Thomas, [and Bertus]
Please don't misunderstand me, I don't want to be a naysayer here.
You both did a very good job.
I like your taste in testing, Thomas.. 😀
So, that leaves the toroidal cores. Are they all from the same manufacturer?
Then still the winding technique is to be better tested / measured..
Ciao, George
Please don't misunderstand me, I don't want to be a naysayer here.
You both did a very good job.
I like your taste in testing, Thomas.. 😀
So, that leaves the toroidal cores. Are they all from the same manufacturer?
Then still the winding technique is to be better tested / measured..
Ciao, George
Inductors?
Why do I have the feeling that the differences in core materials are swamping the effects of the winding techniques? It would be nice if the cores were the same so there would be no question of which method sounds better. Of course this will bring up the issue of why if there are differences.
Roger
Why do I have the feeling that the differences in core materials are swamping the effects of the winding techniques? It would be nice if the cores were the same so there would be no question of which method sounds better. Of course this will bring up the issue of why if there are differences.
Roger
Joseph K said:
No misunderstanding at all! I really appreciated your job! Just, I'm not a technician and have some difficulties to follow your arguments. My contribution to the discussion is not "against" anybody nor to "defend" bertus coils. Just offering facts from fields where I'm more experienced.
Bertus cores are from CWS, while the T106-2 are from Micrometals (are the original coils used by Tripath in TK2350 ev boards).
Why don't you try bertus coils in your amp1? Maybe, me and bertus like the same music and the same distorsions 🙂 .
Ciao
Giovanni
heh,
Just one more point: If you look back at the sim at post #89, and the real curve at post #90, then there is a strong difference, apart from the multiple resonances: the Q of the real circuit is much lower, than the Q in the simulation, at the ~48MHz resonance point. [the attenuation in the real curve is ~50dB, while in the sim it's ~70dB]
Now, I was playing with the sim to get back the real value also for the Q - and I had to raise R3 from .1 ohm, in the fig., up to ~150 ohm!!
Other than .001 ohm DC R value of the wire!! Now, if somebody would help & explain how much of it is due to Core losses, and how much is due to the skin effect ? [probably a nice big part..] And the wire is only .5mm in that coil. A 1mm wire is much worse at high frequencies..
Ciao, George
Just one more point: If you look back at the sim at post #89, and the real curve at post #90, then there is a strong difference, apart from the multiple resonances: the Q of the real circuit is much lower, than the Q in the simulation, at the ~48MHz resonance point. [the attenuation in the real curve is ~50dB, while in the sim it's ~70dB]
Now, I was playing with the sim to get back the real value also for the Q - and I had to raise R3 from .1 ohm, in the fig., up to ~150 ohm!!
Other than .001 ohm DC R value of the wire!! Now, if somebody would help & explain how much of it is due to Core losses, and how much is due to the skin effect ? [probably a nice big part..] And the wire is only .5mm in that coil. A 1mm wire is much worse at high frequencies..
Ciao, George
Nice review Thomas,
I have some feeling that the main difference is caused by the larger core and by that, lower induction in the core. Maybe the core material plays also its role? Curious about what MPP cores will do …
But back to the topic header: How does it compare to the UcD?
😉
I have some feeling that the main difference is caused by the larger core and by that, lower induction in the core. Maybe the core material plays also its role? Curious about what MPP cores will do …
But back to the topic header: How does it compare to the UcD?
😉
Thomas, Thanks for the listening test, and the comparison. This morning I planned to warn you, that there is one thing that isn't tested in the test you wanted to perform, and that is the core brand / manufacturer. Jan's cores and the ones I ordered from CWS have the same colour code (red), but there definitely is a difference in colour in certain kinds of light. Therefore I am pretty sure they come from a different manufacturer.
Joseph, reducing the resistance of the coil from 50 mOhm to 20 mOhm does increase the damping of the amplifier. If there is a choking coil between amp and woofer, you are right: this doesn't help much (for the woofer but not for the tweeter). In the setup I am putting together I have an amplifier for each driver, so each driver is damped directly by the amp.
Pjotr, you wrote:
"If the pulse output of the chip is “perfect distortionless” the energy lost in the coil is subtracted from that perfect output causing distortion. Especially with the output of the Tripath chips where the frequency is largely signal level dependant.
So in a way the varying frequency causes frequency dependant losses causing distortion."
My feeling is that if I had written such words, you would have considered it gibberish.
The simple scheme is that audio frequencies are not at all affected by skin or proximity effects, and RF frequencies are affected, but this effect adds to the inductor impedance, and so helps reduce RF in the output. In this case the RF causes some heating (increased by skin and proximity effects), but this cannot affect the resistance for audio frequencies. By the way, my cores get hardly warm at all, where Jan's coils do get quite warm, and I think this heating is mainly caused by the core material.
Groet, Bert
Joseph, reducing the resistance of the coil from 50 mOhm to 20 mOhm does increase the damping of the amplifier. If there is a choking coil between amp and woofer, you are right: this doesn't help much (for the woofer but not for the tweeter). In the setup I am putting together I have an amplifier for each driver, so each driver is damped directly by the amp.
Pjotr, you wrote:
"If the pulse output of the chip is “perfect distortionless” the energy lost in the coil is subtracted from that perfect output causing distortion. Especially with the output of the Tripath chips where the frequency is largely signal level dependant.
So in a way the varying frequency causes frequency dependant losses causing distortion."
My feeling is that if I had written such words, you would have considered it gibberish.
The simple scheme is that audio frequencies are not at all affected by skin or proximity effects, and RF frequencies are affected, but this effect adds to the inductor impedance, and so helps reduce RF in the output. In this case the RF causes some heating (increased by skin and proximity effects), but this cannot affect the resistance for audio frequencies. By the way, my cores get hardly warm at all, where Jan's coils do get quite warm, and I think this heating is mainly caused by the core material.
Groet, Bert
bertus said:
Hi Bert,
If that was the simple true story more core losses would improve sound also because the core eats up mainly HF energy, NOT audio energy...
The fact is that the coil needs to smooth out the HF content after the chip, that is its duty, not to waste part of it by losses! Not in the core and not in the wire.
From your findings it appears that core losses are dominant here.
Cheers 😉
summary of discovers
Hi Bert, George, Piotr,
could you resume for a poor guy that understand only notes at what point is the solution of the mistery of nice sound of Bert's coils?
Of course it's not interesting to say that it cannot be: they do 🙂 . What I'd like is to improve them. 😎
Would you say that a bigger core in Bert's coil would be better I'd try (I've ready two micrometals T106-2), or try bigger wire or thinner in bank winding.
Ciao
Giovanni
Hi Bert, George, Piotr,
could you resume for a poor guy that understand only notes at what point is the solution of the mistery of nice sound of Bert's coils?
Of course it's not interesting to say that it cannot be: they do 🙂 . What I'd like is to improve them. 😎
Would you say that a bigger core in Bert's coil would be better I'd try (I've ready two micrometals T106-2), or try bigger wire or thinner in bank winding.
Ciao
Giovanni
Re: summary of discovers
Hi Thomas,
Nobody is saying that Bert is on the wrong track 😉 We are discussing the ins and outs of the various details. Any of them has its pros and cons so we need to come to a combination of those details that are optimum. Only trying the real thing will prove what is right and was is less optimum.
The basic principle is to keep coil losses as small as possible, in the core as well as in the wire. Using a bigger core lowers core losses, a better less lossy core material also. A bigger core also gives you the opportunity to do with a single layer with some airspace between the turns. It can be that a single layer of thinner wire performs better than thicker wire with overlapping turns. But only trying will give you the ultimate answer. Anyway non-overlapping turns gives lowest parasitic capacitance.
“Bank wiring” is just a measure if the number of turns do not fit around the circumference of the core to reduce capacitance. But it is better if you can do without it.
Cheers 😉
thomaseliot said:
Hi Thomas,
Nobody is saying that Bert is on the wrong track 😉 We are discussing the ins and outs of the various details. Any of them has its pros and cons so we need to come to a combination of those details that are optimum. Only trying the real thing will prove what is right and was is less optimum.
The basic principle is to keep coil losses as small as possible, in the core as well as in the wire. Using a bigger core lowers core losses, a better less lossy core material also. A bigger core also gives you the opportunity to do with a single layer with some airspace between the turns. It can be that a single layer of thinner wire performs better than thicker wire with overlapping turns. But only trying will give you the ultimate answer. Anyway non-overlapping turns gives lowest parasitic capacitance.
“Bank wiring” is just a measure if the number of turns do not fit around the circumference of the core to reduce capacitance. But it is better if you can do without it.
Cheers 😉
Re: Re: summary of discovers
Hi Piotr,
many thanks to make things easy to understand for me. I was just asking what I could try to enhance Bert's coil. 🙂
If the mistery is in the core, what would be the bigger core with the best material to get 11uH with 1mm wire? I would try it.
Also I'm going to try copper foil coils (see next post).
Ciao
Thomas
Pjotr said:
Hi Piotr,
many thanks to make things easy to understand for me. I was just asking what I could try to enhance Bert's coil. 🙂
If the mistery is in the core, what would be the bigger core with the best material to get 11uH with 1mm wire? I would try it.
Also I'm going to try copper foil coils (see next post).
Ciao
Thomas
soundcheck said:
Hi Klaus,
I've seen your post only now, thanks. Still waiting for an answer from Mundorf for these 😀
An externally hosted image should be here but it was not working when we last tested it.
11uH is the value recommended by Tripath.
If I can say, with your wonder speakers I'd go for a more powerful amplifier than TA2020, its not bridgeable and a second one wouldn't help.
Ciao
Thomas
Re: Re: Re: summary of discovers
Hi Thomas,
To do it right it invokes a lot of calculation, using also some monographs and after that some trial and error is needed either. But it requires at least the knowledge how to calculate a basic inductor.
Here http://www.mag-inc.com/ you can download a lot about the backgrounds and how to calculate and even some software to do the math for you. But maybe this is a bit going too far for you.
I myself have best results with MMP cores for chokes up to 400 kHz to have the lowest losses but I did not try them for Tripath amps. MPP cores consist of 50% iron powder and 50 % nickel powder.
Cheers 😉
<Edit>
Btw, if you have difficulty to get some coils from Mundorf, go to your local speaker DIY shop and buy some 100 uH foil coils. Then carefully unwind them until you have 11 uH. It will be probably a lot cheaper also.
thomaseliot said:
Hi Thomas,
To do it right it invokes a lot of calculation, using also some monographs and after that some trial and error is needed either. But it requires at least the knowledge how to calculate a basic inductor.
Here http://www.mag-inc.com/ you can download a lot about the backgrounds and how to calculate and even some software to do the math for you. But maybe this is a bit going too far for you.
I myself have best results with MMP cores for chokes up to 400 kHz to have the lowest losses but I did not try them for Tripath amps. MPP cores consist of 50% iron powder and 50 % nickel powder.
Cheers 😉
<Edit>
Btw, if you have difficulty to get some coils from Mundorf, go to your local speaker DIY shop and buy some 100 uH foil coils. Then carefully unwind them until you have 11 uH. It will be probably a lot cheaper also.
What about this?
I just fixed 10A, 100kHz, 11uH and 16 AWG wire (this one tuning Current density), because recommended by Tripath.
The size of the core is obtained tuning the desired temperature rise.
Would be worth?
Ciao
Thomas
I just fixed 10A, 100kHz, 11uH and 16 AWG wire (this one tuning Current density), because recommended by Tripath.
The size of the core is obtained tuning the desired temperature rise.
Would be worth?
Ciao
Thomas
More on inductors
Thomas,
This appears to be a power supply output inductor design tool. I don't know how it would relate to the class D filter job. Really the jobs are quite different with no DC current, high frequency blocking and audio passing. I guess the ripple current should be considered the maximum audio current? Certainly it should not get near saturation at full output into 4 ohms. Design center frequency should be 500 KHz for high power amps and probably 750 KHz for the small chip amps. There is a broad range in the chip amps and some do go up to the 1 MHz range.
This could be a place to start though.
Roger
Thomas,
This appears to be a power supply output inductor design tool. I don't know how it would relate to the class D filter job. Really the jobs are quite different with no DC current, high frequency blocking and audio passing. I guess the ripple current should be considered the maximum audio current? Certainly it should not get near saturation at full output into 4 ohms. Design center frequency should be 500 KHz for high power amps and probably 750 KHz for the small chip amps. There is a broad range in the chip amps and some do go up to the 1 MHz range.
This could be a place to start though.
Roger
Hi Roger,
you are right, there is no way with that software to get a core with 11uH, 16AWG 29 turns, for example, tue values of the coil I have on the table. 😡
Ciao
you are right, there is no way with that software to get a core with 11uH, 16AWG 29 turns, for example, tue values of the coil I have on the table. 😡
Ciao
Coil turns?
The data aren't that bad, losses and all but you do need to rerun it @ 500 KHz to check. 9 turns is good in that it would allow the use of multiple strands or Litz wire. Losses would be even lower.
Roger
The data aren't that bad, losses and all but you do need to rerun it @ 500 KHz to check. 9 turns is good in that it would allow the use of multiple strands or Litz wire. Losses would be even lower.
Roger
HI,
When it comes to filter inductors the ripple current is the switching ripple at HF. The max audio current being much lower in frequency is actually considered as DC current. Max current is the two combined, I "think" something like Max DC + 20% for the ripple?
When it comes to filter inductors the ripple current is the switching ripple at HF. The max audio current being much lower in frequency is actually considered as DC current. Max current is the two combined, I "think" something like Max DC + 20% for the ripple?
- Status
- Not open for further replies.