Correct, the whole point of an iron core is reducing size and cost. ERSE I-core 5.0mH is $13. A comparible heavy gauge 5.0mH air core is $50+. I'd rather spend the $37 saved on buying better speaker drivers, which can give a significant audible and measurable reduction in non-linear distortion.
I'm not sure i agree that the tweeter inductor is critical 😕. Tweeter inductors aren't usually required to handle heavy currents, you'd probably blow up the tweeter before even a mediocre iron core inductor ran into issues. That said, tweeter inductors are generally pretty small and therefore cheap. Most hobbyists aren't going to fuss over a couple dollar premium to get a 0.5mH air core inductor for the tweeter.
An iron core is more likely to experience issues in a woofer filter due to the higher current flowing. A lot of woofers also tend to have severe break up nodes too, much higher in amplitude than the tweeter can produce so you'd think that if iron cores produced wide band noise problems, the woofer would be most likely to make it known.
The same flawed logic seems to be applied to capacitors with extremely expensive plastic film caps being considered 'critical' for tweeter filters and crappy electrolytics being delegated to woofer filters. From an engineering perspective the complete opposite is true: the woofer circuit is the one that should be given the better poly caps as it needs to handle higher currents.
Comparing frequency content of signals before and after inductor driving an 8ohm resistive load at around 40W. Result: Low level intermodulation distortion occurs. Much lower than most speaker drivers would produce at 40W however 🙂.
Air cores don't saturate. The result would be exactly the same as what is coming out of my amplifier with a lowpass applied. If someone wants to send me a free 'reference quality' 5.0mH air core inductor with the same DCR as the ERSE inductor i'm more than happy to test though.
I can already do something like that by simply notching out the 133Hz and 1250Hz test tones digitally.
I'm not sure i agree that the tweeter inductor is critical 😕. Tweeter inductors aren't usually required to handle heavy currents, you'd probably blow up the tweeter before even a mediocre iron core inductor ran into issues. That said, tweeter inductors are generally pretty small and therefore cheap. Most hobbyists aren't going to fuss over a couple dollar premium to get a 0.5mH air core inductor for the tweeter.
An iron core is more likely to experience issues in a woofer filter due to the higher current flowing. A lot of woofers also tend to have severe break up nodes too, much higher in amplitude than the tweeter can produce so you'd think that if iron cores produced wide band noise problems, the woofer would be most likely to make it known.
The same flawed logic seems to be applied to capacitors with extremely expensive plastic film caps being considered 'critical' for tweeter filters and crappy electrolytics being delegated to woofer filters. From an engineering perspective the complete opposite is true: the woofer circuit is the one that should be given the better poly caps as it needs to handle higher currents.
Ok here are the recordings.
1. Signal at amplifier. What you are hearing is the signal recorded at the output of my amplifier while driving a 5.0mH ERSE inductor in series with a 8ohm resistor at around 50Vpp (~40Watts) with a two tone signal - 133Hz sine and a 1250Hz sine at a slightly lower level.
amprecording
2. Same as above but with the 133Hz and 1250Hz tones notched out. Some small amount still remains. What you are hearing is the frequency content which is left over - the noise and distortion floor of my DAC (ASUS Xonar DX), Amplifier (modified DSE A2760 / Koda KD260) and ADC (E-MU 0204 USB).
amprecording_notch
3. Signal at after inductor. What you are hearing is the same as from the amplifier but with an effective lowpass filter applied by the inductor+resistor combination (therefore the 1250Hz tone is reduced in amplitude somewhat). You will also hear any frequency content that the inductor adds (non linear distortion, etc).
inductorrecording
4. Same as above but with 133Hz + 1250Hz tones notched out.
inductorrecording_notch
1. Signal at amplifier. What you are hearing is the signal recorded at the output of my amplifier while driving a 5.0mH ERSE inductor in series with a 8ohm resistor at around 50Vpp (~40Watts) with a two tone signal - 133Hz sine and a 1250Hz sine at a slightly lower level.
amprecording
2. Same as above but with the 133Hz and 1250Hz tones notched out. Some small amount still remains. What you are hearing is the frequency content which is left over - the noise and distortion floor of my DAC (ASUS Xonar DX), Amplifier (modified DSE A2760 / Koda KD260) and ADC (E-MU 0204 USB).
amprecording_notch
3. Signal at after inductor. What you are hearing is the same as from the amplifier but with an effective lowpass filter applied by the inductor+resistor combination (therefore the 1250Hz tone is reduced in amplitude somewhat). You will also hear any frequency content that the inductor adds (non linear distortion, etc).
inductorrecording
4. Same as above but with 133Hz + 1250Hz tones notched out.
inductorrecording_notch
Last edited:
Ran the notched recordings through the notch filter a second time for better rejection of the test tones.
amprecording_notch2
inductorrecording_notch2
amprecording_notch2
inductorrecording_notch2
I don't know what to make of that. Your technique includes DAC and amplifier noise. How do you assess audibility?
I still think that only a bridge or null technique is useful here. Easy enough to add a series resistor to one of the two coils being compared to match DCR.
See, the common noise cancels out! The headphones below could be a tweeter, maybe with a series capacitor to protect it. The resistors could be 16R, and the coils anything you have of the same mH value.
I still think that only a bridge or null technique is useful here. Easy enough to add a series resistor to one of the two coils being compared to match DCR.
See, the common noise cancels out! The headphones below could be a tweeter, maybe with a series capacitor to protect it. The resistors could be 16R, and the coils anything you have of the same mH value.
Attachments
By listening to the above recordings that I posted 😕 The majority of noise is from mains ripple on the amplifiers supply rails. You can't expect too much when running an amplifier at half it's rated power continuously. The noise from the DAC/amp/ADC is exactly the same on every recording, disregarding the fact that it's passed through a lowpass filter by the inductor under test. If you listen to the amp and inductor residuals the IMD is obvious.
Lets be realistic here, I use exactly the same equipment to listen to music, so if it can't be audible under a worst case scenario, what chance does it have of being audible under normal listening conditions?
Problem with that is all the components need to be very well matched (<0.1%) to get decent common mode rejection. Otherwise you're back to square one trying to extract meaningful information out of a noise floor.
Last edited:
TMM,
This is FANTASTIC stuff. Thank you so much for doing this, it's great information especially the filtered noise floor measurements. It does a great job of showing how low a level of distortion products are generated by a cored inductor used within the ratings of the part. I am even less concerned with using cored inductors than I was before. I just wish I had the time to join in with some additional measurements. I may try and set something up where I run an identical test signal through a quality woofer at different levels. The noise floor of a microphone/ambient room system will be way higher than electrical test equipment.
I'm perplexed as well by System7's insistence on using a technique that will be much more difficult to use in practice, and apparent refusal to study the data presented for what it is...and how to use it in context with the rest of a loudspeaker system. Just wanted to let you know I definitely appreciate the obvious work you've put into this.
Scott
This is FANTASTIC stuff. Thank you so much for doing this, it's great information especially the filtered noise floor measurements. It does a great job of showing how low a level of distortion products are generated by a cored inductor used within the ratings of the part. I am even less concerned with using cored inductors than I was before. I just wish I had the time to join in with some additional measurements. I may try and set something up where I run an identical test signal through a quality woofer at different levels. The noise floor of a microphone/ambient room system will be way higher than electrical test equipment.
I'm perplexed as well by System7's insistence on using a technique that will be much more difficult to use in practice, and apparent refusal to study the data presented for what it is...and how to use it in context with the rest of a loudspeaker system. Just wanted to let you know I definitely appreciate the obvious work you've put into this.
Scott
You're welcome 🙂. I'm tempted to throw an Usher 8945A on my test baffle and see how much power it takes to reach the same level of IMD. Like you say, I'm imagining that room effects will become a real problem in getting an accurate measurement.
The basic idea is pretty sound, implementation is difficult. What it needs to be implemented practically is a high power variable resistor and high power variable inductor. The variable inductor would need to be air core obviously. Then you'd adjust the variable components until you get adequate common mode rejection. Then you have the issue of accurately amplifying/measuring the differential signal that results.
Is it worth the effort of doing? Nope. What's the point of trying to measure something that is firmly below the noise floor in a typical system? People seem to forget that if something is over 90dB down, and you're listening at 90dB SPL, it is below the threshold of human audibility. Then consider that your average speaker is throwing in an absolute mess of non linear components that are at absolute best 60dB down and it is not even worth worrying about. With that in mind my test pretty much debunks the "Barkhausen Effect" for all practical uses.
Time and money is better spent on things that do matter, like researching better speaker drivers. Buying overkill crossover components eats into my speaker driver R&D budget 😎
i have a problem here, the shunt current is not the same current being listened to. using high value PP can soon land you in the same cost to value propositions as the LPF air / cored inductor tradeoff. I find to get reasonable filter Qs, a small resistor is usually added in series with the cap. easing esr concerns of bipolar electros anyhow.
the tougher Q in my mind is large series caps in front of mid-ranges....
users shopping for cheap large value PP is bound to get you into way more dodgy parts, as the end terminating methods for foil caps is all over the place unless they are dV/dt rated parts! the end connections for electros is more mature and understood.
Last edited:
Are you saying that distortion of shunt currents doesn't matter? They are just as important as series currents. Effectively you are listening to both since shunt components are driven by finite impedance nodes in a passive xo.
Last edited:
who says they are distorted. even if they were what is the relationship to current and frequency? yeah its a LPF the node currents decrease greatly past the XO point.
Last edited:
I don't know, I assumed that was what you were getting at as I can't think of any other reason why you'd point out that shunt currents aren't being listened to?
consider a reasonable 3-way, looking at XO to the woofer, are you saying replacing a 125 uF with PP would be wise investment? and if you do, wouldn't it be better to upgrade the mid series cap to full film cap since most of the current is going to the mids voice coil > https://sites.google.com/site/undefinition/tarkus
BTW by my analysis I'd add a small resistor in series with the 125 uF bipolar if I built this.
BTW by my analysis I'd add a small resistor in series with the 125 uF bipolar if I built this.
Last edited:
TMM, I appreciate you have done something quite detailed and methodical here. The measured noise floor is hugely down with a cored coil. 60-80dB is good. Well, heck, I'm not averse to ferrites myself!
But as with all that Clarity Cap business, I still feel no-one is really answering the question in any real scientific way with crossover components in general.
Troels Gravesen swears the mid range input cap is critical on quality. But a lot of people feel Troels is too much in Jantzen Audio's camp for financial reasons. ODougbo loves his sheet air coils. Some people swear that resistors make a difference too. Joe Rasmussen says that if a capacitor has a series bass shunt resistor in its design, it hardly matters how good the cap is. 😕
I'm starting to think I'll just have to do this myself. Because we need to compare the cheapie to a quality reference component of whatever type to make any progress. Half the trick might be in knowing what to listen for, as is so often the case with audio,
I continue to think that some applications stress a component more than others, especially high Q. Which will magnify quality considerations. Some sort of home audio bridge experiment ought not to be too hard. You can adjust inductance by moving the ferrite slug in and out of the coil, as we did in radio frequency circuits. I wish I still had some of those old decade boxes and variable tuning capacitors we used for adjusting exact values of L, C and R. 🙂
But as with all that Clarity Cap business, I still feel no-one is really answering the question in any real scientific way with crossover components in general.
Troels Gravesen swears the mid range input cap is critical on quality. But a lot of people feel Troels is too much in Jantzen Audio's camp for financial reasons. ODougbo loves his sheet air coils. Some people swear that resistors make a difference too. Joe Rasmussen says that if a capacitor has a series bass shunt resistor in its design, it hardly matters how good the cap is. 😕
I'm starting to think I'll just have to do this myself. Because we need to compare the cheapie to a quality reference component of whatever type to make any progress. Half the trick might be in knowing what to listen for, as is so often the case with audio,
I continue to think that some applications stress a component more than others, especially high Q. Which will magnify quality considerations. Some sort of home audio bridge experiment ought not to be too hard. You can adjust inductance by moving the ferrite slug in and out of the coil, as we did in radio frequency circuits. I wish I still had some of those old decade boxes and variable tuning capacitors we used for adjusting exact values of L, C and R. 🙂
I guess you missed post 142 & 143, he did that esp. for you!
turn your head phones gain way up
BTW youd be hard pressed to hear anything unless your test rig is better and more power than his, or you picked very bad parts indeed.
As with anything it needs to be assessed on a case by case basis.
The current handled by the 125uF cap is non-negligible around the corner frequency - around 3A @ 50W into 8ohm. That's a pretty big ask for a single 38x18mm electrolytic. If I were building the crossover i'd be use 5x22uF+15uF reasonably sized electrolytic caps in parallel. C2 probably needs to be beefed up accordingly. Putting a 50uF electro in parallel with a 15uF poly is pretty damn silly as the 50uF will take the bulk of the current anyway. Putting a small value poly in parallel gives little benefit at all - the whole thing needs to be replaced with poly. 3x22uF electro would probably be better and cheaper than the 50||15 electro||poly combo.
The main concern for me when specifying crossover capacitors is longevity. If the whole speaker system budget is $2000 then i'd have no problem plurging an extra $50 for a few poly capacitors knowing that they will never go bad, unlike electrolytics. If the whole speaker costs $100 then it's a waste of money spending even $20 on capacitors.
Myself included.
Last edited:
at the LPF corner is roughly half power, but then again most folks don't run sine waves parked at 100W into their speakers.
I don't get your multiple smaller electro theory thing for speakers , at least you backed away from pure poly caps here. I do see some benefit for SMPS output filters with lowering esr at HF >> 20KHz.
Multiple smaller value capacitors of the same voltage rating have more surface area, therefore together they have a higher ripple current rating than a single capacitor. They won't heat up as much and therefore will last a lot longer.
Consider the Panasonic FC datasheet (yes, i know it's a polarised cap but the same logic applies)
http://industrial.panasonic.com/lecs/www-data/pdf/ABA0000/ABA0000CE22.pdf
100V, 150uF = 0.917Arms ripple current rating
100V, 22uF = 0.260Arms ripple current rating
Therefore if you used seven 22uF caps in parallel to create a 150uF cap you have 1.82Arms ripple current rating, twice that of the single cap.
When i consider using a poly cap for current handling is when <33uF needs to handle high current. You'd need a lot of small electros in parallel to get any decent current handling. It looks ghetto, takes a while to solder - i'd rather just fork out for a poly cap.
There is a bit of truth to that. For one it makes the ESR of the cap less critical from an audible/tonal perspective. You might be able to get away with a slightly smaller (physically, not value) cap, and/or make it last a bit longer.
I'm not sure if that introduces saturation problems. I need to brush up on my magnetic theory but I believe there is a saturation advantage to having the I-core stick out at each end of the coil. Hopefully someone more knowledgable can confirm.
I would propose some type of center tapped air coil instead, where the center tap is implemented with some kind of wiper.
Last edited:
I ran a EI cored inductor for a short while as the single pole of the bass driver crossover.
It was the recycled microwave oven transformer that I cut open and reused.
The I's were laid on top of a thick paper gasket lying on the up turned E's.
I could hear the growl from the "loose" I's when the inductor AC current rose.
I could also adjust the inductance by changing the gasket thickness.
It did prove that the flux did create a mechanical vibration (the growl).
It did prove that the inductance was adjustable.
It did prove that I will always and only look to active High Pass bass filters in future.
It was the recycled microwave oven transformer that I cut open and reused.
The I's were laid on top of a thick paper gasket lying on the up turned E's.
I could hear the growl from the "loose" I's when the inductor AC current rose.
I could also adjust the inductance by changing the gasket thickness.
It did prove that the flux did create a mechanical vibration (the growl).
It did prove that the inductance was adjustable.
It did prove that I will always and only look to active High Pass bass filters in future.
Cross over
I think on active cross over there are others problems , the main is amplitude on output amplifiers .
But if you try an air core inductor protected by a strong iron shield and
strongly anchored on his support , many of this problems going solved
I think on active cross over there are others problems , the main is amplitude on output amplifiers .
But if you try an air core inductor protected by a strong iron shield and
strongly anchored on his support , many of this problems going solved
Sorry, no. The shield will both act as a core and as a shorted winding, altering the inductance and creating loss. Packing an air-core inductor inside an iron shield will mess it up due to its rather strong interaction with the inductor's field. Unless the inductor is toroidally wound, but then, screening is not necessary anyhow.
- Status
- Not open for further replies.