300w mos-fet amp

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
lets bulid Quasi's great quasi-complementary design.
that gives 350W(rms) to 4ohm. of course with lower rails (and maybe a few modifications) the power will be less, so if 250-300W is enough, you can use 60-70V idle rails.


on the last page of the thread you will find the attached schematic in "pdf".
there is pcb-drawing there, too.

it's recommended to read the whole thread!


(instead irfp450 you can use the cheaper irfp240 in the output)
Hi Rasvan

The amp suggested by edl and ezis666 has been running now for more than a month without any issues. It is very quite when idle and very clean under power.

Last week I took it to a friends house for it's first public audition.

We ran a NAD CD player and NAD receiver as the pre-amp into the power amp driving a set of IMF transmission line speakers (very nice speakers).

Everyone was impressed with the amp and another amp will be built by the host and his son as a project.

If you decide to build it, contact me if you want any advice.

amplifierguru said:
Hi quasi.

Does your amp have a turn on thump? Have you tried to DC couple it to eliminate the electro's from the signal path?


Well yes and no.

The amp boards themselves do when they are first turned on and the woofers do pop a bit.

The completed amp has speaker delay via the protection circuit and this comes in about 4 seconds after the amp is turned on so no thump is sent to the speaker.

You will note that I also a have a soft turn on cct for the powersupply and this lasts about 1.5 seconds. This means that the voltages for the amp modules rise slowly and a speaker delay is required to prevent any problems here.

Just wondering. I used to make PA amps similar to this topology in the 70's but with complementary feedback triples using MJ15003/4. There's probably a few still floating around there they were installed in a lot of clubs and discos.

They didn't us a relay output so I dethumped the turnon by using a long RC timeconstant feed to your T7 which simulated the slow charge-up of the primitive bootstrap which never thumped.

You don't mind having electro's in the signal path?

Have you tried cascoding the input stage/ FETs instead of the bipolars? dead easy with no circuit changes just drop in.
Hi Quasi,

Perhaps you would like to try the cascode as the current arrangement appears quite sensitive to supply line artefacts and there goes low level resolution - "depth" , "space" . Think of it as a simple divider with the Zout of the collector in the input stage. The active load takes even more, but if you raise the collector impedance significantly less supply hash gets in. And there's not significant gain prior to the injection point so feedback won't help it.

Evolution. Don't believe me. Simulate it with a voltage source superimposed on the supply. Better still a BC546, two R's and a small C to GND.
On the thumbnail. Forget the component values (std 1k resistors from quickie) Simple circuit SE resistive output from stage one, next stage gets fraction of supply hash defined by R5 divider with Zc of the first stage. But cascode raises that by orders of magnitude so the R5 fraction is much less , while signal levels remain the same. So stage 2 receives many times less supply hash from the improved divider. The practical circuit uses a current mirror load which effectively increases R5 assuring more supply hash but more gain as well.

You need to envisage the supply line with a sawtooth on it looking to get in and R5 /Zc divider is it . Then it's amplified by all the gain of the high gain Vas stage.

So the current mirror is good if you need more LF gain as the stage 2 miller C will kill it at high freq. But for CLEAN from unreg supplies the cascode AND/OR differential second stage.


  • sh.png
    3 KB · Views: 465
Hey Greg (A'guru)

Ah yes I see. My cascode was a zener locked one and I found this unsatisfactory from a noise point of view. So your idea brings up new (and interesting) discussion. Probably outside the scope of Raszvan's thread though.

Hey Rasvan, if you decide to build the amp you would be able to implement Greg's idea easily by cutting some tracks and adding / changing some components. But maybe you should just build it first and see what you think, then tweak later.

Yes Quasi,

Zener locked cascodes, in my experience, always give higher (THD+noise) than resistive divider and C bypass to ground.

A good test if you have a 2 channel (stereo) amplifier using a common unregulated supply is to drive the other channel with signal and look/listen to what's coming out of your undriven/shorted input channel. Crosstalk - but generally it's coming due to Power Supply intrusion. Go from there for improvements as, if it can be stimulated interchannel, it can probably occur intrachannel.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.