BG NX Hi-Q in super e-cap good enough for VD, VREF ?

sceglar

Banned
2005-03-19 1:34 am
.
Does anyone know if Black Gate NX Hi-Q in the super e-cap config. are low enough in impedance and noise at RF, for decoupling for VD and VREF? (no ceramic bypass)
VA and VD of the d/a chip are powered from the same supply, so the decoupling at VA and VD are essentially in parallel with each other. I was thinking for this reason it may not be a good idea to use the OSCON//ceramic at VD and the black gate at VA(+VREF)

I can use oscon//ceramic for each VD and VA (+VREF), or Black Gate Hi-Q NX in super e-cap for each VD and VA (+VREF).

Can anyone suggest which would likely be the better option?

Thanks,
shane
 

Attachments

  • digital_board_mod.jpg
    digital_board_mod.jpg
    55.5 KB · Views: 1,053

Extreme_Boky

Member
2003-12-07 11:57 am
Try 0.1uF / 50V NX for "on the chip" decoupling. No need for super E cap configuration here.

I also use Vishay MKP 1837 series to obtain completely opposite sound character from Black Gates.

Mixing them together produces by far the best results. Both are very low noise types.

I didn't play with OsCons.

Good luck,

Regards,
Extreme_Boky

PS

You can observe the influence of different capacitors' types / values if you monitor the noise at the ground plane and at the supply pins while using different types. You'll realise that you need LOW NOISE GROUND PLANE to start with!!! Otherwise, you'll be spending your money on expensive capacitors which will change the sound character, but NOT bring any real improvement in lowering the noise at supply pins. So, think low noise ground planes firs, or together with capacitors implementation.
 

sceglar

Banned
2005-03-19 1:34 am
.
Thanks for the info..

From what you're saying it sounds like the non-polar BG's are fine for this application..

I am curious though, as to why you'd need that 0.1uF BG bypass if they are supposedly good for low impedance and inductance into the GHz range (guess I was looking for confirmation on this).

It was the inductance cancelling of the super e-cap that attracted my attention to them.

Working with the caps that I have here at the momment, do you suppose there is anything incorrect about using them as super e-caps without that 0.1uF bypass (of any kind), 44uF total for local decoupling after a CCS fed TL431 shunt reg?.

I'd like to just get in there any try them all, but the tracks on the PCB are thin, and I'd prefer not to run the risk of lifting them, and the insuing botch job that would follow.

regards,
shane
adelaide
 

Extreme_Boky

Member
2003-12-07 11:57 am
I use 0.1 / 50V NX types Black Gates and Vishays on their own. Paralleling does not sound good to me at all. I also remove whatever was there in the first place. I experimented with inductance cancelling E-CAP configs and this always sounded worse to me than single capacitor implementation. Try to keep the leads as short as possible. If you are careful, you can almost achieve 'surface mount length' with these Black Gates as well. You'll need to solder them right at the pins. If you are not experienced, start with second-hand e-bay stuff. You can get pretty nice sounding equipment for around A$300 - A$400. Once experienced, proceed to more expensive units.
Lower capacitance values might be needed around noisy devices like uPC, clocks / oscillators and digital devices in general. Vishay MKP1837 types come in very low capacitance values. Black Gates only down to 0.1uF - but are great! You have to try them all and find those you like. OsCons are very good as well...

Extreme_Boky
 

sceglar

Banned
2005-03-19 1:34 am
.
The data sheet for the ak4524 suggests 10uF, the dac came with 22uF, these are the sorts of values I see used regularly for decoupling following a TL431.

Is 0.1uF going to be enough; the shunt reg and then the 0.1uF, and thats it ?

I could try it, for sure. Solder at the pins like you suggest and not use the pcb tracks, good.

Not so experienced with micro surgery like this, used to tube amplfiers.

Thanks again,
shane
 
Extreme_Boky said:
I I experimented with inductance cancelling E-CAP configs
Extreme_Boky

Inductance cancelling is the biggest joke in electronics for years. I sincerely hope Maxwell does not have access to the internet up there.

To my ears the super-E config sounds very well by the way.

The importance of the variouis decoupling caps depend on the pins' sensitivity - which differs by function and design, and mostly is not indicated in the datasheet.

On keeping the ground plane free from RF currents one may read:

http://www.tentlabs.com/Info/Articles/Supply_decoupling.pdf

cheers
 

sceglar

Banned
2005-03-19 1:34 am
.
I have found reference to the TL431 oscillating if the 'next' cap is less than 5uF, so 0.1uF is unsuitable all by itself.. it needs a bulk filter after the shunt reg.

The smallest suitable e-cap config then using the hi-Q caps is 44uF/6.3V, I wonder about the ESR using a high value like this, even if they are 'super-low'. And your thoughts on the inductance cancelling thing are appreciated.

The article is very interesting it shows the application for seperate IC's. Would there be any advantage in using one ferrite before each decoupling cap for the individual VA and VD pins, within the same IC ?.

TL431 - 22uF hiQ - ferrite - 0.2uF super e-cap for VA (+VREF)
and the same for VD with the branch after the TL431.

AKM recommend VA and VD be powered from same supply, I wonder why that is??..


All suggestions much appreciated.

thanks,
shane
 
sceglar said:

The article is very interesting it shows the application for seperate IC's. Would there be any advantage in using one ferrite before each decoupling cap for the individual VA and VD pins, within the same IC ?.

AKM recommend VA and VD be powered from same supply, I wonder why that is??..


All suggestions much appreciated.

thanks,
shane

Hi shane,

Somewhere in the article I indeed prescribe the use of bead / capacitor combo for every supply pin.

Same supply: Advised because of startup behaviour and voltage difference. Mostly both are specificied. I've always found strictly separating supplies helps in terms of sound quality.
 

sceglar

Banned
2005-03-19 1:34 am
.
Hi Guido,

Thanks, this is great news.

I'll seperate the supplies, no parallel decoupling caps.. no problem!

TL431- oscon - ferrite - ceramic for VD
TL431- BG NX - ferrite - BG Hi-Q super e-cap for VA and VREF.

I'll be sure to consult the data sheet regarding which supply comes up first.. V will be trimmed so there is no mis-match.

Thanks again.

regards,
shane
 
Hi Guido,

Hey I am about to decide that too Shane

Dont´t you also think that you should use a cap after the ferrite in parralel with the ceramic?

I have thought of this for the digital (CS8416->AD1896 VD)

Lm337->CCS->TL431//22uF Oscon->Ferrite bead//22uF Oscon// 0.1uF//0.01uF ceramic

(A CCS for each TL431)

And this for the Dac psu 3.3V VD and 5V VA
Lm337->CCS->TL431//220uF NX-HIQ->Ferrite bead//22uF BGNX-HIQ//100nF BG NX HIQ.


Ferrite bead:
http://www.farnell.com/datasheets/65013.pdf