When I built my current workstation, I assumed I would be able to just
plug in my headphones directly into the sound card like I did with the
previous sound card in the old workstation. I
discovered that the new sound card just didn't have the ability to
drive the headphones beyond barely audible whispers from another room
level. The temporary fix was a 150 watt per channel surround sound
receiver with a headphone plug. A serious waste of electricity and
source of sweltering heat during the summer of 2005. I immediately
started looking for a more appropriate solution.
I built an op-amp-buffered C-moy amplifier on a breadboard, and it also
didn't have the drive that the headphones needed. That and the 14pin
DIP op-amps kept overheating. After that experience I knew I needed
something a lot more serious then a op-amp directely driving headphones. I
extracted some amplifiers circuits from some CDROM drives with the same
results. My C-moy amplifier gave me the false impression that I needed
more then +10dBV of gain, nudging me towards the +23dBV of gain in the
original HPA17 design.
I discovered some antique 8-bit ISA sound cards with speaker plug
outputs, so I started dicing up the cards to isolate there amplifier
circuits. To my dismay, nothing seamed to improve the quality of the
sound card amplifier chips. This in combination with my experience with
the CDROM circuits reinforced my lacking impression of single chip
It was about the winter the of 2005 when I sat down and started making
many different drafts of amplifier circuits, involving independent
output drivers and voltage gain stages. The seventeenth combination was
the dual op amp [midpoint reference buffer, and amplifier stage]
driving a BUF634 as the finals. It was about this time I came across
the PIMETA v1.0, and my gut was telling me to by one then. I held off
to see if I could improve the HPA17 design, and build it cheaply.
The schematic went through fifteen revisions before I started laying
out the circuit boards. The original HPA17 design had dedicated
power rails for each channel, with a dedicated voltage regulator to
drive each channels midpoint-reference. There was just no way to cram
all that onto a single three by four inch circuit board, and I wanted a
Solid ground plane. So a compromise was made, to drive both channels
midpoint from a single voltage regulator, and the power rails were
combined. Resulting in the "M.C.C. HPA17r15" back in 2009.
At this point, I had been using a diced up sound card for four years.
This thing had a THD of 5.1%, a noise level of -40dB, and loved to AM
demodulate every single PCI signal bleeding threw the sound card. I
decided enough was enough as I dropped the HPA17 development and set
out to buy myself a PIMETA. I discovered that the PIMETA V1.0 had been
replaced with a newer one that required surface mount chips. I do not
have the experience with surface mount hand soldering to trust myself
with surface mount chips. Because I hesitated to buy a PIMETA, I now
had no choice other then to finish the HPA17. Until this time the
amplifier was something I was working on during my free-time. The name
of the game from this point on was, "Failure is Not an Option".
Given the new drive to finish the HPA17, I set out to make sure the
main amplifier board was good. I could adjust the input and output RFI
filters cheaply at a later date, the HPA17r16mb had to be right the
first time. After a year and a half of going over every single aspect
the main amplifier board layout, from via placement and size through
signal and power trace proximity, the amplifier was ready for
prototyping. The final "M.C.C.
HPA17r16" design has NEPP listed MIL-PRF-55182 "Established
Reliability" resistors, instead of the
original 1% commercial grade parts. Every single electrolytic capacitor
was replaced with 25 volt ESRL High Reliability electrolytic
capacitors in the design,
solder-dipped, and verified under 24/7 load for over a year to be tin
whisker free. Every single part on the main board was burn-in tested
for at least six months, to verify that it was good to go before being
installed on the final circuit board.