The FPGA verifies the checksum after a load and won’t run if it’s wrong. I can’t tell you what goes wrong with the copy of the data from the SD card into the display processor’s flash ram. Historically (not necessarily in your case) most loads were corrupted on the SD (or USB) drive often caused by the user not using the OS’s eject function. Some more loads were corrupted by crossed files in the file system. We added more checksums to the files to help detect these kind of problems, but I can’t tell you what happens when those checksums fail.
It’s not clear that it’s a failed FPGA load that is the root cause of bad sounding updates. It is the case that rarely the power has to be removed after an OS update to get the sound to be good. That’s probably due to the display processor ram not being initialized the same in a power up situation as a reboot situation.
I’m not trying to shoot down all of the various suggestions as to where the failure is. I’m just saying that historically there were multiple failure modes and now there are fewer of them. Where PS Audio is in working on a better update procedure I don’t know.
Hi Ted, checksums are not actually that reliable as they don’t record which bit pattern is in which memory location CRCs do. I used to write programming algorithms for PALs / PLAs back in the 80/90s and devices would oftain pass checksums e.g. byte 1 might have 20 hex and byte 2 byte have 30, when I fact it should be other way. CRCs take into account memory order.
I was using checksum as a loose generic word, there are multiple levels of (32 bit) CRCs in the FPGA (checked in hardware) and we used a (different) CRC on the installation files proper.
Ah, right, good oh, as you say, tricky to tie down. I don’t know anything about programming modern chips unfortunately, apart from they’re much easier to program
I have had two units to test recently. The second because the courier managed to ding the first DSD Sr unit so I was sent a replacement.
Both have been tested against a Benchmark DAC 3. Both are running V1. The first I updated the bridge with Mconnect after installing Snowmass V1. The second unit I updated the bridge over the network and then did the Snowmass V1 installation
.
I rigorously check that volume is at 100 for all playback. The attenuator is off.
I am pleased to report that both units sound the same. The second unit seemed broken in whilst the first unit sounded like brand new capacitors do and needed a few days to settle. I am familiar with the change in capacitor sound (for capacitors in the signal path) in the first few days as I have replaced capacitors on amplifiers in the past. You need to know what to listen for.
Difference vs reference Benchmark DAC 3.
Obvious differences: Thinner in the lower mid range (I prefer) Treble is 1db lower (no big deal just an observation). Bass is similarly resolving and quite remarkable with both DACs.
Subtle differences: Smoother overall with less attention to transients (less punchy or dynamic). Deeper soundstage as a result. A more analog or vinyl presentation (which just confirms what everybody says about this DAC). The dynamic balance and overall presentation is superb. Although the level of detail presented by both DACs is similar, the superb balance of the DSD Sr allows for more detail and timbre to be heard (as less attention is drawn to punchy transients, sibilance etc.)
I haven’t bothered with Snowmass V2. I will wait to see how this issue develops.
Both DACs are Stereophile Class A+. Both DACs are excellent. I would say that if you want a more vinyl analog style presentation then the DSD Sr has the edge with a row 5-10 presentation. If you want more dynamics and punch and excitement then the Benchmark DAC 3 is more front row.
Wow. The Benchmark DAC 3 sounded quite plain and uninvolving to me. Almost antiseptic and digital. I can think of several DACs I would rather listen to. The top of the list would be the DirectStream but would also include the Schiit Yggdrasil and Gungnir Multibit, the PS Audio DirectStream Jr., and the Metrum Pavane.
Thanks Ted! That was the repsonse I was looking for. I’ve been 100% digital pretty much since I got the DS when it first came out. I got an NPC to digitize vinyl so I could listen to LPs without a preamp, which I did for a while but not for the last year or more. I now have a BHK pre and have planned on hooking up the turntable via the analog outputs of the NPC, but I’ve swapped things around and don’t have a good spot for the turntable at the moment. You have given me a new reason to get to it, i.e., to help with tinnitus (maybe). I wonder if all those die hard vinylphiles do it for their tinnitus? Nah, probably not …
That some folks will feel very strongly one way or the other is of no surprise. It doesn’t make one DAC massively superior to another. They are both Class A+ Stereophile recommended components. Both are good. I have the luxury of both as well as the built-in DAC (9018 ESS chip) in my pre-amp. None of my DACs are crap. All are actually very good. I don’t hear anything particularly wrong with any of them. (Unlike MQA which just sounds wrong)
That said, currently I am really enjoying the DSD Sr!!!
Here is a post from another thread in which the user prefers Snowmass V1 to V2, writing:
In the lucky camp that Snowmass v1 worked perfect for me and my DS/ Bridge II/ Room setup, Must admit with V2 was always waiting for burn in to make it sound better.
Immediately on V1 reinstall the goosebumps are back.
That analog sound is exactly what I love about the Directstream Sr. I had on hand an NAD M51 DAC for about the last 12 months which was also an A+ Stereophile DAC. For the price, the NAD M51 is amazing. Though I think NAD just recently discontinued the model. Just like the Directstream, the M51 has a unique way to do the D to A conversion. Similar to the observations you had with the Benchmark DAC 3, the M51 sounds more digital to my ears. It’s more forward giving a greater sense of detail. But that forwardness in the treble is fake detail. or at least that’s how I would describe it, as that forwardness tries to trick you into thinking there’s more detail than there actually is. Sibilance, or more of it, is what I look for to tell me if a component is trying to trick me into thinking it’s giving more detail. I find excessive sibilance to be a good indicator of this. So while it may sound more “dynamic” because of this, that’s not my cup of tea. The Directstream sounds wholly organic, natural and effortless in it’s presentation. The rest of the components I’ve chosen for my system do the same thing because that’s the kind of sound I want to hear. I can listen for hours on end (and I often do) without any listener fatigue.
So, after coming back from work today and spinning some virtual disks, the sound is dramatically better than it was yesterday! The treble is much reduced back to normal levels, such I may rake my Martin Logan’s forward again to how they were before. The upper bass seems fuller (possibly an effect of normalised treble) and no longer sounds thin.
I have performed no reboots to acheive this, so I don’t think the issue is due to failures in the flash ram to program correctly, or copy to the fpga, but the update process in general upsets the audio quality for a long time, but suddenly clears itself out in a non-gradual way.
I finally tried going back to Snowmass V1 after much skepticism reading this thread.
Surprisingly, I prefer it to Snowmass V2 for many of the same reasons cited here - mids were fuller, the treble was less sibilant - and overall, the musical flow was that much better.
Prior to this I was dissatisfied with V2 and had attempted some speaker positionning adjustments to no real avail. The sound seemed more analytical.
It isn’t expectation bias as I had expected them to sound the same - logic dictates so.
It may be more of the Furutech GTX-D R NCF outlet that I have the distributor on breaking in, though I have had it now for a while.
But whatever it is, I am sticking with Snowmass V1 rather than futzing around and going back to V2.
For the record, I do not use a bridge and have never had any issues installing the firmware. Rolling back to V1 was no exception. I have not had an issue with the DS waking from standby randomly either.
My preceding post originally contained a pasted link to Paul’s December 19th announcement of SMv3 (v3.05) to try to help people find it, but the link disappeared. Maybe if I embed it HERE it will remain accessible. SMv3 appears to be a minor bug-fix to SMv2, so I am guessing it is not going to sound any different than SMv2 (but then again, SMv2 was supposed to sound identical to SMv1).