Can you spot the major fail?


I'm sorry I don't have the schematic. KiCad 6.0 shows only "?" for most components. This is 5 years old nearly. It's a CMoy style headphone amp creating a virtual floating virtual ground instead of a +/- rail. The circuit reinforces that VGND with an NE5532 buffer producing VCC/2, stabilised by the large bulk caps and noise filtered by high resistor values (high impedance = HPF with very low cut off).

The flaw I'm referring to is the VGND buffer being an NE5532 (U3) while the amplification stage are OPA551s. The issue presents when the OPA551s are driven harder than the 20mA limit of the NE5532 providing the ground reference. Current has to flow through back to ground... and AC (audio) signals the current goes in both directions relative to that VGND.

Thing is, it still works fine. The goal of this amp is to have .... always, enough gain budget. I think I settled on a gain of +11 and output current limiting of 33 Ohm balanced with the 220uF and the target headphones of 38Ohm produces a bode plot with sensible parameters, particular focus on the bass end.

I'm not intending on going deaf either. It's the bass I'm interested in, I have headphones that will attempt down to 20Hz and I love that FAT sound. So I need the gain and more importantly the current to drive that bass. 20mA won't do. 100mA is more like it with at 12V drive range. The mid range is usually -5-10db or more and the bass is a shelf at 100Hz +10db.
 
Last edited:
The other flaw is that while using 100K resistors to charge the VGND reference cap creates a new DC LPF and mutes almost all power rail noise, it also means that 220uF cap takes about 5 seconds to charge. The result is harsh DC offset on the output while it's powering up. The OPAs have an enable pin, a comparator on the VGND which enables it is an "next version" upgrade along with a high current VGND amp.
 
Well the SMT connectors will pull of the PCB after a while too? Or are the locating tabs under them actually soldered into the holes? I don't recommend pure SMT connectors for anything that sees regular human interaction like a headphone socket. You can put vias in-pad to improve the strength of SMT mount connectors BTW.
 
  • Like
Reactions: paulca
I have ripped 2 off PCBs in my times, but not on this. The issue is they don't flow well in solder oven profiles and often end up barely sitting on top of the solder rather than sinking into it. These ones where hand soldered. Surface mount pads can be ripped off too.

I have resoldered several ebay/aliexpress modules with surface mount audio or power jacks. Usually when things go wrong, it's one of my first go tos, is check the connector solder under a microscope and redo if suspect.

To be honest the most difficult problem with the custom PCB turned out to be getting it into the case it was designed to fit. I forgot to account of the angular clearance to allow the shaft of the volume pot to go through a drilled hole, and allow the board to drop flush onto it's screw mounts. It went in, but it's not coming back out. There is about -2mm clearance and it's wedged in. In hindsight should have cut a slot for the volume knob and back filled it.

Annoyingly I noticed a power glitch the other day moving cables around. It might not be the surface mount audio jack that causes and issue, but the barrel jack for power.