I had a mystery puzzle in my head....see if anyone who can share his experiences.
For the past two month, I decided modified my 10 years old commercial Audax MTM kit. I did the following steps:
1. New order 4 Audax HM170Z from madisound. (The new Audax HM170Z is surprising much heavier and show much more low freq response)
2. Replace old Audax tweeter with scanspeak HDS tweeter.
These actions end up redesign a new crossover. Out of expectation, it take me 2 month job, almost 20 hour per week and still cannot get sound perfect.
(What I mean sound quality, I mean at least surpass Harbeth M30.1 or Gamut m'inent speaker standard. Since I just had both in house for the benchmark comparison).
After so many hour in experiment and tweaking in the crossover, I find the good sound need
(1). Added many damping resistor in the crossover. For example, need series resistor with capacitor.
(2). Slow down the crossover slope when approach crossover point, then place notch after 2~4 octave of crossover point. For example, if crossover set at 2khz, then it would get much refined sound by added notch at ~8khz for woofer and 600hz for tweeter. These notch also need damping resistor.
Point (1) is similar to Lynn Olson result.
The Ariel, Part II
So far, all my crossover component was done by purphase high street eletronic shop (US $0.5 for a 2uF Metallized Polyester capacitor for example)
So if crossover cap need resistor for dampling, does it matter use expensive cap?
I think so called expensive capacitor is just lower down the ESR. But you need damp any way. Why care ESR?
PS:
In such low frequency, I believe the only matter in capacitor from idea is
(a) ESR (b) ESL (c)distortion of capacaitor
I dont beldieve in (b) and (c) since the distortion for a 200Volt passive Polyester capacitor in such low frequency is less even compared to 0.001% OP AMP. Remember the world best speaker driver only achieve 0.1 % distorion at best.
So does it make sense to purphase expensive capacitor? why?
For the past two month, I decided modified my 10 years old commercial Audax MTM kit. I did the following steps:
1. New order 4 Audax HM170Z from madisound. (The new Audax HM170Z is surprising much heavier and show much more low freq response)
2. Replace old Audax tweeter with scanspeak HDS tweeter.
These actions end up redesign a new crossover. Out of expectation, it take me 2 month job, almost 20 hour per week and still cannot get sound perfect.
(What I mean sound quality, I mean at least surpass Harbeth M30.1 or Gamut m'inent speaker standard. Since I just had both in house for the benchmark comparison).
After so many hour in experiment and tweaking in the crossover, I find the good sound need
(1). Added many damping resistor in the crossover. For example, need series resistor with capacitor.
(2). Slow down the crossover slope when approach crossover point, then place notch after 2~4 octave of crossover point. For example, if crossover set at 2khz, then it would get much refined sound by added notch at ~8khz for woofer and 600hz for tweeter. These notch also need damping resistor.
Point (1) is similar to Lynn Olson result.
The Ariel, Part II
So far, all my crossover component was done by purphase high street eletronic shop (US $0.5 for a 2uF Metallized Polyester capacitor for example)
So if crossover cap need resistor for dampling, does it matter use expensive cap?
I think so called expensive capacitor is just lower down the ESR. But you need damp any way. Why care ESR?
PS:
In such low frequency, I believe the only matter in capacitor from idea is
(a) ESR (b) ESL (c)distortion of capacaitor
I dont beldieve in (b) and (c) since the distortion for a 200Volt passive Polyester capacitor in such low frequency is less even compared to 0.001% OP AMP. Remember the world best speaker driver only achieve 0.1 % distorion at best.
So does it make sense to purphase expensive capacitor? why?
Hi,
I'm curious before we get into good caps, where are your measurements?
If you use something like XSim you can include ESR in the simulation and you can see the effects.
To answer your question about a tweeter, yes it's worth getting a decent capacitor even if you have a resistor in series, but if you do that you should get a good resistor. Well, if you put ANY resistor in series with a driver it should be a good one (Mills, $5-$7 each) in my very opinionated opinion. It's very common for tweeters to have a resistor to adjust the level to match the mid or woofer, so almost any capacitor in line with a tweeter is going to also have a resistor in the way. This is because tweeters tend to have more sensitivity, AND by the time you are done with baffle step compensating a mid-woofer in a 2-way, you've further widened the gap in sensitivity. Almost any discussion you've read about capacitor mods in the tweeter probably had a resistor in series. Not all of course, but my guess would be about 90% of all commercial multi-way speakers have a resistor and a cap. DIY'ers are funny though and may jump through hoops to avoid that.
Once you have properly measured your speakers and designed a great crossover with perfect phase matching, then it's worth getting a good capacitor AND a good resistor. Before that you're wasting money.
I invite you to read the details of my opinion in my blog page about this.
If you are doing all of this by ear though, please don't spend any money except on tools first. If you need to do that as cheaply as possible, research Room EQ Wizard and XSim. If you can spend money I'd suggest what I use, OmniMic+ DATS V2. to measure and XSim for simulation.
To properly simulate a crossover you need the acoustic and electrical measurements. That's where DATS comes in. Also, it has really good capacitor and inductor measurements, including ESR, making your simulations that much more accurate. There are ways of doing the electrical with home grown tools, but what a PITA. DATS by itself is $99. but it's only $50 or something bundled with OmniMic from PE. There's also Pocket CLIO which I have never used but is pretty advanced in features. Once I become a crossover design guru I'll see if I can sell a speaker (maybe even a pair!) to pay for it. Many here on DIY however like the free/homegrown alternatives.
Best,
Erik
I'm curious before we get into good caps, where are your measurements?
If you use something like XSim you can include ESR in the simulation and you can see the effects.
To answer your question about a tweeter, yes it's worth getting a decent capacitor even if you have a resistor in series, but if you do that you should get a good resistor. Well, if you put ANY resistor in series with a driver it should be a good one (Mills, $5-$7 each) in my very opinionated opinion. It's very common for tweeters to have a resistor to adjust the level to match the mid or woofer, so almost any capacitor in line with a tweeter is going to also have a resistor in the way. This is because tweeters tend to have more sensitivity, AND by the time you are done with baffle step compensating a mid-woofer in a 2-way, you've further widened the gap in sensitivity. Almost any discussion you've read about capacitor mods in the tweeter probably had a resistor in series. Not all of course, but my guess would be about 90% of all commercial multi-way speakers have a resistor and a cap. DIY'ers are funny though and may jump through hoops to avoid that.
Once you have properly measured your speakers and designed a great crossover with perfect phase matching, then it's worth getting a good capacitor AND a good resistor. Before that you're wasting money.
I invite you to read the details of my opinion in my blog page about this.
If you are doing all of this by ear though, please don't spend any money except on tools first.
If you need to do that as cheaply as possible, research Room EQ Wizard and XSim. If you can spend money I'd suggest what I use, OmniMic+ DATS V2. to measure and XSim for simulation. To properly simulate a crossover you need the acoustic and electrical measurements. That's where DATS comes in. Also, it has really good capacitor and inductor measurements, including ESR, making your simulations that much more accurate. There are ways of doing the electrical with home grown tools, but what a PITA.
DATS by itself is $99. but it's only $50 or something bundled with OmniMic from PE. There's also Pocket CLIO which I have never used but is pretty advanced in features. Once I become a crossover design guru I'll see if I can sell a speaker (maybe even a pair!) to pay for it. Many here on DIY however like the free/homegrown alternatives. Best,
Erik
Last edited:
Hi Thanks the reply.
I did use OmniMic for frequency measurement and use RF IC design tool for crossover design.
My point is, if "expensive resistor or cap" doesnt mean high linearity but mean less parasitic R, L,C whatever, then it would be total pointless for such expensive component, since these parasitic can be treated as part of crossover design, and you get even some free R or L by these parasitic.
Unless people mean, say bad resistor has bad ohmic contact like slightly rectifier in micro quantum mechanic level or time varing or frequency dependent parasitic ? in audio freuqnecy? Even that, all these still can translate to distortion (harmonic) and can be measured.
Though not measured in person, it is hard to believe these passive component has distortion in the level of speaker driver (60dbc at best).
I did use OmniMic for frequency measurement and use RF IC design tool for crossover design.
My point is, if "expensive resistor or cap" doesnt mean high linearity but mean less parasitic R, L,C whatever, then it would be total pointless for such expensive component, since these parasitic can be treated as part of crossover design, and you get even some free R or L by these parasitic.
Unless people mean, say bad resistor has bad ohmic contact like slightly rectifier in micro quantum mechanic level or time varing or frequency dependent parasitic ? in audio freuqnecy? Even that, all these still can translate to distortion (harmonic) and can be measured.
Though not measured in person, it is hard to believe these passive component has distortion in the level of speaker driver (60dbc at best).
Last edited:
It's not pointless, it's just imprecise.
What do you use to measure driver impedances?
Also, try out XSim, purpose built for speakers, free, very accurate.
Yes, if you use SPICE modelling then you can include all these parameters of this in your design. However You are assuming that everything you hear is in those models. I don't know that is true. You are free to argue otherwise, but I don't argue, I just state my belief.
Also, you need those parameters. For instance, you need to measure parasitic capacitance and inductance in any R that you use. I don't personally have those tools. Best I can measure is ESR in caps and DCR in inductors.
My suggestion to you is to take this out of theory and into the real world soon. Make the best crossover you can, then replace your cap with a Mundorf MKP and Mills resistor. They are the cheapest, and best sounding combination I know of. If you can't hear a difference between that and anything else, then you've achieved bargain nirvana. You will know spending different amounts of money won't help you.
Mundorf MKP run around $4-$10 for average values. Mills $5-$7 depending on the wattage.
Let me know what you think after you've listened.
I think many make the mistake of trying to theorize first, and if they don't like the theory never experiment. I hope you are willing to experiment a little bit and then try to discover a model that explains what you hear as opposed to trying to create a model first and not experimenting until you have one.
By the way, I know a brilliant scientist in photonics who uses this approach and does great things. That is, experiment first and often as opposed to getting stuck at the modelling stage of discovery.
Best,
Erik
What do you use to measure driver impedances?
Also, try out XSim, purpose built for speakers, free, very accurate.
Yes, if you use SPICE modelling then you can include all these parameters of this in your design. However You are assuming that everything you hear is in those models. I don't know that is true. You are free to argue otherwise, but I don't argue, I just state my belief.
Also, you need those parameters. For instance, you need to measure parasitic capacitance and inductance in any R that you use. I don't personally have those tools. Best I can measure is ESR in caps and DCR in inductors.
My suggestion to you is to take this out of theory and into the real world soon. Make the best crossover you can, then replace your cap with a Mundorf MKP and Mills resistor. They are the cheapest, and best sounding combination I know of. If you can't hear a difference between that and anything else, then you've achieved bargain nirvana.
You will know spending different amounts of money won't help you. Mundorf MKP run around $4-$10 for average values. Mills $5-$7 depending on the wattage.
Let me know what you think after you've listened.
I think many make the mistake of trying to theorize first, and if they don't like the theory never experiment. I hope you are willing to experiment a little bit and then try to discover a model that explains what you hear as opposed to trying to create a model first and not experimenting until you have one.
By the way, I know a brilliant scientist in photonics who uses this approach and does great things. That is, experiment first and often as opposed to getting stuck at the modelling stage of discovery.
Best,
Erik
I agree with the good resistor part, which means a functioning resistor, but
I can't with the rest of it. Why do you risk a wrong opinion when it's so easy
to prove that any normal resistor usually found in the XO filters doesn't show
any sign of inductive reactance in the human hearing range?
Last edited:
"Risk a wrong opinion" ??? I could ask that about why would anyone listen to you.
Can you please prove to me that inductive reactance is the only additional resistor measurement worth considering? Can anyone please state we have measured all that could possibly be known ever about resistors or capacitors and their effect on hearing?
I'm certainly not going to advocate that anyone spend $30 on a resistor. I do think however that more people should do their own experiments and suggest an inexpensive one. Oddly, I get attacked for not agreeing to use one set of mathematical formulas or another as ruling all that I hear. This is why I don't want to argue. I'm merely stating my opinion, and suggesting that the curious spend a modest amount of money in experimentation rather than modelling. When I do so some one invariably comes up with whatever their pet model that is guaranteed to encompass all capabilities of human perception and experience as well as the behavior of part X. Let it begin anew.
The idea that an opinion that doesn't match your model MUST risk being wrong (ooooh, dangerous!) isn't science, it's just another opinion. Science is born of and fed by experimentation. Models are always incomplete and added to when we have to. If an inexpensive experiment is attacked, it's not helping to promote science, it's helping to promote a political agenda about an opinion that is written scientifically. The same is true when experience without models and double blind tests are attacked. If I don't doulbe blind test something or have a measurement surely I'm wrong. It's not science, it's elitism to approach knowledge this way.
To "risk being wrong" is the very esense of experimentation and discovery. Now, there's different risks. There's the type of risk that puts lead in children's water. VERY BAD RISK! Then there's the risk that costs $5. Very low risk. So, yes, risk being wrong. Try it, find out for yourself. That's my motto. "Risk BEING WRONG!" unless it costs more than $20.
Then of course there are the quasi-scientists who get all aflutter over non-vetted individuals doing such experiments on their own. Heavens to Betsy! NO! They'll ruin Eastern and Western civilization, and take away the meaning of our diplomas if we let them do that!! << sigh >> Have I made my point? I am sharing my very best advice, and suggesting they spend a little bit of cash on their own to listen and see if they care or not. Is that not a much faster way to discovery than to read or to help create even more of the countless pages of argument over the subject? I should get a medal instead. Some one at least give me a cookie, damnit!
Best,
Erik
Last edited:
Thanks the reply.
I would certainly go Mundorf/Mills experiments as your suggestion. Thanks.
Once I go that, the res/cap value would be fixed and no change anymore.
I find the crossover is so critical and sensitive. Even a 0.5 ohm or 0.1uF show some difference sonically, espically in classical music replay.
I used to have 2.7ohm, 3.3ohm 3.9ohm...as example for many tweak. Even for damping efffect. The sound show different for each value. But for FR measurement, it is very tiny difference. Or one could say, the microphone position has much more variation than that tiny res value change. This make me really wonder the effectiveness of FR measurement or even the entire modelling.
This lead another puzzle in my head. When I tweak a good sound crossover, it is good sound regardless of where I sit and listen, even in distant next room listening. Similarity , when the sound is dry and bad, it is dry everywhere no matter how I position the speaker. But different location and speaker position certainly lead to huge frequency response difference if one put microphone there. Then, by logically thinking, FR measure almost unrelated to sound quality. This bring me back try and error route.....or I am too ignorance to miss anything?
I would certainly go Mundorf/Mills experiments as your suggestion. Thanks.
Once I go that, the res/cap value would be fixed and no change anymore.
I find the crossover is so critical and sensitive. Even a 0.5 ohm or 0.1uF show some difference sonically, espically in classical music replay.
I used to have 2.7ohm, 3.3ohm 3.9ohm...as example for many tweak. Even for damping efffect. The sound show different for each value. But for FR measurement, it is very tiny difference. Or one could say, the microphone position has much more variation than that tiny res value change. This make me really wonder the effectiveness of FR measurement or even the entire modelling.
This lead another puzzle in my head. When I tweak a good sound crossover, it is good sound regardless of where I sit and listen, even in distant next room listening. Similarity , when the sound is dry and bad, it is dry everywhere no matter how I position the speaker. But different location and speaker position certainly lead to huge frequency response difference if one put microphone there. Then, by logically thinking, FR measure almost unrelated to sound quality. This bring me back try and error route.....or I am too ignorance to miss anything?
Last edited:
If you hear good sound outside of the room it's a good indicator your acoustics are good.
I had a very interesting conversation with an ASC rep about this recently at the California Audio Show. We both noticed that good rooms at shows you could tell they were good in the hallways. Invariably the rooms had good room treatment. Unfortunately 80% of the dealers there didn't want to put in room treatment because they felt it sent the wrong image to buyers. And you know what? Most people at the show coudln't really tell. That's why commercial stuff is in the state it's in.Could this be part of what you are hearing?
Let me know what you think of the experiment.
Afterwards I'll send you to a link of some one else on DIY who tried it, but don't want to prejudice you now. Best,
Erik
Oh, in my current crossovers, I have 2 caps, 4.7 and 12uf. I added 0.1uF to the first, made everything too bright, could see it in XSim, and could hear it. Unguided listeners could too.
On the 12uF though, really great change to the midrange with the cap combo. So, yes, it's all relative. See my other thread about the Duelund silver foil caps I was trying.
On the 12uF though, really great change to the midrange with the cap combo. So, yes, it's all relative. See my other thread about the Duelund silver foil caps I was trying.
As for FR, I've realized after looking at what I like and don't like, that FR is not the single most important thing for me. I prefer drivers with very little ringing, and large diaphragm tweeters that get most of the details to my ear and lack evidence of compression. Do that and I'm much more forgiving about the absolute FR. That's just my tastes of course, and I've had to learn about them by seeing the terrible FR of speakers I like.
Adding a series resiztor will increase equivalesnt series resistance and at some degree equivalent series inductance. But a good capacitor is not only low ERS or ESL. In fact some very good and very expensive capacitors don't have lowest ESR ESL.
Beside ESR and ESL there are few parameters very important as well: DF - dissipation factor, loss tangent, Q - quality factor.
And there is far more on a audiophile grade capacitor, like dielectric absorption, dielectric strength, leakage current, piezoelectric effect and much more.
Regards,
Tibi
What you're suggesting is similar to what golden ears do. That's fine for them
and manufacturers, it's not for the truth. All the great scams have some tiny true
parts that serve as a foundation and bait. The same applies to non-inductive
resistors vs. basic (inductive) ones. The truth is that their inductances are not the
same, but the difference is outside of range we are interested in.
If you're positive that there is more to it, than maybe you should contact Vishay
and possibly earn yourself something much more than just a cookie or a medal.
Btw, golden ears aren't that golden at all, otherwise they would hear the negative
impact of the underlined situation.
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.