There has been a lot of talk about using aircore inductors in the output filters of Tripath amps.
It has been my opinion that the aircore is ill suited for this use because it tends to radiate much more RF than a bobbin or toroidal type inductor. It will also likely have a higher DCR than the other types, due to the higher number of turns needed.
At long last I have been able to test the 3 different types of coils on the Sonic Impact board. Below are the results. But 1st a little background.
The Tripath is a switching type amp that switches at frequencies ranging from about 700-1000 kHz. This switched signal is varied in duty cycle following the audio signal amplified. The high frequency switched square wave signal is fed thru an LC low pass filter to recover the audio signal.
The inductors in question are the L of this filter. Most small class-T amps use a bobbin type coil but other types are used or available as upgrades. Which works best? And why?
The goal of the inductors should be to attenuate the ultrasonic switching frequencies, while having as little effect as possible on the audio signal.
How do the 3 types stack up?
Using the stock bobbin type coils as a zero reference, below we see relative levels of RF in radiated and transmitted output form.
Stock Bobbin: (reference)
Radiated 0dB
Output into 8 ohms 0dB
Aircore (Autocostruire)
Radiated + 11.5 dB
Output into 8 ohms -5.6dB
Toroidal (type 61 core)
Radiated +8 to -4dB
Output into 8 ohms -6dB
From the above we can see several things.
1) The aircore and toroidal allow 1/2 the RF sent to the load as the stock bobbin. That’s good.
2) The aircore radiates almost 4 times as much RF as the stock coil. That's bad. The toroidal radiates more and less than stock. That's because the radiation pattern is not symmetrical. More on that below.
What we can not see from the above is that the bobbin and aircore inductors tend to radiate evenly thru 360 deg. parallel to their length. The toroidal radiates the most parallel to its flat side and very little on the ends. A figure 8 pattern.
Also, as the aircore coils tend to radiate so much more than the others, the spaces between the coils can see combined signal levels at least twice as high as elsewhere. The other coils do have combing effects, but they tend to be only 30~40% above non combined. Thus the area between the 4 coils where the electrolytic caps lives is very hot in RF with the aircores.
Another thing not shown is the resulting waveform after filtration. I will try to post a photo later. The toroidal and aircore both give something near a sine wave, while the bobbin shows a strong triangle wave. The triangle wave is evidence that a lot of RF harmonics are left in the signal.
The measurements were done at 12V into an 8 ohm resistive load. Stock 150nF caps across the output terminals.
Radiation was measured with a sort length of wire attached to an oscilloscope probe. Wire was placed 3mm from the coils for measurements.
It has been my opinion that the aircore is ill suited for this use because it tends to radiate much more RF than a bobbin or toroidal type inductor. It will also likely have a higher DCR than the other types, due to the higher number of turns needed.
At long last I have been able to test the 3 different types of coils on the Sonic Impact board. Below are the results. But 1st a little background.
The Tripath is a switching type amp that switches at frequencies ranging from about 700-1000 kHz. This switched signal is varied in duty cycle following the audio signal amplified. The high frequency switched square wave signal is fed thru an LC low pass filter to recover the audio signal.
The inductors in question are the L of this filter. Most small class-T amps use a bobbin type coil but other types are used or available as upgrades. Which works best? And why?
The goal of the inductors should be to attenuate the ultrasonic switching frequencies, while having as little effect as possible on the audio signal.
How do the 3 types stack up?
Using the stock bobbin type coils as a zero reference, below we see relative levels of RF in radiated and transmitted output form.
Stock Bobbin: (reference)
Radiated 0dB
Output into 8 ohms 0dB
Aircore (Autocostruire)
Radiated + 11.5 dB
Output into 8 ohms -5.6dB
Toroidal (type 61 core)
Radiated +8 to -4dB
Output into 8 ohms -6dB
From the above we can see several things.
1) The aircore and toroidal allow 1/2 the RF sent to the load as the stock bobbin. That’s good.
2) The aircore radiates almost 4 times as much RF as the stock coil. That's bad. The toroidal radiates more and less than stock. That's because the radiation pattern is not symmetrical. More on that below.
What we can not see from the above is that the bobbin and aircore inductors tend to radiate evenly thru 360 deg. parallel to their length. The toroidal radiates the most parallel to its flat side and very little on the ends. A figure 8 pattern.
Also, as the aircore coils tend to radiate so much more than the others, the spaces between the coils can see combined signal levels at least twice as high as elsewhere. The other coils do have combing effects, but they tend to be only 30~40% above non combined. Thus the area between the 4 coils where the electrolytic caps lives is very hot in RF with the aircores.
Another thing not shown is the resulting waveform after filtration. I will try to post a photo later. The toroidal and aircore both give something near a sine wave, while the bobbin shows a strong triangle wave. The triangle wave is evidence that a lot of RF harmonics are left in the signal.
The measurements were done at 12V into an 8 ohm resistive load. Stock 150nF caps across the output terminals.
Radiation was measured with a sort length of wire attached to an oscilloscope probe. Wire was placed 3mm from the coils for measurements.
I have tried aircores on my TK2350 and it sounded great, specially in the highs but after 5 min i noticed that the sound got a bit screachy and i found that the heatsink was very hot. so to much RFI and oscilation on the mosfets.
i have to admit it did have a very good sound for that 5 min
cheers
Rudi
i have to admit it did have a very good sound for that 5 min
cheers
Rudi
MrDodo said:
Good to be back! Been busy building amps.
I was afraid you might ask that.
How they sound can be a lot harder to tell than just measuring them, as it is mostly subjective. A good 3 way blind test is in order.
What I hear in my first tests is that the aircore and toroids sound "smoother" than the stock coils, but that is mostly at high power. I think that others have found the same thing. The difference is subtle, but real. I don't know why there is a difference, it can't be that we are hearing those ultrasonic artifacts, can it?
I will continue to listen and there are a few more tests to do that might reveal why they sound different.
FWIW, I have found the bobin coils on the Fenice board to be nice and smooth sounding. They are made by Panasonic, don't know who males the stock coils on the Sonic board.
My 50 cents opinion is that the coil is not really the problem. Take a look at this very realistic estimate:
Stock Coil THD: 0.001%
Air Coil THD: 0.0005%
Dead Time / Switching THD (inside the chip): 0.05% (50-100 times more).
So can you hear difference btween the coils?
Yes! But only as a result of different series impedance, and difference in inductor properties vs. the load of the speaker 'du jour'. To make an absolute judgement of this is not realistic in my opinion.
If you want to design a really good coil for this amplifier, i can recommend RM10 size ferrite with 1 mm air gap, and 3F3 material. On this you wind 99.9% pure silver wires with Teflon flex insulation. All should cost only a few $ a head.
This will give you a low impedance, low THD coil with very good HF properties, and also a low emission. Depending on the inductance, you will get a saturation current of 12 - 25 Ampere. (in the realistic inductance range for Class D amplifiers).
Stock Coil THD: 0.001%
Air Coil THD: 0.0005%
Dead Time / Switching THD (inside the chip): 0.05% (50-100 times more).
So can you hear difference btween the coils?
Yes! But only as a result of different series impedance, and difference in inductor properties vs. the load of the speaker 'du jour'. To make an absolute judgement of this is not realistic in my opinion.
If you want to design a really good coil for this amplifier, i can recommend RM10 size ferrite with 1 mm air gap, and 3F3 material. On this you wind 99.9% pure silver wires with Teflon flex insulation. All should cost only a few $ a head.
This will give you a low impedance, low THD coil with very good HF properties, and also a low emission. Depending on the inductance, you will get a saturation current of 12 - 25 Ampere. (in the realistic inductance range for Class D amplifiers).
Lars Clausen said:
An interesting choice. An aircap coil allows more current before saturation. But it radiates much more than a non-gapped core - an RM10 is said to need 1 inch spacing between coils. More turns are needed too. So the silver wire would be a good idea.
I don't know type 3F3 core material. Can you post specs? Type 61 works well for me - hi-Q in the frequency range we need, and only 11 turns for 10uH! (on the size core I use).
Always looking for better stuff.
http://www.ferroxcube.com/prod/assets/3f3.pdf
3F3 are close equivalent for N49, one famous class D design use PL-9 Samwha ferrite. I used N87 gaped RM7, pretty nice also up to 8A.
3F3 are close equivalent for N49, one famous class D design use PL-9 Samwha ferrite. I used N87 gaped RM7, pretty nice also up to 8A.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.