# Class A vs AB vs D amplifiers

It's not the tweeter I'm worried about. But rather if we had a tweeter that has flat frequency response up to 1MHz, it's my ears that I'm worried about, even if this is ultrasonic.
With so many hypotheticals, you should be worried about basically anything..

.. do you know how high the absorption of ultrasound is in air?

Last edited:
MAB
Measured tweeter current from Hypex NC400, with 5kHz signal frequency and carrier frequency residuals. HF current was 16mA peak. Tweeter heating?

View attachment 373082
You've shown plots like this before, often with much higher peak currents. I'll ask for about the tenth time: can you show exactly how that measurement is set up? My difficulty is that for a typical tweeter inductance (0.03mH) and carrier frequency (400kHz), the tweeter impedance should be something like 75-80 ohms. For a 400mV carrier residual, the current will be 5mA without accounting for the impedance of wiring and cables (which will have both shunt capacitance and series inductance).

If we accept 10mA RMS as the actual number and a voice coil DCR of 5R, the heating from the carrier would be half a milliwatt. That's an awfully small windmill to tilt.

Last edited:
Don't think so. The audio current is much higher than the RF residudal. Therefore the audio current would make the heat.
It is absolutely clear, my measurement is for @John_Siau who was worried about thermal heating as a result of class D switching frequency residuals. I hoped it was clear.

You've shown plots like this before, often with much higher peak currents. I'll ask for about the tenth time: can you show exactly how that measurement is set up? My difficulty is that for a typical tweeter inductance (0.03mH) and carrier frequency (400kHz), the tweeter impedance should be something like 75-80 ohms. For a 400mV carrier residual, the current will be 5mA without accounting for the impedance of wiring and cables (which will have both shunt capacitance and series inductance).
I do not think the current shown was higher in the past. In fact this last plot is from 2018. They are consistent. Always take into account the amp used in the test. It is measured as a voltage drop across 1 ohm shunt in series with the tweeter. BW of the sensor is in MHz. 1V ~ 1A then directly.

I do not think the current shown was higher in the past. In fact this last plot is from 2018. They are consistent. Always take into account the amp used in the test. It is measured as a voltage drop across 1 ohm shunt in series with the tweeter. BW of the sensor is in MHz. 1V ~ 1A then directly.
Could you give some details of the test setup? Resistor placement, wiring, physical layout...? This is a very easy measurement to get incorrect numbers.

MAB
With so many hypotheticals, you should be worried about basically anything..

.. do you know how high the absorption of ultrasound is in air?
Now you made me curious, how high is the absorption of ultrasounds in air? Is there an upper frequency limit for audio?

Now you made me curious, how high is the absorption of ultrasounds in air? Is there an upper frequency limit for audio?
The answer to your question is no. There is only an AUDIBLE upper frequency limit, which I believe is about 22Khz. This applies to humans, of course. As an aside, there is a rumor that the Russians have developed an ultrasonic weapon -- totally inaudible -- that they have deployed against US personnel in Guantanamo Bay, Cuba. It is said to cause piercing headaches and dis - orientation.

Now you made me curious, how high is the absorption of ultrasounds in air? Is there an upper frequency limit for audio?
At the 3e Audio amplifier switching frequency of 600 kHz, 20 °C and 50% relative humidity, the attenuation is 59 dB/m.

Czy KIEDYKOLWIEK został udowodnione, czy jest tylko hipotezą? Czy zmierzyłeś, ile nadajników 400 kHz NAPRAWDĘ dociera do cewki drgającej poprzez przewodnictwo systemu? Zrobiłem zadanie domowe.

Translation by Moderator: Please use English going forward. Thank you. (Bitte sprich Englisch)

“Here is my measurement without comment.
1. Configuration DC48V/200W f=211Hz, 71W at 3.9 ohms
2. 511 kHz 14 mW at 3.9 ohms / with tweeter the impedance will be about 3.2 kOhm, so a very low current flows 18 uW, = 0.000018W 3. Input voltage on TPA3255”

#### Attachments

• 1 Aufbau.jpeg
259.6 KB · Views: 27
• 511kHz.jpeg
172.3 KB · Views: 30
• U ein am TPA3255.jpeg
160.7 KB · Views: 32
Last edited:
The answer to your question is no. There is only an AUDIBLE upper frequency limit, which I believe is about 22Khz. This applies to humans, of course. As an aside, there is a rumor that the Russians have developed an ultrasonic weapon -- totally inaudible -- that they have deployed against US personnel in Guantanamo Bay, Cuba. It is said to cause piercing headaches and dis - orientation.
Yeah the upper human limit I know of But I was thinking if there is any limit to how short a wavelength can get in air because of the size of the air molecules or whatever.
At the 3e Audio amplifier switching frequency of 600 kHz, 20 °C and 50% relative humidity, the attenuation is 59 dB/m.
View attachment 373148
Ah interesting, thanks for the link!

Class D and electric cars suck!

Here is my measurement without comment. TPA3116
1. Setup DC24V/100W f=211Hz, 31W at 3.9Ohm
This small oscilloscope is equipped with a poor probe, but it is sufficient for rough measurements and troubleshooting of audio.

#### Attachments

• TPA3116 Brum.jpeg
205.4 KB · Views: 25
One needs to measure speaker current, not the load voltage.

One needs to measure speaker current, not the load voltage.
oo thanks for the tip, can you explain to me why I should measure the current instead of voltage?

oo thanks for the tip, can you explain to me why I should measure the current instead of voltage?
Because voltage only tells you part of the story. You need the impedance and voltage to calculate power. Here is a typical tweeter. The resonance at 500Hz is mechanical. The rising impedance as frequency increases is the inductance of the voice coil, which has impedance give by:
Impedance = (DCR + 2πfL)^(0.5)
DCR = 4.6Ω
L = 0.05mH
f = frequency (Hz)

The left graph is the spec sheet parameters, with the rising impedance due to voice coil inductance (0.05mH). The right hand graph is what the impedance would look like of the tweeter if the measurement was extended to 1MHz, and there were no additional sources of parasitic inductance. Reality will be even higher impedance than my simulation.

The tweeter is made by Seas.

The claim from Benchmark engineering:
"Class-D out-of-band noise and distortion (above 20 kHz) may interact with the non-linearities of tweeters. This can produce intermodulation distortion (IMD) which can produce noise and distortion at audible frequencies (well below 20 kHz).

Ultrasonic noise and distortion can also create voice coil heating and this can reduce the sensitivity of the tweeter. In severe cases, this could damage a tweeter."

The claim is predicated on multiple layers of maybe. Indeed it may happen if someone invents a magical tweeter with negligible inductance. I think Benchmark knows this very well, the statement doesn't even need measurements to debunk, it fails freshman physics as understood since the 19th century.

edit: typo
edit: corrected formulae, looks like I fail frosh physics too...

Last edited:
What do u think of this comment?

Unlike class-D amplifiers, the AHB2 is noise free over its entire 0.1 Hz to 200 kHz frequency range.
What matters to me is: Can I hear any noise from the amp?

Don't settle for the limited performance of class-D amplification.
AHB2 is an impressive product deserving of our respect but this kind of marketing nonsense is beneath Benchmark, assuming that's where it originates. The modern class D amps are also impressive products deserving of our respect.

If the noise is inaudible then the amp is noise free for my practical purposes. Measuring the differences of degrees of noise freedom may be a way to pass the time for some but it's all the same to me.

pma
Because voltage only tells you part of the story. You need the impedance and voltage to calculate power. Here is a typical tweeter. The resonance at 500Hz is mechanical. The rising impedance as frequency increases is the inductance of the voice coil, which has impedance give by:
Impedance = DCR + 2πfL
DCR = 4.6Ω
L = 0.05mH
f = frequency (Hz)

View attachment 373195
The left graph is the spec sheet parameters, with the rising impedance due to voice coil inductance (0.05mH). The right hand graph is what the impedance would look like of the tweeter if the measurement was extended to 1MHz, and there were no additional sources of parasitic inductance. Reality will be even higher impedance than my simulation.

The tweeter is made by Seas.

The claim from Benchmark engineering:
"Class-D out-of-band noise and distortion (above 20 kHz) may interact with the non-linearities of tweeters. This can produce intermodulation distortion (IMD) which can produce noise and distortion at audible frequencies (well below 20 kHz).

Ultrasonic noise and distortion can also create voice coil heating and this can reduce the sensitivity of the tweeter. In severe cases, this could damage a tweeter."

The claim is predicated on multiple layers of maybe. Indeed it may happen if someone invents a magical tweeter with negligible inductance. I think Benchmark knows this very well, the statement doesn't even need measurements to debunk, it fails freshman physics as understood since the 19th century.

edit: typo
I recorded in position No. 70 the impedance was 500kHz and 3.2kOhm high.

Because voltage only tells you part of the story. You need the impedance and voltage to calculate power. Here is a typical tweeter. The resonance at 500Hz is mechanical. The rising impedance as frequency increases is the inductance of the voice coil, which has impedance give by:
Impedance = DCR + 2πfL
DCR = 4.6Ω
L = 0.05mH
f = frequency (Hz)

View attachment 373195
The left graph is the spec sheet parameters, with the rising impedance due to voice coil inductance (0.05mH). The right hand graph is what the impedance would look like of the tweeter if the measurement was extended to 1MHz, and there were no additional sources of parasitic inductance. Reality will be even higher impedance than my simulation.

The tweeter is made by Seas.

The claim from Benchmark engineering:
"Class-D out-of-band noise and distortion (above 20 kHz) may interact with the non-linearities of tweeters. This can produce intermodulation distortion (IMD) which can produce noise and distortion at audible frequencies (well below 20 kHz).

Ultrasonic noise and distortion can also create voice coil heating and this can reduce the sensitivity of the tweeter. In severe cases, this could damage a tweeter."

The claim is predicated on multiple layers of maybe. Indeed it may happen if someone invents a magical tweeter with negligible inductance. I think Benchmark knows this very well, the statement doesn't even need measurements to debunk, it fails freshman physics as understood since the 19th century.

edit: typo
This is the impedance formula, not what you write

Last edited:
MAB
This is the impedance formula, not what you write
At those frequencies, the DCR can be neglected to determine the current. (X >> R). Then once current is determined, you can use R to calculate heating.

Replies
412
Views
101K
Replies
403
Views
88K
Replies
503
Views
93K
Replies
4
Views
660
Replies
123
Views
6K