[bmwpowere36m3]

Hey guys, I'm still new to the whole understanding electronic circuits but I'm making headway... I'm particularly interested in output coupling caps and their role in DAC's. I've recently iModed my iPod Mini so I can take advantage of the Wolfson WM8711 DAC for the time being.

First question, when typically used coupling caps block DC offset from entering the next circuit (that part I understand)... however are they always implemented like this:

A capacitor in series with the "line" and a resister? In essence the implementation of a coupling capacitors is a high-pass filter, correct? Thus using f = 1 / 2*PI*C*R you can get the -3dB point (Hz). So my next question is how are these values (C & R) determined, are they arbitrary (as long as you get a f3 thats low enough) or is there more?

My understanding is choosing a f3 point that ~10X less than the lowest frequency of interest will reduce phase distortion, i.e. for audio circuit a f3 of ~2 would keep distortions out of the 20+Hz range.

I was reading Walt Jung's, "Picking Capacitors" and in it states that not only keeping a low f3 is important, but also minimizing impedance (ideally 10K or less)... opinions? This impedance I think he's referring to is the paralleled resistance of the input (pot in an amp, RL) to the "pull-down" resistor in coupling cap circuit (Rb).

So for my CMoY (10K pot) I'd need a ~any size Rb, but for a typical amp (50K pot) you'd need ~12K or less. So how would a 47uF & 10K(total R) for a f3 of .34Hz output coupling circuit sound/work? As I'm typically seeing Rb values ~47K and some DIY DAC's as high as 330K.

Now this info applied to my DAC, Wolfson WM8711 which states this as a recommend setup:

I understand C1 and R1, but R2 series resistor... it's only needed for a ham-fisted user you might ground the "line" when in use, correct? Preventing frying the unit. This covers HP filters, do DAC's typically also have LP filters to prevent high frequency oscillations due to over-sampling? Any recommendations?

I'm almost done, honestly, my last question deals with bypassing caps. I'm strictly interested in the ratio, if there is any, used between say two caps. I've read somewhere that the smaller (typically film) cap should be 1/10th or 1/100th the size of the larger cap (typically electrolytic).

That's it, thanks for any help

*edit* I actually did have one more question, where do you guys source Vitamin Q caps, I can only seem to find them on eBay for a "reasonable", i.e. cheaper than some guitar sites which are charging ~10-20 per cap. And what are reasonable prices, I'm looking for some (I think) 0.047uF.

 

[tomb]

This is a lot to put into one post. You might try asking one question at a time in the future. I'll try to give this a shot, though.

Quote:

Originally Posted by bmwpowere36m3 /img/forum/go_quote.gif Hey guys, I'm still new to the whole understanding electronic circuits but I'm making headway... I'm particularly interested in output coupling caps and their role in DAC's. I've recently iModed my iPod Mini so I can take advantage of the Wolfson WM8711 DAC for the time being. First question, when typically used coupling caps block DC offset from entering the next circuit (that part I understand)... however are they always implemented like this: A capacitor in series with the "line" and a resister? In essence the implementation of a coupling capacitors is a high-pass filter, correct? Thus using f = 1 / 2*PI*C*R you can get the -3dB point (Hz). So my next question is how are these values (C & R) determined, are they arbitrary (as long as you get a f3 thats low enough) or is there more?

No, they're not arbitrary. The Wikipedia circuit diagram you reference is a simplistic model. However, for most circuits it's sufficient to give you an estimate. Generally speaking, "R" represents the connected load in this instance.

However, as you correctly surmise later down in your post, there is almost always a resistor to ground connected to the capacitor. This keeps voltage on the cap, which keeps it charged and also helps to bleed down the capacitor when power is disconnected. (Otherwise, stray contact could immediately discharge the cap with a resulting pop or worse.)

In modeling a source input to an amp - there's often a input-impedance-setting resistor, too. This is a resistor to ground in parallel with the pot. It causes the source to see a constant load, regardless of the volume position of the pot. If you are using Tangent's CMoy tutorial, this is "R2" and its value is 100K, in spite of a 10K pot. So, this resistor must also be accounted for in performing your frequency cutoff calculations. Quote:

My understanding is choosing a f3 point that ~10X less than the lowest frequency of interest will reduce phase distortion, i.e. for audio circuit a f3 of ~2 would keep distortions out of the 20+Hz range.

That's what's often said, but I'm not sure everyone agrees with that. More important is that IMHO, -3dB is pretty dang significant for a source component. If you select a point that's -3dB for even 10Hz, the actual response curve will begin dropping at a much higher frequency than even 20Hz. It is audible on high-quality equipment. For instance, here's an example if you actually plug-in the Gain equation itself for the circuit (the fc equation is only valid at the -3dB point):

(Note that the curves above have been corrected for the charge resistors on a BantamDAC and the input impedance-setting resistor on a Millett MAX/MiniMAX.) Quote:

I was reading Walt Jung's, "Picking Capacitors" and in it states that not only keeping a low f3 is important, but also minimizing impedance (ideally 10K or less)... opinions? This impedance I think he's referring to is the paralleled resistance of the input (pot in an amp, RL) to the "pull-down" resistor in coupling cap circuit (Rb).

It's been awhile since I read the Jung/Marsh article on caps, but my guess is that he's referring to total input impedance. The higher the impedance, more noise results. Generally speaking however, anything within 10K to 100K is OK in DIY headphone amps. At 100K, there's more noise, but at 10K you have to use much larger output coupling caps to maintain a reasonable fc. Many of us settle on 50K for that reason.
Quote:

So for my CMoY (10K pot) I'd need a ~any size Rb, but for a typical amp (50K pot) you'd need ~12K or less. So how would a 47uF & 10K(total R) for a f3 of .34Hz output coupling circuit sound/work? As I'm typically seeing Rb values ~47K and some DIY DAC's as high as 330K.

Yep. If you try some values when adding up parallel resistances, the higher the Rb, the less effect on the total input impedance. Lower values of Rb will result in significant lowering of the overal input impedance, meaning that you will have to select larger caps. This is not good, because film caps will give you the best audio, but they become prohibitively expensive and impractical at values above 1 to 10uf. So, the best compromise is what's sought.
Quote:

Now this info applied to my DAC, Wolfson WM8711 which states this as a recommend setup: I understand C1 and R1, but R2 series resistor... it's only needed for a ham-fisted user you might ground the "line" when in use, correct? Preventing frying the unit. This covers HP filters, do DAC's typically also have LP filters to prevent high frequency oscillations due to over-sampling? Any recommendations?

Yes, the series resistor is there for protection of the DAC output. Stray contact before the coupling caps are charged can result in some spikes that will destroy the DAC chip. I've done it on a few occasions with Alien DAC's, unfortunately. As you've noticed, the charge setting resistors on a DAC circuit may have relatively high resistance due to the previous explanation. I haven't run the calcs, but it's possible that the caps aren't fully charged for a half-minute or so upon power-up, due to these values of Rb in a DAC. A series resistor provides a safety valve to protect the DAC chip in those scenarios.
Quote:

I'm almost done, honestly, my last question deals with bypassing caps. I'm strictly interested in the ratio, if there is any, used between say two caps. I've read somewhere that the smaller (typically film) cap should be 1/10th or 1/100th the size of the larger cap (typically electrolytic).

Yeah, that's a rule-of-thumb, but in practice there's very little practicality in using a bypass that's smaller than 0.1uf, IMHO. Stick to values of 0.1uf to ~0.47uf for bypassing. Keep in mind that bypassing is highly subjective and non-predictable. You may find combinations that work well, but others that don't. Generally speaking, for instance, bypassing any power-type electrolytic is probably a good idea in the signal path. However, bypassing Black Gates is often a waste of time. On the other hand, if you can design the circuit components in the RC model so that "C" is a film cap and no electrolytics are used - you are well ahead of the game. Quote:

That's it, thanks for any help *edit* I actually did have one more question, where do you guys source Vitamin Q caps, I can only seem to find them on eBay for a "reasonable", i.e. cheaper than some guitar sites which are charging ~10-20 per cap. And what are reasonable prices, I'm looking for some (I think) 0.047uF.

Umm ... look to the projects in my signature and you may get an idea for sources of Vitamin Q's (Beezar.com).

However, if you are willing to purchase in quantities of 10 or 20 at a time, go to LA Surplus Electronics on ebay.

 

[bmwpowere36m3]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif This is a lot to put into one post. You might try asking one question at a time in the future. I'll try to give this a shot, though.

Yea, I usually hate when people post a ton of threads for a few questions that might be simple answers or answered a ton of times...

Quote:

Originally Posted by tomb /img/forum/go_quote.gif That's what's often said, but I'm not sure everyone agrees with that. More important is that IMHO, -3dB is pretty dang significant for a source component. If you select a point that's -3dB for even 10Hz, the actual response curve will begin dropping at a much higher frequency than even 20Hz. It is audible on high-quality equipment. For instance, here's an example if you actually plug-in the Gain equation itself for the circuit (the fc equation is only valid at the -3dB point):

So what's a good F3 point to choose, 2Hz?

Quote:

Originally Posted by tomb /img/forum/go_quote.gif It's been awhile since I read the Jung/Marsh article on caps, but my guess is that he's referring to total input impedance. The higher the impedance, more noise results. Generally speaking however, anything within 10K to 100K is OK in DIY headphone amps. At 100K, there's more noise, but at 10K you have to use much larger output coupling caps to maintain a reasonable fc. Many of us settle on 50K for that reason.

Got it!


Quote:

Originally Posted by tomb /img/forum/go_quote.gif This is not good, because film caps will give you the best audio, but they become prohibitively expensive and impractical at values above 1 to 10uf. So, the best compromise is what's sought.

How about 22uF BG Hi-Q's (only readily avail size, others too small), bypassed by .22uF VitQ or Russian PIO or bypassed by .022uF VitQ or Russian PIO?

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Yes, the series resistor is there for protection of the DAC output. Stray contact before the coupling caps are charged can result in some spikes that will destroy the DAC chip. I've done it on a few occasions with Alien DAC's, unfortunately. As you've noticed, the charge setting resistors on a DAC circuit may have relatively high resistance due to the previous explanation. I haven't run the calcs, but it's possible that the caps aren't fully charged for a half-minute or so upon power-up, due to these values of Rb in a DAC. A series resistor provides a safety valve to protect the DAC chip in those scenarios.

So you recommend installing it or omitting it to reduce signal degradation?

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Yeah, that's a rule-of-thumb, but in practice there's very little practicality in using a bypass that's smaller than 0.1uf, IMHO.

Why?

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Stick to values of 0.1uf to ~0.47uf for bypassing. Keep in mind that bypassing is highly subjective and non-predictable. You may find combinations that work well, but others that don't. Generally speaking, for instance, bypassing any power-type electrolytic is probably a good idea in the signal path. However, bypassing Black Gates is often a waste of time.

Okay

Oh, and I really appreciate your input.

 

[qusp]

OK i'm interested in this too; i'd start another thread but my situation is so similar as to be almost identical, so I hesitate to start a new thread. I hope the OP doesn't mind. i'm building a capped dock for the same purpose; (decoupling for my DIYMOD and the bantam dac(among others) and i'm using 3.3uf OIMP VCAP (with .22uf TFTF VCAPs coming soon. I'm using it with a pico (input impedance of 50k) I think i'm alright on the corner frequencies 0.9hz (or there abouts) and phase distortion ( 9hz) with those values. I hadn't been going to install those resistors though (although now i'm starting to change my mind) so let me get this straight. (not very good at all at reading circuit diagrams) so I install a the 3.3uf vcap after the lineout (but in this case also after the line in to the VCAP dock) after that cap; on each channel I install a 47k resistor (like a 47k ohm caddock ultra precision film resistor) also connected to ground to protect from voltage spikes if grounded; say by pulling the mini out to slowly etc???? and a 100k resistor like the vishay bulk foil type, figure I go all out with this one because its in the signal path and its paired with VCAPs; no point going vcap and then using a cheap resistor afterwards (where does audio ground come into this one?) this is inline with the lineout and towards the line in of the pico, which protects the souce from voltage drift? is that right?? please excuse my ignorance. i'm actually going to install the .22 or.33 teflon VCAPs after the OIMP (in series not parallel, this is how its done in the RWA VCAP DOCK) I had been going to use them as bypass caps just for something different, but have read that it produces mixed results at best. and this smaller value cap wont happen for a few weeks after I hope to have the dock set up with the OIMP's

thanks guys any info would be much appreciated, but keep in mind that as far as SMD and amp building design is concerned i'm very new to it all; modded a couple of ipods but mainly been making cables, so my maths aren't really up to speed for these purposes. I've been using an excel script to work out the values and corner frequencies needed.

and to the OP I hope you dont mind me butting in a little

 

[qusp]

theres some vitamin Q .22uf caps in the cables and tweaks for sale section here

 

[tomb]

Quote:

Originally Posted by bmwpowere36m3 /img/forum/go_quote.gif Yea, I usually hate when people post a ton of threads for a few questions that might be simple answers or answered a ton of times... So what's a good F3 point to choose, 2Hz?

Probably. If you refer to the chart I posted, the 100K and 50K curves are -3dB at 1Hz and 3Hz, respectively. They barely begin to turn down at 20Hz. 2Hz is definitely in that same region. Note that "100K" and "50K" are simply used for identification in your example. A 1Hz or 3Hz cutoff curve would have an identical shape, regardless of the specific component values (as long as the RC product was the same).
Quote:

How about 22uF BG Hi-Q's (only readily avail size, others too small), bypassed by .22uF VitQ or Russian PIO or bypassed by .022uF VitQ or Russian PIO?

Personally, I wouldn't fool with bypassing the BG's. The 22uf you mention are fine if we're talking NX Hi-Q's and it sounds like we are. VitQ's might work well if you have genuine ones, but the Russian PIO's might be a trifle slow. Either of them will be so much trouble to install due to the tremendous size difference from the little BG's that it's probably not worth the pain. I would be leary of installing any other type of film cap bypass - they probably will have no effect with the BG's. YMMV.
Quote:

So you recommend installing it or omitting it to reduce signal degradation?

Personally, I would go without a series resistor, but that's MHO and YMMV. If you are careful to always plug in the power to the DAC several seconds before connecting it to an amp, things will be fine. Or, having them all connected and powered up at the same time works, too. If you'd feel safer with them, I think 100 ohms is what I've seen mentioned elsewhere.
Quote:

Why?

I was expressing an opinion based on availability and commonality with other circuits by using a readily usable component value. Again, YMMV.

There is supposed to be a logic to bypassing: the film cap "grabs" the high frequency signals down to a "crossover point" determined by its own independent fc. However, that assumes that the film cap is so much faster than the electrolytic that it will carry all the frequencies until that cutoff point is reached. This is not assured in every case. As mentioned, BG's are often fast enough that film cap bypasses have no effect - the film cap can't wrest the high frequencies away from the BG. Other electrolytics and film cap bypasses have varying degrees of compatibility - none of which are predictable or documented except in specific instances where they've been tried first-hand.
Quote:

Okay

Again, I was expressing an opinion that may be biased toward bigger equipment (Millett MAXes). Actually, tiny film caps would be cheap enough, so it might be worth experimenting. You're probably on your own in this size range, though, so be prepared to test several combo's - with BG's alone as an important baseline. BTW, almost any quality film cap - alone - will probably sound better than even the NX Hi-Q BG. It just depends on whether you can find one with a suitable rating for the fc you need and the physical size constraints.

Also, don't neglect the possibility of running several VitQ's or other film caps in parallel (to get the necessary uf rating).
Quote:

Oh, and I really appreciate your input.

Thanks - you're welcome.

 

[tomb]

Quote:

Originally Posted by qusp /img/forum/go_quote.gif <snip> thanks guys any info would be much appreciated, but keep in mind that as far as SMD and amp building design is concerned i'm very new to it all; modded a couple of ipods but mainly been making cables, so my maths aren't really up to speed for these purposes. I've been using an excel script to work out the values and corner frequencies needed. and to the OP I hope you dont mind me butting in a little

Not sure I understand all of that. I believe I remarked in the Aussie-Bantam-Group-Buy thread that if you install output coupling caps offboard, you're sortof on your own. That means you should still attempt to replicate the circuit, meaning you should somehow have those 330K resistors to ground attached to the coupling cap output. That's what's in the BantamDAC circuit, anyway.

As for a Pico, I don't understand why you wouldn't want to use the DAC in that directly. From everything I've heard, it's an excellent DAC. Maybe you got a version that didn't have a DAC? Sorry, but I haven't kept up with the options for that amp.

Series resistors are up to you. As stated up there, I think there's some benefit to purity in the signal. The direct-from-DAC-chip capacitive-coupled circuit is often the biggest advantage with these simple DACs such as the Alien and Bantam. So, my choice (IMHO) would be to go without them, but use some care in connecting the DAC (let it power up for several seconds before connecting the output).

 

[bmwpowere36m3]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Also, don't neglect the possibility of running several VitQ's or other film caps in parallel (to get the necessary uf rating). Thanks - you're welcome.

I guess the problem I'm currently running into is my main headphone amp is a CMoY w/ a 10k pot and it needs to use high C's to maintain a decent F3 level.

Here's a little plot I did, that shows various F3 points w/ respect to Total Impedance/Resistance & Capacitance:

^^^Thus having a 10k pot, no matter the "loading resistance" in the amp the most RL you can have is ~10k and with a reasonably sized RB (~47k)... that'll drop the total resistance to ~ 8.3k. So to maintain at least a 2Hz F3, I need 10uF. Which in a film would be expensive and more over large, like you said.

I wanted to kill two birds with one stone, use this DAC setup for my current CMoY and down the road in a Millet "SS" Hybrid or a MAX. However by that point, maybe I should just look into building a standalone DAC (Gamma1 possibly).

So I guess my final question is, what's a better sacrifice... using a Blackgate and easily getting 10+uF and maintaining a low F3.... OR use standalone (or paralleled) film caps and sacrifice the F3 value.

Would metallized film caps (polypropylene reasonably priced) be better than a BG or really just the PIO film caps?

 

[tomb]

Quote:

Originally Posted by bmwpowere36m3 /img/forum/go_quote.gif <snip>So I guess my final question is, what's a better sacrifice... using a Blackgate and easily getting 10+uF and maintaining a low F3.... OR use standalone (or paralleled) film caps and sacrifice the F3 value. Would metallized film caps (polypropylene reasonably priced) be better than a BG or really just the PIO film caps?

When you're stuck with the 10K input, the 22uf BG NX Hi-Q is the simplest and cheapest way to output quality. You could probably go as low as 4.7uf with little ill effect, but that's still a huge rating when it comes to film caps - paralleled or not.

Actually, even metalized polyester is better than a BG in some circumstances. I'm currently using a 1uf MKT Vishay-Roederstein (metalized polyester) in a BantamDAC and it handily beats the BG NX Hi-Q. I'm inputting to a 50K Alps on a Millett MiniMAX, of course, so the fc is fine.

 

[qusp]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Not sure I understand all of that. I believe I remarked in the Aussie-Bantam-Group-Buy thread that if you install output coupling caps offboard, you're sortof on your own. That means you should still attempt to replicate the circuit, meaning you should somehow have those 330K resistors to ground attached to the coupling cap output. That's what's in the BantamDAC circuit, anyway.

cool well then basically i'm just replicating the circuit outside the dac in both cases of the ipod and bantam. with the ipod however I wasn't planning on using resistors at all until I decided to build the bantam and use the same dock as the caps for the bantam; also I had read elsewhere that it was a good idea, but not with resistors rated that high.. are you with me. so now it seems I have to add a 330k resistor (was planning on just the 47k) in order for it to be compatible with the bantam is that right? will this 330k resistor mean that its not able to work with the DIYMOD output? and is it just protection for swing or does it have some sonic effect as well??

Quote:

As for a Pico, I don't understand why you wouldn't want to use the DAC in that directly. From everything I've heard, it's an excellent DAC. Maybe you got a version that didn't have a DAC? Sorry, but I haven't kept up with the options for that amp.

Yes i've got the amp only pico

since the beginning tere has always been a dac/amp and an amp only version. which is why I was mentioning the corner frequencies above as calculated with the pico (50k input impedance) and the VCAP dock 3.3uf VCAP OIMP. the only other dacs i've got nstalled in my RME Fireface 400 pro audio interface (not really portable, although it can run off firewire power

so Macbook->FF400->redwine signiture 30.2->k1000 is portable if you tink like that

)



Quote:

Series resistors are up to you. As stated up there, I think there's some benefit to purity in the signal. The direct-from-DAC-chip capacitive-coupled circuit is often the biggest advantage with these simple DACs such as the Alien and Bantam. So, my choice (IMHO) would be to go without them, but use some care in connecting the DAC (let it power up for several seconds before connecting the output).

ok i'll leave them out; i'm well practiced at doing that kind of start up and shutdown procedure thing, from playing keyboards in a band, as well as running a little home studio for audio production. the thing with this is that i'm building the VCAP dock mainly for use with my DIYMOD->pico rig but thought maybe I could make it work with the bantam since I was building it anyway. but if modifying it to work with the bantam will lessen the performance with the DIYMOD then i'll forget about that and find some more suitable caps to go with the bantam. i'm using the same hammond case as the mini3 uses for the bantam so I have room for something decent.

thanks again

 

[cotdt]

On mine I used an F3 of 20Hz, it allowed me to use smaller and therefore better components than if I picked a lower F3. Your F3 does not need to be that low.

 

[qusp]

so in essence my question is: can I design this PORTABLE VCAP TYPE DOCK to be used with the DIYMOD AND the bantam dac since the DIYMOD doesn't REQUIRE (its been working fine for years for the users so far) there to be a resistor after the caps between signal and ground. at least i've never seen a dock for an IMOD that has one. my small blackgate dock certainly doesn't and the ALO ones dont either. I'm not sure whether the VCAP docks do or not but I dont think so. so basically; is this 330k value for the resistor for the bantam output circuit going to interfere with the proper operation of the VCAP dock in its task for blocking DC from my DIYMOD without causing there to be a higher corner frequency than say 1.6hz? all the while sounding as good as possible. I dont want 2 good sounding devices when the original idea was to have one great sounding device. if there is a conflict here could it be avoided by placing a switch in the path/ground placing 2 resistors of appropriate values and switching between them as needed as there will never be a need for both the ipod and the bantam to be plugged in at once?. I dunno maybe I should just buy another hammond, save the VCAPs for the DIYMOD and just buy some decent midrange caps for the bantam. how well does the bantam dac compare to the wolfston dac in the ipod? i'm just trying to utilize and economize with the limited funds I have on hand thus the whole idea of doing this in the first place. any help would be much appreciated i'm finalizing my order for the bantam and another order where I would get the appropriate high quality resistors.

thanks in advance

 

[bmwpowere36m3]

How do Sanyo OS CON's sound, I read up that they are actually really nice for DAC's...

 

[Gross]

This seems to be a fairly appropriate place to ask this Q. I work with professional sound systems on a daily basis, dealing with different things like mic and line levels, and specifically phantom power. Mic jacks can have Phantom power on them which is 48v DC on the hot and common, and using the ground as the return. There is a chance that something could be wired incorrectly, or I am dealing with a line jack that is also a mic jack on a mixer, so phantom could be turned on. Having the Coupling caps on the output of my DAC protects me for the most part, but there are occasions where a jack could be wired and there will be 48v on the ground, so I am wondering the effects of having a capacitor on the ground of my dac as well. In essence I would 3 caps, left, right and ground. Would this change the size requirements, or have any other negative effects?

Thanks.

 

[qusp]

Quote:

Originally Posted by bmwpowere36m3 /img/forum/go_quote.gif How do Sanyo OS CON's sound, I read up that they are actually really nice for DAC's...

havent used them; i'm only slowly working my way through the endless supply of different caps. even some that weren't even designed for audio turn out to sound great. mundorf SIO and SGIO are great for a mid-high priced cap. Red Wine Audio use them in their amps and they give a really hefty but detailed SQ. I think they would suit you well, but maybe in the higher end. you'de have to try and get them in bulk. Even mundorf supreme sound great. check'm out