Some people say that use chip reduce component i.e. no level shifter, comparator, and so on so it is faster.
That is totally wrong!
Class D with N-Fets always has similar part: Comparator, Level Shifter, Inverting, bootstrap, Low and High Driver and finally the mosfet. In chip it is packed. You cannot tweak. With discrete, there are many circuit possibility designer can take, from as simple as possible untuk complicated, cascode and so on.
The key is how fast which sometime new parts are faster.
And the faster is the key.
That is totally wrong!
Class D with N-Fets always has similar part: Comparator, Level Shifter, Inverting, bootstrap, Low and High Driver and finally the mosfet. In chip it is packed. You cannot tweak. With discrete, there are many circuit possibility designer can take, from as simple as possible untuk complicated, cascode and so on.
The key is how fast which sometime new parts are faster.
And the faster is the key.
You're not quite right)))
The delay in the comparator and driver stages has a very strong effect. The lower they are- Higher loop gain- the better the sound. Even with silicon transistors.
Last edited:
No No No))))
Use faster)))
LT1713 ultrafast comparator - 7ns
LMG1210 fast gate drive - 20ns
Discrete UcD - 120ns min.
Use faster)))
LT1713 ultrafast comparator - 7ns
LMG1210 fast gate drive - 20ns
Discrete UcD - 120ns min.
Selling price is also concern.
And profit as the most. The designer said fast switching is not the only. Class D no need at 1MHz level. 400kHz fsw is just right. So that overall delay can be covered using ordinary parts.
Sure it is good if you make this only several unit for your self. Or your customer is so premium.
But selling OEM at 100USD per pc is not the case.
And profit as the most. The designer said fast switching is not the only. Class D no need at 1MHz level. 400kHz fsw is just right. So that overall delay can be covered using ordinary parts.
Sure it is good if you make this only several unit for your self. Or your customer is so premium.
But selling OEM at 100USD per pc is not the case.
I know what you mean.
With fsw 400kHz still a bit room for parts delay and overall feedback loop delay.
As post filter feedback, the amplification feedback is done in sonic range.
With fsw 400kHz still a bit room for parts delay and overall feedback loop delay.
As post filter feedback, the amplification feedback is done in sonic range.
Last edited:
Yeah... I believe Bruno have done the best as he spent a lot of his life for UcD.
In fact, no one can do better right now.
In fact, no one can do better right now.
Son, you will go deaf at your parties with loud music!
Thank you mom, I've already had lunch))))
Thank you mom, I've already had lunch))))
Anyway I agree that it is time to move to GaN.
Specially for hi-fi range. If not, other people will take that opportunity to take as leader.
Specially for hi-fi range. If not, other people will take that opportunity to take as leader.
Mr. Kartino;
What is the switching frequency limit of current ucd high/low side gate driver with transistor.
If current ucd self osc. is enough for couple of Mhz what is the point of using gate drivers?
I do not understand why offer gate drivers for GAN.
What is the switching frequency limit of current ucd high/low side gate driver with transistor.
If current ucd self osc. is enough for couple of Mhz what is the point of using gate drivers?
I do not understand why offer gate drivers for GAN.
Last edited:
Switching at appx 400kHz typically is good enough for hifi.
For pro audio usually 250kHz for big amp, 275 to 300kHz for smaller. Some PA amplifier are designed as hi fi size, 200W to 400W which adopt similar circuitry as hifi.
GaN offer fast, less losses and so smaller footprint. Easy manage dead time
For pro audio usually 250kHz for big amp, 275 to 300kHz for smaller. Some PA amplifier are designed as hi fi size, 200W to 400W which adopt similar circuitry as hifi.
GaN offer fast, less losses and so smaller footprint. Easy manage dead time
Last edited:
What is the best class d ,i cant say,i have many different class d amplifier ,you imust have the right combination .
Correct me if i am wrong current ucd comparator delay is more than 100ns. We want much low dead times for GAN, are we have to sacrifice ucd or is there any way make ucd faster?
These are two different things. GaN allows for shorter dead times, which helps reduce distortion. You can still have a feedback loop which will then be more effective since it has less distortion to correct. Of course the fact that ucd comparator delay is long limits the bandwidth, but 100ns translates to, what? 10mhz? That would not be the biggest problem I think.