Coil Crosstalk
I'm interested in some opinions on coil crosstalk as a function of orientation.
We usually see coils mounted axially, and the usual configurations are coaxial,
parallel or perpendicular, each with its own characteristics of crosstalk as a function of distance from each other.
This is often seen as an issue in stereo amplifiers, but is also an issue with regard to printed circuit traces and connecting wires.
I mentioned the use of radial output coils above, where the coil stands up on end. I'm inclined to think that two of these would probably exhibit the same crosstalk behavior between each other as axial mounted coils in a parallel configuration, but I could be wrong.
However, I suspect radial coils might have less crosstalk interaction with PCB traces, since those are parallel to the board surface and run perpendicular to the axis of the coil. This would also be the case for wires connecting to the amplifier board that tend to run parallel to the board.
Any opinions on whether the radial mounting of the coil has any advantages?
Cheers,
Bob
I'm interested in some opinions on coil crosstalk as a function of orientation.
We usually see coils mounted axially, and the usual configurations are coaxial,
parallel or perpendicular, each with its own characteristics of crosstalk as a function of distance from each other.
This is often seen as an issue in stereo amplifiers, but is also an issue with regard to printed circuit traces and connecting wires.
I mentioned the use of radial output coils above, where the coil stands up on end. I'm inclined to think that two of these would probably exhibit the same crosstalk behavior between each other as axial mounted coils in a parallel configuration, but I could be wrong.
However, I suspect radial coils might have less crosstalk interaction with PCB traces, since those are parallel to the board surface and run perpendicular to the axis of the coil. This would also be the case for wires connecting to the amplifier board that tend to run parallel to the board.
Any opinions on whether the radial mounting of the coil has any advantages?
Cheers,
Bob
Hi Bob,
I don't think there should be any cross-talk issues unless the coils are too close. An effective experiment would be to drive one channel at high power into a dummy load and listen to a speaker on the other channel. A cheap, efficient speaker would be perfect for this, we're talking corrugated paper surround kind of old school speaker.
If this does turn out to occur, what level of stereo separation might it be equivalent to? So even if it does occur, program material or cross-talk limits of the amplifier might make the individual case a non-issue.
-Chris
I think you're right about standing the coil perpendicular to the PCB. The same cross-talk will also apply, or close enough depending on any effects from the chassis.
I don't think there should be any cross-talk issues unless the coils are too close. An effective experiment would be to drive one channel at high power into a dummy load and listen to a speaker on the other channel. A cheap, efficient speaker would be perfect for this, we're talking corrugated paper surround kind of old school speaker.
If this does turn out to occur, what level of stereo separation might it be equivalent to? So even if it does occur, program material or cross-talk limits of the amplifier might make the individual case a non-issue.
-Chris
Keantoken,
If you are using the thermoplastic coated magnet wire you can solvent bond the wire. That was on the site I've mentioned twice now. It would be easy, make the coil, tie wrap three or four places and soak in solvent. Perhaps not as good as a baked coil but it is one of their recommendations, solvent type was on the site.
If you are using the thermoplastic coated magnet wire you can solvent bond the wire. That was on the site I've mentioned twice now. It would be easy, make the coil, tie wrap three or four places and soak in solvent. Perhaps not as good as a baked coil but it is one of their recommendations, solvent type was on the site.
Bob,
I'm not sure how it correlates to the smaller coils you'd have mounted on a pcb but in a crossover network we turn any coils perpendicular to each other so they don't interact as they would by putting them parallel. You don't have to worry about the board tracks but i would think it would be similar if he coils are close enough together to have overlapping fields. if you put two coils together and measure the inductance you can easily see how two coils can change the measured value of a single coil. Change the orientation with an LCR meter connected and this is obvious.
I'm not sure how it correlates to the smaller coils you'd have mounted on a pcb but in a crossover network we turn any coils perpendicular to each other so they don't interact as they would by putting them parallel. You don't have to worry about the board tracks but i would think it would be similar if he coils are close enough together to have overlapping fields. if you put two coils together and measure the inductance you can easily see how two coils can change the measured value of a single coil. Change the orientation with an LCR meter connected and this is obvious.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Bondable Magnet Wire
Bondable magnet wire, also referred to as self-bonding magnet wire, is film insulated wire top-coated with a thermoplastic adhesive. When activated, the thermoplastic bonds turn to turn windings to produce self-supporting coils or coils of unusual or difficult configurations. Use of bondable magnet wire may offer advantages over conventional magnet wire in certain winding applications, eliminating the need for bobbins as well as taping or varnishing steps. Activation of the bondcoat may be achieved with either heat, or in some cases solvent, or a combination of the two. Although bondcoats may be added to any conventional film, consideration should be given to the re-softening temperature of the adhesive in that it may not withstand the operating temperature of the higher rated primary insulations.
BONDING COATS
TYPE SOLVENT ACTIVATION BONDING
TEMPERATURE(°C)* SOFTENING
TEMPERATURE(°C)** Polyvinyl Butyral Alcohol 110 - 140 105 Epoxy MEK 150 - 200 130 Polyester MEK 190 - 210 130 Polyamide None 200 - 230 175
* May vary based on wire size and coil design
** most room temperature bond strength loss at this point
** most room temperature bond strength loss at this point
BONDING METHODS
Solvent Bonding:
Some bondcoats can be activated by applying certain solvents during or after the coil winding process. Application of the solvent, usually via saturated wick during winding, causes the bond coat to reflow. The process requires the use of a fixture to hold the coil in place while the solvent is drying. Once dry, the coil should be heated to dry off any residual solvent which might cause long-term coil failure, as well as to complete the bonding process.
Heat Bonding:
All bondcoats can be heat bonded, either by oven-heating or by directing hot air on to the wire during winding. In either case, the principle is to heat the winding slightly above the bondcoat's reflow temperature and then cool it below its rated bond strength temperature. Oven bonding is accomplished by heating the coil for a period of time sufficient to obtain uniform heating throughout the winding, followed by a cooling cycle. Heating time is generally 10 to 30 minutes, depending on the size of the winding. Disadvantages of oven bonding are the longer bonding time as well as the potential need for many winding fixtures. Hot air bonding, though done typically at slower winding speed, has the advantage of the elimination of a secondary bonding operation. This method is cost effective and usually associated with low temperature bondcoats and wire sizes smaller than 34 gauge.
Resistance Bonding:
Resistance bonding is done by applying electric current to the winding to electrically heat it to the proper bond temperature. Bonding voltage and time are dependent on wire size and coil design, and therefore will need to be developed experimentally for each specific application. This method has the advantages of being quick and generating uniform heat distribution. It is typically used for wire sizes heavier than 34 gauge.
When building speakers i find the mechanical stability of the coils to be top priority, then orientation, same goes for capacitors, most are really poor in the mechanical stability respect, most often are the big audiophile capacitors the worst, they are big, loose in the windings and have high(ish) inductance. If you terminate a cross over with resistors you can play music and listen to it, the components simply vibrate an play music, less so if your components are good, Coils must be hard and vacuum impregnated and I prefer stack-foil type caps. I look for good mechanical properties and low inductance, both is most often found in caps with a small physical size.
When I make output coils I wind them over the parallel resistor, I then mix some 5min epoxy, dump the gooed part into a Heatshrink tube and shrink as to seal the goo and coil in.
When I make output coils I wind them over the parallel resistor, I then mix some 5min epoxy, dump the gooed part into a Heatshrink tube and shrink as to seal the goo and coil in.
Agreed, if you are making a “conventional” linear amp with first or second-order compensation (see thought below).
There's been some over-the-top discussion about parasitic oscillation but this palls into insignificance compared to the products of EVIL speaker designers
Not to mention the speaker cable. A reasonable length (say 5 m) could have enough capacitance to make an amp with no output inductor go unstable.
Something to consider: class-D amps with post-filter feedback (e.g. Hypex NCore) do not require another inductor after the feedback loop, and use the output filter to provide two of the compensation poles. It would be fun to try and make a linear amplifier with fourth or fifth-order compensation where two of the poles are provided by a passive LC at the output. The inductor would then be inside the feedback loop and there’d be no worries about frequency response variations or distortions due to the inductor behaviour. In fact, an inductor with a magnetic core could be used (because negative feedback would take care of any related non-linearities) which then would mean no more worries about coupling or radiation via the magnetic field.
Here is what I recommend for an objective bench test for crosstalk between coils.
Build two 2uH coils and parallel each of them with 2 ohms, just as in a real amplifier. Load each of the coils with 8 ohms. Drive the first with a power amplifier at 10V rms. Ground the "input" end of the other coil. Measure the crosstalk voltage across the load of the second coil. A 1mV reading would correspond to crosstalk of -80dB.
This allows the crosstalk as a function of frequency to be measured in the same kind of environment they will be in in actual use. Crosstalk as a function of orientation and spacing can easily be investigated.
Such experiments would also demonstrate what coil geometries are more or less prone to crosstalk.
Cheers,
Bob
Yes, loudspeaker crossovers often take advantage of perpendicular coil orientations. The same will apply to crosstalk between output coils in a stereo amplifier. I've never seen a stereo power amplifier board that had the coils oriented cross-wise to each other.
Cheers,
Bob
I use 1.8 mm diameter wire on my coils (0.5 uH to 1 uH) - I just bought a small roll and standardized on that (kind of like my 0.33 Ohm emitter degen resistors). I often but not always use a cable tie to secure the coil. I must admit I had not given further mechanical stability a thought.
I like the idea of using epoxy or the pre-coated wire Kindhornmsn referenced.
I like the idea of using epoxy or the pre-coated wire Kindhornmsn referenced.
Each channel on my amps is on opposite sides of the chassis. I think orienting them so the inductor ends are not pointing at any of the sensitive electronics is good advice even under these conditions.
I guess one way to check if you hve a problem is just to short the inducor out - while driving a resistive load of course. If you see a difference in distortion look further - assuming of course there are no other issues with the design.
I guess one way to check if you hve a problem is just to short the inducor out - while driving a resistive load of course. If you see a difference in distortion look further - assuming of course there are no other issues with the design.
I don't think it's as much about putting new stuff in the book as much as it is about putting in stuff that everyone needs to know and having all of the modern knowledge in one place. If I had to write a book I wouldn't expect myself to be 100% original, that's just not going to happen.
So maybe Self has a few pages on the output coil and crosstalk (I seem to remember so). Does that mean every future book or article dealing with the topic should reference Self's work? Any engineer could come up with an article on that subject and devise the proper tests for themselves. It's nothing revolutionary or original to apply the scientific method.
If Bob does his own tests and presents his own findings in his own words, then I don't see the problem.
We're not all 100% original, yet we survive...
So maybe Self has a few pages on the output coil and crosstalk (I seem to remember so). Does that mean every future book or article dealing with the topic should reference Self's work? Any engineer could come up with an article on that subject and devise the proper tests for themselves. It's nothing revolutionary or original to apply the scientific method.
If Bob does his own tests and presents his own findings in his own words, then I don't see the problem.
We're not all 100% original, yet we survive...
Hi Bob,
Many thanks.
It isn't any stretch at all to test inside a chassis you are currently building to check out the layout either. The figure you gave how the detected signal equates to a real measurement is also very useful. I do realize that it isn't a calibrated test, but close enough to optimize a layout.
-Chris
Many thanks.
That is very "doable" without needing great power or very specialized equipment. I can do this, but members without a bench full of test gear will need an amplifier for the "dead" coil output.
It isn't any stretch at all to test inside a chassis you are currently building to check out the layout either. The figure you gave how the detected signal equates to a real measurement is also very useful. I do realize that it isn't a calibrated test, but close enough to optimize a layout.
-Chris