Rank: Member
Groups: Member
Joined: 4/21/2007(UTC)
Posts: 59
Location: Minnesota
|
This is a concluding edit to mention that these issues have been resolved, though they are not fully understood. The issues below arose while the Buffalo3Pro/Mercury were running in a test environment, openly mounted to a metal plate. Without solving the issue I went ahead and installed everything in the final compact chassis and then the gain problem never occurred. I now believe that I2C control wiring needs to be very well shielded with clock and data wires the exact same length - and as short as possible. No unpredictable changes to the DAC operation are currently observed. Cheers! Original post: Hi Russ and Brian, Question: Considering the Mercury circuit, would you recommend against using a relay switch to change the gain resistor setting? I'm imagining something like a 2 or 4-state (times 4) "Mini-Darwin" that I could run with I2C (hang a small board from the 8 through-hole resistor pads with a ground plane between relays and Mercury)? If that approach is a non-starter, any other suggestions to adapt Mercury gain 'on the fly'? The overall goal is simply to remain within the upper range of DAC digital volume settings, not to integrate an analog volume control. Background: I'm working with Mercury for a few USB DACs for myself and family, and I'm 'all-in' with I2C control. I'm loving the sound of Mercury via headphones. The music source material I have covers a wide range of inherent gain settings. With stock 60 ohm gain resistors driving balanced 300 ohm headphones, a few old recordings in my collection could reasonably be enjoyed at near full DAC gain. However, the bulk of my music plus online streaming sources require huge attenuation with Mercury's stock 60 ohm gain setting resistors. For example, on many albums OxOa (7.8% of range) in DAC register 27 is the max tolerable volume. Playing with so much attenuation seems to lower SQ. All my eventual DAC users want good headphone performance, but I can't predict the efficiency of their other downstream amplification equipment. There could be significant mismatch between speaker outputs versus headphones driven directly by Mercury. Many TIA for any opinions/ideas! (I DO have an old stereo balanced Darwin!  ) Frank Edited by user Tuesday, April 27, 2021 8:46:42 PM(UTC)
| Reason: Not specified
|
|
|
|
|
|
Rank: Member
Groups: Member
Joined: 4/21/2007(UTC)
Posts: 59
Location: Minnesota
|
A couple more details... 1. The best isolated switch I can find is this: https://omronfs.omron.co...cts/pdf/en-g3vm_61g2.pdf It adds about 1 ohm. Anybody know of a better isolated part? 2. I would probably use two 30 ohm resistors in series and when the relay closed it would short one of them. So at 60 ohms the open relay doesn't factor into performance. But at ~31 ohms, how much of an issue is matching the resistance of the diferentially-paired relays? Would Mercury's balancing servo correct for any small differences? Edited by user Thursday, April 16, 2020 7:17:27 PM(UTC)
| Reason: Not specified
|
|
|
|
|
|
Rank: Member
Groups: Member
Joined: 4/21/2007(UTC)
Posts: 59
Location: Minnesota
|
Revalation to me: At low volumes, attenuation by LMS/Squeezelite 1.8 sounds BETTER than onboard ES9038 attenuation. So I won't worry about tweaking Mercury for now. I had completed a little 'solder-on' board layout and just as a matter of diligence before ordering I directly compared the DAC to the software. ...glad I did... SQ difference is small, but impact on enjoyment is very clear. Edited by user Saturday, April 18, 2020 12:18:35 AM(UTC)
| Reason: grammar
|
|
|
|
|
|
Rank: Member
Groups: Member
Joined: 4/21/2007(UTC)
Posts: 59
Location: Minnesota
|
Greetings, With respect to signal levels I am comparing the 9038 onboard firmware with various I2C configurations. With the onboard firmware and 60 ohm resistors in the Mercury, the command 'speaker-test -c 2 -t sine' results in a peak-to-peak potential of 7.2v (measured by oscilloscope). That seems erroneously different from the desired 2.8v (for 2v RMS) and I have no explanation. Any suggestions what might be affecting mercury outputs in this way? Many thanks in advance for any suggestions! Frank Williams @francolargo PS - I have found that the slightest disturbance caused by measuring the Mercury output will drop the voltage by 75% to ~1.8v peak-to-peak. This originally confused me because it is the same as setting bit 2 of register 15 to 0 (default). However, i2cdump in that state shows no differences from the intended configuration. Yet, in that lower-output state the 7.2v peak-to-peak output is restored by resetting bit 2 of register 15 to 1. Finally, yes, I double checked that the "correct" 60 ohm resistors are in the Merc. Edited by user Wednesday, April 29, 2020 2:00:13 PM(UTC)
| Reason: more measurements
|
|
|
|
|
|
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.
