Interesting, a spiral wound inductor?
With the active (driven) foil on the inside it might be in a way 'self shielding' - what do you think?
True, according to all the reviews it sounded pretty high end too though which wouldn't be harmed by going air. I did note the cool capacitors though - too expensive and big for my DIY.
One thing I didn't see was the active part - is that underneath as surface mount?
Yes I've been reading about this, it does seem the best method of using a core.
Still not convinced it's as linear as air though, plus they're much more difficult for me to make. One of the DIY wins of air is that I can wander into the garage and make them (as long as i can count to 30 or 40 four times so I have to be sober).
Plus there's a lot more air available in the garage than ferrite.. 😀
I'm not 100% certain but it doesn't look spiral wound. It looks like the kind of foil inductors you'd use in a crossover.
FOIL INDUCTORS
It looks like the FETs are clamped to the walls.
Redirect Notice
By definition a ferrite core will never be as linear as an air core, but if you design a gapped ferrite core properly it can be very linear over the operating range of the amplifier. You can almost as easily make your own by using a dremel and cutting wheel to grind out a small gap in a ferrite or powdered iron core.
Interesting info, thanks.
The inductors look spiral to me which is genius. Skin effect and spiral wind makes then very interesting indeed. They sell them as small as 50uH, not down to the 10uH I'd be after but they look eminently DIYable to me and would appear to have significant benefits.
I'll look into this more now.
An externally hosted image should be here but it was not working when we last tested it.
The inductors look spiral to me which is genius. Skin effect and spiral wind makes then very interesting indeed. They sell them as small as 50uH, not down to the 10uH I'd be after but they look eminently DIYable to me and would appear to have significant benefits.
I'll look into this more now.
air core leak emi like crazy, but are immune to saturation and add little distortion
taking 1/3 of the turns off the 50uH coil should net about 20uH. you would have to measure as the inductance increses with coil thickness
taking 1/3 of the turns off the 50uH coil should net about 20uH. you would have to measure as the inductance increses with coil thickness
Last edited:
Not convinced about the EMI leaks yet TBH, I think they just need a little space to work in.
I'll get some better EMI data in the next week or two.
I've had a calculate using the nearest cheapest material - EMI shielding tape 😀
Using 10mm copper tape here: 30M length Conductive Copper Foil Tape, self adhesive, EMI Shielding - UK Stock | eBay
and this calculator
Flat spiral coil design calculator
Wound on a 10mm former it needs 22 turns for 10uH and creates a 13mm dia result.
Not sure if inter-turn capacitance would be low enough.
Wire cross section is only 0.03mm x 10mm = 0.3mm2 but skin effect area is good. (Do we want skin area to be good?).
DC resistance would be 44 milliOhm if I calculated correctly (0.3mm2 => 0.61mm dia wire.
Cheap too at about 33 pence per coil and rigid as the adhesive is built in...
I'll get some better EMI data in the next week or two.
I've had a calculate using the nearest cheapest material - EMI shielding tape 😀
Using 10mm copper tape here: 30M length Conductive Copper Foil Tape, self adhesive, EMI Shielding - UK Stock | eBay
and this calculator
Flat spiral coil design calculator
Wound on a 10mm former it needs 22 turns for 10uH and creates a 13mm dia result.
Not sure if inter-turn capacitance would be low enough.
Wire cross section is only 0.03mm x 10mm = 0.3mm2 but skin effect area is good. (Do we want skin area to be good?).
DC resistance would be 44 milliOhm if I calculated correctly (0.3mm2 => 0.61mm dia wire.
Cheap too at about 33 pence per coil and rigid as the adhesive is built in...
Hi,
I'm not sure these inductors are the best thing you can use for output filtering in a class D. They are designed for crossovers in audio band and
not for PWM carrier suppressing in a class D.
Few reasons I think of:
- Inter turn capacitance are increased, so they can be less effective at suppressing PWM carrier.
If it is spiral than maybe it is done to make it wideband enough for audio, not one decade higher.
- The filter inductor is in series with the channel resistance of the output stage FETs which are normally in the range of 50 mohms, so you don't need uber low resistance in the inductor.
In theory inductor resistance may help improving output stage nonlinearity
at the expense of some dissipation.
- Skin effect, if matters at all in this app, is on your side to suppress PWM carrier.
I'm not sure these inductors are the best thing you can use for output filtering in a class D. They are designed for crossovers in audio band and
not for PWM carrier suppressing in a class D.
Few reasons I think of:
- Inter turn capacitance are increased, so they can be less effective at suppressing PWM carrier.
If it is spiral than maybe it is done to make it wideband enough for audio, not one decade higher.
- The filter inductor is in series with the channel resistance of the output stage FETs which are normally in the range of 50 mohms, so you don't need uber low resistance in the inductor.
In theory inductor resistance may help improving output stage nonlinearity
at the expense of some dissipation.
- Skin effect, if matters at all in this app, is on your side to suppress PWM carrier.
I think too that the capacitance is a very bad thing, I can half that by winding back-to-back tape (doubles conductor and insulation thickness).
The skin effect turns our LC filter into an RLC filter at frequency which would help damping and reduce its level.
Doubling up the tape seems to be the obvious then, if only costs a couple of mm of coil width to 16mm. It also halves resistance which is useful to reduce coil heating.
I think we have a winner 🙂
The skin effect turns our LC filter into an RLC filter at frequency which would help damping and reduce its level.
Doubling up the tape seems to be the obvious then, if only costs a couple of mm of coil width to 16mm. It also halves resistance which is useful to reduce coil heating.
I think we have a winner 🙂
That calculator is for a flat (or planar) spiral wound inductor with round magnet wire, not flat foil.
The inter-winding capacitance with flat foil will be much larger than a single layer inductor with normal magnet wire.
Article about field and circuit models for wound foil inductors: http://downloads.hindawi.com/journals/apec/1976/803052.pdf
Article about designing copper foil inductors: http://powerelectronics.com/mag/Datatronics.pdf
Last edited:
any solenoid will leak, even a torroid leaks at the 360 junction.
opposed solenoids in bucking mode would be verry interesting test though.
its on my todo list.
air core has no barkhausen noise , so maybe hiss go away as found on some designs.
opposed solenoids in bucking mode would be verry interesting test though.
its on my todo list.
air core has no barkhausen noise , so maybe hiss go away as found on some designs.
Any updates?
Great reading!! And lots to learn!!
I have wound air-coils about 10 years ago for a Tripath and it improved sound by a substantial margin (much more transparent yet less harsh = airy yet stable).
Would love to make coils again for my fully active setup using Zoudio boards that produce 50 watt (more like 38Watt) per channel. Could not see any pictures in the thread, and would love to know if the ribbon type finally was superior to the 'sewing-machine' coils in sound and EMI.
I would like to wind 10uH coils (16 of them ;-). If someone has made this value in foil, I would love to hear the exact procedure and measures!
So reposting some pics would be lovely!! (small size and not externally hosted please so next generations can see them too ;-)
Great reading!! And lots to learn!!
I have wound air-coils about 10 years ago for a Tripath and it improved sound by a substantial margin (much more transparent yet less harsh = airy yet stable).
Would love to make coils again for my fully active setup using Zoudio boards that produce 50 watt (more like 38Watt) per channel. Could not see any pictures in the thread, and would love to know if the ribbon type finally was superior to the 'sewing-machine' coils in sound and EMI.
I would like to wind 10uH coils (16 of them ;-). If someone has made this value in foil, I would love to hear the exact procedure and measures!
So reposting some pics would be lovely!! (small size and not externally hosted please so next generations can see them too ;-)
Last edited:
I'd like to have air-core coils for my Zoudio amps too. A couple observations;
- Mine are setup as a simple two-way, with cross in the low, hundreds of Hz range. I assume it's a waste of time / effort for the low-pass outputs...
- How the heck would you remove the existing inductors? Heat the whole board to solder temperature and play "leggo-my-eggo" with a pair of tweezers? That filter cap - a fairly tiny SMD - is right there, obscuring entry of a soldering iron tip big and wide enough to heat that foil. 100% success times 4 (or 8) is required not to destroy the amp.
I suppose you could crush them out mechanically, but that seems like a distasteful approach. As in many of these class-D amplifiers' "cell phone level of compaction, for compaction's sake", no quarter was given for the pooge experimenter. Even jumping them with a wire added over the top, pulling the caps and attempting to connect up a new filter arrangement outboard - how to get a good ground connection? A wire back to each cap's ground pad?
Seems like major surgery any way I look at it. "1st, pull the output connector strip". Yeah, that's going to come right off without taking anything with it...
- Mine are setup as a simple two-way, with cross in the low, hundreds of Hz range. I assume it's a waste of time / effort for the low-pass outputs...
- How the heck would you remove the existing inductors? Heat the whole board to solder temperature and play "leggo-my-eggo" with a pair of tweezers? That filter cap - a fairly tiny SMD - is right there, obscuring entry of a soldering iron tip big and wide enough to heat that foil. 100% success times 4 (or 8) is required not to destroy the amp.
I suppose you could crush them out mechanically, but that seems like a distasteful approach. As in many of these class-D amplifiers' "cell phone level of compaction, for compaction's sake", no quarter was given for the pooge experimenter. Even jumping them with a wire added over the top, pulling the caps and attempting to connect up a new filter arrangement outboard - how to get a good ground connection? A wire back to each cap's ground pad?
Seems like major surgery any way I look at it. "1st, pull the output connector strip". Yeah, that's going to come right off without taking anything with it...
Looking at the AIO4CH board, it seems shorting the existing inductors externally may be possible, with no harm to adjacent components. Removal of the existing caps as well. Simulations tell me the effectiveness of the caps (as a second order filter component and as a continuous conduction mode path for the inductor to charge/discharge current) dont care a bit if there's some extra lead inductance, so a short wire back to the SMD ground pad could actually work.
So one approach might be;
1. Build the new LC network as an outboard fixture, inside a fully enclosed metal box or not.
2. Connect to it through the existing speaker terminal strip - no need to remove.
3. Short the existing inductors, leaving them otherwise in place and undisturbed on the board.
4. Remove the existing MLC SMD caps.
5. Arrange some way to get to board Ground. This could be a wire to each SMD ground pad, through the Power Supply input ground connection, or both.
Until running the simulations, I never knew that the speaker load is the only thing dampening the class D output LC filter. Using the typical 10uH / .68uF values, 4 ohms gives the best behavior. Zero Ohms load and the thing goes sky-high at resonance. TI mentioned in their slaa701a.pdf "LC Filter Design" that allowing such higher Q resonance can be perceived audibly as a "harsh sounding" high end - as well as cause inadvertant amplifier shutdowns.
If there's truth to that, our speaker loads better be well behaved electrically, when using an ordinary class D amplifier. I wouldnt be surprised if a Zobel at the amplifier output terminals could change sonic character. My amps - thanks to Zoudio enabling the programmable filters therein - connect directly to the driver terminals. Still, that's not the same as a resistor in a simulation - but better than a crossover RLC network as load.
So one approach might be;
1. Build the new LC network as an outboard fixture, inside a fully enclosed metal box or not.
2. Connect to it through the existing speaker terminal strip - no need to remove.
3. Short the existing inductors, leaving them otherwise in place and undisturbed on the board.
4. Remove the existing MLC SMD caps.
5. Arrange some way to get to board Ground. This could be a wire to each SMD ground pad, through the Power Supply input ground connection, or both.
Until running the simulations, I never knew that the speaker load is the only thing dampening the class D output LC filter. Using the typical 10uH / .68uF values, 4 ohms gives the best behavior. Zero Ohms load and the thing goes sky-high at resonance. TI mentioned in their slaa701a.pdf "LC Filter Design" that allowing such higher Q resonance can be perceived audibly as a "harsh sounding" high end - as well as cause inadvertant amplifier shutdowns.
If there's truth to that, our speaker loads better be well behaved electrically, when using an ordinary class D amplifier. I wouldnt be surprised if a Zobel at the amplifier output terminals could change sonic character. My amps - thanks to Zoudio enabling the programmable filters therein - connect directly to the driver terminals. Still, that's not the same as a resistor in a simulation - but better than a crossover RLC network as load.
Last edited: