Class D amplifier using SiC Mosfets

[Bigun]

The stuff I do is for personal use too. I don't feel good if it does not outperform commercial. But then it can become commercial.
I jumped from class AB to class D because class H/G uses too many parts.

I am not afraid of a few extra parts - my question is, does class AB sound better to you than class D ?

 

[Viennatom_new]

@ionutgaga: Where can I find the schematic?

 

[Eva]

I am not afraid of a few extra parts - my question is, does class AB sound better to you than class D ?

What class AB? I have serviced class AB from 1980s not showing any crossover distortion at 20khz in oscilloscope, no bias pot, no excess heat. I have serviced class AB from 2000s requiring circuit modifications to eliminate crossover distortion already visible at 1khz.

What class D? I have repaired car amplifiers producing 1% THD at normal volume. I have developed "king of poor" or more likely "daughter of rich and poor" modulator which does 0.005% THD up to 5Khz, 50W.

General criteria is "beware of toys and junk".

But please, lets stop bugging ionutgaga thread.

 

[Bigun]

I apologize of the OT in ionutgaga thread. In compensation I would like to help but the technology is unfamiliar to me. I do like SiC, it has some elegant abilities that I once looked at for making UV sensitive CCD image sensors on account of the large bandgap. I also like them for rectifier diodes in traditional high voltage valve amplifiers. I can imagine that it could be an interesting challenge to impose their use in Class D and then see how to get the best out of them.

 

[NewTesla]

Please note:
- I wish I could have so many resources to waste.
- I'm running into difficulties finding a brand interested in technology far better than yours.
- The system got so corrupt that technology no longer matters for many firms. Company hierarchies became sodomite in nature.
- There is an economic crack like 1929's coming soon. Imitators getting higher priority in funding system is the cause.
- I'm starting to consider finding a job in a different sector meanwhile.

I agree 100% with you. Same experience here from the last 8 years. Me, just another one leaving the ship.

 

[ionutgaga]

This topic was created for opinions and tips, but it seems to have failed. I do not want to become a place of the complaint, or a place where everyone will be proud of their own achievements, but they will not show them. If you have anything to say about the subject, please, otherwise you must leave the ship.

 

[NMOS]

All fixed-frequency modulation brands are imitating each other. Some results are really poor and dirty.

It's not about technology. It's not about specs. It's not about waveforms. These companies won't hire you even if you have better technology at lower cost. They want to keep their structure, with shy engineers cleaning the shoes of selfish executives.

You won't be able to compete in price with something cloned by several dozen factories in China. Innovation was the remaining solution, but innovation requires a company grown by innovation, not by engineers cleaning shoes of executives. Everyone shall clean their own shoes.

its oldfrom 2018 but I can confirm 100 % this

 

[U130421]

I got 2.2mhz switching 84 volts with IRFB4212 mosfets. the high side drivers get hot at that frequency but the output at 20khz at 1v is equivalent to class A. OPA2134 provides the correction signal.
overload capability decreases with frequency.
only reason i am not using class D is the amp goes into a sx-3700 and causes low level noise on FM broadcasts as it is 5" from the tuner.

How much dead time you applied ?

 

[kASD]

Ionutgaga is eating tomatoes, hence cannot reply right now.

 

[kipman725]

To inject some optimism I have done things with SiC that would be simply impossible with Si. For example 100kHz, 100kW 1000V DC bus full bridges. They shine in higher power applications where Si MOSFETS with >600V rating tend to have high Rdson and charge. For classD this would 20kW+ per channel amplification. There is some demand for such amplifiers as even the k20 is not powerful enough to fully utilise some of the largest subs. However I am not in a position where I have access to the resources to develop such a beast.

*As for tips (from my usage of CREE/Wolfspeed devices)
1. Make sure you take the gate negative to fully turn devices off (-5V)
2. 18-20V gate voltage required
3. You may have to to add gate capacitance to stop Dv/Dt induced turn on
4. Be prepared to burn a lot of power in the gate drive

 

[Viennatom_new]

kipman, for ROHM, you can only have about -2 to -3V negative. -4V abs max rating! Agree about the 18V plus.

 

[kees52]

All fixed-frequency modulation brands are imitating each other. Some results are really poor and dirty.

It's not about technology. It's not about specs. It's not about waveforms. These companies won't hire you even if you have better technology at lower cost. They want to keep their structure, with shy engineers cleaning the shoes of selfish executives.

You won't be able to compete in price with something cloned by several dozen factories in China. Innovation was the remaining solution, but innovation requires a company grown by innovation, not by engineers cleaning shoes of executives. Everyone shall clean their own shoes.

It,s like politics Eva. switching fast is most limited by design, and pcb, normal Si mosfets in multilevel switcing are faster then a good designed pcb board.

 

[ionutgaga]

Ionutgaga is eating tomatoes, hence cannot reply right now.

...

 

[kees52]

...

Because I am just still busy with the autobias circlotron who go start when wether get bad I did some sims with class d multilevel.

Not easy, but i think because of very fast parts available class d gets much better, even without feedback i did get low distortions. Pics are with feedback, three and 8 level.

Nice design, do it work nice? have you youtube movie with sound>?

regards