Selecting MOSFETs for Class D amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello,

I am currently building a class D amp of my own design.

The modulator is done and seems to work reasonably well using the following parts:

TL084 op-amp configured as a triangle generator
LM311 comparator
Discrete pre-amp

Switching frequency ~100kHz.

However, I need to select some MOSFETs for the output stage. The amp will use a single 12 volt supply and a full bridge output.

What MOSFET parameters should I be looking for to suit my design? I don't want to waste money so I want to choose my MOSFETs carefully before I buy them (but I also don't want to spend more than I have to, given the relatively low power levels of my amp), and hopefully leave some room for expansion (I might want to increase the switching frequency in the future).

Also, how can I drive the FETs from the logic level comparator output? Dedicated MOSFET driver? What about using a totem pole?

Is there a reason why no-one uses BJTs for this sort of task?

I'm interested in how experienced amp builders approach these challenges.

Thanks in advance.
 
Last edited:
Switching frequency ~100kHz.

-> Quite low, but theoretically still enough for full audio range. To have at least some decent ripple attenuation you might choose a 4th order lp filter, which more or less takes away the possibility to feedback after the filter - bad.

What MOSFET parameters should I be looking for to suit my design?

-> depends on output power / load impedance ....
most important is low Rdson, low gate charge and fast diode recovery. At Vbus = 12V, you can get away with 20 V mosfets!!

Also, how can I drive the FETs from the logic level comparator output? Dedicated MOSFET driver?

-> integrated mosfet hi and low side driver (costs usually about nothing, a buck or so)

What about using a totem pole?

-> also possible to build it in discrete part level but more of a hassle

Is there a reason why no-one uses BJTs for this sort of task?

-> more than one reason.
 
Thanks Tom.

I have selected the following FETS - going by the model numbers they seem to be intended as a matched pair, although there are a few differences in the numbers between the N and P channel devices - I'm not sure if this will be OK in practice?

P channel: IRF9Z24NPBF - 172mOhm RDSon - Gate charge: 19nC
Turn on time: 13 nS Rise time: 55nS Turn off time: 23nS Fall time: 37 nS

N channel: IRFZ24NPBF - 70mOhm RDSon - Gate charge: 20nC
Turn on time: 4.9 nS Rise time: 34nS Turn off time: 19nS Fall time: 27nS


It's the difference in speeds that really concern me - I assume I can compensate for this by adding dead time on the high side switches?
 
Last edited:
Yes, I realised this was a bit too low when I built the thing - I have gone up to 500kHz now.

I did use the FETs mentioned and some drivers by Microchip - don't have the part numbers offhand. Seem to be coping for now and not too expensive, may go all N-channel when cost or RDSon becomes an issue :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.