I set to 200KHz high pass at this point. But it can easily be changed. I don't think it's a good idea to use anything active right at the input, you have to put a higher divider ratio to keep it from ever clipping at all condition because clipping of audio frequency will generate odd harmonics that can extend to the pass band of any filter and be detected as RF oscillation. You want to have the passive filter as the first stage to attenuate the amplitude before driving any active device.
To moderator
Maybe it's time to move this discussion to a separate thread. When I responded to Mr. Cordell, I did not expect this become a life of it's own.
Last edited:
Not to mention a elevated noise gain at the Fc.
Shibasoku used 9th order Chebychev in the 725D analyzer.
The filter is unusable for extremely low THD measurements.
The filter exhibits some 20dB of noise gain at 20kHz.
Perhaps. The filter is for digital audio use. The noise, in the THD mode called analysis mode, is removed by the 725's discrete DSP. The noise from the 20KHz makes THD+N mode almost impossible. However there are 30KHz and 100KHz band limit filters as well that don't elevate the noise. I'm still scratching my head on this 20KHz filter though.
Well, the DSP still has to work in a Gaussian way. No brick walls without a lot of ringing. Are there any test points that you can chase the signal through? I guess you would have visited them a long time ago. An impulse test might show the answer.
You may well be troubleshooting a feature that they had to fix elsewhere. Still, it sounds like a wonderful instrument.
-Chris
You may well be troubleshooting a feature that they had to fix elsewhere. Still, it sounds like a wonderful instrument.
-Chris
The only purpose of the filter is the attenuate the audio frequency so it does not swamp the detector and only detect the high frequency oscillation. The only requirement is to attenuate as much as possible. We don't care about distortion, ripple, noise or anything, just need to get rid of the audio signal.
In the simulation circuit in post #6784, I found a 22uH for L1, 850mA saturation current that resonance freq of 100MHz. B82462G4223M EPCOS (TDK) | 495-1992-1-ND | DigiKey This will be a good one for the first inductor. Even though it poops out at 100MHz, The attenuation of audio frequency should be extreme high already. Using two of the 0.01uF 100V 0805 or 1206 ceramic cap in parallel for C1 should give low enough parasitic. I think we don't even need any divider at the front.
In the simulation circuit in post #6784, I found a 22uH for L1, 850mA saturation current that resonance freq of 100MHz. B82462G4223M EPCOS (TDK) | 495-1992-1-ND | DigiKey This will be a good one for the first inductor. Even though it poops out at 100MHz, The attenuation of audio frequency should be extreme high already. Using two of the 0.01uF 100V 0805 or 1206 ceramic cap in parallel for C1 should give low enough parasitic. I think we don't even need any divider at the front.
Hi Alan,
I think we know that. David and I were discussing something else entirely.
The ripple in the stop band may cause you grief. The stop band cuts quickly, then lower down will recover with a ripple in the response. Check this before blindly building a filter like a Chebychev filter. Since 100 MHz is well outside your stop band, the inductor bandwidth isn't important up there. It does it's work down in the audio band. As far as parasitic capacitance or inductance goes, you needn't worry about having to use surface mount parts. The normal leaded capacitors work just fine. Look inside any 100 MHz 'scope and see what I mean. Earlier ones used leaded parts. The new ones at 100 MHz use surface mount parts because they are less expensive to build with.
So, relax. Build your filter and test it out before feeding the output to any poor defenseless integrated circuits. All you need is your signal generator and a 'scope to play with this. Have fun and push the books aside for a change.
-Chris
I think we know that. David and I were discussing something else entirely.
The ripple in the stop band may cause you grief. The stop band cuts quickly, then lower down will recover with a ripple in the response. Check this before blindly building a filter like a Chebychev filter. Since 100 MHz is well outside your stop band, the inductor bandwidth isn't important up there. It does it's work down in the audio band. As far as parasitic capacitance or inductance goes, you needn't worry about having to use surface mount parts. The normal leaded capacitors work just fine. Look inside any 100 MHz 'scope and see what I mean. Earlier ones used leaded parts. The new ones at 100 MHz use surface mount parts because they are less expensive to build with.
So, relax. Build your filter and test it out before feeding the output to any poor defenseless integrated circuits. All you need is your signal generator and a 'scope to play with this. Have fun and push the books aside for a change.
-Chris
Yes, it is only for digital audio tests. Not THD+N but FFT (DSP). Analog would use the 30-100KHz filters... IIRC, there is also a position for adding your own filter characteristics.
THx-RNMarsh
Option switch and your own plugin. There is a back plane connector free for the option board.
Anyway dis Bob's' thread.
Compact output coils
We’ve all seen power amplifiers with large, impressive-looking output coils. Some are as large as 1 inch in diameter and just as long, and sometimes constructed with wire AWG as large as #14. It is intuitive to think that these are important for the best audio quality. Some have argued that the use of an output coil degrades the sound and they advocate not using an output coil at all. Others have argued that an output coil is OK as long as its inductance value is no more than 1-2 uH and that the coil is of high quality and not in proximity to ferrous materials. Some also argue for large-AWG coils to preserve the very large damping factor that is often achieved with solid state amplifiers.
I’ve been giving this some additional thought lately and would like to play the Devil’s Advocate and perhaps stir up some discussion. I will question some of the arguments and assumptions made. Are large output coils really necessary, or even desirable? I’ll test the water by seeing if I can make the case for smaller coils made with smaller AWG wire. Let’s assume that the target is a 2uH coil.
First is the issue of coil resistance and damping factor (DF). Large AWG is usually necessary to get coil resistance down into the teens of milli-ohms or lower. Some advocate 10 mohm or less in the quest to chase big DF numbers. I think this is a bit mis-guided. Even a 40 mohm coil will only limit DF to 200. Although I am an advocate of high DF, a value of 200 should be plenty, especially in light of the reality of speaker cable resistance.
Some may express concern about power dissipation in such a coil. It will be in proportion to the coil resistance as a fraction of the load impedance. If a 40 mW coil is loaded with 4W, then the coil will dissipate 1% of the power. If the power output is 400W into 4 ohms, the coil will dissipate 4W. This is adequately small and will only be an issue in continuous sine wave testing anyway.
Some may have a knee-jerk reaction against pushing 10’s of amperes through a couple of feet of #22, even if only briefly. No problem. It is all about heat. Tiny bond wires in power transistors do it all the time. But they are inside the feedback loop; the output coil is not. Might there be non-linearity associated with the current density in the cross-section of the coil wire? Are there any physicists in the house?
Here I’ll advocate for coils that are less than about 0.5 inch in diameter and wound with 18-22 AWG wire, and which are no longer than about 0.5 inch. Here are some examples of 2uH coils that obey this criteria:
1. 14T, #22 on 0.5 inch, length = 0.35 inch: 29 mohm
2. 21T, #22 on 0.375 inch, length = 0.55: 33 mohm
3. 18T, #22 on 2 MOF 3W resistors, 2 layers, length = 0.25: 27 mohm
4. 19T, #18 on 0.5 inch, length = 0.77: 16 mohm
5. 14T, #18 on 0.5 inch, 2 layers, length = 0.3: 13 mohm
These designs are from inductance calculators, but a few examples have been wound and seem reasonably close to prediction.
Coil #3 can be mounted as a radial component. One layer can run up and the second layer can run back down. The two resistors on which it is mounted can be connected in series at the top end. I wonder if radial mounting geometry might reduce magnetic interactions with wires and traces running parallel to the surface of the board. In fairness, a 2-layer coil takes more labor to build.
A smaller coil takes up less precious board space. Moreover, a smaller coil of the same inductance exhibits a more compact magnetic field than a larger coil. This reduces cross-talk into and from nearby circuitry, and into the other channel in a stereo amplifier. Interaction with nearby ferrous materials might be reduced as well.
Large-AWG coils are not all they are cracked up to be when skin effect is considered. For 14 AWG, the skin depth equals the radius of the wire at only 6.7 kHz. At this frequency, the ac resistance due to skin depth is about equal to the dc resistance, doubling total resistance. In contrast, the skin depth frequency of 22 AWG is well above the audio band at 42 kHz. Does this matter? I don’t know.
Larger-AWG (#14 - #18) coils are usually self-supporting, while smaller-AWG coils are on a former or even a large resistor. The self-supporting coils can be subject to magnetically-induced vibration and resonance when high current is passed. I wonder what the mechanical resonance frequency of such coils might be. Have any of you heard an output coil (or anything else) “sing” in an amplifier under load when driven with a sine wave? Not sure if I’ve seen coil dope on such coils, but it might be a good idea. Smaller coils on formers can easily be doped and encased in heat shrink, greatly reducing the possibility to move or resonate. Is signal-induced movement within the coil an issue deserving concern in high-end audio? I don’t know.
OK, I’m ready for some tomatoes. Let’s hear some pro and con opinions.
Cheers,
Bob
We’ve all seen power amplifiers with large, impressive-looking output coils. Some are as large as 1 inch in diameter and just as long, and sometimes constructed with wire AWG as large as #14. It is intuitive to think that these are important for the best audio quality. Some have argued that the use of an output coil degrades the sound and they advocate not using an output coil at all. Others have argued that an output coil is OK as long as its inductance value is no more than 1-2 uH and that the coil is of high quality and not in proximity to ferrous materials. Some also argue for large-AWG coils to preserve the very large damping factor that is often achieved with solid state amplifiers.
I’ve been giving this some additional thought lately and would like to play the Devil’s Advocate and perhaps stir up some discussion. I will question some of the arguments and assumptions made. Are large output coils really necessary, or even desirable? I’ll test the water by seeing if I can make the case for smaller coils made with smaller AWG wire. Let’s assume that the target is a 2uH coil.
First is the issue of coil resistance and damping factor (DF). Large AWG is usually necessary to get coil resistance down into the teens of milli-ohms or lower. Some advocate 10 mohm or less in the quest to chase big DF numbers. I think this is a bit mis-guided. Even a 40 mohm coil will only limit DF to 200. Although I am an advocate of high DF, a value of 200 should be plenty, especially in light of the reality of speaker cable resistance.
Some may express concern about power dissipation in such a coil. It will be in proportion to the coil resistance as a fraction of the load impedance. If a 40 mW coil is loaded with 4W, then the coil will dissipate 1% of the power. If the power output is 400W into 4 ohms, the coil will dissipate 4W. This is adequately small and will only be an issue in continuous sine wave testing anyway.
Some may have a knee-jerk reaction against pushing 10’s of amperes through a couple of feet of #22, even if only briefly. No problem. It is all about heat. Tiny bond wires in power transistors do it all the time. But they are inside the feedback loop; the output coil is not. Might there be non-linearity associated with the current density in the cross-section of the coil wire? Are there any physicists in the house?
Here I’ll advocate for coils that are less than about 0.5 inch in diameter and wound with 18-22 AWG wire, and which are no longer than about 0.5 inch. Here are some examples of 2uH coils that obey this criteria:
1. 14T, #22 on 0.5 inch, length = 0.35 inch: 29 mohm
2. 21T, #22 on 0.375 inch, length = 0.55: 33 mohm
3. 18T, #22 on 2 MOF 3W resistors, 2 layers, length = 0.25: 27 mohm
4. 19T, #18 on 0.5 inch, length = 0.77: 16 mohm
5. 14T, #18 on 0.5 inch, 2 layers, length = 0.3: 13 mohm
These designs are from inductance calculators, but a few examples have been wound and seem reasonably close to prediction.
Coil #3 can be mounted as a radial component. One layer can run up and the second layer can run back down. The two resistors on which it is mounted can be connected in series at the top end. I wonder if radial mounting geometry might reduce magnetic interactions with wires and traces running parallel to the surface of the board. In fairness, a 2-layer coil takes more labor to build.
A smaller coil takes up less precious board space. Moreover, a smaller coil of the same inductance exhibits a more compact magnetic field than a larger coil. This reduces cross-talk into and from nearby circuitry, and into the other channel in a stereo amplifier. Interaction with nearby ferrous materials might be reduced as well.
Large-AWG coils are not all they are cracked up to be when skin effect is considered. For 14 AWG, the skin depth equals the radius of the wire at only 6.7 kHz. At this frequency, the ac resistance due to skin depth is about equal to the dc resistance, doubling total resistance. In contrast, the skin depth frequency of 22 AWG is well above the audio band at 42 kHz. Does this matter? I don’t know.
Larger-AWG (#14 - #18) coils are usually self-supporting, while smaller-AWG coils are on a former or even a large resistor. The self-supporting coils can be subject to magnetically-induced vibration and resonance when high current is passed. I wonder what the mechanical resonance frequency of such coils might be. Have any of you heard an output coil (or anything else) “sing” in an amplifier under load when driven with a sine wave? Not sure if I’ve seen coil dope on such coils, but it might be a good idea. Smaller coils on formers can easily be doped and encased in heat shrink, greatly reducing the possibility to move or resonate. Is signal-induced movement within the coil an issue deserving concern in high-end audio? I don’t know.
OK, I’m ready for some tomatoes. Let’s hear some pro and con opinions.
Cheers,
Bob
I don't want to give theory, I can only tell why I ended up with what I have. I use thick wire of 0.065" diameter ( I can't find the roll, I know is either 16 or 14 gauge). I would it on an AAA battery for 14 turns.
I want to use as heavy as possible. I do have experience with speaker cable and I experimented and end up with 4 pairs of 12gauge monster type cables per speaker. I could and my wife could actually hear the difference when adding more pairs until the 5th. So I am die hard heavy gauge fan when anything comes to connecting to speaker. I feel the wire of that inductor is at least 12" long if you pull it straight, I sure won't want to have expensive big speaker cable just to be bottle necked by the inductor. Even with big wire, I still want to make it as short as possible.
I want to use as heavy as possible. I do have experience with speaker cable and I experimented and end up with 4 pairs of 12gauge monster type cables per speaker. I could and my wife could actually hear the difference when adding more pairs until the 5th. So I am die hard heavy gauge fan when anything comes to connecting to speaker. I feel the wire of that inductor is at least 12" long if you pull it straight, I sure won't want to have expensive big speaker cable just to be bottle necked by the inductor. Even with big wire, I still want to make it as short as possible.
I remember building some Dyna Stereo 120s, which had quite large diameter output coils
around the output blocking caps (it was a single supply design).
Bender-120.ULT Rebuild Kits Page
yes... across the board the output inductor is too small wire gauge.
It may also be easier and maybe better location to place it near the amp's speaker binding posts.
THx-RNMarsh
Hi AJT,
This could turn out to be a lengthy discussion too!
Hi Bob,
Excellent topic! Very rarely talked about.
Excellent topic! Thanks for bringing it up.
-Chris
This could turn out to be a lengthy discussion too!
Hi Bob,
Excellent topic! Very rarely talked about.
I'm of the same mind there ... There is a balancing act between necessary but competing design requirements. It's all about how that balance comes out.
... and there is the balance and rational behind it. I would accept that BTW.
I would cautiously agree with you. I have serviced amplifiers where the coil has been cooked, some from the amp going DC. You just know the speakers weren't very happy about it either!
I'm not a physicist, but I do know that heating copper wire will raise the resistance. Will this matter considering the crossover and voice coils will change even more since they are longer lengths of copper wire? Another decision to make.
Sounds good, and practical. My choice would be #5, a larger gauge, two layer design. I think it delivers the best compromise of board space, IR losses and cost.
Great pints to consider ... I mean points ...
However, the 22 ga wire is much higher in resistance than the larger ga wire to start with. When then does the HF losses equal the losses at 22 ga. You also have skin effect on the speaker wire, cross-over components and ... the voice coils. Where are the biggest losses and how great are they? This might be valid if you're really thinking a 22 ga wire coil is on the table. I dropped it from consideration a while ago.
Probably. But what is the maximum displacement? Probably not very much, and a little silicone (in lengthwise strips) would damp them.
Higher in the larger coils due to stiffness to weight ratio. However, coil vibration can be damped. Also, the power transformer probably puts out more noise, and the output coil vibration occurs in sync with loud music passages - you wouldn't hear them anyway.
No, the ceramic capacitors were louder.
I have. Normally they use magnet wire which comes pre-coated.
But heat shrink is expensive, and it will block air flow to cool the coil. The smaller coils would require more air cooling than the larger ones as more heat leaves via the leads in large ga coils.
Sure, if only to conclude it isn't an issue. This issue is easy to solve with damping material that does not block too much air flow. Silicone. Lovely stuff.
Excellent topic! Thanks for bringing it up.
-Chris
perhaps Bob can devote an entire chapter or at the very least an appendix...
all the while, i was thinking that the output coils were no big deal till now...
i guess you learn something new each day....
AFAIR, Daniel Meyer also suggested this in in Tigersaurus 250 watter build...
6 turns of #16 magnet wire on the body of one of the large filter caps.
ant those caps were at least 2.5 inches in diameter...
An externally hosted image should be here but it was not working when we last tested it.
for minimum coil resistance one needs a coil diameter to coil length ratio of roughly 4.
Any coil that is longer than ~25% of the coil diameter will have a higher resistance.
Using 13.92T in 3 layers on a 8.26mm diameter bobbin that is 3.93mm wide one gets 2.0005uH and 11.3milli-ohms for 516.2mm of wire in the coil. the overall diameter will be 15.34mm. One would have to add a bit more length and resistance for the tails.
This is arrived at using Wheeler's formula: L = 7.87M²N² / 3M+9B+10C, in nH when dims are in mm
where M= mean diameter, N= Turns, C= coil height, B= coil width.
M+C = overall diameter, M-C = bobbin diameter
The numbers are a bit odd since I adjusted dimensions until the Inductance L near enough equaled 2uH.
0.7mm wire results in 0r0201
1mm wire results in 0r0113
1.2mm wire results in 0r0085
1.5mm wire results in 0r0059
1.8mm wire results in 0r0044
These are the minimum resistances for an air-cored copper coil of 2uH
Longer coils will have more resistance.
Adding on a correction for skin depth will generally add less resistance than changing to a long coil.
Any coil that is longer than ~25% of the coil diameter will have a higher resistance.
Using 13.92T in 3 layers on a 8.26mm diameter bobbin that is 3.93mm wide one gets 2.0005uH and 11.3milli-ohms for 516.2mm of wire in the coil. the overall diameter will be 15.34mm. One would have to add a bit more length and resistance for the tails.
This is arrived at using Wheeler's formula: L = 7.87M²N² / 3M+9B+10C, in nH when dims are in mm
where M= mean diameter, N= Turns, C= coil height, B= coil width.
M+C = overall diameter, M-C = bobbin diameter
The numbers are a bit odd since I adjusted dimensions until the Inductance L near enough equaled 2uH.
0.7mm wire results in 0r0201
1mm wire results in 0r0113
1.2mm wire results in 0r0085
1.5mm wire results in 0r0059
1.8mm wire results in 0r0044
These are the minimum resistances for an air-cored copper coil of 2uH
Longer coils will have more resistance.
Adding on a correction for skin depth will generally add less resistance than changing to a long coil.
Last edited:
This I will follow, from all the advice that I could found (no coil, 1uh...4uH and other) I selected 1uH and I went for low resistance and good (impressive ) looking the resulting coil was 20mm diameter and created from 6mm2 (AWG 9+-) the resistance is about 15mOhm. Do I need to be ashamed... or proud... as I said before I will follow this.
) looking the resulting coil was 20mm diameter and created from 6mm2 (AWG 9+-) the resistance is about 15mOhm. Do I need to be ashamed... or proud... as I said before I will follow this.
Last edited:
Hi Richard,
Yes, I understand and respect your opinion favoring larger wire gauge - and you indeed may be right. You are certainly not alone.
But what is your reason for this technical preference? For example, are you basing this on the issue of coil resistance alone, as to provide DF greater than 200, or is it something else?
Cheers,
Bob