Ultra Amplifier with JFET input and Lateral MOSFET out

It is fine, nothing to change. I am having similar thoughts to Brain, run the output from the raw DC and the upconvert the driver stage by a voltage doubler followed by a regulator. The front end needs be clean.
 
Last edited:
Brian and Nico! I agree with both of you, and as far (which is not very far) as I understand we ar nitpicking at this point.
The HV is well regulated, and in my case, probably includes a cap multiplier, My IPS will most happily accept an additional small voltage drop from the multiplier, while the VAS is the stage that needs the extra voltage (super regulated, that is). Will there be any audible or measurable difference at all?
 
  • Like
Reactions: U102324
As a complete amateur I am really pleased to have conversations with audio erudites who knows what you are talking about! Thank you for spending your time.

Brian: Yes, I will make Gerbers tomorrow when I am sober and awake. I usually make PCBs at NexPCB, but my last order stalled as they asked for all sorts of peronal data, VAT numbers and what not. Can anyone recommend a cheap manufacturer? I have plenty of time but not that much money, hehe.
 
The improved PSRR (from Cap-X) shouldn't improve performance. Rather, it helps it from degrading if the supply isn't clean.

For the voltage boost, in sims I found at least 3V to be of benefit. For some reason, my models for SC3503/SA1381 responded well to the boost. TTC004B/TTA004B didn't seem to care. Not sure if this is simply a simulation quirk or a reality.

There's two reasons to boost the IPS/VAS rails from my perspective:
  1. To allow for a filter (Cap-X or RC filter) to improve PSRR without loosing voltage.
  2. To allow the MOSFETs to swing closer to the rails which increases Class AB output potential without actually raising the OPS rails. The main reason this is a benefit is when running a large Class A region where dissipation is a concern. You can run a bit more Class A for the same Class AB output potential.
Once you decide to boost, the more the better. But the benefit declines as you keep increasing. I think somewhere between +5V and +10V is a good target. I choose 5V to 6V. The AMB B24 uses a 10V boost.

Keep in mind the voltage rating of the input FETs. In my version using LSK489/LSJ689, they are rated for 50V. Taking into account PSU regulation and mains regulation, I'd try to keep the IPS-VAS rails are 45V or less. You have different devices, so you'll need to assess this yourself.

Regarding power levels, I targeted for 30V main rails because that provides 40W of class AB output. But more importantly, the output in the first 25W is very clean. Distortion seems to rise more above this point. I can't see needing more that 25W, so this works for me.
 
  • Like
Reactions: chermann and m0rten
Brian, I know this is not a military project but can your simulator automatically step all components under worst case condition and tolerance spread, say over voltage as well and a failure analysis. If I could I would because that will already beat any design I have ever seen on DIY for integrity and design with a near 100% probability of success. It all depends if it will do it automatically, else don't do it you will be and old man at the end.
 
LT Spice doesn't have a Monte Carlo analysis that I'm aware of. At least not an automated one. I think you need to use fancier professional simulators for those bells and whistles.

I try look at the design and theorize which components might need precise values. Then adjust them manually to confirm my suspicions.
 
  • Thank You
Reactions: U102324
I don't know if I am going to break a DIY golden rule, but my suggestion would be for Morten to get pricing for the prototype and then give us his banking detail so that we can help him finance it, Even a dollar some times help, lets call it crowd funding. I wanting to sell kits as well then Brian and Morten select their share for either a kit or fully built product. Open a bank account BRIATEN and the business kicks off.