Sonic Frontiers SFD-1 MK2 Vintage DAC Measurements

Read all about this DAC here: https://www.superbestaudiofriends.o...dac-for-750-that-stomps-most-modern-dacs.508/

I'm happy I now have modern instruments to measure this DAC, just out of curiosity. To summarize, I once owned this DAC, sold it, and regretted doing so. @Carlos CPA mentioned he was going to put a souped up one for sale at special member prices when he saw me mentioning my regret. I thought the price he was offering was too low, so I threw in a little extra. (Point is that we have a great community here. We have members who care about each other to the point of things not really being about profit or greed, keeping things in the community to deserving people who as passionate about stuff).



Anyway, this Sonic Frontiers SFD-1 MK2 looks to have the level 99 upgrade treatment based on the BOM from
https://partsconnexion.com/. For those who do not know, Chris Johnson who started Sonic Frontiers decade ago now heads up pcx. This specific unit sounds more resolving, more neutral, and less syrupy than the one I owned before which only had the level 1 upgrade and possibly an older iteration of the UltraAnalog DAC module.

The UltraAnalog modules were a thing back in the day, being in competition to various R2R ladder chips from BB. Those who heard the Moth DAC at the California meets years ago may be familiar with its core sound. I know a few others have purchased DACs with UltraAnalog modules. Some big names in pro gear back in the day used them too. I believe UltraAnalog glued together some off the shelf chipses to get "20-bit" performance. The module output voltage.

 

Here be a photo of the module. I don't know if the -C025 means anything.


I found the datasheet and have attached it below as well. The SFD-1 has a tube analog output (6DJ8 type) stage, so it will be unlikely to meet any published specs. Still, I am curious.

Note the Pacific Microsonics PMD100 digital receiver. This receiver had a cult following not too different from the TDA1541, although nowhere as popular because it's a digital receiver chip, not a DAC chip. The PMD100 supported HDCD, which encoded extra bits up to 20 in maybe the last few LSB (not sure, taking a guess) while maintaining compatibility with ordinary CD / Redbook (in this case some noise). I have a few HDCDs, but when it became clear that HDCD was going nowhere (think of the now defunct MQA as HDCD part deux), Microsoft bought them. In typical Microsoft fashion, they had no fucking clue what to do with it. I wonder who was in charge of HDCD at Microsoft at that time? ;-)

 

Sonic Frontiers SFD-1 MK2
0dbFS
1kHz 24-bit, no-dither, AES input
AP HPSA cheat mode ON
Balanced Output into 200k-ohms



Actually, not bad! Quite outstanding considering tube output and era (1990s). Let's take a look at -30, -60, and -90dbFS next.

 

-30dbFS


-60dbFS

 

Sonic Frontiers SFD-1 MK2
-90.3db (flipping that one LSB)
16-bit out from AP

This is a really good result for 16-bit, especially back in the day where most CDPs and DACs couldn't even make a semblance of a stair step.


For reference, this is the result from the Rega Planet CDP that I owned back in the day that came out years later toward the late 90s. Credit: https://www.stereophile.com/content/rega-planet-cd-player-measurements

 

Now let's take a look at 24-bit performance. 1000Hz signal. -90dbFS


Not quite 20-bit performance as advertised. Both the PMD100 and UltraAnalog datasheets say 24-bit and 20-bit. At 20-bits we should expect to something closer to a sine wave instead of one more steps to the stairs!

However, I bumped into this by accident. The bits increase as we go up in frequency! Sometimes a simple suite of four tests / graphs doesn't get us everything! Real science is about discovery, not applying litmus tests / Excel bar graphs.

24-bits. 3000Hz signal. -90dbFS


24-bits. 5000Hz signal. -90dbFS


EDIT: I realized, it's probably the analog reconstruction / anti-aliasing filter as we move up in frequency.

 

Here is the "linearity" test using 16-bits and 1kHz. Actually, it's the bandpass level sweep with 1/24 selectivity at the generator frequency. Can confirm that we are indeed getting the full 16-bits, but DAC accuracy about +/- 1db off toward the last LSB. Subjectively, the SFD-1 doesn't resolve fine detail like more modern DACs

 

THD ratio vs level
at three different frequencies
90kHz bandwidth


Very Important Note: results will vary depending upon tubes.

 

1kHz at high bandwidth (500kHz)


Checking higher frequencies... (given the experience with this situation: https://www.superbestaudiofriends.o...ue-science-and-exploration.13584/#post-411401)

10kHz at high bandwidth (500kHz)

 

Output at higher levels has more distortion in the highs.

 

Let's take a look at white noise, rolloff, and analog/digital filter performance. Measurement is 500kHz bandwidth.


White noise again, but 44.1kHz. Basically flat to 20kHz with sharp rolloff right after. No softpoo early rolloff here. Note squiggles because this is averaged noise.

 

For the 0.1db peepers:

Sonic Frontiers SFD-1 MK2
Used Telefunken EC88 tubes
Frequency Response
Balanced Output
0dbFS

 

I do, ripped from CDs. But no measurement / test signal type tracks.