lGl-2, continuing "hybrid madness" - no GNFB class A

Status
Not open for further replies.
Jump to Latest
#61
Not to rain on anyone's parade, but the circuits on page 1 have an awful lot of parts, which degrades reliability. For Hi-Fi I would replace the plate resistors in the front end with current sources (or is it "sinks"), part for the better linearity and higher gain, but mostly for the power supply hum rejection.

I was once heavily into solidstate power amp design and I've always felt that how the amp handled clipping, Rf or digital noise on the input and general mis-use should be a top priority. Also turn-on/off transients, especially if they ever get direct coupled to a tweeter in a multi-amp'd system (where an active crossover is used). I'd put a passive Rf filter right at the input, so you never ask the circuit to do something it can't do well. Maybe a 1K in series and a 1nF polypropylene cap to ground (-3dB at 145kHZ assuming a very low source impedance around 100R).
 
#62
Bob Richards said:
Not to rain on anyone's parade, but the circuits on page 1 have an awful lot of parts, which degrades reliability. For Hi-Fi I would replace the plate resistors in the front end with current sources (or is it "sinks"), part for the better linearity and higher gain, but mostly for the power supply hum rejection.

I was once heavily into solidstate power amp design and I've always felt that how the amp handled clipping, Rf or digital noise on the input and general mis-use should be a top priority. Also turn-on/off transients, especially if they ever get direct coupled to a tweeter in a multi-amp'd system (where an active crossover is used). I'd put a passive Rf filter right at the input, so you never ask the circuit to do something it can't do well. Maybe a 1K in series and a 1nF polypropylene cap to ground (-3dB at 145kHZ assuming a very low source impedance around 100R).
Click to expand...

Hi Bob,

The ones at page 1 were just early concepts, for discussion and exchange of ideas 😉
The one built and tested in the end is at page 4 - >HERE<
Number of parts / reliability dependency is arguable. In a well-designed circuit parts don't break without a reason. I also like current sources, but in some cases their influence on circuit performance may be not as high as one would imagine. I always use RF filter at the input, but I tune it to higher roll frequencies, otherwise it influences the square wave response.
You are absolutely right with regards to clipping and other mis-use behavior - I always test my circuits on most of the possible troubles, including highly capacitive loads.

Cheers,
Valery
 
#74
jwilhelm said:
That's what I assumed but thought I better check. Thanks. Does the extra supply voltage help? I installed 400 volt caps everywhere thinking I might want to try turning it up a bit.
Click to expand...

Extra supply voltage lowers that "famous" 2-nd harmonic a bit. I did exactly the same - used 400V-rated caps and then tried to raise the supply all the way up to 350. Well, 300 seems to be the optimum.
 
#75
I finally got a chance to connect these input boards to Slewmaster output boards. I'm running a 45 volt supply through a bulb limiter. I didn't have a chance to experiment long but I wasn't able to get enough current to set up any bias current in the outputs. Is this because of the lower voltage supply?
 
#76
Hi Jeff,

I have checked with my build - output buffer quiescent current is 5.8mA with +/-70V rails. Running it with +/-45V rails gives us 5.7mA, which is still OK.

We need to make some measurements to understand what's going on there.
I would start with voltage drop over R18 (schematic from post #35).

Cheers,
Valery
 
#78
Yes, no problem. You can even connect both output pins and servo together without any resistors - it will work anyway. The standing current is set by CCS, so it will stay the same. Well, resistors will just make the measurements easier...
 
Status
Not open for further replies.
Top Bottom