How many hours of play time so far? Fully broken in already?
Have you located the low pass filter at output? In the 1021 removing it led to the biggest improvement in perceived highs. If for some reason you absolutely need that filter try replacing the horror ceramics with something decent.
The ceramic cap was the first thing I kicked out, reminded me of these same terrible ceramic disc caps at the output of countless CD players to filter some frequencies in bat regions.
I put some good Siemens film caps now directly at the output, not sure if it does any good or bad now.
Played maybe around hundred hours or so, probably still breaking in.
I had the idea to try some Elna Silmic at Vref, they are available at 6.3 V/1000µF. They have the characteristics the Dam is lacking, but maybe they are too "slow".
I put some good Siemens film caps now directly at the output, not sure if it does any good or bad now.
Played maybe around hundred hours or so, probably still breaking in.
I had the idea to try some Elna Silmic at Vref, they are available at 6.3 V/1000µF. They have the characteristics the Dam is lacking, but maybe they are too "slow".
I have a question, I have the latest version which is supposed to be able to be powered by DC, but how is that possible because how can you bypass the bridge rectifier on the dam?
At 100hrs it is fully broken according to my low standards 🙂 Have you tried omitting the output filter caps altogether? Seems to work best for me.
I would not try Silmics if the highs are already wanting.
I would not try Silmics if the highs are already wanting.
DC has the miraculous ability to pass through bridge rectifiers 🙂 Yes, you can also physically remove the rectifier, but i think most of us have not.
I always perceived the Silmics as a little recessed in the highs and with fatter low regions, which would fit quite well then.
And I can easily omit the 1,2 nF caps, will try tomorrow.
And I can easily omit the 1,2 nF caps, will try tomorrow.
if I choose to use dc and use the DIYINHK 0.8uV Ultralow noise DAC power supply regulator +-9/12/15V 1.5A*x2 module, using the 4 pin power socket on the dam which is really only 3 pin would the 0 volt position be minus and the others 2 pins be plus?
Could we still need to reduce the 0.1R series resistor to 0.01R?
Mods | moreDAMfilters
Note especially simulations on bambadoo's implementation of factory mod + 4000uF (per rail). Could we still be getting 0.4mV at 100Hz and a simple 0.01R 0603 replacement would lower that by a factor of 10? Higher frequencies seem unaffected.
Would be great if someone more experienced could help clear up this issue 🙂
Mods | moreDAMfilters
Note especially simulations on bambadoo's implementation of factory mod + 4000uF (per rail). Could we still be getting 0.4mV at 100Hz and a simple 0.01R 0603 replacement would lower that by a factor of 10? Higher frequencies seem unaffected.
Would be great if someone more experienced could help clear up this issue 🙂
Okay this is something everyone should read if not already (good info in the comments too):
Vref musings: the final frontier? | moreDAMfilters
I wish someone has reposted this earlier... Looks like the 0.01R low-res mod is important, though no simulation result is posted for rev.2(=rev.>2 - 3*100uF) + >4800uF. Limitations are discussed to my satisfaction as well in the blog post...
WSL0603R0100DEA Vishay / Dale | Mouser
Vref musings: the final frontier? | moreDAMfilters
I wish someone has reposted this earlier... Looks like the 0.01R low-res mod is important, though no simulation result is posted for rev.2(=rev.>2 - 3*100uF) + >4800uF. Limitations are discussed to my satisfaction as well in the blog post...
WSL0603R0100DEA Vishay / Dale | Mouser
Last edited:
Soren, have you had a chance to measure other shift registers? The voltage ripple potentially caused by LVC595A internal resistance seems to overwhelm any subtle distinction in resistor tolerance.
4V for 16bits: LSB=61uV
0.3R 2.7mA in the LVC: 810uV ripple
I hope TotalDac used better shift registers otherwise I would feel really sorry for their customers... even if the sound is great
4V for 16bits: LSB=61uV
0.3R 2.7mA in the LVC: 810uV ripple
I hope TotalDac used better shift registers otherwise I would feel really sorry for their customers... even if the sound is great
Last edited:
And this is the reason i have never believed resistor tolerance between 0.01% and 0.05% makes any difference. Not that it matters anymore.
Don't see how this relates to TotalDac: a good sounding resistor is a good sounding resistor all else being equal. Having compared a lot of resistors i consider the z-foils to be at the pinnacle. Too bad power dissipation is low and the price is unfriendly, or i'd be using them everywhere 🙂
Don't see how this relates to TotalDac: a good sounding resistor is a good sounding resistor all else being equal. Having compared a lot of resistors i consider the z-foils to be at the pinnacle. Too bad power dissipation is low and the price is unfriendly, or i'd be using them everywhere 🙂
Sure resistors can still make a difference. Perhaps it's a different distortion coming from ripple vs non-linearity.
Or perhaps we should all be thankful that 10bits is perhaps all humans can enjoy? Maybe a better(-hearing) species would appreciate delta-sigma a lot more
Imagine the artistic possibilities...
Last edited:
That's no specialty shift register... I doubt TI had in mind TotalDac when they made the chips
Still I think we deserve an answer from Soren. Even if he believes his responsibilities end when music is produced, there is the question of whether our recent mods are beneficial. They certainly sound a lot different and should measure better in ripple. Shall we draw the conclusion that rev.3/4/5, dam1121, etc. are still flawed and need modding to sound right?
Last edited:
So, you need confirmation from the original designer because you cant judge if your mods was "good" or not? You seem torn between tech and audiophilia?
You found the old discussions - I think it is there that you will find your answers to your questions rather then Sören.
One benefit, and the reason I got a 1121, is that the payload is re-clocked after the fpga - a big mistake in the 1021. Still, I would love to know the buffer/PLL algo.
//
You found the old discussions - I think it is there that you will find your answers to your questions rather then Sören.
One benefit, and the reason I got a 1121, is that the payload is re-clocked after the fpga - a big mistake in the 1021. Still, I would love to know the buffer/PLL algo.
//
No, the LVC595 is the best fit. Other parts will have the same issues with the bonding wires.