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

Jump to Latest
#1,706
SA2016: using lateral mosfets EXC10N20 EXC10P20

... very difficult to get the multi pair exicons stable. Perhaps my beta test layout (socketed gate resistors; long wires to gate) is the problem. Increasing the gate resistors from 220R to 330R helps a little bit but does not cure the oscillations.

1kHz small signal output 1W at 8R load resistor during tests:
The oscillations start (frequency around 21mHz) if I increase the bias higher as 100mA per pair. The THD+N increases from a good 0.002% up to 0.015% and the oscillation is viewable dancing on the 1kHz sinus.

Anybody with experience using those EXC10N20/EXC10P20 mosfets - just to be sure - could the 21MHz oscillation be from the Exicon's or do they tend to oscillate with higher frequencies?

Thanks in advance, Toni
 
#1,707
SA2016: using lateral mosfets EXC10N20 EXC10P20

... got it. 🙂

6 x RC combination (100pF + 47R) from gate to drain as short as possible - subtle 21 MHz oscillations are gone. 🙂

Some THD+N values @ +/- 42V regulated power supply, bias ~ 85mA per device, 8R resistive load:

1k@70W 0.00060%
20k@70W 0.0048%

1k@50W 0.00063%
20k@50W 0.0046%

1k@50W 0.00078%
20k@50W 0.0041%

1k@1W 0.0019%
20k@1W 0.0026%

SNR referenced to 70W@8R
103.5 dB wideband, 111dB bw 80k, 117dB bw 20k, 120.5dB A-weighted

SNR referenced to 1W@8R
85 dB wideband, 94dB bw 80k, 99dB bw 20k, 102dB A-weighted

Slewrate > 100V/µS symmetric

BR, Toni
 
#1,708
SA2016: using lateral mosfets EXC10N20 EXC10P20

test setup ...

BTW: Bias is rock stable from cold to hot.
 

Attachments

  • sa2016_test_setup.jpg
    sa2016_test_setup.jpg
    509.9 KB · Views: 601
#1,709
Do the RCs allow you to reduce the gate stoppers?

I've seen 20MHz and 100MHz oscillation on the bench, but I'm not sure it was even coming from the Exicons. All I had to do was touch the heatsink or volume knob and it would change.
 
Last edited:
#1,710
Gate resistor look symmetric. So, how about Ciss between N & P channel?! I wonder about that when Exicon sells their new generation MOSFET, ECX10N20/ECX10P20 - Not suffix ''R'' like old version.

I usually use Rg = 220R for P channel, 330R for N channel and attached with 100pF caps from Gate to Drain.

BTW, Nice slew rate.

Regard,
 
#1,712
Walkalone said:
Gate resistor look symmetric. So, how about Ciss between N & P channel?! I wonder about that when Exicon sells their new generation MOSFET, ECX10N20/ECX10P20 - Not suffix ''R'' like old version.

I usually use Rg = 220R for P channel, 330R for N channel and attached with 100pF caps from Gate to Drain.

BTW, Nice slew rate.

Regard,
Click to expand...

Thx! Your Rg values are well known here from an application note for ALFETs where N-channel Ciss is 500pF and P-channel Ciss is 730pF.

But the newest datasheets from EXICON ecx10(n|p)20 show the same Ciss and transconductance for N- and P-channel.
I have posted a table with newest datasheets Ciss values here: EXICON Ciss/Rgate calculation table
True or a copy and paste error? Thats why I have choosen 220R for N- and P-channel gate resistors.

BR, Toni
 
Last edited:
#1,714
astx said:
...
But the newest datasheets from EXICON ecx10(n|p)20 show the same Ciss and transconductance for N- and P-channel.
I have posted a table with newest datasheets Ciss values here: EXICON Ciss/Rgate calculation table
True or a copy and paste error?
...
Click to expand...

This must be a datasheet error. Still slightly unstable using RC 100pF+47R and Rgate all 220R. After changing N-channel to 330R stability for a bias from as low as 30mA per device to 300mA is mostly given.
Due to still some invisible instability (my spectrum analyzer range is 40MHz) I have added extra 27pF caps from gate to drain. Now no signs of instability viewable. High output power distortions worsens a little bit with this extra 27pF (THD20k@70W from 0.0035% to 0.0045%).

With this extra cap I have tried Rgate as low as 180R for N-channels and Rgate 120R for P-channels. Stable so far - no sign of instability if bias will be increased from 30mA to 300mA. THD as above.

Next step:

  • removing 27pF caps and
  • try to tune the RCs to get some bandwith and thus THD performance back...

BR, Toni
 
#1,716
Yes - maybe one with good contacts to exicon can write an email to clarify the details?

Just have tried with 220pF/33R snubber.
Doesn't cure the high bias with suddenly increased THD problem. Rgate N channel increased from 180R to 330R and P channel from 120R to 220R with this snubber doesn't help either.
Tomorrow I remove the snubber and will try different G/D caps - starting with 100pF as you are using.

A stable configuration can be achieved using:

  • Gate stopper N/P channel: 330R/220R
  • G/D RC snubber 100pF + 47R
  • G/D capacitor 27pF
BR, Toni

P.S.: The new datasheet contains this text: "Enhanced oscillation suppression in multi-device applications". 😱 What the hell was needed with earlier devices from exicon to get them stable?
 
#1,718
I can't help feeling the main problem is the distance of the O/P stage from the previous stage. IMHO, 50mm is too far but ju..uust doable with multiple O/P devices in a 200W 8R amp.

With a more compact layout, its always less necessary to use the 'snubbers'. Applies to all topologies I've played with.
 
#1,719
kgrlee said:
I can't help feeling the main problem is the distance of the O/P stage from the previous stage. IMHO, 50mm is too far but ju..uust doable with multiple O/P devices in a 200W 8R amp.

With a more compact layout, its always less necessary to use the 'snubbers'. Applies to all topologies I've played with.
Click to expand...
how should the three wires be dressed?
All tightly bundled, or all separated, or one separated and two twisted?
Which one is separate?
 
#1,720
kgrlee said:
I can't help feeling the main problem is the distance of the O/P stage from the previous stage. IMHO, 50mm is too far but ju..uust doable with multiple O/P devices in a 200W 8R amp.

With a more compact layout, its always less necessary to use the 'snubbers'. Applies to all topologies I've played with.
Click to expand...


The L-MOSFET amplifier itself is stable. No signs of instability viewable.
The V-MOSFET variant (IRFP240/IRFP9240) of this amplifier is rock stable with 150R gate stopper resistors. The pcb layout is nearly the same!

The subtle higher distortions start when the bias will be increased to above 100mA per device and hence the interaction between the OPS devices will increase too.
I think this is more a problem of the too less distance between the OPS devices. The gate stopper resistors are less as 5 mm in distance to the OPS devices plastic body - but how to get shorter?
This text from the datasheet
"Enhanced oscillation suppression in multi-device applications"
​
implies that problems using multiple pairs of EXICONs may arise.
Note: with only one pair of devices installed there wasn't any oscillation viewable. I've bought my lot of EXICONS in spring 2016 but the date code seems to be 2015 (S1522/S1535). Maybe they have send me devices which don't have this enhanced oscillation suppression?

BTW: haven't we found out during development of the high performance SA2014 amplifier that the distance from IPS/VAS to OPS doesn't play such a big role (speed of light) if correctly compensated. My 8 pairs SA2014 has a distance from OPS to VAS of minimum 250mm and maximum of 600mm due to the many devices!


BR, Toni
 
You must log in or register to reply here.
Top Bottom