What's the benefit of Bypass Capacitors?

[Deleted member 65]

My latest audio aquisition the Musical Paradise MP-D2 DAC equipped with a tube & capacitor analogue output stage.

Replaced stock tubes prior first power up with Russian NOS tubes. Aiming to replace original capacitors with Mundorf Silver/Gold Oil (on order).

Awaiting the Mundorfs, installed Jantzen Silver Z-caps 0,1uF as bypass capacitors with the standard Obbligato gold 2,2uF caps.

Haven't yet listened to my new setup. All powered up and breaking in if that's needed at all.

My query/question/advice is: What would the value of Bypass Capacitors in my application/DAC, if at all? Read that it may bring "air", "extended highs" ... aso

 

[amirm]

The purpose of a bypass capacitor is to provide local energy reserve right where it is needed and shunting of high frequencies to the ground. The power supply has such capacitance but it is too far away to work the same way. Digital circuits for example in the front-end of a DAC for example switch fast and can create spikes on the power supply line without bypass cap. And there can also be oscillation.

Note that the connection to sonic improvement is another matter than the pure functionality of the device using bypass caps. So what I say doesn't necessarily explain the attributes people attach to it.

 

[Speedskater]

Are we all on the same page as to what a 'bypass' capacitor is?

 

[Deleted member 65]

Are we all on the same page as to what a 'bypass' capacitor is?

Not sure, trying to learn. I've installed them in parallel with the Obbligatos (output caps) seen in picture above.

 

[Cosmik]

As Amir says, it is there to provide a small, local, low impedance reservoir of power right next to a circuit or component - similar to a small rechargeable battery. The wires that bring the power to your circuit are inductive and resistive, so as a sub-circuit or component dynamically varies its current draw, unless there is a bypass capacitor, the power supply voltage across it may vary. This will also affect other nearby sub-circuits hanging off the same power supply connections. Digital circuits for example, draw very large, brief, pulses of current when they switch.

A bypass capacitor across the component could be seen as 'filling in' the power supply current during those pulses (and will then effectively recharge itself from the power supply for the rest of the time). It is good practice to place a bypass capacitor across every integrated circuit or sub-circuit in a design.

Larger capacitors (such as the large electrolytics in an analogue DC power supply) are more inductive than smaller capacitors, meaning they would not be as good at bypassing the current spikes of a digital IC, say, even if they were placed close by. By paralleling larger and smaller capacitors it is possible to get the best of both worlds in any application.

 

[RayDunzl]

Where do the replaceable capacitors in the picture reside in the circuit?

Bypass/Decoupling - maintaining a DC voltage and shunting variations to ground

or

Coupling - passing the AC musical signal toward the output terminals while blocking the DC component?

Example:

 

[Deleted member 65]

Where do the replaceable capacitors in the picture reside in the circuit?

Bypass/Decoupling - maintaining a DC voltage and shunting variatiohttp://www.head-fi.org/t/802557/musical-paradise-m2-d2-ak4490-balanced-dac-reviews/60ns to ground

or

Coupling - passing the AC musical signal toward the output terminals while blocking the DC component?

They are installed in parallel with the Obbligatos (the output caps after tubes prior output terminals) i.e installed in the same physical terminals as the Obbligatos.

Someone doing this with same DAC as I have:

http://www.head-fi.org/t/802557/musical-paradise-m2-d2-ak4490-balanced-dac-reviews/60

Shit, I should have continued my career as Radar Engineer within the Swedish Army instead of switching dealing with crap computers ... ;-) would have most likely understood all of this then.

 

[Speedskater]

The Obbligatos are audio coupling capacitors, but not DC supply bypass capacitors.
So the Silver-Z caps would also be audio coupling capacitors (in parallel).
Their major affect would be to drain your wallet.

 

[Deleted member 65]

The Obbligatos are audio coupling capacitors, but not DC supply bypass capacitors.
So the Silver-Z caps would also be audio coupling capacitors (in parallel).
Their major affect would be to drain your wallet.

No big drain, apart from that any electrical impact on the output signal compared to: 1 2,2uF vs 1 2,2uF & 0,1uF in parallel? Possible to measure difference, if any?

 

[DonH56]

Coupling or decoupling, the usual purpose of multiple different-valued capacitors in parallel is to provide a better (lower) broadband impedance. Capacitors (C) have parasitic resistance (R) and inductance (L) that makes them self-resonant at a frequency determined by the L and C values. Above that frequency, the capacitor no longer acts like a capacitor, but looks inductive instead. That means it is less effective as a coupling or decoupling capacitor. Larger capacitors generally self-resonate at lower frequencies, but offer a lof of charge storage. Higher-frequency noise is generally lower in amplitude, and the impedance of a capacitor is inversely proportional to frequency (higher frequency equals lower impedance -- up to the self-resonance frequency) so smaller capacitors take over as the big ones "fade away".

Below is data for 1 uF and 0.01 uF chip caps. Note that the 1 uF impedance curve changes slope around 5 MHz, becoming inductive instead of capacitive. The 0.01 uF capacitor is good all the way up to about 200 MHz. A number of different values in parallel provides adequate decoupling from essentially DC to 10 GHz in my final design. Shift the values and frequencies a few orders of magnitude and you can see the applicability to audio circuits.

 

[Deleted member 65]

Coupling or decoupling, the usual purpose of multiple different-valued capacitors in parallel is to provide a better (lower) broadband impedance. Capacitors (C) have parasitic resistance (R) and inductance (L) that makes them self-resonant at a frequency determined by the L and C values. Above that frequency, the capacitor no longer acts like a capacitor, but looks inductive instead. That means it is less effective as a coupling or decoupling capacitor. Larger capacitors generally self-resonate at lower frequencies, but offer a lof of charge storage. Higher-frequency noise is generally lower in amplitude, and the impedance of a capacitor is inversely proportional to frequency (higher frequency equals lower impedance -- up to the self-resonance frequency) so smaller capacitors take over as the big ones "fade away".

Below is data for 1 uF and 0.01 uF chip caps. Note that the 1 uF impedance curve changes slope around 5 MHz, becoming inductive instead of capacitive. The 0.01 uF capacitor is good all the way up to about 200 MHz. A number of different values in parallel provides adequate decoupling from essentially DC to 10 GHz in my final design. Shift the values and frequencies a few orders of magnitude and you can see the applicability to audio circuits.

View attachment 5674

Thx, sort of supports the idea of adding more "air", "treble", "extension" ... if paralleling two audio output caps with different values (high & low).

 

[DonH56]

Yah, "sort of"... The self-resonant frequency of most audio coupling caps is way, way above the audio band. There are other things that can happen with capacitors, like hysteresis and voltage nonlinearity, but like a lot of things most of them happen at a level well below audibility, at least for me. That said, there are plenty of people with better ears than I who claim to hear things I can only imagine.

 

[RayDunzl]

there are plenty of people with better ears than I who claim to hear things I can only imagine.

Mine are definitely of the "low pass" variety.

Always have been, at least since age 8 or so when I discovered the anomaly via a test LP 20-20k sweep.

However, I find I can't even imagine higher than I think I can hear. I would relate that to trying to imagine a color I couldn't see, were my eyes deflicted with some color blindness.

What's your Day Job, Don?

 

[DonH56]

I started repairing TVs and stereos when I was 14, worked doing that to pay for the last couple of years of HS and on through college plus working in the EE lab and as a TA. Odd jobs along the way ranged from working on a dairy farm to running with a wheat harvest crew (OK to MT) to maintenance and horse crew for a college camp plus playing trumpet and running sound boards for a while (not a lot of money but was fun). Worked for several audio stores and built a few audio thangs along the way. After graduation started working with high-speed data converters (ADC/DAC) and support circuitry, RF circuits, and such (all transistor-level full custom analog IC designs) for nearly 30 years (up to being Director of a small company doing R&D stuff) before getting laid off and finding a job first in customer support then doing analog IC validation (for the same company; I got the job offer quickly and with a couple of kids in college didn't want to move so grabbed it but it has worked out well). Most of my career has focused on mixed-signal analog circuits from hundreds of MHz to many GHz (up to 300, briefly) with a bit of digital filter stuff (not really DSP as we know it today) along the way. Radar, space (worked on the radar in the space shuttle plus some satellite stuff), lots of different things. These days I do analog IC validation for computer stuff (mainly SAS/SATA controllers) along with various SI/PI (signal integrity/power integrity) tasks. I like the breadth though it makes for long hours at times. Having a pretty solid design background is helpful as I can liaison among designers, test engineers, and customers.

 

[fas42]

Thx, sort of supports the idea of adding more "air", "treble", "extension" ... if paralleling two audio output caps with different values (high & low).

As Don points out, intelligent paralleling can be very effective - but the higher the frequency you're trying to impact the harder it gets; absolute precision in how the additional capacitance is brought to bear is essential, otherwise the benefits will be close to zero, or could make the situation worse - there may be a resonant impedance if wrong combinations of values are used. So, don't do this in a haphazard manner; know exactly what the benefit of the additional capacitor is, for a good outcome.

 

[tomelex]

Looks like they have allowed you to obsess over these by making it so easy to pop in different ones. Try any kinds you want although electrolytics or tantalums are polarity sensitive and you should consult the manual or manufacture of that unit on which way round they need to be connected. They gave you a "tone control" by swapping capacitors instead of turning a bass or treble knob. This could be the path to audio hell for you . If you change the actual value of the capacitance enough you will hear it, small changes, such as adding little itty bitty bypass caps would not be audible 99% of the time if the circuit is designed correctly to start with.

 

[tomelex]

Yah, "sort of"... The self-resonant frequency of most audio coupling caps is way, way above the audio band. There are other things that can happen with capacitors, like hysteresis and voltage nonlinearity, but like a lot of things most of them happen at a level well below audibility, at least for me. That said, there are plenty of people with better ears than I who claim to hear things I can only imagine.

that is soooo subtle, whether you meant it that way or not my man!

 

[tomelex]

I started repairing TVs and stereos when I was 14, worked doing that to pay for the last couple of years of HS and on through college plus working in the EE lab and as a TA. Odd jobs along the way ranged from working on a dairy farm to running with a wheat harvest crew (OK to MT) to maintenance and horse crew for a college camp plus playing trumpet and running sound boards for a while (not a lot of money but was fun). Worked for several audio stores and built a few audio thangs along the way. After graduation started working with high-speed data converters (ADC/DAC) and support circuitry, RF circuits, and such (all transistor-level full custom analog IC designs) for nearly 30 years (up to being Director of a small company doing R&D stuff) before getting laid off and finding a job first in customer support then doing analog IC validation (for the same company; I got the job offer quickly and with a couple of kids in college didn't want to move so grabbed it but it has worked out well). Most of my career has focused on mixed-signal analog circuits from hundreds of MHz to many GHz (up to 300, briefly) with a bit of digital filter stuff (not really DSP as we know it today) along the way. Radar, space (worked on the radar in the space shuttle plus some satellite stuff), lots of different things. These days I do analog IC validation for computer stuff (mainly SAS/SATA controllers) along with various SI/PI (signal integrity/power integrity) tasks. I like the breadth though it makes for long hours at times. Having a pretty solid design background is helpful as I can liaison among designers, test engineers, and customers.

Your Hired!!

 

[watchnerd]

I vote this DAC as "most likely to induce neuroses and inflame audiophilia nervosa".

Tube rolling is bad enough. Combined with capacitor rolling, and then all the different inputs....so many combinatorial variables it's a tweaker's dream / nightmare.

 

[Ken Newton]

I vote this DAC as "most likely to induce neuroses and inflame audiophilia nervosa".

Tube rolling is bad enough. Combined with capacitor rolling, and then all the different inputs....so many combinatorial variables it's a tweaker's dream / nightmare.

"Dream/nightmare." I think that captures it. A large portion of audiophiles seem perfectionist by nature, often, to a fault. If not obsessing over 'perfecting' an audio system they would be obsessing over 'perfecting' something else. It's simply in their nature. Audio gives them a constructive, if expensive, outlet for their obsessive nature. There are others, however, who seem to experience an therapeutic benefit from tweaking/fiddling with their HiFi. Doing so seems to function as a healthy diversion from life stress. The act of fiddling about with their systems is fun for them.

I recall long ago reading an audio magazine interview with Fabio, remember him, he of the long blonde hair? It turns out that Fabio is an audiophile (who knew?!), and reported enjoying the activity of simply changing the expensive cables of his expensive system. I don't recall whether he claimed to hear any difference though. Apparently, he simply liked to toy around with the devices and equipment. I've no ridicule for him though, as I suppose that's no different from someone who enjoys the activity of working on automobiles, for example. In both cases, for them, the journey is at least as enjoyable as is the destination.