Thanks for the heads up fish.

Used the BD139 and BD140, standard stuff.

I actually tried using a rod saw at one point, but the blade was too dull. I had pilot holes drilled at each of the four corners, and I figured as long as I kept it square with an external guide I would be fine as I moved down. It worked for the most part I guess  I think if the damn thing had a speed control I would have been fine. But it only goes at high RPM and that becomes hard to control.

anybody building a M3 should forgo the decade old opamps in the Part List and go with something more modern.  the ADA4627-1BRZ, ADA4898, and OPA1611 are easy recommendations - ancillary gears and ears apply.  keeping L/R opamps the same and switching up the G channel will change the sonics as well.  i would recommend the SOIC-8 adapters from Leckerton or Brown Dog.  up the O11 V-out to get close to the max rating of the opamp you want to roll.  30-31V gives you a lot of opamps to choose from.  socket R10 and D5 in the O11 to make sure your V-out is under the max rating of the opamp. 

socket the gain resistors.  the stock 11x gain is too much for sensitive HPs, but has less NFB. 

the Par-Metal 20-series 12x12x3 is a decent chassis for the price.  i personally dislike the black anodize, but like the natural anodize and black painted chassis.  the 3" tall chassis gives you extra room for taller heat sinks and plenty of ventilation.  the stock 1.5" heat sinks are good for 150mA of quiescent current bias.  i would recommend at least 100-110mA of bias.  if going above 150mA, you may want to consider a TX with more VA like 25VA or more and taller heat sinks.  if going with a single chassis build like the 12x12x3, a custom wound shielded/encapsulated TX from SumR is cheap insurance for another $40 or so.  Richard is a total pleasure to deal with A+++ customer service.

In my opinion, even good opamps don't save the Mini 3, it has some weird stuff happening with timbre and noise, as if there's some mess up in the topology.
The M3 on the other hand is excellent regardless the opamp choice.

for the B22, i would would recommend trying higher quiescent current bias.  you may want to give yourself some more TX power headroom and go with a slightly larger VA than recommended. 

the stock quiescent current is 120-160mA.  i upped it to 225mA and felt it sounded a bit too "tooby" and diffuse with uncontrolled, sloppy bass and softened note attack - but i really liked the midrange emphasis that both good and bad tube amps are known for.  i personally prefer 190-200mA.  you may not.  those that feel the B22 to be too reserved and analytical sounding i think would find it to be much less so with more midrange presence and a more forward soundstage with a slightly higher bias.  YMMV.

for a single chassis B22, like my own, i would strongly recommend a shielded/encapsulate TX from SumR and a ground-loop breaker.  my own B22 has been through various rebuilds, board configurations, and mods over the past 5 years, but is now in a 16x12x3" black painted alodine Par-Metal Series-20 enclosure.  2 boards, single O22, 100VA encapsulated/shielded SumR TX, stock 1.5" heat sinks, and Goldpoint SA.  ample ventilation with stock heat sinks for a higher quiescent current bias.  even with a gain of 5x, it's very quiet with the 102dB/mW sensitive AT AD2000 and dead silent through the full sweep of the SA with my other HPs. 

high impedance signal runs, RCA inputs to pot and pot to board inputs, should be shielded - twisted pair with shield with shield tied to input ground on the boards.  after dicking around with numerous wanky wire over the years, i've settled on Belden.  this shielded wire is a dream to work with and what Ti Kan recommends: http://www.mouser.com/Search/ProductDetail.aspx?R=8451_010100virtualkey56610000virtualkey566-8451-100 .  for power runs inside chassis and signal output wire i stick with 20awg like this:  https://edeskv2.belden.com/products/index.cfm?event=printPrev&pnum=8523&ut=English .  also available from Mouser.  for a 2 chassis build with an umbilical, do what TK recommends and go with a larger gauge wire for the umbilical.  even with an aluminum RFI-rejecting chassis, i've found it still pays to shield the high impedance signal runs.

the B22 doesn't like to have it's outputs shorted.  the TRS plug/jack will short a channel to ground very briefly when inserting and removing the plug.  it's best to treat it like an 18W speaker amplifier.  the O22/B22 has no current limiting circuit.  i've never had an issue, but you read that there have been some rare failures because the outputs are shorted.  i think this is largely due to people having the music cranked and/or using power hungry phones and not following the following the following practices.  it's basically a non-factor if you: turn down the volume all the way and pause the source before insertion/removal, or add a mute switch on the outputs.  the former is good practice no matter what amp you're using.  no output device whether it be MOSFET, BJT, or opamp likes to drive a dead short to ground.  also, an open-frame TRS jack allows you to adjust the pressure of the contacts by slightly bending them to tweak the force of insertion/removal to avoid slow/clumsy action that can cause a longer dead-short.