Caps for bass fix (ECN0022)

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Caps for bass fix (ECN0022)

Werner Almesberger
Joerg and I discussed the choice of capacitors for the bass fix on
#gta02-core. Some conclusions:

- as explained in ECN0022, the capacitor should have about 100 uF

- tolerance is only a minor concern, given that parts from the same
  reel are likely to be similar

- the capacitor should be rated at 4 V or higher

- ceramic is unsuitable because of harmonic distortions

- Tantalum seems to be the best choice

- due to the frequency-dependent ESR of tantalum caps, the ESR should
  be low (<< 1 Ohm)

- component size and height are critical. The tallest component we
  have inside the GTA02 main can is C1707 with a height of 1.6 mm.
  We can place the capacitors outside the can, but given that we now
  have plenty of room inside, that may not be optimal.

Here's a document that compares various capacitor technologies:

http://industrial.panasonic.com/www-data/pdf/ABE0000/ABE0000TE1.pdf

Particularly the ESR chart on page 5 is interesting.


With this data, I went to Digi-Key and did a search with the
following parameters:

- SMT, in stock, lead free, RoHS compliant
- available as cut tape
- 47 uF ... 120 uF
- 4 V ... 6.3 V
- ESR <= 200 mOhm
- case EIA 3528-21 or smaller

The leaves a choice of products from AVX, Kemet, and Nichion.

Possibilities for inside the can, maximum height 1.6 mm:

- AVX TCJT686M004R0080, 68 uF, 4 V, 20%, 80 mOhm, EIA 3528-12
  Digi-Key 478-3480-1-ND, USD 2.58/1
- Kemet T520T107M004ATE150, 100 uF, 4 V, 20%, 150 mOhm, EIA 3528-12
  Digi-Key 399-5288-1-ND, USD 1.55/1
- Nichicon F320J476MAA, 47 uF, 6.3 V, 20%, 180 mOhm, EIA 3216-16
  Digi-Key 493-3149-1-ND, USD 0.84/1

Note the smaller footprint of the Nichicon part. It could probably
still be considered a viable 3rd source, though.

Outside the can, height unlimited but trying to minimize footprint:

- AVX TCJA107M004R0200, 100 uF, 4 V, 20%, 200 mOhm, EIA 3216-18
  Digi-Key 478-3453-1-ND, USD 2.48/1
- Kemet T528I686M004ATE200, 68 uF, 4 V, 20%, 200 mOhm, EIA 3216-10
  Digi-Key 399-5243-1-ND, USD 4.85/1
- same as above

So it seems that we should plan on using a 3528 footprint, if possible.
If we can place it outside the can, that would be even better.

Comments ?

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Joerg Reisenweber
[Werner Almesberger Sa  15. August 2009]:
> Joerg and I discussed the choice of capacitors for the bass fix on
> #gta02-core. Some conclusions:
[...]
> Comments ?
>
> - Werner
>

2 * 47uF ~= 100uF
sigma-ESR = ESR/1 / 2
even better
3 * 47uF ~= 150uF
sigma-ESR = ESR/1 / 3
;-)

cheers
jOERG


_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

signature.asc (204 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Werner Almesberger
Joerg Reisenweber wrote:
> 2 * 47uF ~= 100uF
> sigma-ESR = ESR/1 / 2
> even better
> 3 * 47uF ~= 150uF
> sigma-ESR = ESR/1 / 3
> ;-)

Hmm, interesting idea. Let's see ... 47 uF, ESR <= 500 mOhm,
4 V or 6.3 V, ...

Naw, still about the same set of choices. Even if I include 33 uF.

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Al Johnson
In reply to this post by Werner Almesberger
On Saturday 15 August 2009, Werner Almesberger wrote:
> Joerg and I discussed the choice of capacitors for the bass fix on
> #gta02-core. Some conclusions:
>
> - as explained in ECN0022, the capacitor should have about 100 uF

> - ceramic is unsuitable because of harmonic distortions

I'll qualify this in case anyone gets the wrong impression. High-k dielectrics
like X7R have extremely high distortion levels. Low-k dielectrics like C0G
have exceptionally low distortion, but aren't available in values bigger than
about 220nF.

> - Tantalum seems to be the best choice

Distortion from tantalums is typically >10x that of a low cost polar aluminium
electrolytic. What makes them the best choice?

Cyril Bateman's articles on capacitor distortion are well worth a read.
Article 6 covers capacitors in the range we're looking at here.
http://uk.geocities.com/cyrilb2@.../capacitors_2.html
http://uk.geocities.com/cyrilb2@.../downloads_5_files/capsound6.pdf


_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Werner Almesberger
Al Johnson wrote:
> Distortion from tantalums is typically >10x that of a low cost polar
> aluminium
> electrolytic. What makes them the best choice?

All the Aluminium caps I can find in the 47-150 uF range are *huge*,
even the smallest one being 7.3 x 4.3 x 1.8 mm. That's more than
three times the surface of a 3528-12, or five times its volume :-(

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Rene Harder
In reply to this post by Al Johnson
Al Johnson wrote:

> On Saturday 15 August 2009, Werner Almesberger wrote:
>  
>> Joerg and I discussed the choice of capacitors for the bass fix on
>> #gta02-core. Some conclusions:
>>
>> - as explained in ECN0022, the capacitor should have about 100 uF
>>    
>
>  
>> - ceramic is unsuitable because of harmonic distortions
>>    
>
> I'll qualify this in case anyone gets the wrong impression. High-k dielectrics
> like X7R have extremely high distortion levels. Low-k dielectrics like C0G
> have exceptionally low distortion, but aren't available in values bigger than
> about 220nF.
>
>  

I agree, High-K ceramic caps have higher harmonic distortion that C0G
types. One source of harmonic distortion is that for XR7 and Z5U
dielectrics the capacitance varies with amplitude (and of course
frequency). For XR7 around 30% and Z5U around 70% [1]. Now, you can
limit this distortion with reducing the voltage swing or using caps with
higher maximum voltage. However, increasing the maximum allowed voltage
will increase the size of the cap quite a bit.
Another source of harmonic distortion is the piezoelectric effect of
High-K dielectrics. If you apply mechanical vibrations (e.g. sound
waves) to the capacitor you will get voltage spikes across it which
leads to distortions of your signal.

>> - Tantalum seems to be the best choice
>>    
>
> Distortion from tantalums is typically >10x that of a low cost polar aluminium
> electrolytic. What makes them the best choice?
>  

Capacitors with aluminum electrolyte are way to big with same
capacitance/voltage, tantalum seems to be the better choice here. I
found this benchmark [2] where AVX measured noise and THD+N with several
different caps for input and output coupling. The THD+N for ceramic caps
is really high compared to the other ones but AVX coupled an acoustic
exciter to the caps to observe and measure the piezoelectric effect.
(see Figure 4). According to this benchmark low ESR tantalum caps are
the second best choice for audio coupling. I should mention, that
surprisingly polymer tantalum caps which have a low ESR exhibit the same
noise than standard tantalum caps.

Why don't we use the 100uF caps from the big-C buzz fix (C4306/4307
GTA02-A7) so we don't have any component change. It's a Tantalum Solid
Electrolytic Chip Capacitor [3] (TLJA107M006R0500)

And at the end we are just designing a cell phone with music playback
ability and no high class audio amplifier so i would not expect an
outstanding THD at all. ;-)


[1] http://my.execpc.com/~endlr/ceramic.html
[2] http://avxta.com/docs/techinfo/tantbench.pdf
[3] http://avxta.com/docs/Catalogs/tlj.pdf


Rene

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Werner Almesberger
Rene Harder wrote:
> Why don't we use the 100uF caps from the big-C buzz fix (C4306/4307
> GTA02-A7) so we don't have any component change. It's a Tantalum Solid
> Electrolytic Chip Capacitor [3] (TLJA107M006R0500)

Are 500 mOhm ESR still okay ? If yes, that looks like a decent choice
as far as the electrical characteristics are concerned. It's too tall
(1.8 mm) for going into the can, though. We'll have plenty of space in
the can while the space outside the can probably remains largely the
same. (Depends a bit on how we're doing with the Calypso and WLAN
replacement/reshuffling.)

My non-electrical criteria are, in this order:

- can be sourced: compatible (*) parts from at least two different
  vendors in stock at Digi-Key and available in small quantities
  (minimum order <= 10)

- low profile: if possible, low enough to fit inside the can (height <=
  1.7 mm)

- smallest possible surface

(*) A small tolerance of capacity or an imperfect footprint match are
    okay. If we select a height <= 1.7 mm, the second source must be
    <= 1.7 mm too.

> And at the end we are just designing a cell phone with music playback
> ability and no high class audio amplifier so i would not expect an
> outstanding THD at all. ;-)

Yeah, we're not exactly building the B&O phone ;-)

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core
Reply | Threaded
Open this post in threaded view
|

Re: Caps for bass fix (ECN0022)

Al Johnson
In reply to this post by Rene Harder
On Thursday 20 August 2009, Rene Harder wrote:
> Al Johnson wrote:
> > On Saturday 15 August 2009, Werner Almesberger wrote:
> >> Joerg and I discussed the choice of capacitors for the bass fix on
> >> #gta02-core. Some conclusions:
> >>
> >> - as explained in ECN0022, the capacitor should have about 100 uF

> >> - Tantalum seems to be the best choice
> >
> > Distortion from tantalums is typically >10x that of a low cost polar
> > aluminium electrolytic. What makes them the best choice?
>
> Capacitors with aluminum electrolyte are way to big with same
> capacitance/voltage, tantalum seems to be the better choice here.

Werner's comparison of sizes demonstrates this nicely. 'Nothing better will
fit' is always a good reason.

> I found this benchmark [2] where AVX measured noise and THD+N with several
> different caps for input and output coupling. The THD+N for ceramic caps
> is really high compared to the other ones but AVX coupled an acoustic
> exciter to the caps to observe and measure the piezoelectric effect.
> (see Figure 4). According to this benchmark low ESR tantalum caps are
> the second best choice for audio coupling. I should mention, that
> surprisingly polymer tantalum caps which have a low ESR exhibit the same
> noise than standard tantalum caps.

THD+N is next to useless as a predictor of audio quality. It's the measure
that caused so much trouble with the early transistor amps because it fails to
distinguish between relatively pleasant 2nd order harmonics and unpleasant
high order harmonics, among other failings. It's the character of the
distortion that makes X7R entirely unsuitable for audio use rather than its
level; just compare the distortion spectra.

> Why don't we use the 100uF caps from the big-C buzz fix (C4306/4307
> GTA02-A7) so we don't have any component change. It's a Tantalum Solid
> Electrolytic Chip Capacitor [3] (TLJA107M006R0500)
>
> And at the end we are just designing a cell phone with music playback
> ability and no high class audio amplifier so i would not expect an
> outstanding THD at all. ;-)

True, but why make it worse than it has to be?

> [1] http://my.execpc.com/~endlr/ceramic.html
> [2] http://avxta.com/docs/techinfo/tantbench.pdf
> [3] http://avxta.com/docs/Catalogs/tlj.pdf
>
>
> Rene



_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core