The problem is if I increase R6 I get huge distortion. 300 Ohms gives me (only) about 1%...😕
Thank you.
Gave it (CCS under top driver) a quick try with LSK170. Shows some promise. I had overshoot before, but clean rise and fall now on square waves. Some HF noise, but no time to track that down. The 170 is not ideal, because current doesn't level out until about 10V G/S. Need better part, or wait a see what LC's fix is.
Sheldon
With K30ATM and BF245A had tested only. But mine's LAT and low driver bias suffices. Check if your input BJTs still run about the same Ie, and if the amp's output tracks sine wave well symmetrically when cranking it up on the scope. Say at 30V pk-pk. About HF noise, see if its acting up so to revisit comp caps or input filtering 'cause the CCS boosts its speed. Will some NTC find enough temp to react when bundled with two small BJTs running 30mW each? We must get to know which exact type works for LC first. Still the CCS is fixing the drivers bias stability not the input drift. But in a bridge everything interacts non the less.
Not in the spice working example #2025. In reality one gets to be a trimmer to counteract the output bias imbalance and offset.
What's the optimum bias across drivers and laterals?
Those of you who built it (the lateral fets version) how does it sound? 🙂
Those of you who built it (the lateral fets version) how does it sound? 🙂
Last edited:
Look a bit back, bias has been discussed. Short answer, 150-200mA. For some comments look early in the thread, Nico was talking a bit 'bout his version you contemplate.
So for 180mA bias what would be the value of this resistors? 😕
Forgive my poor knowledge of electronics design, I'm reading Sedra & Smith - Microelectronic Circuits - for learning but I'm still very in the begining.
Run .op and manipulate R23 until you will get what you need through the LATs. In reality R23 better be a trimmer. The simulator won't have exactly the same Vbe for the drivers or Vgs for the LATs as each one ranging practical part. Just make Nico's as given but with trimmer R23, trimmer R6 and see if it will bias up and work stable on you. You gotta get hands on to experience how it moves... In mysterious ways that is.😀
With Nico's values for the resistors distortion is huge. The sugested above values work like a charm! Why did Nico used the values that give worst distortion?
One more thing: could I use 2SK1530/2SJ201 instead of 2SK1058/2SJ162 ?
They are not lateral but close.
They are not lateral but close.
Only laterals. What Nico showed are real values biasing real parts. Those which worked best in LT Spice are biasing models. Shows how sensitive it is non the less, because those values I gave for the good sim aren't really far. You got to see when hands on and use trimmers on crucial ones until you set it right as I said. Those are just my general opinions trying to help out though, I am not Nico and maybe the right answers are different, remember. In general this is a practical thing. You make it, you evaluate bias and stability on your layout, you try for good offset. You gotta have generator and scope absolutely. Not recommended as novice's project. If in doubt, get official boards and make F5, next of kin best thing. Just my plan B suggestion if things go wrong, not trying to disappoint you.
You're not disappointing me, I thank you for your advice. 🙂
This is an interesting thread. I'll be following it and when a final stable design is achieved I'll build it. 😉
Thanks.
Hi PaulPT,
There is no absolute GO/NO-GO condition in analogue electronics and tweaking practically is more often than not the final stage of the design. Besides, I think that DIY audio guys (myself included) places far too much emphasis on harmonic plots from simulators.
It is possible to optimise every component using a simulator and get really silly numbers. Problem is one eventually believes that it will be the actual performance of the product - it never is.
I am still using my lazy cat to this day, everyday in my office. Even with the assumed high level of simulated distortion, its sound remains absolutely magic.
There is no absolute GO/NO-GO condition in analogue electronics and tweaking practically is more often than not the final stage of the design. Besides, I think that DIY audio guys (myself included) places far too much emphasis on harmonic plots from simulators.
It is possible to optimise every component using a simulator and get really silly numbers. Problem is one eventually believes that it will be the actual performance of the product - it never is.
I am still using my lazy cat to this day, everyday in my office. Even with the assumed high level of simulated distortion, its sound remains absolutely magic.
- Status
- Not open for further replies.