GLA (good little amp) "survivor"

Hello Ostripper
greetings waiting to see your pcb layout so i can learn something before starting my pcb
warm regards
andrew lebon

I've been "bouncing" around the forum , looking for EVERY leach and triple OP
thread , the present course I've set is the best.

(Attachment 1) is my progress (perfection can't be rushed :D)

Red= V+ Blue= V-
Green=ground
yellow=output signal/nfb
dark blue/brown= driver signal returns.

Each section of the amp is boxed in and has it's own separate area.
I'm down to just the small signal section so I can really make things small
and tight.
It still looks like 76 X 153mm will be sufficient for 2 pair OP.
Output stage can be scaled with 25mm extra PCB per output pair. Make
the whole PCB 250mm and you have a 6 pair "big" amp.

Or , split the board at the main njw0302/0281 drivers -
the same design can be modified in any configuration.
I'll have the "front end" done tomorrow.

PS - small signal section will have optional separate PS's AND it's
own separate "quiet" ground that terminates to the main star.

OS
 

Attachments

  • GLA1.2pcb.JPG
    GLA1.2pcb.JPG
    208.8 KB · Views: 631
Last edited:
:):D

smilies work...

let's see if my artist's work goes through...
Hey , it works.

I'm down to the 2 differentials - I'll do them "face to face" ,
for thermal stability.
This amp has way more components than the honey badger , luckily ..
they are smaller ones. :cool: Way cool - diya works again....

OS
 

Attachments

  • almost there.JPG
    almost there.JPG
    174.3 KB · Views: 462
What "tilt" is this supposed to remove?
Treble: I probably can't explain it right; however, the 10R-220u-1k has inflated the effective ESR of the 220u by 10R, making a bad cap simulator and causing a treble droop and the 0.1u is not enough of a fix for that because it probably should have gone parallel with the 10R instead of where its at.
It would be quite puzzling if the existing 1k wasn't already enough groundlift and it would be utterly confounding if adding 10R more helped; so, at post #60, I didn't groundlift the inverting input.
Bass: Also, that RC (gain divider's groundside leg) value is higher than 0.444hz, which is too high to successfully mimic a dc reference, so for the groundside leg of the gain divider, I would prefer a larger cap (bigger than 220u), a higher value resistor or both--whatever it takes to get that RC pitch down below 0.444hz for clearer cleaner bass. link to RC calculator
P.S.
Here's the level: That RC (gain divider's groundside leg) could be 0v-220u||220u-1K-Inverting input. Problem solved!
But maybe it makes a new different problem and on that I have no idea.
 
Last edited:
By andrewT -Are those rail diodes 1n540x size (3A)?

MUR460-E3/54 Vishay Semiconductors | Mouser

Use these for OP flyback + rail apps. - always available/reasonably priced....

Dan , the 10R groundlift is the 0V reference for BOTH the inverting and non-inverting
inputs. The 5-10R + double diode is "standard fare" for DIYA solid state amps.
Look around !
This amp is so good at negotiating DC offset , you could even use a mismatch mje350/
KSA992 as the input pair and STILL have <10mv DC ! (no offset trimmer needed - below 1)
The GLA is totally different than a "blameless" or leach.
I will be using a 220u @35V NP for the DC cap , BTW.

PS - assume you did have a "less than perfect" ground at the input with
the differential referenced to either side of the 10R.
A .1V "bad ground" would equate to over 3V DC at the output.
OS
 

Attachments

  • offset1.jpg
    offset1.jpg
    81.9 KB · Views: 1,128
It's done !

1. - Is the board finished.
2. - Is the signal/power in color - blue =V- red=V+ green=ground(s) yellow=
signal....

3. Is the schema that corresponds to the layout (hopefully -99% ?).

HD pcb bitmap (2100x900) is in the Zip file attached. It should scale to
76mm X 178mm.

OS
 

Attachments

  • GLA1.2PCB.JPG
    GLA1.2PCB.JPG
    668.6 KB · Views: 1,111
  • GLA1.2PCBvoltages.JPG
    GLA1.2PCBvoltages.JPG
    310.5 KB · Views: 1,033
  • GLA1.22.jpg
    GLA1.22.jpg
    161.4 KB · Views: 1,186
  • HD-GLAboard.zip
    68.1 KB · Views: 338
Hi Ostripper
greetings thanks for sharing your pcb design in the final schematic
whats the max supply voltage is 80 volts dc +/- ok for 10 pair output stage
warm regards
andrew lebon;)

LOOK below!!!
If you can source the MG6330/MG9410 outputs , a 5 pair amp (I'm making it now- below), can
match at least 8 or more pairs of MJL21193/4's.

Andrewlebon , you must tell me what offset you measure so I can decide
whether a offset trimmer is worthwhile to include on the level shifter
circuit. My original amp (Sansui) only has 6mv offset , but it uses a DC servo.

I was just looking at some moon audio/Classe' power amps , they look just
like the amp below. Use 14ga copper wire on top of the output rails with
a wire jumper between each output like below. Classe' used a GOLD plated
bus bar - oh boy ... copper will do :p .

EDIT- 80V is just fine - amp (front end) is regulated - all devices are good to 120V (or much more) !
OS
 

Attachments

  • Frankenamp.JPG
    Frankenamp.JPG
    280.5 KB · Views: 1,090
Last edited:
90V rails with 5 pair.

Amp is supply voltage independent.

Distortion is .004%/.008% -10/20K at the levels below.

You might either need a preamp or increase the amps global gain factor.
If you do THIS , you might need to recalculate the lag compensation
(R8 and C5) and run a loop gain analysis to check the phase margin.

I needed almost 4V input to get that 160V p-p waveform.Amp clipped at 85.5V
peaks... :eek:
OS
 

Attachments

  • 90V rails.jpg
    90V rails.jpg
    147.1 KB · Views: 427