Well since I clearly spent a fair amount of time searching and reading posts maybe you could just point us to the findings from time to time?....
I’m not sure if there’s a searchable answer to the question you’re referring to. It seems my problem is closest to bambadoo’s, but I really couldn’t find out if he ever decided which mod is the best. Maybe it’s just really hard. From what I can glean, the simulations run might be in favor of low res mod in terms of low frequency ripple size, but there could be doubt about its accuracy. And if some lowish ripple (that in all likelihood is in the later products) can improve or imbalance the sound by so much, what are we to think about the ripple caused by the shift register? Why wasn’t the “extreme mod” recommended to everyone? Also I read about a bunch of transistor mod but have no idea if they’re still relevant. I was under the impression that Soren implemented the best mods in later revs. Perhaps you’re the best person to answer some of these question?... Thanks a lot!
Last edited:
To be honest, I don't recall where every sub-discussion is located. So it would take me as much time as for you. Searching this thread is the best advice I can give.
Re the clock - a flash from a LED every time a change is ordered would be a good help to check if ones implementation is OK. I have suggested to Sören that a spontaneous printout on the serial would be as helpful - but no respons. Maybe he thinks it's a threat but I see it as providing the user with the means to get the most of his products - should help him. Trivial to implement.
//
Re the clock - a flash from a LED every time a change is ordered would be a good help to check if ones implementation is OK. I have suggested to Sören that a spontaneous printout on the serial would be as helpful - but no respons. Maybe he thinks it's a threat but I see it as providing the user with the means to get the most of his products - should help him. Trivial to implement.
//
Last edited:
These mods were around improving an opamp used as a regulator. I don't believe opamps are used in this manner in current versions.
It's fine... I guess you stopped working on your dam1021 after you got the dam1121s...
Someone followed up with the question of whether this would be a "universal recommendation" for newer revs but received no reply from Soren. I take it from the silence that it's more or less the case (though still not sure how important the BC transistor mod for the opamp regulator is).
Soren, can you share simulation results on the vref mod? Given that it's unlikely for new improvements to be discovered on the vref, the main question seems to be whether the low-res (0.01R rather than 0.1R) mod is worthwhile and how much low-ESR capacitance is needed. The fact that we don't know and have conflicting advice on whether to mod rev.3/4/5 should be concerning to anyone still considering dam1021 (and perhaps dam1121). An open discussion on the best mods, however, would put the worries to rest as we understand what can be reasonably accomplished with some efforts from the users. The vias you left us make things very easy if the right caps are sourced and there are plenty of support here on performing the mod for varying experience levels. I know you have other priorities, but there are substantial and justified concerns left open by the earlier simulations and measurements from Paul and others. I hope that in the end, we can all enjoy the sound of dam1021 with the knowledge that we've performed the reasonable mods needed for this wonderful dac to perform at its potential - not suboptimally, both in theory and practice.
Last edited:
ynmichael, perhaps i am mistaken but i think your questions about resistor values also refer to the old opamp based regulator circuit. From what i understand the current dacs use low noise ic regulators and not zener referenced opamps.
The reason you read a lot more about improving the 1021 is the newer dacs still don't have a similar market penetration and also, presumably, because they offer a respectable performance without any mods.
The reason you read a lot more about improving the 1021 is the newer dacs still don't have a similar market penetration and also, presumably, because they offer a respectable performance without any mods.
So the low-res mod simulations no longer apply to rev.4/5?... Are there any details on the vref circuit changes in this thread? I thought only the SMD resistors and caps were changed in rev.2 for the vref circuit, rest of the opamp 4V circuit kept intact, and more SMD output caps added in rev.3+.
Thanks for refreshing the community's collective memory on this detail!
Let's hope for a reply from Soren! I can check the physical chips too
For the opamp change? I couldn't find anything on it... He also stopped posting late 2015 so maybe the newer changes (though I believe Soren didn't say anything about replacing 4V opamp with some low-noise IC...)
Surprisingly hard to take a good photo, but judging from the appearance I don't believe that the vref circuit has been completely replaced from rev.1.
Also did some more searches in the thread; it seems serious discussion on vref abruptly ended around page 300-350. Notably, DIYBras proposed a number of interesting improvements to vref but concluded, I believe, that Soren's factory mod was better for the elimination of peaks in vref impedance. zfe contributed the most complete set of measurements but none was based on the 0.01R low-res mod that Paul proposed. Soren, you said in the thread that you ran simulations before deciding on the vref improvements. If you don't have time to run more with Paul's low-res mod could you at least post the original simulation results? Thank you.
Also did some more searches in the thread; it seems serious discussion on vref abruptly ended around page 300-350. Notably, DIYBras proposed a number of interesting improvements to vref but concluded, I believe, that Soren's factory mod was better for the elimination of peaks in vref impedance. zfe contributed the most complete set of measurements but none was based on the 0.01R low-res mod that Paul proposed. Soren, you said in the thread that you ran simulations before deciding on the vref improvements. If you don't have time to run more with Paul's low-res mod could you at least post the original simulation results? Thank you.
Attachments
Last edited:
I'm selling finished boards which I believe are designed fine, you're welcome to modify it as you want to, but I will in general only comment if something is incorrect, or some modification may cause damage, or if a comment is for the greater good. Sorry, but I can't comments on all design details, or provide simulations files, or schematics, I have other things to do.
I understand, but based on the information we have it appears that the LVC595 can contribute 800uV of ripple, and rev2 stock vref 400+uV based on Paul's simulation. I understand that in rev.3+ you added 300uF per rail, but we have no data on how that will affect the system. Paul's further simulations are mostly based on his low-res mod. He justified the mod by showing a simulation of bambadoo's 4000uF + factory mod config which resulted in 408uV ripple at 100Hz. When low-res mod (0.01R) is added, it drops to 48uV which is very consistent with the ripple at higher frequencies. Like I mentioned last night, the LSB of a 4V 16bit ladder is 61uV. I hope these appear to be valid theoretical reasons to worry about stock performance even on later revisions. Empirically, we have at least three reports of significantly improved sound quality when more vref capacitance is added. So perhaps there is slightly more to be said if the goal is to get the system to sound as good as it reasonably can, and no one is suggesting batteries everywhere. There is plenty of reason to believe that some well-calculated and relatively simple mods can result in significant SQ improvements. This does not have to mean that the latest versions do not sound great.
We don't expect to understand the full circuitry and we certainly respect your time. I believe that most parts of the system are designed to near perfection which makes possible the great results that everyone is getting despite several inherent technological limitations in the system. But based on the theoretical and empirical evidence of the improvements to be gained in the vref circuit even in the newer versions, I cannot help but believe that there is some significant improvements to be had with simple mods (caps and an SMD resistor perhaps), both theoretical and sound-wise. All of this is a request from the user community for you to help point us in the right direction since you are the only person with the technical expertise and system knowledge to do so. If it were only a problem of capacitance, trial and error would be the best way to go, but things appear to be more complicated here to require the predictive power of simulations. Again, this is us asking for help...
Here's Paul's simulation result again for your reference: Vref musings: the final frontier? | moreDAMfilters This post-dates hifidurino's comprehensive review on the vref mods, and vref discussions here ended around page 350 (Sep. 15).
Last edited: