Samsung HD841 coupling caps

[duncan]

I ran into something a few days ago that puzzles me.

As I mentioned in my (rather windbaggy, sorry) introduction thread (http://www.audiokarma.org/forums/showthread.php?t=142536), I worked for some time on modifying a Samsung HD841 DVD/SACD player and ended up disappointed that I couldn't get it to sound as good as a cheaper, and more easily tweaked, Sony NS775V.

After posting that, I pulled the Samsung apart "one last time" and looked around. The analog out circuit uses two coupling caps per channel, one before the op-amps and one after. I'd measured a DC offset of something over 2 volts across the first cap and had chosen a good polar replacement (Nichicon Muse KZ). I hadn't thought much about it. But this time I replaced them with cheapo (15 cent) Panasonic SU bipolars, and was surprised to hear the player finally come to life and start singing.

My wife consented to blind A/B the two players and we found she too now likes the sound of the Samsung over the Sony. Didn't need any burn-in time or anything, it was just pretty obvious right away.

Thing is, it doesn't make sense to me. If there's a significant DC offset across a cap, shouldn't you use a polar there? :dunno:

I still hate the Samsung's user interface. urg...

By the way, if anyone wants to start tweaking on SACD players for cheap, both of these models are pretty good bargains on the used market right now - home theater people seem to be dumping them for the later and greater video features, which I for one don't care at all about. :thmbsp:

[cetoole]

Probably more to do with the capacitors than the application here, but you didnt say what values (capacitance and voltage) you used. Funny though, as I have read reasonably good reviews of the Muse KZ and bad ones of the Panasonic SU. Maybe try Muse ES, which are bipolar also, and quite inexpensive. I use these in my headphone amp for output coupling (Millett Hybrid Max), though bypassed with PIOs, and like the sound I get from it. Also, in these small values, Black Gate NX isnt unreasonably expensive. You probably would pay slightly more for slow shipping than two suitable caps. Also, while it is true that the DC voltage is fixed, the AC isnt.

Below is a picture of how I did up my Samsung HD841 a while back, including modified outputs, swapping capacitors in the power supply, and some damping. My favorite output is the Swenson modded one, which is those two largish axial capacitors. Those are both 8uf MKP caps I bought cheap through surplus, labeled "Ultracap". Probably not the end all in capacitors, but then again, neither is the TI DSD1608 DAC feeding them. The new outputs are directly wired into new RCA jacks I installed, and I used 200pf mica capacitors I had laying around to do a bit of low pass filtering. I also swapped the opamps buffering and filtering the L/R DAC outputs, both front and back (one dual opamp per side) with AD823, replaced the input coupling cap for the front L/R with an Elna RFS Silmic II from Digikey, 47uf 25v I believe, and shorted the output coupling caps for those channels, since, thanks to the bipolar supply, there wasnt really any DC offset here.

http://img.photobucket.com/albums/v323/cetoole/modded/ModdedHD841.jpg

[duncan]

Thanks for posting the picture.

The original caps were 22uf polar, both before and after the opamps. What I'd first used in their place was a pair of 33uf/25v KZ before and 33uf/6.3v(?) Panasonic SU after. The big improvement came when switching to identical Panasonics on both sides of the opamp. (OPA2107)

I haven't been willing to do the Swenson because of WAF frankly - it's not just my system, and I don't want to skip the mute circuit and then have to tell the missus you have to remember to keep the volume down when you turn this thing on so as to avoid innocently whacking our speakers. I might revisit the unit sometime, try other coupling caps or do a fresh mod a different way on the 5.1 fronts. But we are enjoying it as it is for now.

Edit - I will have to check the DC across the output coupling caps; hadn't occured to me to bypass them as you did.

[cetoole]

I know what you mean about the Swenson mod not being the most friendly setup. It definitely does make things more difficult. I dont make it a point to turn the volume down before powering the HD841 up or down, but there is a quite noticeable thump. The unit's auto power down "feature" doesnt help matters.

I have given some thought to replacing the mod with a JFET or at least JFET input buffer that is DC stable, with an AC coupled input. It would be simple to do, allow the cap to be downsized to ~0.1uf, which means I could use some of those K72 russian caps I have in my closet, or Vitamin Q, and build a new muting circuit without worrying about damaging the DAC. I would hesitate to short the DAC voltage output to ground, even with some series resistance, but even a simple buffer can be made that this will be safe. I just won a lot of 25 reed relays on ebay, which I will be using for muting circuits in the future. Add enough series resistance to limit the short circuit current to a save amount, which shouldnt be too much resistance, and probably could be put in series with the relay. Doing this would give much lower output impedance, but also make the mute not 100%. You could also make the muting relay series to the signal.

One of the buffer circuits I have considered that seems like a good candidate is a modification of AMB's JISBOS circuit, biased much lower and using small signal output transistors, to be more appropriate for driving a kohm range load than headphones. Also should add a LPF at the input. You can see it here: http://www.diamondstar.de/jisbos/jisbos_design.html. Also, one could use a dual matched JFET, one half configured as a source follower, the other as a CCS biasing it, and have equal source resistors. Linear Systems' LSK389 would be as good a choice as anything for this. Higher output impedance, but a much simpler circuit. Erno Borbely discusses several topologies for simple JFET buffers in some of the articles he has on his website.