Always. I'm still a direct couple "purist" because even overdriven I don't trust OTs due to the complex reactive traits it inherent to it, it's just a big mess of potential imperfections I don't want to deal with, not that it isn't good sounding. Same with caps, I use duelund cast caps anytime I use caps at all because I still have them from my original MK6 mod. I'd rather just not wonder if the caps are making a difference by either using the best or none at all, so far in my experiences caps have had a significant effect on the sound whenever I try a cap somewhere of lesser quality, not that my results are conclusive to all circumstances, I'll eventually do some tests when I get around to it. The massive improvement in sound I heard when upgrading to duelunds in my MK6 is the reason I became interested in electronic design to begin with so it left quite an impression. As for direct drive producing more distortion then OT, it really just depends on the design.For a standard source follower, yeah the distortion would be considerable depending on the load. I've spent most of my time designing direct drive output stages because I'd rather have the input stage producing my distortion and after endless hours in spice I've got a handful that should in theory produce next to 0 distortion, like between -120db and -200 db peak distortion according to spice depending on the design. No matter what parasitics I introduce or components I use or even changing component values to a large extent in spice the results are basically the same meaning the designs are solid, some to a higher extent then others. I've only built two of these types of designs so far but they sound flawless, I would need a distortion analyzer to confirm the actual THD but even if it was 10,000 times worse then spice suggests it would still be above average performance. (one of the designs stubbornly registers as 0% distortion in spice no matter what I do to it, a good sign) It's all about error correction via non-negative feedback schemes, NFB is a subtractive process and depending on how it's used it can degrade the sound, not a variable I want to add into the mix. However feedforward and other error correction schemes can let you have your cake and eat it too, there's many ways to approach this and when done right the performance is amazing. At the end of the day I think distortion is a good thing and very desirable but I don't want untamed distortion to dominate my designs. After I get the time/money to test all my current distortion free output stage designs I'm going to switch things around and look into distortive output stages and distortion free input stages, I hear that's a popular concept in japan right now. I can see the appeal due to the rising harmonic content with power of tubes under load which is the opposite of how solid state parts function and in line with how sound through air functions, this is likely one of the big reasons why tubes sound so natural over many solid state designs. I'm not convinced. I too use mu followers for the outputs of my input stage most of the time, it's just sensible all around, constant current on AC, constant voltage on DC, low output impedance. Low output impedance meaning able to drive reasonable loads without loading down the tube at all. Pair this with the fact that my output stages have practically no input capacitances and no current draw and you have a tube that is purely amplifying voltage under ideal conditions, I've heard differences in sound with different tubes here in their linear regions. This alone is not conclusive evidence but I've read too many blogs or threads from people with amps using hybrid mu followers and different tubes that all have the similar ultra linear looking curves on the scope yet having their own distinct sound of their own. On the other hand MrCurwen is the only person I've seen claim otherwise. I'm putting my hat in with the majority so far. Now that I finally have a solid prototyping set-up I'll be able to do a lot more testing later this summer when I get the time.