I am doing measurements in the built amp so including input rc filter and output coil and zobel.
Will do
Crackling appeared after input went higher than 0.7vac and not only while turning the volume pot.
Touching the heatsink reduces this issue so I will ground it.
I am not using C17
An ungrounded heatsink (or not directly decoupled to the PCB in some way) I've found often causes some sort of strange RF effect, so I always pay special attention to it.
Often touching the heatsink will reduce noise simply because when your body is grounded, it becomes a shield, and otherwise it tends to radiate whatever noise hits it back over it's entire surface area. So for instance as your finger approaches the circuit the noise will rise because your body is radiating noise, but when you touch the chassis the noise will suddenly drop, because your body is grounded and now acting as a shield. This can be distinguished from cases where touching something actually causes oscillation or stops it.
Often touching the heatsink will reduce noise simply because when your body is grounded, it becomes a shield, and otherwise it tends to radiate whatever noise hits it back over it's entire surface area. So for instance as your finger approaches the circuit the noise will rise because your body is radiating noise, but when you touch the chassis the noise will suddenly drop, because your body is grounded and now acting as a shield. This can be distinguished from cases where touching something actually causes oscillation or stops it.
He has no High Pass filter on the input of the amplifier.
The NFB becomes the high pass filter and this is a poor way of re-designing the amplifier.
There could be a high pass filter in the source, or the source could have a DC blocking capacitor that is acting as a high pass filter, but it's not in the DCB1.
After increasing the gate to drain caps placed in all output mosfets the amp became stable and now I can play it loud (really loud) with all the input preamps I have available.
Sound lost all traces of sybilance and bass remains deep strong and detailed.
I will now build the second amp so I can listen in stereo and will play with compensation until I find voicing suitable to my system.
Off course the new board will have all mods suggested by the group.
Soon will report back with new subjective listening impressions.
Sound lost all traces of sybilance and bass remains deep strong and detailed.
I will now build the second amp so I can listen in stereo and will play with compensation until I find voicing suitable to my system.
Off course the new board will have all mods suggested by the group.
Soon will report back with new subjective listening impressions.
PS: I am sure it was Miib that sugested these gate to drain small caps during the development of my first mosfet amp ( http://www.diyaudio.com/forums/solid-state/258549-assemblage-power-amp-23.html#post4101192 ).
They tend to reduce highs (the same way as miller caps do in the VAS) but to less extent.
They tend to reduce highs (the same way as miller caps do in the VAS) but to less extent.
Just finished second board including mods as suggested and it worked right from the start.
Stability is no more an issue here.
Although it sounds fast with really good bass, the highs are slightly forward so I would like to reduce those.
I made some measurements that indicate a rise in gain starting at 10k... not fenomenal but enough to be heard.
Gain is up for 1dB at 20k and 3dB at 60k.
Can I reduce high freq gain by using a cap over the FB resistor ? IMO it can increase FB at high freq so gain will be reduced but will it create instability issues ?
Stability is no more an issue here.
Although it sounds fast with really good bass, the highs are slightly forward so I would like to reduce those.
I made some measurements that indicate a rise in gain starting at 10k... not fenomenal but enough to be heard.
Gain is up for 1dB at 20k and 3dB at 60k.
Can I reduce high freq gain by using a cap over the FB resistor ? IMO it can increase FB at high freq so gain will be reduced but will it create instability issues ?
Attachments
Many years ago I worked for an outfit called ILP Electronics that made amplifier modules. One used power MOSFETs and worked fine. But when we had to change the types because of discontinued product we had a lot of trouble with rf oscillation and found that it was due to the tab of the device now being the source connection instead of the drain of the previous device. This was with the heatsink earthed/connected to 0V. As there were many customers for this module, it had to work with the heatsink floating or earthed. I seem to remember that beads were used on the leads. I hope this advice might be useful.
Funny you'd mention that -- I just experimented with this in my Honey Badger today. The gain is flat with my Honey Badgers, but the sound changes quite a bit if the "LC cap" is added across the feedback resistor. Without the cap, the sound is very lively, but also a bit on the "bright" side. If the cap value is too high (5pF or more), the amp sounds boring. For even bigger values, the whole things starts ringing and oscillating, resulting in smoke and blown fuses. Oh, and the square waves didn't look the same in simulation and on the scope (so I just ignored the simulation). Conclusion: just give it a try, start with small cap values, and check the result on a scope.
I use 220p G-S on N-channel lateral parts and 18p in series with 22R for P-channel laterals, though it's really pretty arbitrary. The rationale behind a much higher Cgs for the N-channel parts is to slow their rise times to better resemble those of the P-channel ones.
Cool!
Yes, you can. It's a very standard stabilisation technique. Small values of capacitance will have a big effect in this spot, as they're effectively multiplied by the gain of the amp.
I have been playing with compensation and now I reached a sweet spot.
I am using TMC 120 2k2 500 and no cap over the feedback resistor.
Going from 1k8 to 2k2 in the TMC did remove sibilance while retaining good low freq attack and speed. (Not as good as with 1k8 but fair).. good compromise.
Now building timer and output protection for DC.
Will post final sch and pics latter.
Thank you all for your support.
I am using TMC 120 2k2 500 and no cap over the feedback resistor.
Going from 1k8 to 2k2 in the TMC did remove sibilance while retaining good low freq attack and speed. (Not as good as with 1k8 but fair).. good compromise.
Now building timer and output protection for DC.
Will post final sch and pics latter.
Thank you all for your support.
Been fiddling with feedback and found that with a little bit more I get more linearity and loose the spicy highs... so now I am using 22k for feedback resistor as in the sch in attach.
Now sounding BIG and highs almost acceptable... would like to make it sound BIG but sweet at the same time...
As more feedback helps, I believe the trick is to reduce distortion in the high freq so I would like to try more degeneration in the IPS.
Now I am using LSK170BL 5mA Idss running at 3.7mA (seems to me the sweetspot) and Rs are 22 ohm.
With these Rs I have -1.9mV offset.
I would like to rise these Rs to 100 ohm but simulation indicates a rise in offset.... (to 3mV)... is it normal ?
Would the 100ohm Rs reduce High freq distortion significantly ?
Please have a look at the latest sim....
Now sounding BIG and highs almost acceptable... would like to make it sound BIG but sweet at the same time...
As more feedback helps, I believe the trick is to reduce distortion in the high freq so I would like to try more degeneration in the IPS.
Now I am using LSK170BL 5mA Idss running at 3.7mA (seems to me the sweetspot) and Rs are 22 ohm.
With these Rs I have -1.9mV offset.
I would like to rise these Rs to 100 ohm but simulation indicates a rise in offset.... (to 3mV)... is it normal ?
Would the 100ohm Rs reduce High freq distortion significantly ?
Please have a look at the latest sim....
Emitter resistors (AKA degeneration) on the LTP will probably increase offset and distortion, but that doesn't mean they won't improve the sound. I think you will just have to try it. So far your experiments seem to indicate that harmonic distortion is not what is causing the problem.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.