I wanted to post my listening notes for the capacitor options I've listened to over the past few weeks. I'm including the listening test I did for the last few days with the Nichicon 470uF output caps soldered in place instead of temporarily attached (I believe this cleaned up the signal a bit). I don't know why I ever bothered using the wires and gator clips. I found that with them removed and the main caps soldered in, it was much easier to just pressure fit the different bypass caps in place. I did this from the SMT side of the board, where there was tons of clearance. I was surprised, but the parts made good contact just placed in the holes. These impressions were based on listening with Sennheiser HD 598, which have an impedance of 50 Ohms.
1000uF OSCON - no bypass
360uF OSCON - no bypass
-less mids and bass
470uF Nichicon AK - no bypass
-sharper transient attacks, most transparency
-most treble details and midbass texture
-most balanced sound
---after soldering in the Nichicons---
470uF Nichicon AK - no bypass
-even treble throughout vocal region
-decent treble clarity, hint of graininess
-I wish there was a bit more HF extension
-tightest, most coherent overall sound
470uF Nichicon AK - 1uF WIMA MKS
-more treble, smoother
-best high treble of all combos tried
470uF Nichicon AK - 2.2uF WIMA MKS
-more high treble than no bypass, but less than 1uF bypass
-vocals full-bodied and pleasantly forward in mix
-still less tightness/coherence
Take it all with a grain of salt. These are all sighted comparisons. But after considerable back and forth sighted testing, my current favorite is NO bypass cap. Even though both bypasses brought forward the treble nicely, I found that they added a very subtle "something funny" or perhaps a null (?) in part of the vocal range. It almost sounded like a phase issue, where there is cancelling or disagreement rather than coherence. Maybe I was hearing impedance anomalies at the low corner frequencies of the bypasses...I don't know, or maybe I'm just distinguishing the "hand off" between electrolytic and polyester. I envy you guys with 250 ohm cans...that puts the corner frequencies of your bypass caps WAY lower, perhaps getting such issues out of the MOST sensitive range of human hearing, the 2-5kHz "vocal presence" range, which is where they occur for me with my 50 ohm cans.
OTHER BYPASS OPTIONS
Larger WIMA MKS
One option for me is to use a larger WIMA MKS. X already thought of this of course
Even the 6.8uF MKS (14mm tall) will fit in my tin with the way I mounted my board so low! I'm going to try 3.3uF, 4.7uF, and 6.8uF. 6.8uF would reach down to 470Hz! I don't think I would hear the subtle "hand off" down that low. Still, I am a tad skeptical about this strategy, since according to WIMA, the bigger the cap, the worse the dissipation factor at high frequencies.
See here: http://www.wima.com/EN/mks2.htm Hopefully this increased dissipation will be inaudible, but I think I can already hear more high treble from 1uF vs. 2.2uF, so we will see if this is a trade-off with higher capacitance. Have you tried any of the 4.7uF size yet, X?
Smaller WIMA MKS
Another option would be to push my bypass corner frequency above the delicate vocal region, closer to 10kHz. I could accomplish this with a .33uF bypass. I could still use a MKT but I'm looking at some other options too. WIMA themselves do not actually recommend MKS caps in the signal path for audio circuits, though I know that they are often used this way with success anyhow...see pg. 7: http://www.tawelectronics.com/wima/WIMA%20Audio.pdf
Smaller Vishay MKT
This 0.33uF Vishay MKT cap would fit. Still polyester. I don't know how these sound compared to WIMA MKS. If I'm reading the graph on its datasheet correctly, it has a dissipation of 0.01 @ 10 kHz, a smidgen better than the WIMA spec which is .015. But Vishay also provides more precise measurement data, so it's probably insignificant. Anyhow, this comparison ranks them a tad higher than MKS: http://www.humblehomemadehifi.com/Cap.html If fact, he actually mentions good coherence in his impressions. They are slightly different cap series though. He compared MKT1822 and MKS4, with wider lead spacings and perhaps other differences. I'll be comparing MKT370 and MKS2 series.
Smaller WIMA MKP
Hey, why stick with Polyester if I can fit a Polypropylene of the same .33 uF value? I think I can cram this WIMA MKP2 in there, but Mouser calls it a "RF Microwave Film Capacitor" and its datasheet does not mention bypass as an application, nor are they mentioned in the "WIMA for Audio" doc. Hmm. What's the difference between MKP2 and MPK4...just size I hope?
Smaller VISHAY MKP
This is a long shot for me, but I've drooled over the Vishay MKP polypropylene caps a bit, which are very highly recommended in the Humblehomemadehifi resource. Sadly, I think the only one I can fit would be this puny .12uF cap. Not sure if I'd hear anything at all, since the -3db point (corner frequency) with my 50 ohm cans would be way above human hearing range at 26kHz. Still, for folks with higher impedance headphones, this could be a worthwhile upgrade experiment. Corner frequency is down to 5kHz for 250 ohms.
Please share your thoughts if you have experience with any of these caps!
Thanks for the detailed listening impressions of the various setups. Be warned that not soldering in the bypass cap may not give proper results. The HF's being transmitted by the bypass depends on a good solder joint for low impedance. I am surprised you get anomalies with bypass installed and say it doesn't sound as good. This is probably due to improper connection. Think of it as the worst case of a cold solder joint! Bad for sound quality.
I tried the Silmic 100uF and 1uF Wima in controlled AB testing with an AB switch box and could not hear the difference - except for loss of a small amount of bass extension due to 100uF vs 390uF size.
I think you should solder one of the 1uF Wima's in place and listen again. Use the Nichicon AK that you like so much. It should sound balanced with nice mids and highs.
So for now - I am not sold on Silmics vs Oscons.