A Lateral MOSFET Power Amplifier Design Version 1.1

Thank you all for your encouraging comments 🙂

i dont like the glitch on the rising/falling edges
It looks stable, but like ctrlx, I am concerned about the glitch
I'm just wondering if the glitch isn't the Mosfet gate capacitance in response to fast rise/fall times.

I have to say that all parts are totally new, except for three of them. The capacitor C901, 1uF/63 VDC of the DC Servo section, wasn't available at Mouser so I removed the one from the previous board, the board that has it's resistor smoking in the preceding test at 4 Ohms impedance. Also, with the very high price of the MOSFETs, I reused the two MOSFETs from the same board. I don't have new MOSFETs anymore. They all have been used at least once. Does this glitch could be a result of a bad MOSFETs that would have suffer from the test when the resistor make smoke? Maybe... As for the glitch, I already have started to build the second board. I will continue it and use it for more tests. Again I will have to reuse the two MOSFETs of the second preceding board but these didn't suffer from nothing yet.

When you box it up, you will have another set of challenges to keep it quiet, but this will be achievable I think with no problems.
I already have put the board into the box with quiet results! I cannot ear any hum, just a tiny white noise, hard to hear even in a quiet room with my ear glued to the speaker. And as for the music, I am impressed already! The Amplifier is connected to two used B&W V201 speakers, 4 Ohms, that I bought just for testing this amplifier.

You should be limiting the amplifier bandwidth with the front-end filter set to between 200 and 300 kHz.

1719743004057.png


The front-end cut-off filter here is 3.29 MHz - about 10x too high.
So I guess all the designs from the books we read need revisions! My first amplifier was a disaster cause the design proposed in the book wasn't working at all from the beginning. And this one has a wrong filter setting... 🙂

At least we learn while we read lol...

As an update, the schematic you pasted look's like one of my preceding schematic without the HBR R4 resistor. Here is the latest schematic attached below.

So if I wanted to keep the 220 Ohms low resistor at the input, what about 3,3 nF for C1 ? I already have ten new TDK 0,0033uF in stock, that would result to a 219,222 kHz cut off frequency.
 

Attachments

Excellent that you have these results on the complete amplifier 🙂

The glitch will not be because you have reused some parts.

Set your input bandwidth to between 200 and 400 kHz - 220 Ohms and 3.3nF is 220 kHz which is perfect. If you then redo some of your square wave tests, my bet is the glitch will be gone and I'd say you are in the ball park.
 
my bet is the glitch will be gone and I'd say you are in the ball park.
And we haaaave a WINNERRRR!
LOL

No glitch at all!

Every tests below 470nF are perfect square wave. Here is 8 Ohms with 100nF as an example:

307-8_Ohms-100nF.png


Then here are all the impedance values in parallel with the worst case capacitor, 2,2 uF:

010-Inf_Ohms-2.2uF.png
010-Inf_Ohms-2.2uF.png


110-100_Ohms-2.2uF.png
110-100_Ohms-2.2uF.png


210-30_Ohms-2.2uF.png
210-30_Ohms-2.2uF.png


310-8_Ohms-2.2uF.png
310-8_Ohms-2.2uF.png


410-4_Ohms-2.2uF.png
410-4_Ohms-2.2uF.png


510-2_Ohms-2.2uF.png
510-2_Ohms-2.2uF.png




3KHz is low, use a 10KHz signal 1Vpp with 8 ohm plus 100p to 2u2 in parallel,
I also did the tests at 10 kHz you suggested. Here they are, 8 Ohms in parallel with nothing, then with all capacitor values from my Jig:


600-8_Ohms.png
600-8_Ohms.png


601-8_Ohms-100pF.png
601-8_Ohms-100pF.png


602-8_Ohms-470pF.png
602-8_Ohms-470pF.png


603-8_Ohms-1nF.png
603-8_Ohms-1nF.png


604-8_Ohms-4.7nF.png
604-8_Ohms-4.7nF.png


605-8_Ohms-10nF.png
605-8_Ohms-10nF.png


606-8_Ohms-47nF.png
606-8_Ohms-47nF.png


607-8_Ohms-100nF.png
607-8_Ohms-100nF.png


608-8_Ohms-470nF.png
608-8_Ohms-470nF.png


609-8_Ohms-1uF.png
609-8_Ohms-1uF.png


610-8_Ohms-2.2uF.png
610-8_Ohms-2.2uF.png


Nest step is to put this baby into it's box, then revert the 220pF at the input of the other channel for a 3,3nF, and then listen to it and enjoy it!

Again, many many many thanks to all who supported me in this project. I cannot say how much I appreciated your help and comments. Special thanks to Bonsai! ;-)
 
next test to do is clipping performance at 10KHz.
Here are two captures. The first one is at 750 mVrms (1,061 Vpeak) at the input. The second one is at 800 mVrms (1,131 Vpeak) at the input. In the second test, I could start to smell the amplifier heating. The heat-sink was hot but I could easily let my hand on it without sensing exaggerated heat.

750mV-10kHz.png


800mV-10kHz.png



And as a curiosity, I compared my Rotel RB-1582 MKII with my DIY Amplifier...

With no connection at the input, in the headphone, the RB-1582 MKII has a tiny white noise. My DIY amplifier has a tiny hum over a tiny white noise, but the sum of both were lower then what I could hear on the RB-1582 MKII !

In the speakers with no connection at the input, my DIY Amplifier is totally quiet, I cannot hear anything with my ear close to both speakers. With the RCA cables connected, I could hear the same noise as in the headphone if I put my ears close to the speakers. But sitting in front of them, it is totally quiet.

These are with the same inbox setup cables as before. The two RCA input connectors are still on each side of the speakers terminal. The nest upgrade would be to have a new rear panel made with the two RCA connectors as close as possible one from the other. A futur project upgrade.

I also wanted to compare these 10 kHz square wave tests. My DIY Amplifier seem better than the RB-1582 MKII. Here are the results for the worse capacitors. I omitted

these vlaues as they were to similar : 8_Ohms-100pF.png, 8_Ohms-470pF.png, 8_Ohms-1nF.png, 8_Ohms-4.7nF.png

The left side is the Rotel RB-1582 MKII and the right side is my DIY Amplifier. In the 2,2uF test, I kept the Rotel RB-1582 MKII in the background (in Red) as a reference.

8_Ohms.png
8_Ohms.png


8_Ohms-10nF.png
8_Ohms-10nF.png


8_Ohms-47nF.png
8_Ohms-47nF.png


8_Ohms-100nF.png
8_Ohms-100nF.png


8_Ohms-470nF.png
8_Ohms-470nF.png


8_Ohms-1uF.png
8_Ohms-1uF.png


8_Ohms-2.2uF.png
8_Ohms-2.2uF.png



I am happy ! 🙂
 
The Rotel has a lot of overshoot even with very low C values. I think the Rotel amp doesn’t have an input BW limiting filter, or it is set very high. The controlled rise/fall times on your amp are IMV excellent and the behavior into capacitive loads exemplary. Keep in mind, unless you are driving ESLs, it is very unlikely you will see loads of 2.2 uF, but I always recommend testing this.
 
you need to drive the amp with more input, so that the sine top/bottom becomes flat
then take a picture and turn down input level

Good Morning,

Here is a screen capture with 1 kHz signal, no load, on the left, and 1 kHz signal, 8 Ohms load, on the right. I am not sure if I injected the same level at the input seeing that my full attention was on the capture button and my left hand was on the Input dial to be ready to lower it after the capture.

Capture d’écran du 2024-07-02 07-06-48.png


And here is a screen capture with 10 kHz, no load, on the left and 10 kHz, 8 Ohms load, on the right:

Capture d’écran du 2024-07-02 07-07-05.png
 
With no connection at the input, in the headphone, the RB-1582 MKII has a tiny white noise. My DIY amplifier has a tiny hum over a tiny white noise, but the sum of both were lower then what I could hear on the RB-1582 MKII !
Addendum :
These tests was before I tested the amplifier with a square wave. As soon as I started the square wave tests, I realized that I inverted my speaker connections at the board terminal on the left channel. Yep, I did that!

So I have redone the headphone tests with no input this morning and got better results than yesterday. The tiny hum was du to the invert connection. Now, I cannot hear anything on the right channel and a lower hum+white noise on the left than yesterday. And since my 20 years old hearing tests at the hospital, my left ear has always been more discriminating than my right.
 
at 1KHz you have nice clean clipping, but at 10KHz you have rail stick/latchup.
some amps also have an oscillation burst with the rail stick/latchup into/out of clipping but yours seems ok in that regard.
you can sometimes get rid of the rail stick/latchup with a "baker diode" across the vas, or just leave it as is.
 
you can sometimes get rid of the rail stick/latchup with a "baker diode" across the vas, or just leave it as is.
I had made 5 boards and got one extra for a total of 6. In a near future I intend to reused parts from one board of the preceding version 1.0 and built a new board 1.1 with them to experiment with the pull-down option on the DC Servo and made further tests. I also want to test the version 1.0 board with my new Jig to see it's specifications compared with the new version 1.1