Alien DAC v1.1 Construction Thread - Page 87

As long as you have caps in there, and don't parallel them with a dead short, you won't defeat the circuit protection (i.e., DC blocking) role of the coupling caps, so play around to your heart's content. If you are still nervous, just check the DC offset on both channels with your DMM each time.

Those are the .18uF Vit Qs from Beezar.

Tom, Your consistent persistence and ability to voluntarily and gladly help out here doesn't cease to amaze me. I'm like, <half your age and am usually too tired to do anything else but what I have to, lately. muchos kudos man!

Calling me an old man, huh? OK - I admit it - I am definitely part of Ingwe's Team OAF.

Seriously, thank you for your kind words. I've been busy with the Alien DAC, too - out of necessity. As mentioned over in the SSMH thread, I blew out a couple of my Aliens - the ES version being one of them. I needed quite a few sources for the Atlanta Meet held back on the 5th, so I replaced my ES Alien with a slightly different version in the same Penguin tin. It's packed in pretty good. I had to pop out the lid some, but a little strategic bending and the lid is still quite secure. What you see there are six 0.22uf VitQ's in parallel - three per channel.

'Course, I'm coming to this late - FallenAngel and some others have been using appropriately sized film caps and removing the electrolytic output caps long before I tried this one. The question was whether paralleling several VitQ's would have the same effect. Dsavitsk revealed to me lately that he's been paralleling VitQ's successfully and thought they sounded better than one big one. So, I thought I'd try this - it worked out very well. There seems to be no lack of bass at 0.66uf per channel hooked to a MAX, while the mids and highs are superlative. Quite simply, it's the finest sounding Alien DAC I've built.

Hi TomB,

Very cool build -- I was eying a tin of about that size myself tonight, wondering if I could fit some VitQs in with the Alien...

Did you remove the electrolytics at CL/CR and replace them with the VitQs? Do they provide enough DC protection?

Thanks,

__Roy

Wow, you really out Vitamin Qed me. That looks sweet.
What value are you using for R15/16? With a standard input impedance of 50k it seems you'd want something in the >=100k range. Even still you end up with a cutoff freq of about 10Hz, but I would imagine that is OK.

EDIT: I see Jeff (Glass Jar) uses 332k - that should do it.
EDIT 2: That'll do 5.5Hz cutoff - should be fine

Yes and yes.

Please let me warn you though - this was not easy packing them in that tin. I essentially had to bend the tin sideways at the output jack and then screw the ring down until it popped into place, then bent the tin back. Even then, I had to bend the solder lugs back almost at right angles so that it would fit between the edge of the tin and the VitQ's. The bottom three VitQ's connect to the CL pads from the bottom up, while the top three VitQ's connect to the CR pads from the top.

Thanks!
Yep - it's all stock except for the VitQ's at CL and CR. (Well, I used 1000uf for the power cap and 330uf everywhere else, but had tested those values earlier on another Alien.)

Ack, I meant to post that in the MHM thread :P nonetheless, you've contributed here too though. Not any less worthy of praise are quiet a few other good folk around here. Pars, Error401, and others.. Thaks for making this place what it is!

I sort of thought maybe you posted that because of all the Vitamin Q's floating around here lately. That, or either Pars' or Error401's first name is Tom, too.

At the risk of asking too much, really:

* DC offset. I confess, I think I understand what this would be (+/- signal moving wave into all + or - voltage?), but I am not certain and have no idea how to measure it or what the cause can be and why the coupling caps prevent it. Obviously this is a pretty basic audio circuit issue and perhaps there is a place on line already that explains this?

* Replacing the CL/CR electrolytics. What general guidelines can anyone offer about replacing these with film caps (VitQs, etc.) while preserving the DC protection the electrolytics offer. I'm currently bypassing the stock 33uF electolytics that came with the glass jar kit with .18 1000V VitQs. TomB has replaced them entirely with 3x(.22uF in parallel)=.66uF (correct math?) on each channel VitQs. Is there a range I should stay inside to experiment with here? I have some nice 1uF and 2uF 600V film caps from other projects, could I try them without risking the amp I connect to? Or are they way too high capacitance?

Again, thanks in advance for your generous and ongoing support for beginners on this forum.

__Roy

DC offset is a quiescent state measurement (i.e., no signal present, as in no music playing, etc.). It is simply a static measurement of the amplification/output section. It is measured using a voltmeter (DMM) between the left output to ground, and the right output to ground (each channel separately).

Capacitors block DC... unless the cap has failed, it will not pass a DC voltage. Capacitors will pass AC voltages (music, etc.). This is the basic property of capacitors that makes them useful in circuits.

I don't want to get into cap wars too much, other than to say that in terms of the actual device approaching the ideal model, an electrolytic cap is awful, particularly when compared to most film caps. Most people think that film caps sound much better than electrolytics when used as coupling caps. Alot of tube amp builders are actually using film caps as the PSU caps as well (an amp with NO electrolytic caps in it). You could try your 1uf and 2uf film caps without worry.

DC offset refers to a constant voltage offset on the signal. Rather than swinging around 0V like an ideal signal, there is a constant offset which brings the centrepoint of the AC signal up. If your amplifier doesn't block it, the DC component will be amplified too, and driven into your headphones/speakers. Since it doesn't represent audio it's unwanted, and can harm the headphone/speaker by driving a constant current through it - this displaces the cone and can heat the voice coil enough to damage it.

The use of coupling capacitors fundamental electronics issue, and goes back to the basic function of a capacitor. Whenever you place a capacitor in series like these coupling capacitors, you end up with an RC circuit formed by the series capacitor and the input impedance of the circuit. This RC circuit is a high-pass filter with a 6dB/octave rolloff below the breakpoint - DC is completely blocked. To put it simply, capacitors pass high frequencies and block low frequencies. The larger the capacitor, the lower it goes before it starts to block the signal.

'AC coupled' circuits block DC and only pass the AC portion of a signal. 'DC coupled' means that there are no DC blocking elements in the circuit.

If you want to learn more, read any electronics textbook [/quote]

The DC protection will be present with any capacitor, since the RC filter impedance increases with decreasing frequency at 6dB/octave, and since DC represents effectively an 'infinitely low' frequency, the impedance of the filter will be infinite as well no matter the value, blocking the DC.

You can calculate what value is appropriate if you know the input impedance of your amplifier and how low you want to have a good response. The point at which the filter as a -3dB response (called the breakpoint) is given by:

Where R is the input impedance of your amplifier and C is the capacitor value in Farads (so divide your uF value by 1,000,000). You probably want the -3dB point to be about an octave below the lowest frequency you want to keep to ensure flat response. If you don't know the input impedance, 10K is a conservative guess that should work well with pretty much anything. The lower it is though, the larger caps you'll need, so film caps become unwieldy and electrolytics are used, like here.

DC offset occurs because the typical power for IC circuits and PC's is +5VDC. Audio is AC; the signal "swings" between an upper voltage and a lower voltage. Ideally, that lower voltage is negative with pure AC. However, the way the PCM chip does it is by swinging between 62% of that +5V and 50% of the +5V. What that means is the bottom of the signal wave is ~+2.5VDC. Hence, there is 2.5VDC offset before anything is connected.

Capacitors block DC, while fully passing AC. This allows us to use capacitors on the sound output of the PCM chip to zero out that +2.5VDC and allows only the AC music signal to pass. There's a lot of devil in those details, though - the caps must be charged first (DC offset is still present until then), caps have their own sound signature they impart to the music signal, and caps form an RC circuit with the resistance of the load - this causes the bass frequencies to be cut off.

As mentioned above, the output capacitors form an RC circuit with the load to which it's connected. Select a capacitor with a rating too low and you will lose some of the bass frequencies.

You need to know the input impedance of your amplifier in the case of the Alien DAC. Most of the time, that's the impedance of the volume pot. For instance, in the case of the Millett MAX, a 50 Kohm pot is used for the volume control (ALPS RK27). The equation for determining the bass cutoff is:

f = 1/(2*pi*C*R), where f is the frequency in Hertz, C is capacitance in farads, and R is in ohms.

Rearranging to solve for "C", we have: C = 1/(2*pi*R*f).

Let's say we want to be safe and want a cap that will give us at least 10Hz. The reason we want to do this is because the filtering starts at frequencies above that. "f" is only where the drop is down -3dB. So, solving for "C", we might have:

C = 1/(2*pi*50000*10) = 0.3 x 10E-6, or 0.3uf.

This is saying that a 0.3uf capacitor on the output of the Alien, connected to an amp with a 50K volume pot, should have a -3dB drop in frequency response at 10Hz.

OK - hope that helped.

PRE-EDIT: I see that we've had Pars and error401 post before me. I typed all of this in the meantime, so I'll post it anyway. Maybe it will help to have different perspectives on the same subject.

You guys are wonderfully generous. Thanks to all three of you.

__Roy

Alright, I already have a question. In a circuit like the Alien DAC, why would you bypass CL/CR instead of replacing them? This is assuming I have correctly understood "bypassing" to mean connecting another capacitor in parallel. I see that different knowledgeable folks have done both things.