Sonic Frontiers SFD-1mk2 Vintage DAC for $750 That Stomps Most Modern DACs

I've only run the DAC via XLR out due to my preamp only having XLRs. I did use the DAC's RCA out for direct recording to a tape deck. Tapes came out very good.

 

I asked the same thing to Chris Johnson the designer and he said that it should sound same because SFD-1 only has 1 Ultra Analog chip and is single ended in operation compared to SFD-2 which is fully differential.

 

I always have trouble reading these color codes. In theory, there is a different spacing between bands to tell you which end has the first band, but so often the band spacing is equal. 3.7G seems really high. Could it be 127 Ohm, 0.1%, 15 ppm reading from the other end?

BYW, amazing zoom on those shots.

Jac

 

Yes, that is the other alternative. In looking at the color bands more carefully, it looks like the brown band is slightly closer to the red band (as compared to the orange band is to the purple band). So 127 Ohm seems likely to be correct.

 

Hello, I'm rather new here but I recently acquired an SFD-1 mkII myself and came across this thread. I'm specifically curious about bypassing the 6dB attenuation done by the PMD100. From what I've seen it involves shorting pin 19 to 22 in order to feed pin 19 the +5 volts (High in TTL) to switch the attenuation off. Has anyone here attempted this mod? I want to know if there is more to it than just shorting the pins like desoldering another pin or changing out a component.

Silver

 

Can you provide links to where you saw this?

 

Isn't the 6db attenuation only used for HDCD decoding?

 

This link mentions the chip specifically in the "upgrade" section of the page.
https://www.soundstagenetwork.com/wrkman07.htm

The manual for the PMD100 mentions how pin 19 affects how attenuation is done on the chip. A low input (0V) tells the chip to do -6dB attenuation digitally for non-HDCD content. A high input (+5VDC) tells the chip not to do this, which causes another pin designated for gain to output a signal when HDCD content is played (so the dac in question can apply 6dB in the analog section). The latter part is unimportant since HDCD is mostly a dead format, and I just want the unit to play music at normal levels all the time (the levels the SFD says it outputs for only HDCD). I've seen a few pictures online of wires shorting pins 19 to 22 as well.
http://tech.juaneda.com/download/PMD100.pdf - PDF of manual, pg. 4 shows the pinout. pay attention to pin 19 (SCAL) and 22

The above manual mentions that the attenuation is for non-HDCD content. This is to ensure HDCD content is of the same nominal levels as normal CDs. The reserved top 1-bit for HDCD dynamic peaks is unnecessary for modern usage, and bypassing it should make the SFD-1 mk2 operate at close to standard voltages.

Silver

 

Thanks, Silver. Its good information.

When I first got my SFD-1, I worried about the low output, but then, over time, I completely forgot about it. Given the age of these dacs and the unavailability of the chips if something went wrong, you might consider living with it for a while rather than take on this tweak.

By the way, make sure to roll some tubes. They make a significant difference in sound quality.

Jac

 

Just took delivery of my SFD-1 MkII SE+. Unfortunately, the input board was dislodged during shipping and the COAX- lead broke at the solder joint (photo taken after I pushed the board back into place and secured the optical connector to the rear faceplate).



It works if I hold the wire down onto the contact, so everything else is fine. I started using a solder sucker to try and clean out the via, but it's pretty hard going. Grabbing the wire underneath with pliers or tweezers immediately pulls heat away and the joint cools down. This is the best I could do taking a photo of the underside of the PCB after a few tries with the iron and sucker. There isn't a lot of room to work under there.



I remembered I had some luck repairing the pot on my MCTH by using tweezers to push old wire through the via while heating with the iron, so I'll give that a go after work tonight. Worse case, I should be able to get a decent connection just soldering the wire to the "pad", rather than having it pushed properly through the via.

Does anyone have any tips or caveats they can share?

 

Use flux and flow some solder on the stuck wire and then you should be able to gently pull out the wire.

 

Agreed. When I updated my SFD1 recently, I found the original solder on the PCBs hard to heat and manipulate as compared to Cardas Eutectic. I had to increase temp on my iron and keep it there a bit longer than typical. Use of new solder/flux helped. Good luck.

 

I already rolled a matched pair of Telefunken ECC88s. I had the pair sitting around for a while, very low hours on my old Mjolnir 2 that I sold a while back. I gotta say, they bring out just how good the dac is. I will for sure explore more options because color can be enjoyable, but these probably show me the closest to what the SFD would sound like if it was solid state.

The seller shipped them with a pair of Sovtek 6922 and I did give those a listen before swapping. Warm and fuzzy mids and sweet treble. This unit definitely takes well to tube rolling which excites me for messing with it's tone at some point.

Austin

 

Imma noob to vintage, so the lesson I have learned is: be gentle with old stuff.

I used some flux and flowed some solder on top. It melted more easily, and I was able to start pulling some of the old wire out the bottom. But the top of the via came off. Hmmm ... I kept going at it with the iron and when I thought the solder was melted, I pulled downwards on the wire with the needle-nose, possibly (no, definitely) a bit too firmly. The inside of the via and connected trace came away, so it was hanging below the PCB. Double hmmm ....I trimmed off most of the wire still soldered to the via with a side cutter. Luckily, the trace/via was kinda spring loaded upwards. So I tinned the wire, stuck it through the hole, and was able to solder it to the end of the trace. Plugged everything in and it works. Not pretty if you look really closely (and know where to look), but it's working so I'm happy. I'll throw it in a carry-on next time I fly through Toronto (wtf knows when that will be) and have Parts ConneXion put in a new input board.

So initial impressions with AES:

• Wider, deeper stage
• Laid back presentation (overall feel, not FR), yet also articulate and engaging
• Not particularly resolving, but the music is all kind of "there"
• Cymbals a bit tizzy on some tracks
• S/PDIF sounds thinner in comparison
  
• FR is possibly a bit warmer than BF2, but it might just be the open feeling giving that impression
  ... new toy syndrome and all that, YVMV

I'm running @Xecuter's old Valvos that came with it. Might pop in the pair of EH 6922 I have on hand later after I get a proper feel for the sound to see what changes occur.

I tried some HDCD rips I found in my collection, but the light doesn't come on. foobar2k detects them so I know the files are encoded. The HDCD light comes on at power up, so I know it works. I guess HDCD isn't working on my unit. Not a big deal. It's definitely quieter than my BF2, so that 6dB bypass @silverbot01 mentioned above is definitely of interest.

Something else a bit strange. I made up an AES XLR cable and connected both AES and S/PDIF at the same time out of my DEQ2496 using this impedance matching splitter. With only the S/PDIF coax cable connected, everything works fine. When I have both connected, AES is fine but I have to put some sideways/inwards pressure on the coax connector for it to work, almost like proper contact isn't being made. But the coax works fine when plugged in on its own, so it's weird. Not a big deal in the end, as I'll be feeding it AES and the coax will go to the BF2. Curious though.

 

@gixxerwimp, For HDCD make sure that the digital data doesn't get altered at all. The bits must remain the same for the HDCD to be decoded. If I apply any amount of digital attenuation to my PMD100 based DAC the HDCD will be defeated, as it becomes a different data stream.

 

Let us know if you contact Parts ConneXion about the availability of a new board. I'm guessing that they don't have any new boards, but could "fix" it in a way similar to your effort.

Nothing wrong with your fix, by the way. The only risk is that the trace, if not down on the board, may not be supported and vibration could cause it to break in the future. A future fix, if it should break again, is to follow the trace to the next solder pad on the board, remove the wire from it's current location, and solder the wire piggy-back to the the next solder pad.

If a new board wasn't available and you wanted a total fix, the board isn't that complicated. It would be easy to recreate the design in something like KiCad and have Osh Park or equivalent make you a new board.

 

Doh! Of course, it seems totally obvious now that the attenuation and EQ applied by the DEQ2496 would wipe the the HDCD encoding from the stream. Thanks for pointing this out.

The trace is about 2mm wide and fairly thick. I was the complete opposite of gentle when I pulled it off, and I don't think there's much chance of it breaking due to vibration over time. It's almost like a flat wire already. Thanks for the ideas on possible future fixes though. I doubt I'll be contacting PC any time soon, but I'll definitely share any info I get.

I realized that I could try the two single Telefunken E88CCs I have for my MCTH, even though both are somewhat noisy. Turns out the noise is much less noticeable in the DAC. Initial impression is they sound smoother and more even overall, perhaps a bit more spacious. And the treble doesn't have the slight tizz of the Valvos. Channel balance is pretty good, even though they're two random singles.

 

I have the same experience regarding noise. Tubes that were noisy in other applications were quiet in the dac. I suspect that they are using the tubes as an IV conversion and possibly as a buffer. Both of those applications have low gain and the feedback probably controls noise. Just guessing here. I don't have any real circuit knowledge or expertise.

I tried a range of tubes including some Mazda ECC88 (said to sound similar to Amperex) and Siemens ECC88 (similar sound family to Valvo). My favorite ended up being some 6H23N NOS Russian tubes. A pair of Vokshod (rocket logo) 6h23n came in the box with the dac and have proved the best synergy with my system and the dac. They sound clean and neutral, but with a little liquid warmth. Great bass, too. The Reflektor version is very close and I have a pair of those as back up. I haven't tried them, but 6n1p Vokshod are said to be as good or better and are available cheap!

Ok. I have to ask. I have been noticing the saying in your signature for a few years. I know what a gixxer is. Are you willing to explain "professional tricycle rider"?

 

Just curious about your mention of the 6n1p--I was under the impression that these were dangerous to use in the SFD because of the larger current draw. Does anyone have experience actually using these successfully? I'll admit I'm too afraid to try.

 

Excellent Catch! I was not aware of that issue, but you are right, the datasheet shows basically twice the current. If it was something I built myself, I always overbuild the power supplies, it wouldn't be a problem. But on the SFD-1, you probably shouldn't take the chance.

And no, I haven't tried it.