When you put a 10 khz square wave into a class D amp , garbage comes out. Square wave tests have been the benchmark of good audio performance. That explains why many concerts bring on listening fatigue. That effect where the sound seems really good at first, but then you want leave after half an hour. Class D is cheap and light weight , that is the entire point of it. I have tested some good class AB designs to full power with 10 khz square wave with clean output. Of course, zobel networks must be disconnected for such testing. The double die lateral mosfet amp in this example is on 10 khz , scope at 20v/div. note the rise time. This could not be done with class D even a fever dream.
Attachments
What has a square wave to do with audio??
If you do what every amp tester does, low pass input filtering at say 30kHz when testing with square waves, class D is just as fine as class A or AB.
BTW on your scope pic, your amp is slew rate limiting on the fallng edge. It doesn't do this well, but it is unreasonable to require it anyway so no harm done.
Jan
If you do what every amp tester does, low pass input filtering at say 30kHz when testing with square waves, class D is just as fine as class A or AB.
BTW on your scope pic, your amp is slew rate limiting on the fallng edge. It doesn't do this well, but it is unreasonable to require it anyway so no harm done.
Jan
ah I love it when a class g discussion turns into a religious discussion about class D…’does it sound cold? Well it is called ICE’
Back to class G. I don’t think they got a bad rap but indeed better parts, class D and expense has relegated them to an ‘interesting variant’. I love my hitachi ha-5300, it is getting a full restore soon
Back to class G. I don’t think they got a bad rap but indeed better parts, class D and expense has relegated them to an ‘interesting variant’. I love my hitachi ha-5300, it is getting a full restore soon
They are an interesting variant because class G (H) are completely scaleable, within the limits of available loudspeaker drivers. Class D is still NOT scaleable. A 50 ns difference in propagation delay across a stack of 10 output transistors is a don’t care for class G. It will cause class D to go thermonuclear. The solutions are either to make it electrically small (has practical limits) or lower the switching frequency (impacts fidelity). The switching power supply has the same issues, but lowering the switching frequency below 100kHz is practical. The better ones just run lower frequency as the current demand goes up - and do that to stay in resonant mode. They are not cheap and not available on Aliexpress. You can buy a 3KVA toroid (or two 1.5kVA Anteks) and 22000uF/250V caps for less money. And the intermittent load (one second on five off, for hours) you can put on that without complaint is staggering.
The best super power amps for touring concert use are not class D, but class TD. They are infinitely scaleable, as long as you don’t count on them to put out full power at 20 KHZ (tracking turns off and they run AB and would overheat QUICKLY). For practical frequency/energy distribution they are fine. Tweeter drivers can always use a smaller class D amplifier, running off practical +/-80V supply. Three 200V TO-220’s in parallel (per leg of the bridge) is all that’s needed, and you CAN switch those on and off simultaneously enough, even at 400 kHZ.
The best super power amps for touring concert use are not class D, but class TD. They are infinitely scaleable, as long as you don’t count on them to put out full power at 20 KHZ (tracking turns off and they run AB and would overheat QUICKLY). For practical frequency/energy distribution they are fine. Tweeter drivers can always use a smaller class D amplifier, running off practical +/-80V supply. Three 200V TO-220’s in parallel (per leg of the bridge) is all that’s needed, and you CAN switch those on and off simultaneously enough, even at 400 kHZ.
Three TO-220s in parallel? Infineon has 650 volt GaN parts with 30-odd amps continuous rating / 60ish amps peak, < 0.1 ohm ON resistance, and 10 nanosecond switching time. And they come in P and N flavours:
https://www.mouser.com/datasheet/2/196/Infineon_IGT65R045D2_DataSheet_v01_00_EN-3538345.pdf
https://www.mouser.com/datasheet/2/196/Infineon_IGLT65R045D2_DataSheet_v01_00_EN-3518088.pdf
I'd bet many class-D amps will be GaNified within a few years, especially if Infineon come out with 250 volt parts.
https://www.mouser.com/datasheet/2/196/Infineon_IGT65R045D2_DataSheet_v01_00_EN-3538345.pdf
https://www.mouser.com/datasheet/2/196/Infineon_IGLT65R045D2_DataSheet_v01_00_EN-3518088.pdf
I'd bet many class-D amps will be GaNified within a few years, especially if Infineon come out with 250 volt parts.
I use 4 of these class H 700W @ 8 Ohm modules for my subwoofers, and I've tested them on a set of Magico Q7 too - no problem with audio quality at all, compared to Dan D'agostino mono blocks.
https://groundsound.com/hpa2k.php
Second time was a pair of B&W 800 - the ones with a pair of 10" woofers. These never seemed to sound right - bloated and too "warm". Then with these class H modules, there was suddenly a lot more control in comparison with the Lyngdorf power amp. The owner immediately bought a set and didn't look back.
These modules also seems to "eat" up difficult loads in passive filters and several parallel coupled drivers.
Class D comes closer and closer, but often include fan-noise and coil-whine, if you want high wattage and sensible prices.
Loving these class H modules - they just power anything you throw at them - at any level you wish 😍
https://groundsound.com/hpa2k.php
Second time was a pair of B&W 800 - the ones with a pair of 10" woofers. These never seemed to sound right - bloated and too "warm". Then with these class H modules, there was suddenly a lot more control in comparison with the Lyngdorf power amp. The owner immediately bought a set and didn't look back.
These modules also seems to "eat" up difficult loads in passive filters and several parallel coupled drivers.
Class D comes closer and closer, but often include fan-noise and coil-whine, if you want high wattage and sensible prices.
Loving these class H modules - they just power anything you throw at them - at any level you wish 😍
Last edited:
54 milliohms of ON resistance, heat sunk through the damn PCB? Thats how class D goes thermonuclear. Move the decimal point one or two places and you get somewhere. That puts you in 200 volt silicon territory. If they were to do a paradigm shift to 600 volt supplies and 16 (maybe 8) ohm subwoofers GaN would be appropriate. Operate in the sweet spot, get the CURRENT down. Probably still want a TO-220 or two, to pull out the 40 watts of heat. Even at 99% efficiency these power levels require heat sinking.
The second part's package has the heat slug on top (TOLT package) instead of the bottom so you can put a real heatsink on it. I thought the first part (TOLL) had the same arrangement, mea culpa.
And you're absolutely correct, dissipating much more than a watt on a PCB is a fine way to get an overcooked taco shell. Oh well, maybe Infineon will come to their senses.
And you're absolutely correct, dissipating much more than a watt on a PCB is a fine way to get an overcooked taco shell. Oh well, maybe Infineon will come to their senses.
They don’t expect the dissipation profile of a class D H-bridge running a 4 or 2 ohm load at 10kW peak, 1kW average. Directly driving speakers with those things efficiently would require higher impedance.
The GaN I was used to designing RF power amplifiers in just kept getting more efficient as load impedances at the drain went up. Unfortunately, the passive losses went up the higher the transform ratio, when nulling out more and more capacitance at the same time. So you basically hit a wall trying to scale those up, too.
Power conversion applications are more forgiving, as you CAN lower the switching frequency when you keep adding zeros to the power. Solar farms don’t run up at 400 kHZ, as the output frequency needs to be 60 Hz. Packages can be bigger (with more source inductance which is a death knell for class D audio), and they can be spread all over hell and gone.
The GaN I was used to designing RF power amplifiers in just kept getting more efficient as load impedances at the drain went up. Unfortunately, the passive losses went up the higher the transform ratio, when nulling out more and more capacitance at the same time. So you basically hit a wall trying to scale those up, too.
Power conversion applications are more forgiving, as you CAN lower the switching frequency when you keep adding zeros to the power. Solar farms don’t run up at 400 kHZ, as the output frequency needs to be 60 Hz. Packages can be bigger (with more source inductance which is a death knell for class D audio), and they can be spread all over hell and gone.
no they are not wrong they are entirey correct. ive heard the same complaints from everyone who has reviewed them, that they are not up to par with a good class A or AB amp and my ears tell me the same thing. they ARE NOT wrong. class D simply just is not as good plain and simple no way around it bud
problems with 2 ohm loads 100% depends on the class of operation per the time in history it was built lol. i have listened to class D and they generally SUCK. thats why i dont trust them yet because even the allegedly “good” class D stuff SUCKS to my ears. it just doesnt have the musicality a PROPER built amplifier has. class D has never been there and still isnt. all class D has a gimmicky sound
no class D beats a good class A amp. literally zero class D amps beat a GOOD class A amp lol
even steve guttenberg said a midfi class AB denon PMA amplifier sounded better in the bass and midbass than the new technics hypex module integrated amplifier and denon absolutely BLOWS
a thread about class G brought more class D d riders than had i made it a class D thread i swear
Well, what happens is class G was only considered a fad and not a serious long term solution. In either the hi-fi or professional camps. The 3-ppm purists dismiss it out of hand because of the commutation noise. It’s always worse than feedback theory predicts in high power applications because most of it ends up coming from electromagnetic induction and ground contamination. Switching currents on that level with an electrically large layout is simply going to radiate. Not enough to be a FCC violation, but enough to induce -80dBc into the amplifier’s INPUT. Hi-if users simply will not live with that, despite the fact that it is completely inaudible when cranking up Van Halen or Led Zep.
Class D is completely free of these artifacts (but can have others that IMO are worse). Making the power supplies too small can make them misbehave - very audibly. That’s not really class D’s fault - you do that with AB, G, or H and the same $*** happens. Pro audio users (the touring companies, installers) have abandoned class G entirely because of weight and power consumption. They aren’t going back to iron pigs. They just live with the increased cost of initial purchase and replacement because of the overall bottom line which includes operating costs.
So where does that leave class G? Mostly in the hands of amateurs. And a few nutcases bent on building 2000+ watt amplifiers, but who do not possess the skills, patience, or NRE dollars required to DIY the same in class D. Although the last such amplifier I made was actually only 150 watts. A home brew subwoofer plate amp for a 10” sub for my office…..
More big ones are on the horizon - but first I have to finish building this house. I don’t have the $500k that it would take to hire it all out, any more than the same money to buy a modern ~50kW PA.
Class D is completely free of these artifacts (but can have others that IMO are worse). Making the power supplies too small can make them misbehave - very audibly. That’s not really class D’s fault - you do that with AB, G, or H and the same $*** happens. Pro audio users (the touring companies, installers) have abandoned class G entirely because of weight and power consumption. They aren’t going back to iron pigs. They just live with the increased cost of initial purchase and replacement because of the overall bottom line which includes operating costs.
So where does that leave class G? Mostly in the hands of amateurs. And a few nutcases bent on building 2000+ watt amplifiers, but who do not possess the skills, patience, or NRE dollars required to DIY the same in class D. Although the last such amplifier I made was actually only 150 watts. A home brew subwoofer plate amp for a 10” sub for my office…..
More big ones are on the horizon - but first I have to finish building this house. I don’t have the $500k that it would take to hire it all out, any more than the same money to buy a modern ~50kW PA.
Please stop embarrassing yourself publicly with your extreme ignorance of basic facts about amplifiers.
Here is measurement of Lyngdorf MXA-8400 class D amplifier on 2-ohm load:
https://www.stereophile.com/content/lyngdorf-audio-mxa-8400-8-channel-power-amplifier-measurements
Over 500 W continuous on 2 ohms! From class D amplifier!
Another example: JMF HQS 7001 class D amplifier (from https://www.stereophile.com/content/jmf-hqs-7001-monoblock-power-amplifier-measurements):
Over 750 W continuous on 2 ohms! From class D amplifier!
Not everyone. Here is a review of Lyngdorf class D amplifier, including comparison to Benchmark AHB2 (spoiler: no difference!):
https://www.stereophile.com/content/lyngdorf-audio-mxa-8400-8-channel-power-amplifier
That’s just not that big an amplifier. The 200 watt per channel class is where class D is in a sweet spot. The voltages can be supported by 200V silicon, and you can force the required currents through a low inductance transistor package. When neither of those is true anymore you start having to make compromises.