Hi Vunce.
Great news. You may also want to keep the input wires far from mains, rails and speaker wires and twist every pair of wires.
Thanks Shaan.
Now that I’m satisfied with the noise reduction, I can proceed with final wiring and tidy up the spaghetti mess of wires.
Maybe, it can get even better!
UPDATE: Oscillation With Renesas MOSFETs - A Much Better And Easier Solution
Hello members.
While the previously posted solution here did solve the issue of parasitic oscillation with Renesas MOSFETs, during further tests with four channels of V4H I did manage to make one of the channels oscillate even with the capacitors installed. Some hair was pulled subsequently and the channel became stable with 100pF instead of the previously mentioned 270pF! Obviously no two MOSFETs are same and different ones would need different value of gate-source capacitor.
Anyway, I didn't stop but kept on trying to get to something that would stop me from dropping Renesas forever.
And I have good news.
--------------------------
The Solution:
R27,R29 and R28,R30 i.e. the gate resistors (initially 330R and 220R), all increased to 560R (and gate-source capacitors removed).
--------------------------
Now there is absolutely no parasitic oscillation in any of the channels, that tested 8 pairs of old and newly bought non-matched Renesas MOSFETs, and are perfectly stable for any signal level. Unlike the capacitor trick this solution has zero footprint, is symmetrical and works consistently. So I will update the original design with this one. If anyone is facing the same problem with 2S MOSFETs then please replace the 330R and 220R resistors with 560R. If biasing and offset trimming was done already then don't worry, DC operating conditions won't be affected after installing the four resistors.
Thanks and cheers.
shaan
Hello members.
While the previously posted solution here did solve the issue of parasitic oscillation with Renesas MOSFETs, during further tests with four channels of V4H I did manage to make one of the channels oscillate even with the capacitors installed. Some hair was pulled subsequently and the channel became stable with 100pF instead of the previously mentioned 270pF! Obviously no two MOSFETs are same and different ones would need different value of gate-source capacitor.
Anyway, I didn't stop but kept on trying to get to something that would stop me from dropping Renesas forever.
And I have good news.
--------------------------
The Solution:
R27,R29 and R28,R30 i.e. the gate resistors (initially 330R and 220R), all increased to 560R (and gate-source capacitors removed).
--------------------------
Now there is absolutely no parasitic oscillation in any of the channels, that tested 8 pairs of old and newly bought non-matched Renesas MOSFETs, and are perfectly stable for any signal level. Unlike the capacitor trick this solution has zero footprint, is symmetrical and works consistently. So I will update the original design with this one. If anyone is facing the same problem with 2S MOSFETs then please replace the 330R and 220R resistors with 560R. If biasing and offset trimming was done already then don't worry, DC operating conditions won't be affected after installing the four resistors.
Thanks and cheers.
shaan
Attachments
Hi Shaan,
For the past two days I’ve been enjoying my fresh new V4H.
Today, after several hours of playing when the preamp volume was decreased to low level listening the V4H started to act strange. The left channel would drastically reduce in loudness, almost muted. When the volume was turned back up no problems at all, it has only happened when the volume gets turned down low. Then I read your last post that oscillation could still happen even with the 270pF caps I have installed. First, is this a symptom of that very problem? Should I stop using the amp until this is fixed?
Regards,
Vunce
For the past two days I’ve been enjoying my fresh new V4H.
Today, after several hours of playing when the preamp volume was decreased to low level listening the V4H started to act strange. The left channel would drastically reduce in loudness, almost muted. When the volume was turned back up no problems at all, it has only happened when the volume gets turned down low. Then I read your last post that oscillation could still happen even with the 270pF caps I have installed. First, is this a symptom of that very problem? Should I stop using the amp until this is fixed?
Regards,
Vunce
Hi Bambadoo,
I’m using the Mosfets Shaan supplied.
Hi Prasi,
I’m using my AKSA Lender preamp which has a stepped attenuator, no carbon wiper here. This has not happened with the other amps I use. I will dig in again today to swap resistors as per Shaan’s suggestion.
I’m using the Mosfets Shaan supplied.
Hi Prasi,
I’m using my AKSA Lender preamp which has a stepped attenuator, no carbon wiper here. This has not happened with the other amps I use. I will dig in again today to swap resistors as per Shaan’s suggestion.
Attachments
So I’ve been using the V4H constantly, approximately two days, since R27-R30 were increased to 550r. I’m happy to report: No funky behavior! I’ve used different preamps with high gain, low gain and no gain(B1 buffer). In my system the B1 works very well with the V4H.
Thank you Shaan for sharing this fine project
HAPPY NEW YEAR!!
Vunce
Thank you Shaan for sharing this fine project
HAPPY NEW YEAR!!
Vunce
PeeCeeBee V4H Revision 1 - Stability and Speed Update
Hi members.
I'm happy to announce that the PeeCeeBee V4H has been updated to Revision 1. This update will add the following to the amplifier:
A: Guaranteed stability with either Renesas or Exicon MOSFETs.
B: Gate resistors same as PeeCeeBee V4 i.e. 100Ohm (or even lower!).
A:
Instead of trying to make the different constants equal i.e. add gate-source capacitors, I have added new constants to the MOSFET parasitics that vary with signal. Small drain-source capacitors added to each MOSFET to shunt the inherent and variable drain-source capacitance. This eliminates all output stage parasitic oscillation due to MOSFET mismatch/interaction.
B:
Instead of increasing gate resistors to stop oscillation, I have added a constant load to the VAS. Small capacitors from VAS collector to Ground added so that the parasitic inductance of the copper traces between VAS and MOSFET gates are shunned at high frequencies. This eliminates all parasitic oscillation due to VAS-MOSFET interaction and allows 100Ohm or smaller gate resistors without any stability issue.
In my test setup I paralleld 100ohm gate resistors with the existing 560ohm, resulting into around 85ohm to each gate. I have tested the modded amplifier with and without load, from zero signal to clipping. Squarewave tests have been done without input RF filter and output Zobel filters. All tests passed with zero oscillation. Slew rate increased from around 40V/uS(with 560ohm) to 100V/uS which is very handy for a high headroom amp such as this one.
FAQ:
Q. What are the dimenstions of the new design?
A. All mounting holes remain at the same position as before and PCB size has changed 0 mm.
Q. Can these mods be done in the existing Rev 0 boards?
A. YES. I will upload full tutorial shortly.
Q. How does it sound after the Revision 1 mod?
A. Just like a PeeCeeBee -tight lows, stunning mids and sparkling highs. THD figures remain unchanged.
Q. Why so late?
A. I wish my first tests earlier last year showed the oscillation. I think it was ultimately good that the oscillation showed up, V4H is now a much more stable amplifier.
Hi members.
I'm happy to announce that the PeeCeeBee V4H has been updated to Revision 1. This update will add the following to the amplifier:
A: Guaranteed stability with either Renesas or Exicon MOSFETs.
B: Gate resistors same as PeeCeeBee V4 i.e. 100Ohm (or even lower!).
A:
Instead of trying to make the different constants equal i.e. add gate-source capacitors, I have added new constants to the MOSFET parasitics that vary with signal. Small drain-source capacitors added to each MOSFET to shunt the inherent and variable drain-source capacitance. This eliminates all output stage parasitic oscillation due to MOSFET mismatch/interaction.
B:
Instead of increasing gate resistors to stop oscillation, I have added a constant load to the VAS. Small capacitors from VAS collector to Ground added so that the parasitic inductance of the copper traces between VAS and MOSFET gates are shunned at high frequencies. This eliminates all parasitic oscillation due to VAS-MOSFET interaction and allows 100Ohm or smaller gate resistors without any stability issue.
In my test setup I paralleld 100ohm gate resistors with the existing 560ohm, resulting into around 85ohm to each gate. I have tested the modded amplifier with and without load, from zero signal to clipping. Squarewave tests have been done without input RF filter and output Zobel filters. All tests passed with zero oscillation. Slew rate increased from around 40V/uS(with 560ohm) to 100V/uS which is very handy for a high headroom amp such as this one.
FAQ:
Q. What are the dimenstions of the new design?
A. All mounting holes remain at the same position as before and PCB size has changed 0 mm.
Q. Can these mods be done in the existing Rev 0 boards?
A. YES. I will upload full tutorial shortly.
Q. How does it sound after the Revision 1 mod?
A. Just like a PeeCeeBee -tight lows, stunning mids and sparkling highs. THD figures remain unchanged.
Q. Why so late?
A. I wish my first tests earlier last year showed the oscillation. I think it was ultimately good that the oscillation showed up, V4H is now a much more stable amplifier.
HNY!
The Rev1 mod is kind of a brute force. Through these tweaks we shunt the inherent parasitics of the board and MOSFETs and introduce to the amplifier our own parasitics that are well defined and doesn't vary with power / load / signal level / rail voltage / temperature, and fast amplifiers absolutely love well defined values.
The Exicon MOSFETs are less in need for the mod than Renesas ones, but will have the same benefits which are stability and slew-rate increase.