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

Member
Joined 2011
Paid Member
ASTX LABS - SA2017 VMOS IXYS - Measurements

Noise measurements with closed case. Nearly the same results as on testbed with regulated supply and without case. :)

Output µV noise is a bit lower if I short circuit the inputs instead of using 600R resistor.

BR, Toni
 

Attachments

  • sa2017_noise.png
    sa2017_noise.png
    126.7 KB · Views: 809
Member
Joined 2011
Paid Member
ASTX LABS - SA2017 VMOS IXYS - Measurements

Heavy overload with connected speakers. Amplifier has very good control and there are no signs of oscillation. Rail sagging good viewable. This amp is all but not a treble speaker killer! ;)
Will move the amplifier now to my home cinema and change SA2017 IXYS status from testing to production ready. :)

BR, Toni
 

Attachments

  • sa2017_speaker_overload_hameg_20181205161632.jpg
    sa2017_speaker_overload_hameg_20181205161632.jpg
    89.1 KB · Views: 755
Last edited:
Member
Joined 2011
Paid Member
You are welcome!
And yes - the VP7723D is old but good. Of course not as good as a newer AP which is way too expensive for diy.

The automatic modes via GPIB bus are working very good. I'm using gpib tools/libs/kernel modules for linux and perl as scripting language.

One channels full plot takes about 40 minutes to check (fetching THD and noise data as presented the latest posts), but is running full automatic. :)

Have fun, Toni
 
... got today new MyDOA V4.1 prototype boards for the preamp ...

Have fun, Toni

MyDOA v4.x?? What changes from your v3.1 design?

I have built your v3.1 and it works and sounds great! I have used your v3.1 DOA successfully in two DIY projects:

#1) In a legacy audio preamp that include switchable baxandall tone controls and ...
#2) At the heart of a headphone amp.

I find it very competitive with John Hardy's 990C implementation of the classic Deane Jensen circuit.

In fact I think that I might well prefer your DOA over the 990C as I find it less fatiguing during long listening sessions.
 
Member
Joined 2011
Paid Member
Dear Carl,

- your results are very close to my findings. IMHO it is extremely important for diy'ers if one can repeat the results! Many thanks for spending your time testing the MyDOA V3.1! :) It would be very nice if you could post some pictures of your setup!

The MyDOA v4.1 is a rework of the pcb to make room for some small modifications:

  • allow adding a series resistor in TMC compensation for better GM and PM at unity gain
  • room for a VAS diode for reducing sticking on overload
  • allow TPC compensation with simply soldering the TMC resistor R7 to ground using a dedicated gnd-pin
  • allow replacing the current mirror trim-pot using fixed resistors (added 2 extra pins)
  • higher output currents to drive 75 - 100R loads
Now I can start testing the modifications using the new pcbs... but no need to replace the MyDOA v3.1. The MyDOA v3.1 is well tested and stable at unity gain and above.

The new pcb I have mostly designed for one reason: to speed up the assembling, because I can easy solder 10, 20 or even 30 pieces in one step due to the tab and v-routed pcbs in 2x5 col/row.
The new tower amplifier will get active crossovers for bass/mid/treble and needs many DOAs...

BR, Toni
 
Last edited:
Member
Joined 2011
Paid Member
Off topic: currently I'm evaluating "ngspice" an open source circuit simulator which runs native under Linux (my preferred OS).

If interested you can simulate my latest amplifier project (with IXYS power mosfets) using NGSPICE . Download the necessary files here.
(see post #29 and 30)
The results are from one (!) simulation run with a script that automates all tasks like tran stepping, Tian loop gain, GM, PM, FFT, THD, average power dissipation, SOA of selected Transistors ...

Attached the schematic and an example ngspice simulation result.

Have fun, Toni
 

Attachments

  • sa2015_v_mosfet_ixys_single_schematic.png
    sa2015_v_mosfet_ixys_single_schematic.png
    257.8 KB · Views: 609
  • Screenshot_20190106_205122.png
    Screenshot_20190106_205122.png
    200.7 KB · Views: 642
Last edited:
Member
Joined 2011
Paid Member
ASTX LABS - SA2017 VMOS IXYS - Measurements

Just wondered how good the SA2017 IXYS module performs (VAS diode and overcurrent protection in situ) if bias is reduced to 100mA (+/- 42V regulated supply):

1W@8R:

THD+N 20k without bandwidth limit (!) 0.006%
THD+N 1k without bandwidth limit (!) 0.0051%

THD+N 20k with bw limit 80k 0.004%
THD+N 1k with bw limit 80k 0.0022%

Attached a picture with a GPIB-bus driven hardcopy of the distortion residual spectrum (ex VP7723D using HP3585A) and the measurement setup.

BR, Toni
 

Attachments

  • IMG_20190116_175114.jpg
    IMG_20190116_175114.jpg
    496.1 KB · Views: 633
  • sa2017_low_bias_1w_8r_20k_distortion_residual.png
    sa2017_low_bias_1w_8r_20k_distortion_residual.png
    6.6 KB · Views: 624
Last edited:
Se'as Toni,

thanks for your reply.

The problem was that the data sheet says nothing about the source-case connection for the ECW20N(P)20. Only for the -Z or -S types, which are indeed source connected.

However, Exicon has replied to my question in the meanwhile and confirmed that also the ECW20N(P)20 (without any suffix) have source-connected cases.

Thanks again and best regards from NÖ,

Martin
PS.: Bought ECW20N(P)20-S last week, delivered today. :)
 
Member
Joined 2011
Paid Member
...
Attached a picture with a GPIB-bus driven hardcopy of the distortion residual spectrum (ex VP7723D using HP3585A) and the measurement setup.
... a new member for my measurement "park" ...

E.g.: a Cbc measurement of Sanken 2SC6145AY using 10mV/100kHz test signal. (Cjc space charge capacitance at vBC=0V)

BR, Toni
 

Attachments

  • astxlabs_new_member_of_measurement_device.jpg
    astxlabs_new_member_of_measurement_device.jpg
    148 KB · Views: 387
Last edited: