has anyone seriously considered why you'll never find those fancy cascode stuff in commercial amps priced below say 3000$?
and has anyone seen an audio IC that has such fancy stuff?
No?
Why not?
Its because of yield. Those cascodes may or even will require adjustment that means rework and that lowers the margin, in the case of IC the yield.
Its only worth consideration if the DIYer has plenty of time and enough equipment that could buy a 3000$ commercial amp combined.
You'll have to have plenty of prefessional know how to design a PCB such that those cascodes will not oscillate.
and has anyone seen an audio IC that has such fancy stuff?
No?
Why not?
Its because of yield. Those cascodes may or even will require adjustment that means rework and that lowers the margin, in the case of IC the yield.
Its only worth consideration if the DIYer has plenty of time and enough equipment that could buy a 3000$ commercial amp combined.
You'll have to have plenty of prefessional know how to design a PCB such that those cascodes will not oscillate.
Youre very creative, with some experience youll make a very good designer, think the above part carefully through again.
Then try simming it.
Youll cure to a large extent another problem of the design as well as improve its performance by a couple oders of magnitude.
You can find such a design in a commercial product, its been 13 years I sold the design to them. Offcourse no input transistor cascodes though, that just messes some other parameters up. If youre going to use those cascodes be sure to place a cap across those small 10ohm resistors though.
It's not...Higher impedance M-Ohm would make it so...ideal CCS better option. difference is like bouncing a ball on a soft wooden floor and on solid concrete. One rattles the building and is felt everywhere. Both floors bounces the ball one just bounces the ball better..
No I didn't mean resistor should be at this spot, not at all, I said it is doing what can be expected from resistor nothing more. CCS were present in this spot before and were replaced with resistor because the simplicity. Not a good choice since distortions are way to high. As I said CCS wil be incorporated here. Will post some ideas about my solution regarding this.
hehe yes obviously would be much better. First of all I had to replace two pin zener on PCB and put instead something small enough to fit in between neighbouring parts. Secondly TL431 were at my hand in my stock, so logical thing happened.
Replace a zener and current supply resistor with bunch of parts is not what is desired here. Our aim is always to make it in a simple manner, small parts count with perfect performance. I promise I will design new PCB, SSA BIGBT HP style to start GB when we'll perfect out complete schematic. At this moment I think we know a lot more how to improve SSA, give attention to its specific parts to improve its performance by a couple orders of magnitude, as homemodder said.
P.S. One important rule for combined zeners and CCS, they all have to be temperature independent.
Last edited:
SSA BIGBT HP PCB is among perfect ones. No HF mistake, no mistake at all for given sch. We'll have to improve some parts of the sch for the next SSA PCB and I am willing to design it myself, no problem.
TO-126 not needed here, we are talking about very high output impedance CCS supplying current from 15 V stabilized rail to input pair emitters, all together 0,12 W of power dissipation on one serial BJT (14,4 V voltage drop at 8 mA). Should be made out from app. three active parts plus few passives and completely temperature independent.
CCS made out BF245 has scary current vs. temp curve. Bias consequently cannot rely on such big temp dependency.
3k3 2W can stay if zeners will be substituted with active ones, only if you want to make "never connected" power supply.
In that correlation it is excusable of course but probably not canceled out bias variations in a whole range. Did you thermally coupled input BJT-s and these two fets? Maybe tightly coupled these four would be enough to stabilize complete front-end bias?
Last edited:
Used thermal grease between input parts, then bound with shrink wrap. NTC was bonded to the top of the pair with epoxy. Seems to work well. In my case, it was undercompensated with the 220R series resistor (running 10mA through driver, no driver cascode, so heat output greater than LC's driver). With no series resistor it might be very slightly over compensated.
Now I await the findings on bridge supply zeners and CCS.
Sheldon
Hi Sheldon
Nice progress you've made.
So NTC is working properly, glad about that. If it is still undercompensated with 220 ohm, than try a little lower value ie. 180 ohm. Without serial compensation resistor it is always overcompensated.
The decision to wait for zeners and CCS is also very wise, I'm working on it.
Nice progress you've made.
So NTC is working properly, glad about that. If it is still undercompensated with 220 ohm, than try a little lower value ie. 180 ohm. Without serial compensation resistor it is always overcompensated.
The decision to wait for zeners and CCS is also very wise, I'm working on it.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.