SMT passives for audio

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: SMT passives for audio

Post by JR. »

SUBass wrote:Well...I got the caps in..

AVX Niobium Oxide
Nichicon Aluminum Polymer
Panasonic FP in both 6.3V and 16V
Murata Aluminum Polymer

Both the Murata and AVX appear to be environment sensitive as they were shipped in sealed bags with dessicant and a humidity gage to see how much humidity they were exposed to. That makes me a little uneasy about using them in a piece of gear that will be used in non-controlled environments. That's a real pity too because they were my favorites.
Unless you see something on the data sheet saying that I wouldn't make such an assumption. It may be to insure the parts will solder well after sitting, or some more innocent explanation.
Each one of the caps changed the spectrum and dynamic response of signal going through it. It's mainly a subtle difference, but in spending time with each cap I could distinctly pick out their subtleties. I have a specific application that I tired these out in, coupling from a buffer to a couple of preamps in my bass rig. At first I populated a cap or two on the buffer board and started playing. Swapping different caps to different pres, and comparing to no cap at all as well as bass straight into the amp. After getting a feel for the caps, I tracked a DI bass track into my DAW and setup a reamp to feed the buffer. That way I could put each cap into the signal path, track, then sit back and listen more critically.
If subtle differences are audible they are measurable.Do you have any data to support your observations.Spectral changes can be measured with analyzers, dynamic changes can be parsed out with tone burst testing or similar benchmarks.
Going through the list...

The AVX rolled off some bottom (perhaps need a bigger value) and compressed the signal a bit. It also had a slightly forward midrange around 2.5k. It's possible that the apparent midrange boost is really due to the attentuation of the lows. Signal through the AVX didn't seem as "relaxed" as signal without a cap. It did make things seem "tighter."

Nichicon Al Polymer. This was a really interesting cap. It tended to preserve transients better but made things "sit back" a bit. Again...There was a seeming shift in really low end of the sound...But it was a "tighter" "smoother" sound.

Panasonic FPs. As should be expected...There was not detectable sonic difference between the 6v and 16v parts. They sounded the same. The transient response and midrange/top end of the FP was pretty good. The bottom end was a little weak though. This again makes me think the value of 47uf is a little low.

Murata Al Polymer. This was perhaps my fave of the group. Many times when I was blind AB'ing I picked it as my preferred sound over the no cap path. It sacrificed the bottom like the other caps but didn't seem to do so as much as the others. It also did the same "smoothing/tightening" thing that the Nichicon did as well as a bit of the mid boost of the AVX. The transients were a slight bit smeared which gave the no-cap path a slight edge.

I'll pull the exact model of all the caps if anyone is interested.
not yet... With that dramatic of an audible difference I would be tempted to do some null tests between cap A and cap B to better quantify and qualify the actual differences.

How does one go about calculating the appropriate size for a coupling cap?
1/(2 x pi x R x C)

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: SMT passives for audio

Post by mediatechnology »

1/(2 x pi x R x C)
What JR said. Javascript calculator:

http://www.muzique.com/schem/filter.htm
With that dramatic of an audible difference I would be tempted to do some null tests between cap A and cap B to better quantify and qualify the actual differences.
That's how I would go about looking for that needle in the haystack.
Andy Peters
Posts: 5
Joined: Sun May 25, 2008 12:16 pm

Re: SMT passives for audio

Post by Andy Peters »

SUBass wrote:Both the Murata and AVX appear to be environment sensitive as they were shipped in sealed bags with dessicant and a humidity gage to see how much humidity they were exposed to. That makes me a little uneasy about using them in a piece of gear that will be used in non-controlled environments. That's a real pity too because they were my favorites.
The reason for shipping the parts in a vacuum-sealed bag with dessicant is because the packages may absorb moisture. All IC and passive packages can absorb moisture; the extent to which this occurs depends on many factors.

What happens on an IR-heated reflow process is that the absorbed moisture can boil off, breaking the package or busting IC bond wires or doing other non-obvious (with close inspection or in-circuit test) problems.

If the parts are hand soldered, the packages won't see the same thermal stresses as they see in the oven.

If parts are intended for machine attachment and they've been out of the vacuum bags for longer than the suggested times, the simple fix is a low-temperature bake-out which safely eliminates any absorbed moisture. Then the parts can be placed as usual.

-a
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: SMT passives for audio

Post by mediatechnology »

I think I want to make a capacitor null tester.

As JR and I have both discussed elsewhere, the terminal voltage developed across a coupling capacitor is usually very small until low frequency cut off is reached. At low frequencies, the reactance begins to become significant relative to the load resistance and the terminal voltage rises. Mid band, and at low frequencies if the capacitor is properly sized, the voltage is vanishingly small. When terminal voltage becomes high, known distortion mechanisms exist, one of which is reverse bias of polarized caps. In speaker-level crossovers for example, film or bipolar caps make a great deal of sense.

Usually, the terminal voltage of properly-sized coupling caps is very, very small. In one test I did, I paralleled a coupling cap with back-to-back diodes. There was no measurable increase in distortion unless the frequency was low and the cap seriously undersized. Diodes provided an ~600 mV threshold; a typical reverse-biased electrolytic starts to look like a reverse biased diode around 1.25V IIRC.

(Dried out old caps do produce terminal voltage and make huge sonic differences as any MCI owner can attest.)

Though the terminal voltage is typically very small, what does it look like? Can something be made to measure what people hear?

A JR said earlier, a null tester might be a good tool to look for that THD needle in the haystack. I used it for the headphone amp to overcome generator THD limitations.

One easy way to make one is use a THAT1240 diff amp/line receiver with one input connected to the generator and the other to the right-hand side of the cap under test, "Cx." The THAT1240 sees whatever Cx terminal voltage that develops as a differential component with the generator - and it's THD - appearing in common mode. The common mode gets nulled out by a good -80 dB or so. But, I think whatever's down there in the differential mud would be buried by noise in the jig. (Unless of course it was a really bad cap.)

So why not use a THAT1510/INA217 mic preamp to add 60 dB gain (perhaps less) to dig a little deeper? (In this particular case I'd use the INA217. It's internal diff amp is trimmed to provide typically 80 dB of CM vs. the 1510's typ. 40-50 dB.) Basically the circuit would have one input connected to the generator and Cx with the other preamp input connected to the right-hand side of Cx. 10K bias resistors provide load for the generator and test load for Cx. For mid band measurements, where the terminal voltage is low, the differential (preamp) gain could be made quite high. The common mode gain would be around -80 dB or more using a INA217. As the measurement moved lower to the cutoff frequency, the preamp Rgain would need to be increased.

I may sketch or cobble something up and check a few caps using Visual Analyzer's FFT.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: SMT passives for audio

Post by JR. »

It might be interesting to revisit this in light of the many capacitor technology changes (like low ESR etc) since decades ago. I don't recall the details but I may have had some quibbles with the testing Jung et al did in Audio Amateur back in the day... IIRC they were chasing DA which mainly shows up with asymmetrical loading (like in S/H) but not in typical audio applications. (In fact DA in typical audio applications, looks like smaller RCs in parallel with the large C, not very significant or IMO audible for DC blocking application).

The best solution is just to avoid large terminal voltages, and similarly, avoid high signal currents, that can raise issues with other non pure C terms inside the capacitor.

I don't know if you've followed Sam Groner's work in this area. I think his current preference is to parallel two electrolytic caps, back to back, with one by definition reverse polarity. On his bench, he reports a first order cancellation of some distortions. I don't know if this potential distortion cancellation between similar caps, complicates the null testing, when comparing two caps. This null test needs to be well designed. Are we talking about nulling vs. an ideal C (say polystyrene or teflon in a high performance servo configuration), or vs a straight wire?

There seem to be a handful of applications where caps are commonly used that involve low impedance and/or not-low currents. It might be interesting to test the new batch of caps to come up with "least bad" solutions for those popular applications.

I will admit publicly, In have used far more blocking capacitors (especially at Peavey) than my purist side was comfortable with, because the marketplace would dismiss any product with clicks and scratchy pots, no matter how sweet the audio path was when playing music. So the customer is always right, even when they're wrong.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: SMT passives for audio

Post by mediatechnology »

JR. You might want to go back and re-read my post. Not because you missed anything but it may have changed due to editing. (There's some combination of keystrokes that will completely nuke a post and I've done that a lot lately. So I edit/re-edit/submit out of fear of losing it.)
The best solution is just to avoid large terminal voltages, and similarly, avoid high signal currents, that can raise issues with other non pure C terms inside the capacitor.
Exactly. Do that and you can put all kinds of non-linearities in the signal path across it (such as diodes) or stuff in series and they won't get noticed.

Yes, a piece of wire compared to a cap. Not an ideal to a DUT.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: SMT passives for audio

Post by mediatechnology »

I don't know if you've followed Sam Groner's work in this area. I think his current preference is to parallel two electrolytic caps, back to back, with one by definition reverse polarity. On his bench, he reports a first order cancellation of some distortions.
JR It's been awhile since I followed Sam's work in that area but wasn't his focus on really large value Cs for use as Cgains in mic preamps? At the low impedances there I think that's one area where terminal voltages do develop and the caps being so large have all kinds of issues. One I looked at was RFI ingress through the case. The balance in stray C from the terminals to the case was very imbalanced. I played with series bi-polar connections and shielding with the variable Rgain in between the caps. I looked at this due to some rectification of hi-band VHF I witnessed producing DC offsets. Caps in mic preamps are critical both at the input and Cgain. I think there's still a lot to be learned in that area.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: SMT passives for audio

Post by JR. »

mediatechnology wrote:
I don't know if you've followed Sam Groner's work in this area. I think his current preference is to parallel two electrolytic caps, back to back, with one by definition reverse polarity. On his bench, he reports a first order cancellation of some distortions.
JR It's been awhile since I followed Sam's work in that area but wasn't his focus on really large value Cs for use as Cgains in mic preamps? At the low impedances there I think that's one area where terminal voltages do develop and the caps being so large have all kinds of issues. One I looked at was RFI ingress through the case. The balance in stray C from the terminals to the case was very imbalanced. I played with series bi-polar connections and shielding with the variable Rgain in between the caps. I looked at this due to some rectification of hi-band VHF I witnessed producing DC offsets. Caps in mic preamps are critical both at the input and Cgain. I think there's still a lot to be learned in that area.
While I wouldn't expect huge terminal voltages in the mic pre gain leg (same voltages pretty much as the inputs), I have run these down to single digit ohms so surely low impedance, and perhaps several mA of peak current. The phantom blocking is another classic PIA. To tie a neat bow on this, the other potentially difficult applications IMO, is in series with an output, where current and loading can be more than low current.

So thinking in terms of Z.. Phantom blocking caps need to be clean wrt <200 ohms. Mic pre gain legs need to be happy working down to, single digit ohms, and output blocking caps, say 600 ohms.

No problem... BTW, Sam's back to back trick, can't be used in the phantom blocking application, for obvious reasons.

Of course, who needs caps in any of these applications? :lol:

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: SMT passives for audio

Post by mediatechnology »

While I wouldn't expect huge terminal voltages in the mic pre gain leg (same voltages pretty much as the inputs), I have run these down to single digit ohms so surely low impedance, and perhaps several mA of peak current.
These are often polarized and 6.3V. They can see amps of fault current and during said fault significant terminal voltages >>6.3V often in reverse polarization. I never looked at how they degraded 'cause my focus was on the silicon. Sam's configuration, back-to-back, might protect the caps since it will look like back-to-back diodes during a fault and clamp the forward voltage to 1-2V. During normal operation of course they do see little terminal voltage at LF since they follow the inputs.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: SMT passives for audio

Post by JR. »

I'm not sure how protective the reverse breakdown path would be. Most of my research into this was with parts installed wrong in production and trying to predict failure. I recall one 10uF@16v or perhaps 25V, that was installed backwards in a circuit with 7.5V across it. IIRC they didn't fail outright, and I don't even recall if I reworked all the boards. I know I tried accelerated burn in to force a failure and couldn't.

JR
Cancel the "cancel culture", do not support mob hatred.
Post Reply