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

A word or two to the wise. If you are trying to keep an SFD-1 alive as long as possible and it still has the original electrolytic capacitors, now is the time to replace all of the electrolytic capacitors.

I started my upgrade after more than a year of procrastinating. When I removed the big power supply caps, I found 3 that had leaked. The worst one had eaten under the solder mask and gotten down to the bare copper. Luckily, I caught it in time and the copper was OK.

So don't wait. The wrong failure could be a big problem.

 

I am gonna be contrarian and go w sell the sfd-2 and get the sfd-1 upgraded... never, just the 4 big coupling caps. if you're handy with a soldering iron, it's quite ez to do on your own.

The sfd-1 excels at tone, midrange, big big convex soundstage, and... being kinda dirty sounding and it's my feeling—never having heard an upgraded one—the dirtiness would be a lil less w upgrades. I like me some lofi dirt tho... Kinda miss this dam stoopid DAC lol.

 

Yes, it does sound less dirty with upgrades. Can't confirm it is specifically the caps cuz I had PartsConexion do all the recommended upgrades but I imagine they're an important factor.

 

Curious what capacitors they used. I have had a few 25 plus years old Adcom dac and no one has this issue. A bit dry and raised esr may be but not leaked caps.

 

Which are these on the board?

 

I have a data point about sound of the SFD-1 mkII before and after upgrades for what it's worth. Last Fall I bought the unit that @RestoredSparda had up for sale and it already had the output caps upgraded from the Solens to Mundorfs. So I can't comment on what difference that makes and it could be significant.

However when I bought the unit it came with a box of PE approved parts ready to install. All new audio grade electrolytics, replacements for the Wima poly caps, some upgraded resistors, etc. I think this was to turn it into a legit SE version, or whatever would max it out. I listened to the DAC for about a month and really enjoyed it. I would agree with how @caute described the sound. I even thought it sounded good with the stock Sovtek tubes...until you try something better.

Then I began tearing it down to do the upgrades, life got busy with work. I invested in a desoldering station, and I finally finished replacing EVERY original cap and got it reassembled about a month ago. I can't say that when I powered it back up it was much different than I remember other than maybe it was a bit cleaner sounding. But the output cap replacement could be significant, I'll just never know. The main thing is that now I have a DAC that I know all the caps have been replaced and possibly good for another 30 years .

 

The caps appear to be Panasonic. They don't put their name on caps, but I recognize the funny little "M" symbol for Matsushita. The difference between 25 and 30 years could be significant. Also, most caps are rated for 2000 hours. At 30 years, 2000 hours is only 11 minutes per day "power on" time. I don't see this as a failure on the part of the cap, rather luck of the draw after so much time.

The caps that leaked were in the C201,C202 and C101,C107 pairs. Both are lower voltage sections of the dac power supply. They all have modern, rubbers seal bottoms.

One thing I wonder about it temperature. I have replaced all of the electrolytics and coupling caps and have the board outside of the case powered up and sitting on some feet to let everything settle in. I have been checking temperature with an infrared thermometer to make sure everything is OK. Everything is in an acceptable range, but things are definitely warm. The hottest area is around the tubes. The tubes themselves register 70 deg C and the heatsink between them is at 60 C.

Near the transformers, I notice that the caps are pretty close to heatsinks on multiple regulators and transistors. Those heatsinks run between 40 and 50 C. The top of the SFD-1 is well ventilated including slots over both the tubes and the power supply area, but the only air coming in the bottom is below the tubes and a long way from the power supply caps. Just a thought.

 

Ah, but just wait until you retire and have time to waste on making things betterer.

I'm doing my upgrade in 2 maybe 2.5 steps. I just finished step 1, replacing all of the power supply caps. Step 2 will be subjectively comparing output caps and the 0.5 step is trying a few alternative resistors in the tube/output section. Based on experience with other projects, I don't expect a lot of change from the resistors, but I do from the output caps. My dac came to me with the Mundorf Evo Oil output caps. I have a range of caps that will work, including a different size of the Jantzen Superior that @caute tried.

Now let just hope that the irreplaceable chips last that long.

 

@lehmanhill Did you ever get to swapping the output DC blocking caps? The default choise seems to be those Mundorf Evo Aluminums but there are lots of other choices out there ($$$) as well. I recall you said that capacitance value shouldn't really matter.

 

That's next. My SFD-1 came with the Mundorf Evo Oil. I'm only going to change the RCA out caps because I have more choices in the parts bin if I only use 2. Plus the XLR out on the SFD-1 isn't really better than the RCA IMHO. I've got some old Russian caps to try. Also Audyn, Mundorf Supreme, Miflex, Auracap, and the Jantzen I mentioned above. Outside of the Supremes, nothing really expensive, but a range of character for sure.

 

Noob question.

So if you replace 2/4 of those caps, rca vs xlr notwithstanding, and you happen to actually use the xlr out, part of the signal outputs through the two different set of caps right?

 

Yep, that's what would happen. I don't know if the sound would average between the two caps or if it would be just awful.

 

There is this upcoming DAC from Linear Tube Audio that looks interesting for us.

https://www.lineartubeaudio.com/products/aero-dac-digital-analog-converter

 

I'm pretty sure I would go the DIY route, but I've been looking into other commercial options. The Linear Tube looks interesting. I also found Lab12 from Greece that looks pretty good using obsolete Phillips chips in parallel with a tube output buffer.

https://www.lab12.gr/product/dac1-reference/

It's available in North America too.

https://fidelisav.com/products/dacs/dac1-reference

But I think I might be more likely to try the Audial S5. They have made some nice improvements since the earlier S dacs.

https://www.audialonline.com/s5/

 

I think most would agree that the SFD-1 with the Mundorf Evo Oil sounds pretty good with the right tube. I was curious if I could find an output cap that I liked better than the Mundorf. After all, cap rolling is similar to tube rolling in that you can get some useful changes in character and clarity. My SFD-1 came to me with the Mundorf Evo-Oil caps, so no Solen caps.

Here comes the typical weasel words, my system, my ears, and my mess between my ears, YMMV. I'm also a big believer in synergy and, in this case, that can be a bigger deal because there are synergies within the SFD-1 between the tube, output resistor, output cap, and the bypass cap in addition to system synergies. In all of these subjective evaluations, I used 6N23P Rockets tubes and Mills MRA-05 wire wound output resistors (R21, LR21). The Mills came with my SFD-1.

Bypass Cap

Before we get into alternate caps, lets talk about the bypass cap. You may have noticed the small oval cross-section caps running right next to the output caps. Those are bypass caps and run in parallel to the output caps. The idea was to use a smaller value cap to help out the higher frequencies, but in my experience this doesn't work in every case. There needs to be a synergy between the 2 caps to get an improvement with bypass caps. My first evaluation was to compare the Mundorf Evo Oil with and without the bypass cap. After all, the bypass cap was chosen for the Solen so there is no guarantee that it helps the Mundorf or any other cap. The differences were small, but I found the Mundorf without the bypass cap had a little more clarity and punch. No need for the bypass caps, so out they came. All further cap evaluations were without bypass cap.

As I mentioned earlier in this thread, these output caps are used to remove the DC voltage that is required to bias the tubes, so when combined with R19/R20, they are a simple hi-pass filter where the corner frequency of this filter with the 3.3 uF cap is just under 0.1 Hz. In my experience, this is a lower corner frequency than necessary. The reason this is relevant is that larger value caps are physically larger and harder to fit, plus more expensive. And the best reason of all is that I have a range of 1.0 uF caps I could try. The corner frequency of the filter with a 1.0 uF cap is about 0.3 Hz and I never heard any issues with bass, so I think 1.0 uF is fine. Actually, I often use 1.5 Hz as a corner frequency for a DC blocking filter which would mean a 0.22 uF if anyone wants to experiment.

I pulled 12 different caps out of the parts drawer with the idea of trying them, but most were too big or had leads that were too fat for the hole in the pcb. In the end, only 4 are worth talking about.

Getting solder out of the hole

As an aside, the SFD-1 pcb is thicker than most and that makes it harder to get the old solder out of the hole. The technique that I found worked was remove the cap, heat the remaining solder to molten, and use a wooden toothpick to push the solder out of the hole. Once the solder cools, gently remove the toothpick and use solder wick to soak up the little hill of solder you will find on the opposite side of the board.

Alternatives to the Mundorf



The Easy Choice – Auricap XO – 1 uF, 200Vdc

As a start, it is the only cap other than the Mundorf that fits horizontally on the board, although I suggest putting a bit of Blue-tack or similar to locate the cap body on the pcb.

The sound of the Auricap XO is clearly better than the Mundorf. You don't notice it until comparing it to a better cap, but the Mundorf has a little bit of haze between you and the music. With the XO, the haze is gone and the clarity is much improved without causing the sound to be overly detailed or harsh. I find the XO to be a very neutral sounding cap. The soundstage is bigger and especially deeper. Instrument location is more stable and separated. Bass is both deeper and remains tight. The Mundorf isn't awful for dynamics, but the XO is little better there too. Considering it's about the same price as the Mundorf, there is really no negative if the choice is between those two caps. I would pick the XO everytime.



At the time of writing this the Auricap XO, 1 uF, 200Vdc cap were available at:

Partsconnexion – USD 16.82 each

hificollective – Euro 21.33 each

The other 3 caps have issues fitting on the pcb, but they can be handled with a little extra work. When mounting horizontally, they have to be above the board with leads looping back to the holes. Or you can mount them vertically (soldier style). Caps are expensive, so I am only changing the output that connects to RCA output. The balanced output of the SFD-1 isn't that great anyway, IMHO.

Jantzen Superior Z – 1 uF, 800VDC

The Jantzen is also better than the Mundorf for clarity and it is even more dynamic than the XO. But the Jantzen is a little bit too strong in high-mid and treble frequencies. That makes the sound a little uneven and sometimes a touch harsh in my system. They might add a touch of clarity to a system that was a little on the warm side. And the difference to the XO isn't so great that I would remove the Jantzen to replace with the XO. But if I didn't already have the Jantzen, I would choose the XO as the more neutral cap.

At the time of writing this Jantzen Superior Z, 1uF, 800 VDC were available at:

Partsconnexion – USD 16.25 each

Hificollective – Euro 13.46 each

The Budget Choice – Cornell Dubilier 940 - 1 uF, 600 VDC

A big caveat here. I only had the 850 VDC version of this cap and the evaluation below is based on that cap. That said, the 600 VDC version is available, is slightly smaller, and has a 1.0 mm lead diameter that will fit in the hole in the pcb. The big assumption here is that the 600 VDC version would sound the same or very close.

This cap also makes the haze go away. In fact, it has the best clarity of any of the caps in this evaluation. On the other hand, it is less dynamic, both macro and micro, than the other caps. I would personally choose this cap over the Jantzen for it's lack of harshness, but I think the XO is worth the extra money and would only choose the Cornell Dubilier if I needed the lower price.

At the time of writing this Cornell Dubilier 940, 1 uF, 600 VDC were available at:

Mouser (US) - $6.02 each

Cost No Object – VHAudio V-cap OIMP – 1 uF, 250 VDC

At more than twice the price of XO, they better sound good. But it's worse than that. The current price is 50% off because this model is discontinued and being superseded by a newer model. The new model 1 uF is almost 6x the XO price. The new version is the ODAM series which might stand for “Oh Damn, that's expensive.”



That said, the OIMP is good enough that I put them in my SFD-1 and ordered a few more to try in another project. It does all the stuff the XO and others do, excellent clarity, big soundstage with a touch more depth, clear and stable instrument location, neutral but almost liquid sound, and tight deep bass.

The thing that is special is the layered sound. I have heard other people talk about layered sound and never really understood what they meant, yet that was the first thing that came to mind when I listened to the OIMP. To try to explain, it means a very good dynamic range or put another way, you can hear the big loud passages without distortion, but you can still hear the quiet stuff like a finger sliding on a guitar string or the in-breath of a singer and those quiet passages have their own loud and soft notes too. I can't really give the OIMP all the credit. The SFD-1 provides the basis for it all and the Rocket tubes are probably the source of this magic, but the OIMP uncover that like opening Newt Scamander's briefcase.

At the time of writing this V-cap OIMP, 1 uF, 250 VDC were available at:

VHAudio - USD 38.99 each

https://www.vhaudio.com/v-cap.html

Mounting the Jantzen, Cornell Dubilier 940, or OIMP caps



Feel free to mount these anyway you like. I chose soldier style mounting with the cap body nearer to the output side for more space between the cap and those hot tubes and heatsink. I moved LC20 to the back side of the board so the output cap would fit. The OIMP have a ferrule on the lead that sticks out, so a spacer was used to keep the ferrule above the pcb.

Output Resistors

My SFD-1 came with Mills MRA-05 wirewound resistors in the R21 and LR21 positions. I bought the Amtrans AMRG resistors that had been used in SE updates and evaluated them in comparison to the Mills. I found the Amtrans to be a little softer and slightly veiled compared to the Mills, so I kept the Mills. This could easily be a system level synergy issue, so it may not apply to other systems.

I hope this is helpful. I'm glad to have my SFD-1 back together so I can just enjoy it.