I wrote distorted, not clipping.
As I said, oscillations.
On +-52V I measured 117.8W on 8R load.
The version I built was with Hitachi mosfets 2SK135/2SJ50 with no problem of any high frequency oscillation. The version with Exicons was built by some others and Filkomen spent quite some time to debug it.
I helped him by phone with some suggestions what to try. What was first thought that cascode could be reason for high frequency oscillation and some changes there helps but not in all cases.
Next step was to remove capacitors connected parallel to the VAS emitter resistor. That helped timios with his amp, but all simulations showed that the amp should work as it was designed.
What was different between those two version obviously where Exicon laterals used here. First step was to increase gate stoppers and it helps but not at all power supply voltages. Adding RC between Gate and Drain on each lateral Mosfet solved all problems, now it works with no HF oscilaton even with VAS capacitor in place. By the way, lowest THD20k was with 100nF.
Final touch was alu plate fixed over lateral and grounded.
Filkomen tried to use KSA992/KSC1845 instead 2N5551/5401 and it works but my simulation did not show lower distortion, but replacing KSA992/KSC1845 with 2SA1381/2SC3503 showed lower distortion but with worst harmonic distribution in simulation.
Damir
I helped him by phone with some suggestions what to try. What was first thought that cascode could be reason for high frequency oscillation and some changes there helps but not in all cases.
Next step was to remove capacitors connected parallel to the VAS emitter resistor. That helped timios with his amp, but all simulations showed that the amp should work as it was designed.
What was different between those two version obviously where Exicon laterals used here. First step was to increase gate stoppers and it helps but not at all power supply voltages. Adding RC between Gate and Drain on each lateral Mosfet solved all problems, now it works with no HF oscilaton even with VAS capacitor in place. By the way, lowest THD20k was with 100nF.
Final touch was alu plate fixed over lateral and grounded.
Filkomen tried to use KSA992/KSC1845 instead 2N5551/5401 and it works but my simulation did not show lower distortion, but replacing KSA992/KSC1845 with 2SA1381/2SC3503 showed lower distortion but with worst harmonic distribution in simulation.
Damir
Attachments
After some listening, amp started to have distorted sound. I've spend a few days to track down the problem and the problem was RF noise no being decoupled to GND. Solution was 10nF ceramic caps from input RCA GND to chassis.
After that, amp sounded great, but on some specific frequencies it sounded weird. I have measured it with Tektronix 2225 and everything was fine. That, I've decided to measure it on Tektronix TAS250, just to be sure and saw oscillation.Guess my 2225 needs some refresh and calibration.
Oscillation occured only above +-50V. On the attached picture, it was +-50.3V. On +-55V, sine wave wasn't recognizable. But how? Amp works and sounds good.
I started to track down the problem and solution was to tighten the Exicon laterals with aluminium profile on the Exicons (see picture). The key is that everything needs to have a good GND connection. After that, oscillations were there, but they were reduced. Next thing was to change gate stoppers from 270R/330R to 560R/680R. After that, no oscillation occured even on +-60V.
And last thing, 1uF caps C8/C11, bypass caps on VAS, with this modifications can be used with no oscillation. They were never the problem, problem were Exicon mosfets.
So, if you want to build this amp and use PSU over +-50V you need to make this mods, but on +-45V (safe margin from +-50) it is good to go with original schematic from post #1.
Good day to you all.
In the first photo.
Connect earth cable(yellow) direct to chassis!
This is the safe ground!
Connect the power supply 0v via gng lifter to chasis.
Hi Damir, i hope that this will be the final tweak.
I mean the most critical.
Regards.
Thimios
Hi Damir, i hope that this will be the final tweak.
I mean the most critical.
Regards.
Thimios
I found similar behaviour with my latest class-G amplifiers (see A small Class-H Lateral MOSFET amplifier).
My usual recipe with the Renesas devices is 220p G-S on the 2SK1058 and 22p in series with 22Ω G-S on the 2SJ162, with 100Ω in series with the gates. Up to now that's just been something that's worked, that has very little effect in simulations, so I've just repeated the same thing with each amp without really thinking about it. Indeed I generally don't bother simulating that bit.
With my latest amp though (Exicon ECX10N20 and 10P20), I get varying levels of HF oscillation on the positive side of the waveform with lowish load resistors (about 6Ω and under) and healthy input level. I found that a finger on the 220p cap shut it up.
Now bear in mind that I don't use Exicon models (I didn't even know any existed, so simply use the Hitachi/Renesas models), so I imagine if I had a decent model for the Exicon devices I would have seen that.
In practice, removing the 220p cap from the Exicon N-channel gate solves the oscillation, with nothing else required. I need to properly explore this before I'm going to call it fixed though.
What is your PSU voltage? Bear in mind that here, there is no oscillation at all on +- 45VDC, even on +-49VDC (I said 45 for safe margin). But on +-50VDC oscillation starts and it is getting worse with higher voltages.
janusz / suzyj
Try these more recent lateral mosfet models from Ian Hegglun - Website
Direct download from Ian's googledrive - Mosfet Models TXT
Diyaudio thread - Better power mosfet models in ltspice
Try these more recent lateral mosfet models from Ian Hegglun - Website
Direct download from Ian's googledrive - Mosfet Models TXT
Diyaudio thread - Better power mosfet models in ltspice
I used Ian models in my simulation and it did not show any instability with 330R/270R gate stoppers and no cap between gate and drain, but real amp had instability problems.