In the sense that alien abductees "demonstrate" to themselves and others that they were abducted and anally probed. Really, performing a good experiment isn't that hard- I'm not sure why people resist doing so.
The IV's Zin is simply,
(impedance of the feedback network + open loop opamp output impedance) divided by open loop gain.
With an oldskool opamp, its open loop output impedance is a few ohms, since we're usually looking at a standard emitter follower output stage. With a Rail-to-rail opamp, its open loop output impedance is an inductor. YMMV depending on the opamp, if it has output-inclusive compensation, its bias current, etc.
The feedback network is usually a R//C, I guess everyone knows the impedance of that.
If the DAC is current out, it will have a rather high impedance. ES9018 will have about 800R. There is some capacitance in parallel, of course.
Anyway. Let's run a dumb sim. It is a bog standard IV with two NE5532s.....
The curves show the differential and common mode input impedance, and the output, for 3 values of the Cap, zero, 100n, and 1u.
Obviously, on the common mode front, nothing happens. Since the two inputs in the differential pair move in unison, it doesn't matter if they are connected or not.
In differential mode, we can see the classic IV input impedance : very low at LF since we have lots of open loop gain, then it flattens at around 10 ohms from 100k to 1M as the cap in the feedback network takes over the resistor, then we run out of open loop gain and it becomes more or less uncontrolled.
The addition of the cap reduces the input impedance at HF and tweaks the lowpass transfer function too. Depending on the value, it can be clean (more or less) or make a nice peak.
If the DAC outputs its pulses in a nice symmetric way, the cap will absorb them and the opamp will have a much easier time, not having to use a lot of slew rate.
So to explain the effect, I'd either bet on HF filtering, or with a large cap, on the huge peak in the frequency response being audible. With some serious abuse on the cap value, you can get a 20dB peak at 10-15 kHz, certainly some people could notice that....
Attachments
As expected, the cap is resonating wth the inductive impedance of the virtual ground at the opamp input. With a big enough cap this will be audible. It will sound different from a simple low pass filter elsewhere in the chain, as that will miss the resonance.
If there is no opamp but instead an I/V transformer then the cap could be resonating with the leakage inductance. Similar outcome.
As I often say, it is easy to make a circuit sound different; to actually improve it you have to understand it better than the original designer.
If there is no opamp but instead an I/V transformer then the cap could be resonating with the leakage inductance. Similar outcome.
As I often say, it is easy to make a circuit sound different; to actually improve it you have to understand it better than the original designer.
It does not sounds different. There is no alteration what so ever for the sound and its fidelity (very low end, middle, very high end audio spectre), and this is quite strange, when a cap of 1µ is in the signal path.
The only what happen when attaching this cap so as described, is an increasing of the precision in placement (perceived location) of the sound elements in the audio environment (soundcene). The sound elements are better placed in space, and therefore the soundcene become wider/larger, it get a better volume.
This is all about the "effect", but not changing in the sound elements qualities.
The only what happen when attaching this cap so as described, is an increasing of the precision in placement (perceived location) of the sound elements in the audio environment (soundcene). The sound elements are better placed in space, and therefore the soundcene become wider/larger, it get a better volume.
This is all about the "effect", but not changing in the sound elements qualities.
All good postulations/sims.
Ok, in your simulations what kind of cap (internal RLC values ?) can cause such a peak with ES9018 for example, and what kind of peaks could be expected with real world typical PP or other 1uF solid caps (and BP electros for interest sake) ?.
Sure.
The first claims are of audio 'improvement'.
Of course that is a loose statement, but then further claims of ''improved soundstage'' also.
This also a loose statement and dependant on the experience/expectations of the listener/reviewer and the equipment used in the evaluations.
Either way, experimenters are reporting favourable subjective changes (no subjective negatives yet ?).
Anybody know of a Chinese cheap eBay ready built board available that is ripe for this kind of hacking/experimentation in order to try this observation/assertion ?.
Dan.
OK, so you ARE claiming it changes the sound. With this description in hand, you should easily be able to set up an ears-only experiment to determine if that change is real or imagined.
Don't underestimate human imagination; I'm pretty aware of the sensory pitfalls of not controlling listening evaluations for unconscious bias and preconception, but there's no way I'd claim a difference like that without doing a real ears-only test, since my imagination and susceptibility to unconscious bias and preconception is just as human as anyone else's.
It depends on the input impedance of the IV, so schematics would be needed...
All good, I/we understand that English is not your first language.
Ok, that increase in precision of depth of sound imaging, and L/R placement imaging translates to lower overall jitter and lower overall intrinsic harmonic/imd noise, including ULF/VLF sourced noise/uncertainty.
I have mentioned this before...a 1F/40R supercap across the DAC OPS supply is stated to improve subjectives.
I have also mentioned that this is an 'infinite' (close to 0Hz) Zobel.
So maybe there is a resonance thing going on and the Zobel helps ??.
Dan.
Sure.
Are there eBay/Amazon boards using other current output Dacs that are easily/cheaply available that are worth experimenting with.........anybody ?.
Dan.
Last edited:
Since a zobel can only damp a resonance which has an impedance peak above the resistor value in said zobel, I wonder what your 40 ohm ESR cap is damping... surely there is no peak above 40 ohms in your power supply ?... Cause if there is, you have another problem... Unless it's 40 mOhms ?
Not my DAC, discussion directly with JR, I could have stated wrong supercap ESR value, but think not IIRC.
Maybe/likely Zobel effective series R parameter is worthy of experiment....there is a very wide variation in supercap internal series resistance values according to data sheets.
Dan.
This is a nice illustration of the understanding of sound perception...
And this in an era where all the worlds' knowledge about perception and technology is a only few Google keystrokes away.
Jan
The 2.2nF caps now become a free parameter to flatten the response which will be op-amp dependent. You want an apples to apples experiment to separate different effects or IMO you are just wasting time.
vs.Coris said:
It sounds different.Coris said:
Which of your two conflicting statements should we accept as your opinion?
Let us assume you really meant the latter statement. This makes three claims:
1. 'Better' positioning - as you probably can't know what the 'correct' positioning is I assume you mean 'tighter' positioning. This could result from the two stereo channels being more alike in amplitude and phase.
2. wider stereo image - which implies a change in channel cross-talk.
3. 'Better' volume - unclear what this means as volume is easily adjusted to suit the listener.
So adding the cap improves channel balance and reduces in-phase crosstalk (or introduces anti-phase crosstalk). This should be measurable and potentially explicable.
don't forget that frequency shifts in the 3-8KHz range can effect perceptions of height and depth.
There is no contradiction, but only an expression issue/lack to define exactly what about. Therefore is enough important that one experience by himself this effect, to avoid such descriptions issues.
I meant "better volume", as better space (the volume of space if I can say so...), not referring to the volume knob on the device.
Better positioning in the soundspace, I mean a more precise location of the sound sources. The sound sources have nothing to do with the speakers. The speakers as sources become just transparent, and one hear and can locate quite precise the sound sources/elements/instruments in between the speakers.
Well, this about the location of sound elements is quite relative. I can not know exactly what was the position of the instruments at the recording, unless this information is provided by the producer. Without this cap in place, the sound elements are quite diffuse distributed into the soundscene. With this cap in place that "diffusion" disappear, or is much reduced, and the location of the elements become more precise.
I meant "better volume", as better space (the volume of space if I can say so...), not referring to the volume knob on the device.
Better positioning in the soundspace, I mean a more precise location of the sound sources. The sound sources have nothing to do with the speakers. The speakers as sources become just transparent, and one hear and can locate quite precise the sound sources/elements/instruments in between the speakers.
Well, this about the location of sound elements is quite relative. I can not know exactly what was the position of the instruments at the recording, unless this information is provided by the producer. Without this cap in place, the sound elements are quite diffuse distributed into the soundscene. With this cap in place that "diffusion" disappear, or is much reduced, and the location of the elements become more precise.
Last edited:
Thanks for clarifying.
So we are left with better channel balance and a modification of crosstalk.
Both are unlikely to arise from just adding a cap at this point. A cap is likely to make no difference or reduce channel balance (unless quite tight tolerance). Crosstalk could arise from the size of the cap (electric field coupling to the other channel) or extra ground currents - but in both cases the most likely outcome is in-phase crosstalk which will narrow the stereo image width.
So we are left with better channel balance and a modification of crosstalk.
Both are unlikely to arise from just adding a cap at this point. A cap is likely to make no difference or reduce channel balance (unless quite tight tolerance). Crosstalk could arise from the size of the cap (electric field coupling to the other channel) or extra ground currents - but in both cases the most likely outcome is in-phase crosstalk which will narrow the stereo image width.
I think I may precise (once again) that this cap is to be placed over the +/- outputs phases of a DAC output, or more general on the +/- poles of a differential signal, which represent in this case one stereo channel. So, there are two caps for the L/R stereo channels.
I just wonder what it may happen if the same approach is to be applied to a multichannel configuration. There is a quite different configured sounscene.
I just wonder what it may happen if the same approach is to be applied to a multichannel configuration. There is a quite different configured sounscene.
- Status
- Not open for further replies.