The LC+Zobel filter used on most Class-D boards discussed in this forum are independent of the amp they are on, and the effect on the sound they have is entirely dependent on your speakers. The filter on most boards is designed for speakers with DC Resistance and Voice Coil Inductance that are far different than the most common speakers. Matching just my zobel to my speakers has made a dramatic difference in my board's sound.
Here is what you do:
Measure the DC Resistance and Voice Coil Inductance of your tweeter (or look it up in the manuf docs).
Then choose one of the standard output inductors and use the following equation to see if the corner frequency of the filter will be acceptable. If your output is single ended, use the actual value of the inductor, If your output filter is BTL, divide the inductor value by two. Choose one of the higher inductor values, like 33uH, since most manufacturer's choose too low for the average speaker (or you can try your current inductor's value to see if you can keep it).
f=1000*(0.2251*Rs)/Lf
Remember to change the value of the inductor from uH to mH before putting it into this equation, then change it back to uH to get the final value. If the cutoff freq is more than 35,000Hz you can keep the inductor, and not too high (I don't know yet what too high is).
Then solve for C:
For Single ended filters:
Cse=1000000*(0.1125/(Rs*f))
For BTL filters:
Cbtl=2*1000000*(0.1125/(Rs*f))
This will give you the capacitor for the output filter (not the zobel, don't confuse the two).
Zobel:
For the zobel you can use any online zobel calculator such as:
http://www.carstereo.com/help/Articles.cfm?id=36
Just plug in your tweeter's DC Resistance and Inductance (in mH)
Here is what you do:
Measure the DC Resistance and Voice Coil Inductance of your tweeter (or look it up in the manuf docs).
Then choose one of the standard output inductors and use the following equation to see if the corner frequency of the filter will be acceptable. If your output is single ended, use the actual value of the inductor, If your output filter is BTL, divide the inductor value by two. Choose one of the higher inductor values, like 33uH, since most manufacturer's choose too low for the average speaker (or you can try your current inductor's value to see if you can keep it).
f=1000*(0.2251*Rs)/Lf
Remember to change the value of the inductor from uH to mH before putting it into this equation, then change it back to uH to get the final value. If the cutoff freq is more than 35,000Hz you can keep the inductor, and not too high (I don't know yet what too high is).
Then solve for C:
For Single ended filters:
Cse=1000000*(0.1125/(Rs*f))
For BTL filters:
Cbtl=2*1000000*(0.1125/(Rs*f))
This will give you the capacitor for the output filter (not the zobel, don't confuse the two).
Zobel:
For the zobel you can use any online zobel calculator such as:
http://www.carstereo.com/help/Articles.cfm?id=36
Just plug in your tweeter's DC Resistance and Inductance (in mH)
If your tweeter is entirely resistive, no inductance like many planars, you should drop the zobel entirely.
Once you have designed the output filter that matches your soeakers, you can move it from board to board, just remove the current output filter on the new board. I am going to put my output filter on a daughter board and design a way to easily attach it to new boards.
Great thread jyoung!
http://content22-foto.inbox.lv/albums184230424/roxis86/12-11-2011/T1-output-filter.jpg
This is the output filter of hifimediy T1 amp with sta510 chip. I dont understand what c3 and c4 caps are doing there? These are 0.1 ceramic caps!
Yes I will remove zobel for planar B&G NEO3 tweeter.
Can I remove zobel for mid driver too?
Its low inductance Vifa NE149w driver.
Le: 0.14 mH
Re: 3,64 Ohm
Speakers is tri-amped. Midrange driver and planar tweter conected directly to amps output without passive crossover.
http://content22-foto.inbox.lv/albums184230424/roxis86/12-11-2011/T1-output-filter.jpg
This is the output filter of hifimediy T1 amp with sta510 chip. I dont understand what c3 and c4 caps are doing there? These are 0.1 ceramic caps!
Yes I will remove zobel for planar B&G NEO3 tweeter.
Can I remove zobel for mid driver too?
Its low inductance Vifa NE149w driver.
Le: 0.14 mH
Re: 3,64 Ohm
Speakers is tri-amped. Midrange driver and planar tweter conected directly to amps output without passive crossover.
About C3 & C4, I don't fully understand all of output filter design. Hazarding a guess, I would say they are rolling off the highs. Probably because the output filter and zobel cause a peak. If you are going to design your own output filter, definitely remove them.
About the mids, I haven't given enough thought to all the ramifications of output filters for mids. Again, hazarding a guess, the effect of the output filter is high enough that it probably won't have a great effect on your results. Also, if you are using a second order or greater analog low pass, or greater, on your mid, and if you moved the filter to immediately after the amp, you should be able to get rid of the entire output filter and zobel, since all the output filter does is get rid of the switching frequency and the mid, and woofer too, low pass will do that for you. You don't want the switching frequency running down the speaker wires, hence the move of the filter. If you are using an active system, I think it might be best to leave the output filter in since the switching frequency messes with electronics.
About the mids, I haven't given enough thought to all the ramifications of output filters for mids. Again, hazarding a guess, the effect of the output filter is high enough that it probably won't have a great effect on your results. Also, if you are using a second order or greater analog low pass, or greater, on your mid, and if you moved the filter to immediately after the amp, you should be able to get rid of the entire output filter and zobel, since all the output filter does is get rid of the switching frequency and the mid, and woofer too, low pass will do that for you. You don't want the switching frequency running down the speaker wires, hence the move of the filter. If you are using an active system, I think it might be best to leave the output filter in since the switching frequency messes with electronics.
With my humble respect, please be careful if you do this. From the TP2050 datasheet: An RC network, or “Zobel” (RZ, CZ) should be placed at the filter output to control the impedance “seen” by the TP2050 when not attached to a speaker load.
Hi jyoung, can I measure the tweeter inductance just with an inductance meter or it needs to be measured at a particular frequency (eg. 10kHz)? Also, in your formula for the output inductor, Lf is the inductance of the coil and 0.2251 a fixed value?
Thank you very much.
Thank you very much.
Darn, I forgot that, thanks! I retract my statement about removing the zobel entirely. I will have to rethink and see if I can find a way to minimize the effect of the "zobel" when the tweeter is purely resistive.
An inductance meter is fine. The inductance of voice coils changes little over the audible range, it is the resistance that changes with frequency as a result of the inductance and that is what the zobel corrects.
Yes, 0.2251 is a fixed value. It is actually a combination of PI and other constants.
Yes, 0.2251 is a fixed value. It is actually a combination of PI and other constants.
Class-D on a purely resistive load. We can reduce the impact of the zobel by increasing the value of the resistor. It would be easy to test the effect of the zobel simply by removing it while the speaker is connected and listening to the difference. We could also reduce the value of the capacitor, though I wouldn't go to an extreme.
What do you think kristleifur?
What do you think kristleifur?
More about the value 0.2251. This value is specific to a Butterworth filter. That seems to be the most common output filter used in Class-D. You would use a different value for Linkwitz, or Bessel. There isn't a reason that I have run into to use anything but a Butterworth config.
I've just hooked up my all new output filter that matches my speakers. It looks like something from Frankenstein's lab, and I am probably ruining my neighbor's radio reception.
The sound is just what I expected, much duller sounding than the original amp, and that's a good thing (if you are in love with the Sure sound, this tweak may not be for you). The reason I am not giddy is that I've been listening to hyped music so much that my ears need to adjust.
I have to ask myself, are all the highs there, and that is a yes. I can hear "air", and acoustic instruments like violin sound more natural than I have experienced on my system. Sibilance sounds more natural. I pay a lot of attention to sibilance because I hear it every day and if the system's highs aren't clean sibilance will sound harsh, if they aren't balanced it will show up in the "shushes" vs "tsssses". Shushes have lower frequencies, and Tssses have higher highs.
The background is nicely black and the sound stage is huge. The sound may not be as open as it should be but I am not sure yet. However, the sound is much closer to what I hear in movie theaters, just with more air; probably because in theaters we are a lot further away from the screen and some of the highs are absorbed by the air.
We can adjust the highs by changing the capacitors in the output filter to cause peaking, or rolloff. I need to do a little experimenting to know which way to adjust the capacitance to get more or less highs.
I have an Audio Research 150.2, one of AR's entries into Class-D, coming tomorrow. I bought it for reference since I knew as I get closer to neutral sound, I would need it. And I can easily resell it on Audiogon. Acoustic Research has a reputation for natural sound, and the 150.2 has gotten rave reviews.
The sound is just what I expected, much duller sounding than the original amp, and that's a good thing (if you are in love with the Sure sound, this tweak may not be for you). The reason I am not giddy is that I've been listening to hyped music so much that my ears need to adjust.
I have to ask myself, are all the highs there, and that is a yes. I can hear "air", and acoustic instruments like violin sound more natural than I have experienced on my system. Sibilance sounds more natural. I pay a lot of attention to sibilance because I hear it every day and if the system's highs aren't clean sibilance will sound harsh, if they aren't balanced it will show up in the "shushes" vs "tsssses". Shushes have lower frequencies, and Tssses have higher highs.
The background is nicely black and the sound stage is huge. The sound may not be as open as it should be but I am not sure yet. However, the sound is much closer to what I hear in movie theaters, just with more air; probably because in theaters we are a lot further away from the screen and some of the highs are absorbed by the air.
We can adjust the highs by changing the capacitors in the output filter to cause peaking, or rolloff. I need to do a little experimenting to know which way to adjust the capacitance to get more or less highs.
I have an Audio Research 150.2, one of AR's entries into Class-D, coming tomorrow. I bought it for reference since I knew as I get closer to neutral sound, I would need it. And I can easily resell it on Audiogon. Acoustic Research has a reputation for natural sound, and the 150.2 has gotten rave reviews.
I tried changing the capacitance in the output filter. I have two 0.1uF paralleled to get the 0.2uF I calculated for the output filter. I removed one of them reducing the capacitance to 0.1uF in each branch. The sound opened up nicely and sounded more like the un-modded Sure, brighter than natural. I would say that we have good control over the sound through changes in the capacitance in the output filter and the resistor in the zobel.
inductance
Why not using smaller value output inductors but bigger caps?
Like 5 uh coils, 1uf caps?
If your planar tweeter is purely resistive, it would be better to use as small coils as possible. If you add 30 uh coils its looses its advantage against dome tweeters.
Planars, electrostats and other non inductive speakers even sounds better with low inductance speaker cables, like braided cat-5.
Why not using smaller value output inductors but bigger caps?
Like 5 uh coils, 1uf caps?
If your planar tweeter is purely resistive, it would be better to use as small coils as possible. If you add 30 uh coils its looses its advantage against dome tweeters.
Planars, electrostats and other non inductive speakers even sounds better with low inductance speaker cables, like braided cat-5.
Lenta, do smaller inductors sound better? Remember, I'm new to Class-D so I am ignorant of fine details. I am relying on my knowledge of physics, and speakers to carry me here.
I choose 33uH to keep the corner frequency low since I'm not sure how high I can go without letting through too much switching frequency. Remember, I am trying to match the output filter exactly to my speakers (4.6 ohms) unlike most manufactures who go with the cheapest components and end up designing for 1.5 ohm speakers (see my post "Class-D Output Filters - A Case Study") and getting a very hot high end on 4 to 8 ohm speakers. I am shooting for the most natural, transparent sound I can.
My next experiment is trying 15uH inductors. My calculations show that on my speakers this will result in a corner frequency of 138,061 Hz. Anyone know if this is too high?
For 8 ohm speakers, the cutoff frequency will be twice as high as for 4 ohm.
I choose 33uH to keep the corner frequency low since I'm not sure how high I can go without letting through too much switching frequency. Remember, I am trying to match the output filter exactly to my speakers (4.6 ohms) unlike most manufactures who go with the cheapest components and end up designing for 1.5 ohm speakers (see my post "Class-D Output Filters - A Case Study") and getting a very hot high end on 4 to 8 ohm speakers. I am shooting for the most natural, transparent sound I can.
My next experiment is trying 15uH inductors. My calculations show that on my speakers this will result in a corner frequency of 138,061 Hz. Anyone know if this is too high?
For 8 ohm speakers, the cutoff frequency will be twice as high as for 4 ohm.
Don't know more than the fact that resistance goes up with temp. I don't know of a counter for that, but there may be one.
Let me clarify:
Using 33uH with 8 ohm speakers will result in twice the cuttoff frequency of 4 ohm speakers. The cutoff frequency I get for 33uH inductors and my 4.6 ohm speakers is 62,755Hz, and for 8 ohm speakers the cutoff raises to 109,139Hz. For 15uH inductors the cutoff freq for 8 ohms goes up to 240,106Hz.
Using 33uH with 8 ohm speakers will result in twice the cuttoff frequency of 4 ohm speakers. The cutoff frequency I get for 33uH inductors and my 4.6 ohm speakers is 62,755Hz, and for 8 ohm speakers the cutoff raises to 109,139Hz. For 15uH inductors the cutoff freq for 8 ohms goes up to 240,106Hz.
What I was trying to suggest is fine tuning a zobel network is not a very good idea, at least not very universal. Thermal modulation spoils it to a degree.
The other thing is you'd like an amplifier to provide two things:
1. Flat frequency response
2. Damping (=low source impedance)
Zobel is mitigating the first if fine tuned for load, but doesn't help much with the latter.
The other thing is you'd like an amplifier to provide two things:
1. Flat frequency response
2. Damping (=low source impedance)
Zobel is mitigating the first if fine tuned for load, but doesn't help much with the latter.
Ok, I understand now. 🙂
I agree with everything you say in principal. In designing crossovers for speakers we would alway use the zobel values that the equations give us and never tamper with them. Zobels are typically used with midranges and changing them will negatively effect the crossover point and meshing of the two drivers.
Class-D is a special case. Here we are not using a zobel in crossing from one driver to the next, it's function is to eliminate the switching frequency as much as possible. The Zobel is there to help the output filter by eliminating the effect of the tweeter's inductance. Now, once I have my theoretically exact output filter and zobel in place and the sound is not quite what I want, I am left with the question of what can I do now. The frequency response is obviously not flat to my ears and if I am unable to flatten it, I will choose a slightly euphonic sound instead. Since changing the zobel here will not unbalance the meshing of the sound of two drivers but simply change the shape of the output curve very slightly, I decided to use the zobel as a subtle EQ (remember, I have already corrected for the gross mismatch the manufacturer's zobel had with my speakers, so a little tweaking is very minor by comparison). I am getting good results from this but it is not easy; EQ never is. Changing the capacitance in the output filter accomplishes the same thing, causing either peaking or rolling off as the case may be.
I agree with everything you say in principal. In designing crossovers for speakers we would alway use the zobel values that the equations give us and never tamper with them. Zobels are typically used with midranges and changing them will negatively effect the crossover point and meshing of the two drivers.
Class-D is a special case. Here we are not using a zobel in crossing from one driver to the next, it's function is to eliminate the switching frequency as much as possible. The Zobel is there to help the output filter by eliminating the effect of the tweeter's inductance. Now, once I have my theoretically exact output filter and zobel in place and the sound is not quite what I want, I am left with the question of what can I do now. The frequency response is obviously not flat to my ears and if I am unable to flatten it, I will choose a slightly euphonic sound instead. Since changing the zobel here will not unbalance the meshing of the sound of two drivers but simply change the shape of the output curve very slightly, I decided to use the zobel as a subtle EQ (remember, I have already corrected for the gross mismatch the manufacturer's zobel had with my speakers, so a little tweaking is very minor by comparison). I am getting good results from this but it is not easy; EQ never is. Changing the capacitance in the output filter accomplishes the same thing, causing either peaking or rolling off as the case may be.
- Status
- Not open for further replies.