Current state of the art Class D?

[Dave Zan]

...The plot in your latest post looks essentially minimum phase, just from eyeball estimate...

To clarify, it's the first plot that looks minimum phase (by eyeball estimate)
The phase in the second plot is the one that looks like a measurement error to me.
There seems no reason why it should flatten out then curve back up after 10 kHz.
It makes no sense as a true delay and looks inconsistent with the frequency response and minimum phase.

Yep. Me too.

That makes three of us.

It's been biblical since we last communicated, what with Pestilence after Fire and Toxic Air, how have you been?

Best wishes
David

 

[kartino]

I started the thread for technical interest so I am sorry too (probably more than you )
Thank you for the comments, the current state of the art is excellent so maybe there's not much more for people to add.

Best wishes
David

I have done some development for Pro Audio. There are still plenty room exploring the details. Typical 800W to 3000W, half bridge, fullbridge, UcD or double feedback. Fully Discrete or IR based. I learn class D from 2005, and I am extremely active do trials. Making big amps is very interesting for me. I do trial based on papers or topology revealed. It was very difficult to get detail circuitry. Thanks to LTSpice! Which make it happened. We can make working amp from scratch. You may check my work in youtube channel: kartino surodipo.

The two gentlement are PHD and they are extremely good. It will need long years from now for another good man. Just like Stanley from Crown. I follow both gentlement here in this forum.

 

[Kurt von Kubik]

Interesting thread.
However sorry to say that this thread does not reveal technical comparison rather than impression from hearing which is not common in class D section.

Both compared have different output rating. A small rating amp will get impression as thin and flat. While big amp will always thick and full. It is always like that.

I am a diyer. I have made many amps. Or hear many chip amps. I made both UcD and double feedback with IR based topology. Both are almost equal. You can make a very good sound quality. Low gain indeed can improve sound quality.

Today most designers have push class D capability to limit. Almost no room more for improvement. Basically with post filter feedback, it can push sonic quality to limit. In this 2020 year, there are many almost perfect components for switching speed. In class D, high speed is not the key, but high speed on an off to get minimum dead time is the key. A switching speed +/- 400kHz almost meet highest quality of class D. Higher speed will be risk of distortion from dead time.

So state of the art class D will be:
1. Post filter feedback/double.
2. Low gain
3. Switching speed +/- 400kHz

Hi Kartino

Technical comparison can be done from datasheets, but that reveals very little of the actual sound of the amps.
I´ve compared nCore 500 OEM to ICEPower 1200 AS2.
My choice would be the ICE amp. It simply has a more realistic sound, and I wasn´t anyway near clipping on any of those amps.
And class D mostly has the very usefull feature, that the last watt is even better than the first one.
Anyways I do not agree, that big amps outperform smaller amps.
The voltage in high power amps will soon enough cause problems.
IMO you have a sweet spot arround 200 Watts @ 8 Ohm for class D give or take.

I´m a bit pussled about the twin loop designs you mentioned.
Twin loop to me means ICEPower. They stick to their twin loop design.
One of ICEPowers very best engineers later on, designed a tripple loop design for TC electronics. This design took class D even further.
Putzeys designed the low gain types, which I do not fancy at all.
Back in time Harris had some twin loop amps, but they were as i recall not selfoscillating.

As mentioned earlier I´m a proud member of a small group, who are crazy enough to do designs of our own.
We´ve figured out some of the matters you´ve mentioned for modern class D.
One of the very special achievements is constant delay throughout the audioband, which is unusual for most amps.
This doesn´t mean, that there is no phaseshift, but it means that the phase is linear, which again leads to very convincing 3D if recorded.
On top of that the Zout is ludicruisly low even at 20KHz. In the output point of the PCB we are talking about µOhms.
The more classic measurements are just arround what is possible give or take a little. We can´t better Putzeys, but we have full gain (30 dB).

Fs is about 500 KHz, where some kind of sweet spot lies, we are using a higher order modulator, and a shrewed device to turn on and off of the output devices, if I may say so

 

[drjlo]

I really like how my IceEdge 1200AS2 amp sounds on normal speakers, but it cannot drive the interface box for Raal SR1a ribbon headphones. These 'phones represent a very low impedance 0.2 Ohm load, and the interface box uses a bunch of resistors to allow speaker amps to drive it, along with providing baffle step compensation.

My run-of-the-mill 125 wpc class AB Marsh amp can drive this box+Raal no problem, but the supposed 600+ wpc 1200AS2 (with the stock smps) cannot drive it beyond a whisper, going into clipping way before reasonable volume level.

I didn't get to try the Raal when I had NCore amps, so I don't know if this is a Class D problem of not being able to provide current or just a problem with ICE.

Any suggestions?

RaalInterface by drjlo2, on Flickr

Raalcombo by drjlo2, on Flickr

 

[cowanaudio]

It's been biblical since we last communicated, what with Pestilence after Fire and Toxic Air, how have you been?

G'day again Dave

It doesn't stop does it? Apart from a few small adjustments it's been business as usual. We've all worked right through, my wife is doing her bit from home. Feel free to organise a meetup at some point for a listen to some interesting speakers.

All the best

William Cowan

 

[pinkfloyd4ever]

I just saw this thread title from the homepage and didn't bother reading through the whole thread, so someone may have already mentioned this...but I just listened to the most recent Darko.Audio podcast on Spotify.

He was interviewing Bruno Putzeys who designed the Hypex modules. He just released a new Class D module of a new design under the name PuriFi. The modules are available for DIYers to purchase here:

EVAL1 - PURIFI

This may be old news to those who frequent this section of DIY Audio. DIYing Class D amps is something I've not explored...yet.

 

[kartino]

Hi Kartino

Technical comparison can be done from datasheets, but that reveals very little of the actual sound of the amps.
I´ve compared nCore 500 OEM to ICEPower 1200 AS2.
My choice would be the ICE amp. It simply has a more realistic sound, and I wasn´t anyway near clipping on any of those amps.
And class D mostly has the very usefull feature, that the last watt is even better than the first one.
Anyways I do not agree, that big amps outperform smaller amps.
The voltage in high power amps will soon enough cause problems.
IMO you have a sweet spot arround 200 Watts @ 8 Ohm for class D give or take.

Noted, sorry was missed the watts
I was misunderstanding that ICE is bigger.

I´m a bit pussled about the twin loop designs you mentioned.
Twin loop to me means ICEPower. They stick to their twin loop design.
One of ICEPowers very best engineers later on, designed a tripple loop design for TC electronics. This design took class D even further.
Putzeys designed the low gain types, which I do not fancy at all.
Back in time Harris had some twin loop amps, but they were as i recall not selfoscillating.

If you remember there very long long time ago, discussion on how the "shape" signal comparator is made. So that all class D work same as very old version. Only feedback modified and of course better parts contribute a lot.

So that there was people comparing the shape signal of UcD comparator. Said that it is not triangle and possible made unlinear comparator.

The other side by double feedback, it contain prefilter feedback, made a very old and proven simple method in providing a sharp triangle feeding to comparator. Second feedback post filter will give pure sinus feedback, better correction and high dumping. This is just from my trial. Again I am not PHD. As from measuring did not reveal any big different, so we did make a hear test where dynamic characteristic can be demonstrated. And yes, double feedback has very good sound. But post filter feedback cannot be lower than prefilter, as when start it will trigger amp become "UcD" generate oscillation faster than prefilter feedback but in very low speed about 65kHz. This will damage LPF and speaker. So in double feedback, this is the challenge. It need a circuitry to delay connection of post filter feedback only after post filter feedback make oscillation. Maybe someone can help. I am thinking use small DPDT or with JFET. Currently I use dummy load 66 ohm 20W ohm to prevent above.

As mentioned earlier I´m a proud member of a small group, who are crazy enough to do designs of our own.
We´ve figured out some of the matters you´ve mentioned for modern class D.
One of the very special achievements is constant delay throughout the audioband, which is unusual for most amps.
This doesn´t mean, that there is no phaseshift, but it means that the phase is linear, which again leads to very convincing 3D if recorded.
On top of that the Zout is ludicruisly low even at 20KHz. In the output point of the PCB we are talking about µOhms.
The more classic measurements are just arround what is possible give or take a little. We can´t better Putzeys, but we have full gain (30 dB).

Fs is about 500 KHz, where some kind of sweet spot lies, we are using a higher order modulator, and a shrewed device to turn on and off of the output devices, if I may say so

Aha so I got the point that less phase shift is better circuitry, means better design. I was confused because phase shift is very small and not so related to sonic quality.
So that, I believe shorter circuitry and less internal cob or any capacitive, low impedance will be the key. Probably that is why ICE ise its own chip. And, avoid bigger LPF with higher switching speed also the key. Yes for small amp, 200W +/- a 500kHz is ideal, meet good parts today. But I don's agree with 1MHz design. My opinion, it is less benefit as 500kHz is very good enough and to maintain distortion of switching dead time.

 

[kartino]

I really like how my IceEdge 1200AS2 amp sounds on normal speakers, but it cannot drive the interface box for Raal SR1a ribbon headphones.

I believe the ribbon is too capacitive and change way of the LPF of the amp work.
I suggest you add dummy load resistor parallel to the speaker. 8-16 ohm equal watt to the speaker. And yes it will consume the power output also, but I believe you can hear sound expected from the amp.

 

[kartino]

I just saw this thread title from the homepage and didn't bother reading through the whole thread, so someone may have already mentioned this...but I just listened to the most recent Darko.Audio podcast on Spotify.

He was interviewing Bruno Putzeys who designed the Hypex modules. He just released a new Class D module of a new design under the name PuriFi. The modules are available for DIYers to purchase here:

EVAL1 - PURIFI

This may be old news to those who frequent this section of DIY Audio. DIYing Class D amps is something I've not explored...yet.

Indeed it is old news
And we are talking about it now

 

[kartino]

Greetings from third world style

Finished UcD Xlite Fullbridge, optimized for fullrange - YouTube

Test UcD Pleci 2x100W 8 ohm - YouTube

Finished UcD Xlite Fullbridge, optimized for fullrange - YouTube

Some my design made by Wahyu Eko Romadhon in his channel:

D2K Fullbridge OUT 98VAC/4OHM Class D Amplifier - YouTube

 

[KaffiMann]

This may be old news to those who frequent this section of DIY Audio. DIYing Class D amps is something I've not explored...yet.

As it seems right now, if you want to reach a certain level of performance, the DIY of class D is mostly limited to ready built modules. So not overly difficult.
Top contenders are IMO (in no particular order): Icepower, Pascal, Purifi.
These three have different market approaches (like how Pascal goes after the PA segment while Purifi seems to be chasing "High End" Audiophile), but performance is more or less compareable between the three.

I am still a little bit one the fence about Hypex, no doubt they make great products, but are they in the same tier as of today?

 

[kartino]

Putzeys designed the low gain types, which I do not fancy at all.

And for this I disagree
The UcD design is good already as it is, with standard gain. What Bruno did with NCore, now Purify is only try to push at very small room for improvement.

Regards,
kartino

 

[drMordor]

So state of the art class D will be:
1. Post filter feedback/double.
2. Low gain
3. Switching speed +/- 400kHz

You forgot the GaN transistors, that will allow the amplifier to operate over 1 Mhz

 

[kartino]

Yes however it will less benefit IMO.

 

[drjlo]

I believe the ribbon is too capacitive and change way of the LPF of the amp work.
I suggest you add dummy load resistor parallel to the speaker. 8-16 ohm equal watt to the speaker. And yes it will consume the power output also, but I believe you can hear sound expected from the amp.

I don't know if you saw the photos I posted, but these are ribbon headphones and already comes with interface box with resistors inside to simulate an 8-Ohm load.

Problem seems to be IceEdge being unable to deliver enough current through the resistor network to the 0.2 Ohm ribbon headphones when far less wattage class AB amp can do it no problem.

 

[kartino]

Just finished designing my another new amp design born.
Simple solution, 2nd feedback post filter taken after relay contact, sensing PWM from core.
This is for PA Amp, as most of my design are for Pro Audio.

 

[kartino]

I don't know if you saw the photos I posted, but these are ribbon headphones and already comes with interface box with resistors inside to simulate an 8-Ohm load.

Problem seems to be IceEdge being unable to deliver enough current through the resistor network to the 0.2 Ohm ribbon headphones when far less wattage class AB amp can do it no problem.

You can simply check with Ohm meter how much resistivity value.

 

[xrk971]

I agree +1. My TPA3255 with PFFB and state of the art layout and nice inductors makes for a super amp. Cannot tell it is turned on with ears pressed to the speaker cone. Very quiet. Here is a recent review of if by Redjr:

Class D Amp Photo Gallery

THD measured is 2x less than TI’s own factory PFFB version for same reference condition at 10w output.

Noise floor is below -125dB:

Why equate the cost of the chip to the overall performance? The silicon used in the Purifi amps won't cost much different to 4-8 dollars either.

At low power levels and with post filter feedback the TPA amps are as good as the others. Better than some and worse than others but certainly within the same league.

 

[kartino]

XRK,

It is even better circuitry. A small class BD, fully symmetric bridge, and post filter.

 

[Kurt von Kubik]

I don't know if you saw the photos I posted, but these are ribbon headphones and already comes with interface box with resistors inside to simulate an 8-Ohm load.

Problem seems to be IceEdge being unable to deliver enough current through the resistor network to the 0.2 Ohm ribbon headphones when far less wattage class AB amp can do it no problem.

The ICEPower 1200 AS2 can deliver 30A, of your headphones need even more than that, i think they might be broken.
The ICE modules all have protection circuits of any kind, så something must go wrong in the HP set-up.