2stageEF high performance class AB power amp / 200W8R / 400W4R

Jump to Latest
astx

astx

Member
Joined 2011
Paid Member
Off Topic: CNC control PC milled with mini cnc

Thimios: a small HIFI2000 case and front panel has been milled to carry the mini cnc OS running on a raspberry pi4 4GB. Inside:

  • 12V/60W power supply
  • passive cooled rpi4 + diy milled control board
  • 12V to 5V converter for rpi4
  • 12 V output for a lubricant pump.
  • 12 V output for a electro-pneumatic valve to control cooling lubricant
  • 230V switched output to control vacuum cleaner
BR, Toni
 

Attachments

  • IMG_20200614_175502.jpg
    IMG_20200614_175502.jpg
    419.9 KB · Views: 715
  • IMG_20200624_171236.jpg
    IMG_20200624_171236.jpg
    469.3 KB · Views: 695
astx

astx

Member
Joined 2011
Paid Member
SA2021 New Class H Prototype working

SA2021 New Class H Prototype working like a charm.

Fixed missing capacitor for low voltage rails. Added a 1000uF per low voltage rail input.

Now it's time to increase the high voltage up to 80V for power tests.
The 20k THD+N doesn't change significant during takeover to high voltage rail - the tests was done with +/- 20V low voltage rail and +/- 31V high voltage rail.
The THD+N 20kHz stays below 0.006% up to 18.8V RMS on 8R load.
The first watt 20k THD+N is 0.0019%
Bias is 240mA total (120mA per TTC/TTA pair).

At the scope you can see the output and below the gate voltage at p-channel mosfet IXTH48P20. Clipping behavior is like SA2014.
I'm very impressed about the good performance. The simulation predicted such good values, but if the real world results are the expected ones, it is always a good day for drinking beer - cheers
:cheers:

Well done DaveZan and Astx! :D
More test data will follow.

Have fun,
BR, Toni

P.S.: the attached input stage is the original SA2014 ips/vas pcb as you can see.
 

Attachments

  • sa2021_prototype_working.jpg
    sa2021_prototype_working.jpg
    764.7 KB · Views: 737
Last edited:
T

thimios

Member
Joined 2009
astx said:
Thimios: a small HIFI2000 case and front panel has been milled to carry the mini cnc OS running on a raspberry pi4 4GB. Inside:

  • 12V/60W power supply
  • passive cooled rpi4 + diy milled control board
  • 12V to 5V converter for rpi4
  • 12 V output for a lubricant pump.
  • 12 V output for a electro-pneumatic valve to control cooling lubricant
  • 230V switched output to control vacuum cleaner
BR, Toni
Click to expand...

astx said:
SA2021 New Class H Prototype working like a charm.

Fixed missing capacitor for low voltage rails. Added a 1000uF per low voltage rail input.

Now it's time to increase the high voltage up to 80V for power tests.
The 20k THD+N doesn't change significant during takeover to high voltage rail - the tests was done with +/- 20V low voltage rail and +/- 31V high voltage rail.
The THD+N 20kHz stays below 0.006% up to 18.8V RMS on 8R load.
The first watt 20k THD+N is 0.0019%
Bias is 240mA total (120mA per TTC/TTA pair).

At the scope you can see the output and below the gate voltage at p-channel mosfet IXTH48P20. Clipping behavior is like SA2014.
I'm very impressed about the good performance. The simulation predicted such good values, but if the real world results are the expected ones, it is always a good day for drinking beer - cheers
:cheers:

Well done DaveZan and Astx! :D
More test data will follow.

Have fun,
BR, Toni

P.S.: the attached input stage is the original SA2014 ips/vas pcb as you can see.
Click to expand...
Excellent work as usual!
That's why i wish to have a mini cnc!;)
Nice work to your new amplifier too!:cool:
A very nice laboratory!!!
Congratulations Toni!:cheers:
I have SA2014 ips/vas pcb.May be it is time try to populate those.
What is the Pi4 dac?
 
astx

astx

Member
Joined 2011
Paid Member
Last edited:
astx

astx

Member
Joined 2011
Paid Member
The pro using this Class H rail tracker is:


  • Very low THD increase during the smooth transition from the low to high voltage rail because the inner power transistors have mostly constant Vce up to HV rail.
  • The power dissipation happens mostly on the outer power transistors which in this case are high power V-MOSFETs. 2 of them can e.g. dissipate a maximum of 1000W!
  • The inner transistors can have high bias to get the first watts in Class A without dissipating too much power.
  • The boot strapping of the MOSFET gate voltage allows to use the full HV rail voltage. Clipping starts at < 3V difference to HV voltage.
BR, Toni
 
astx

astx

Member
Joined 2011
Paid Member
Theoretically yes. But some resistors need to be recalculated.
Of course using higher LowerRailVoltage (LRV) the idle power dissipation will increase without positive effects. The final Amplifier will be equipped with 4 TTC/TTA pairs and hence the total idle bias is about 500 - 600mA. Setting the LRV higher as 20V is IMHO unnecessary waste of energy.


BR, Toni
 
astx

astx

Member
Joined 2011
Paid Member
SA2021 New Class H Prototype - Measurements

... now we have +/- 62V HRV and +/-20V LRV ...

Attached simulation using ngspice vs real word measurements.

Code: 
Fourier analysis for output:
  No. Harmonics: 10, THD: 0.00240604 %, Gridsize: 500, Interpolation Degree: 1

Harmonic Frequency Magnitude Phase     Norm. Mag Norm. Phase
-------- --------- --------- -----     --------- -----------
 0       0         -0.0136   0         0         0        
 1       2e+04     56.33     -2.56     1         0        
 2       4e+04     0.000825  134.5     1.464e-05 137.1    
 3       6e+04     0.0003591 -173      6.374e-06 -170     
 4       8e+04     0.0004111 13.43     7.298e-06 16       
 5       1e+05     0.0004215 171.9     7.483e-06 174.5    
 6       1.2e+05   0.0003035 -20.6     5.388e-06 -18      
 7       1.4e+05   0.0004832 127.1     8.577e-06 129.6    
 8       1.6e+05   0.0003059 -62.6     5.43e-06  -60      
 9       1.8e+05   0.0005116 95.96     9.081e-06 98.52    

LRail      power : 5.67535 W
HRail      power : 328.838 W
Total      power : 334.513 W
Output     power : 198.315  W
Power dissipation: 136.197 W
#---------------------------------------------------------------------
# loopgain results for plot ac3
#---------------------------------------------------------------------
GM : -21.9729
PM : -106.288
3db : 2.27938E+07
0db : 2.71911E+06
  dc gain : 116.629
The scope pictures Gate voltage is not exactly aligned on the y axis!
The shown waveform is exactly as the simulated one.
Measured 20k THD+N bw 80kHz is about 0.0019% using 8R load
Output noise:

  • full bw 132uV
  • 80k bw 60uV
  • 20k bw 33uV
  • A-w: 20uV
SNR vs 200W output power


  • 11 0dB full bw
  • 116 dB 80k bw
  • 122 dB 20k bw
  • 125 dB A-w
SNR vs 1W output power

  • 86 db ful bw
  • 94 dB 80k bw
  • 98 dB 20k bw
  • 102 db A-w
BR, Toni
 

Attachments

  • sa2021_gate_voltages_vs_output_39V_rms_measured.jpg
    sa2021_gate_voltages_vs_output_39V_rms_measured.jpg
    226.8 KB · Views: 520
  • sa2021_gate_voltages_vs_output_39V_rms_simulation.png
    sa2021_gate_voltages_vs_output_39V_rms_simulation.png
    322.6 KB · Views: 457
Last edited:
astx

astx

Member
Joined 2011
Paid Member
SA2021 New Class H Prototype - Parts / Ideas

Important parts arrived today - a lot of IXYS V-Mosfets.

New and relatively cheap hall sensors made me think about a new variant of amplifier SOA protection which doesn't have an impact on sound quality:
Scan current, voltage and heatsink temperature of the power Mosfets/BJTs and calculate SOA using a microcontroller. A fast enough microcontroller can do this several 10k times a second. On overload condition switch of the input (audio relais) and/or power down the amplifier.
​
Whats your opinion?
BTW - Index in first post has been updated ... are you something missing there?


BR, Toni
 

Attachments

  • sa2021_bjt_mosfets_arrived.jpg
    sa2021_bjt_mosfets_arrived.jpg
    282.2 KB · Views: 516
You must log in or register to reply here.
Top Bottom