SystemD_2kW, any interest for an open design?

Hi Weller,
the key words are:
-Control theory
-PID
-High order loop gain, I would call it an PI2D2 loop gain control

Whereof the B side of the U201 forms
- proportional gain defined by R218+R219 and R221
- early limited integrating gain defined by C208, R_Add201 and R221

The second PI you will find in the second gain stage around Q205 and Q206, where R223, R224, R233 and C217 are defining the second PI gain.

The two D portions of the loop gain control are placed in the feedback paths:
-C210&C211
-C_Add201
 
Many thanks for your answer.

I'm trying to redesign the PCB so that is fits on my demand.

The PCB should be SMD as far as it can be. Only some film capacitors in the audio path and components that should be its size because of the heat dissipation.

In the supply section there are some voltages generated. What was the choice of using LEDs (D406 and D407) and OpAmps (U401) in stead of simple 7805 and 7905, and zeners for +/-1.6V? Noise issues?
 
Yup, I had chosen the supplies with LEDs and Opamps for noise reasons.
Temperature drift and tolerances are of course worse than with 7805 etc., but also of lower importance.
Another option would be sallen key supplies as I am using it in the Gainboard of the LiteAmp.
Sallen key supplies also can be done using precision references and OPamps.
But I think this would be overkill.

Layout is half of the game in classD, especially when dealing high power.
Going for a new layout is a hot topic.
Performance
Reliabilty
EMI

You could set up one sample with the original PCB in order to have a reference for measurements and checking wave forms under all load conditions.
 
....oohps, somehow I seem to be a Dinosaur. Don't have a youtube account.

Waveforms:
They are unpleasantly spread througout the thread, but cover lots of different questions including, dynamic hysteretis control, or overcurrent shut down, or and gate drive etc...

Regarding speaker output waveforms, you will find very demanding artificial wave forms in posting #424.
And boring sinewaves including clipping in posting #431.
 
No surprises.

The first screen shot shows approx. 2kW into 4R.
The output voltage of 256Vpp translates to 91Vrms, which results in more than 2kW power.
Scale: 50V/grid (1:10 probe)

The second screen shot shows 2.3kW into 4R, close to clipping.
The output voltage of 272Vpp translates to 96Vrms, which results in 2.3kW power.
Drop of fs becomes visible.
Scale: 50V/grid (1:10 probe)

The third screen shows 2.6kW into 4R, with massive clipping.
Due to rail sagging the peak voltage is even slightly less than at 2.3kW.
Increase of power is related to trapezoidal wave shape instead of a undistorted sinusodial shape, which lifts the RMS voltage without high peak values.
Scale: 50V/grid (1:10 probe)

The measurements have been done manually with a sinusodial input signal.
The popular burst measurements with 300ms/700ms power/break would slightly lift the limits, but not much.

Hi, a small question, this amp is powered by +/- 82Vdc rails?
 
ChocoHolic,

What scopes did you use for the waveforms ? One is a USB scope. What is the one you used in your images of its screen ?

The USB scope is the cheapy USB2100.
The other one is the old Iwatsu TS-8500, 500MHz.
For building and getting it going almost any USB scope will be fine.
A fast scope is only necessary for the development and judging switching wave forms and taming HF-resonances.


Also,

Since the design is fully documented, can we buy such an amplifier on Ebay from Chinese sellers :D ?
Most likely it is unattractive for commercial sellers.
Assembling the PCB and mechanical set up is no fun for mass production. Design focus was clearly on DIY for few pieces.
 
Above +/-85V reliability will decrease.
At +/-100V multiple components will touch their absolute maximum ratings,
or already slightly exceed their absolute max ratings due to switching transients. Caps, MosFets, diodes, IR-chip...
+/-120V would be a reckless starting point in order to make a youtube video 'how I killed my amp'.
 
Disabled Account
Joined 2005
Hi Coco,

90Vdc supply should be fine to drive most common 18" sub.
I am not a scientist like you :) but my fullbridge UcD discrete and fullbridge from IR based design both work very well. That is with 200V mosfet. With 250V mosfet have been tested as well at.115Vdc without problem.

Sometimes people too afraid working in the limit
But really a lot.of people are waiting a monster from you. :)