continually diminished together with the capacitor's impedance rise.

The parallel resistor won't improve damping, because the 220nF capacitor in series to the resistor would effectively decouple it for the entire audible band.

[wengkapre]

I have some pc speakers in the office, I might try this one

[rascal101]

CMIIW, but wouldn't the lower part form a first order high pass filter with an F3 of about 98-147Hz for a driver that exhibits 6 to 4 ohms of impedance at this frequency (typical "8 ohm nominal" speaks these days). If I got this right, then the lows would be rolled off and the driver damping would be continually diminished together with the capacitor's impedance rise.

The parallel resistor won't improve damping, because the 220nF capacitor in series to the resistor would effectively decouple it for the entire audible band.

The impedance of the capacitor diminishes over frequency right? When you have a C in parallel with an RC, it is like having a fixed resistor with a value of 10 and then it is paralleled with a capacitor which has decreasing value say it starts at 5 then its least value is 1. If you plot the values you will note that it has decreasing total impedance, yes?

I think that your analysis for the series RC is not right. Your analysis is only effective if the RC circuit is in parallel with the load wherein it functions as a low pass filter. When you have series RC, your R just forms a voltage drop (for leading edge peak voltage reduction) and your C is effectively shorted when there is increasing frequency. As you mentioned with F3 from 98 to 147Hz you can think of the C as a short starting at that frequency band. If the input resembles a square wave, the voltage formed at the C looks like a spike then when it achieves peak voltage it ramps down. Of course with higher C the ramp down is slower. Without the R the peak voltage is higher. The resulting voltage is a dV/dt or a differential. This is unlike that of a low pass filter where you have an integral or like a ramp.

[Stagea]

But how will the resistor even work when the capacitor in series to it is already showing a very high impedance?

If the combined impedance of the driver and the shunt resistor would result to about 20 ohms (16.1 ohms driver + 3.9 ohm shunt for example), the 220nF capacitor will only pass frequencies beyond 36kHz (F3 point) to the resistor. Above this frequency, the resistor is effectively running in parallel with the 270uF capacitors (the ones in series, hooked back to back). Even at these frequencies, most of the current would be passed on through the larger caps, because this path has a lower ESR (no resistor in series).

Below 36kHz, current will flow almost entirely through the 270uF caps, up until these caps exhibit impedance rise (98-147Hz in the prior example). In effect, wouldn't that still make it a high pass filter?

[rascal101]

But how will the resistor even work when the capacitor in series to it is already showing a very high impedance?

>> The RC circuit works at very high frequencies which is the basis for RC snubbers. Further, the voltage to your RC circuit is the same as your larger capacitor.

If the combined impedance of the driver and the shunt resistor would result to about 20 ohms (16.1 ohms driver + 3.9 ohm shunt for example), the 220nF capacitor will only pass frequencies beyond 36kHz (F3 point) to the resistor. Above this frequency, the resistor is effectively running in parallel with the 270uF capacitors (the ones in series, hooked back to back). Even at these frequencies, most of the current would be passed on through the larger caps, because this path has a lower ESR (no resistor in series).

>> Yes you are correct. Most of the current will pass through the larger capacitor. However, the voltage across your larger capacitor is practically clamped by the resistor R as the series C is approaching very low impedance.

No. It is not due to the lower ESR why current is mostly passing at the larger capacitor. It is simply because of its lower impedance. Even if ESR is 3.9ohms the total impedance of the RC circuit is still higher.

Below 36kHz, current will flow almost entirely through the 270uF caps, up until these caps exhibit impedance rise (98-147Hz in the prior example). In effect, wouldn't that still make it a high pass filter?

>> Both the large C and RC snubber are high pass filters. Only difference is the RC works at higher frequencies than the large C. You are correct, from the impedance perspective the RC circuit has higher total impedance hence there is less current. However, the basis for filter class is voltage not current.

[rascal101]

There is an online material - Calculating Optimum Snubbers by Jim Hagerman of Hagerman Technology. For those with interest, please take time to read it. What I'm trying to achieve to control high frequency transients is best shown on figure 4 as below.



Below is what I want to avoid.

[rascal101]

Just to clarify on my previous statements.

I think that your analysis for the series RC is not right. Your analysis is only effective if the RC circuit is in parallel with the load wherein it functions as a low pass filter. When you have series RC, your R just forms a voltage drop (for leading edge peak voltage reduction) and your C is effectively shorted when there is increasing frequency.

>> Above refers to the RC circuit.

As you mentioned with F3 from 98 to 147Hz you can think of the C as a short starting at that frequency band. If the input resembles a square wave, the voltage formed at the C looks like a spike then when it achieves peak voltage it ramps down. Of course with higher C the ramp down is slower.

>> Above refers to the larger C.

Without the R the peak voltage is higher. The resulting voltage is a dV/dt or a differential. This is unlike that of a low pass filter where you have an integral or like a ramp.

>> Above still refers to the larger C without the snubber resistor so the 220nF snubber capacitor is now in parallel with the larger C resulting to 135.22uF.

Guys, smaller speakers have higher min frequency operation so 100 - 150Hz should be fairly acceptable. If you want to try it with bigger speakers use higher value capacitors. Further, it doesn't have to be increasing the value of the capacitor you can also parallel a smallish value choke which should effectively end issues on low frequency performance  

My point is rather than throw questions try the circuit first then ask questions later. You can make it external. If it works for you then fine. If it doesn't you can play around with the snubber resistor and capacitor values. There are ways to make it work   Also, you can PM so I can calculate it for you  

The snubber circuit posted here is just a guide for you. You can be creative and add your own circuit or components. You do not have to be limited to what I have posted 

[Stagea]

Just to clarify on my previous statements.

I think that your analysis for the series RC is not right. Your analysis is only effective if the RC circuit is in parallel with the load wherein it functions as a low pass filter. When you have series RC, your R just forms a voltage drop (for leading edge peak voltage reduction) and your C is effectively shorted when there is increasing frequency.

No, it is for a series connection. Granted that your main high pass filter (the big C) is taken out of the equation, the resistor will also act as a load, hence the voltage drop across it at ultrasonic frequencies (when the small C impedance is low). It's running a high pass filter on the resistor and the driver.  However, when trying to listen to audible frequencies, the voltage drop would almost entirely be on the capacitor (in fact the impedance presented would probably be so high that the speaker output should only be barely audible if the big C is out). How would this work when the big C path already introduces a much lower impedance path in the first place?

Therefore, in my opinion it will still work as a high pass filter.

[Stagea]

>> Above still refers to the larger C without the snubber resistor so the 220nF snubber capacitor is now in parallel with the larger C resulting to 135.22uF.

Guys, smaller speakers have higher min frequency operation so 100 - 150Hz should be fairly acceptable. If you want to try it with bigger speakers use higher value capacitors. Further, it doesn't have to be increasing the value of the capacitor you can also parallel a smallish value choke which should effectively end issues on low frequency performance  

If you take out the resistor, Capacitors in parallel will exhibit summed capacitance. 220nF + 270uF = 270.22uF.

No. It is not due to the lower ESR why current is mostly passing at the larger capacitor. It is simply because of its lower impedance. Even if ESR is 3.9ohms the total impedance of the RC circuit is still higher.

Sorry, I used the wrong term. I meant impedance, as we are talking about AC signals.

Both the large C and RC snubber are high pass filters. Only difference is the RC works at higher frequencies than the large C. You are correct, from the impedance perspective the RC circuit has higher total impedance hence there is less current. However, the basis for filter class is voltage not current.

Precisely what I was saying, it is a high pass filter. I never said that current was the basis, I was just saying which path current would take (since these are parallel paths). Obviously, the voltage differential measured from the ends of the C and RC segments would be the same, since these are paralleled circuits (C segment and RC segment). Current will largely flow through the C segment, because it introduces a much lower impedance essentially all of the time.

[JojoD818]

Sorry, I used the wrong term. I meant impedance, as we are talking about AC signals.

It seems that some of the defining moments of this discussion has mixed AC and DC analysis. I humbly suggest that we stick with AC analysis for no audiophile/hobbyist in his right mind would feed and listen to DC on their speakers.

[rascal101]

@stagea

There are two 270uF capacitors in series so the total capacitance is 135uF. Adding the 220nF, total capacitance becomes 135.22uF 

[rascal101]

It seems that some of the defining moments of this discussion has mixed AC and DC analysis. I humbly suggest that we stick with AC analysis for no audiophile/hobbyist in his right mind would feed and listen to DC on their speakers.

Agreed.

[rascal101]

No, it is for a series connection. Granted that your main high pass filter (the big C) is taken out of the equation, the resistor will also act as a load, hence the voltage drop across it at ultrasonic frequencies (when the small C impedance is low). It's running a high pass filter on the resistor and the driver.  However, when trying to listen to audible frequencies, the voltage drop would almost entirely be on the capacitor (in fact the impedance presented would probably be so high that the speaker output should only be barely audible if the big C is out). How would this work when the big C path already introduces a much lower impedance path in the first place?

Therefore, in my opinion it will still work as a high pass filter.

The RC circuit does two things, peak voltage reduction and damping of signal. I don't understand why damping high frequency signals is not audible when added with the 20KHz audio band. This is precisely what is happening with tubes and their second order harmonics. Further, there is already scientific study that humans can perceive very high frequencies.

Think of it this way, a high frequency signal say 100Khz is added to a low frequency signal say 100Hz. The amplitude of the high frequency signal is at 0.1% of the 100Hz signal. Compare this signal to a plain 100Hz. Which is more audible? Then compare this to a "controlled" 100Khz signal. Which will sound better?

Yes you are correct the smallish C has very high impedance at low frequencies. It only begins to take effect at very high frequencies. Precisely what the RC snubber is for. It's not after low frequency.

[Stagea]

I've been saying this all along, but I will put it in very simple paragraphs.

1. The big caps already pass the ultrasonic frequencies, because it is a High Pass filter. It lets higher frequency signals pass.

2. The "snubber" will not damp ultra high frequencies, because it passes it too (it's not shorting them out, like in a configuration parallel to the driver). It's a High Pass filter when in series with the driver.

3. Even if the "snubber" started to "work" at those very high frequencies, the big caps are already passing those ultra high frequencies to begin with... without a series resistor at that. That makes the RC pretty redundant in this design (unless you can get significant inductance out of those capacitors at these frequencies).

Lastly, I never said that those ultrasonic frequencies did not matter. That is being disputed by many people, but I'm not taking sides with this because I actually have music recordings that supposedly have those ultrasonic frequencies (not plain old CDs). I also buy gear that supposedly can reproduce them.

[Stagea]

@stagea

There are two 270uF capacitors in series so the total capacitance is 135uF. Adding the 220nF, total capacitance becomes 135.22uF  

Sorry, I forgot about that. I was thinking a single C.

I also computed those F3 points using 270uF. It should've been based on 135uF.

[rascal101]

I've been saying this all along, but I will put it in very simple paragraphs.

1. The big caps already pass the ultrasonic frequencies, because it is a High Pass filter. It lets higher frequency signals pass.

2. The "snubber" will not damp ultra high frequencies, because it passes it too (it's not shorting them out, like in a configuration parallel to the driver). It's a High Pass filter when in series with the driver.

3. Even if the "snubber" started to "work" at those very high frequencies, the big caps are already passing those ultra high frequencies to begin with... without a series resistor at that. That makes the RC pretty redundant in this design (unless you can get significant inductance out of those capacitors at these frequencies).

Lastly, I never said that those ultrasonic frequencies did not matter. That is being disputed by many people, but I'm not taking sides with this because I actually have music recordings that supposedly have those ultrasonic frequencies (not plain old CDs). I also buy gear that supposedly can reproduce them.

Please just read the material on passive RC snubbers. I guess you have not spent time. Also read about RC time constants. You can PM me for further questions. If you have the time, you can check on your cellphone charger on the output rectifier diodes. You can also check the temperature of your output rectifier diodes with and without the RC snubber.

[rascal101]

To illustrate the effect of the RC snubber vs frequency in terms of impedance and compute the corresponding output voltage for a fixed load of 8ohms and an input voltage of 1.00V. Please refer to below:



Notation used are as follows:

C = large capacitor (fixed at 135uF)
Rs = snubber resistor (variable)
Cs = snubber capacitor (fixed at 220nF)

The tables are as follows:

Table 1: Rs = 3.9ohms
Table 2: Rs = 10ohms
Table 3: No snubber

Observations:

1. The total impedance Z with increasing frequency is lower for Tables 1 & 2 (with snubber) compared to Table 3 (without snubber).

2. The output voltage for a fixed load of 8ohms is higher for Tables 1 & 2 (with snubber) compared to Table 3 (without snubber).

3. From the Summary, the output voltage difference between Tables 1 & 2 compared to Table 3 decreases with increasing frequency. Output voltage difference is higher at lower frequencies and lower at higher frequencies based on % delta1 and % delta2

Based on the above data, there is improvement in using RC snubber (on top of the large capacitor) as compared to using a large capacitor alone.

[JojoD818]

Please just read the material on passive RC snubbers. I guess you have not spent time. Also read about RC time constants. You can PM me for further questions. If you have the time, you can check on your cellphone charger on the output rectifier diodes. You can also check the temperature of your output rectifier diodes with and without the RC snubber.

Akala ko ba we will stick with AC analysis? Ayan ka na naman, DC yan example mo eh.

Anyway, it's obvious that the technical posters in this thread have been trained more or less, in the same technical discipline and art of electronics which you have.

Having said that, it is the very reason why this thread has many questions and technical challenges because from where I sit, a snubber can only "snub" whatever component (passive or active) it is paralleled with. In other words, it will only dampen (your chosen term) whatever it is across with, which in this case, is the series C and not the Load speaker.

I believe this is also the supplemental point of these following statements.

1. The big caps already pass the ultrasonic frequencies, because it is a High Pass filter. It lets higher frequency signals pass.

2. The "snubber" will not damp ultra high frequencies, because it passes it too (it's not shorting them out, like in a configuration parallel to the driver). It's a High Pass filter when in series with the driver.

3. Even if the "snubber" started to "work" at those very high frequencies, the big caps are already passing those ultra high frequencies to begin with... without a series resistor at that. That makes the RC pretty redundant in this design (unless you can get significant inductance out of those capacitors at these frequencies).

[rascal101]

Hehehe ... the pulsating DC looks like AC

You're right the snubber only works on the devices where it is in parallel with.

Anyway, nice discussion  

[rascal101]

I did not expect a Php 50.00 suggestion can merit such discussion