simple AC-coupling a THAT1512 / That1510 output - cap value?

long-winded intro:
I bought a fancy converter for doing podcasting at work. The metric halo 2882. Looks amazing right? I thought it would be the perfect piece of gear for our mics. But then I discover, after purchasing, the preamps only provide 40dB of gain. That won't work because all we have are dynamic mics, and the voices are entering the computer much too quiet.

The budget is mostly busted on the converter purchase. So my only option I can think of is to buy a behringer preamp box for $99 - the thought of which pains me to no end. Then I remembered the so-called $5 mic pre.

simple question:
I want to try the most simple schematic from the 1510 datasheet., figure 2, reproduced below. What do I use for AC-coupling the output? (I assume just a polarized electrolytic, but what value?) And is that coupling capacitor even necessary if I build these with fixed gain? I'm assuming the metric halo has some kind of coupling of it's own. Any thoughts?

Other than the connection to the protection di-
odes, the 1510/1512 input pins are connected only to
the bases of their respective input devices. For
proper operation, the bases must be provided a
source of dc bias that will maintain the inputs within
the IC’s input common-mode range. Two different
schemes for this are shown in Figures 2 and 3. Fig-
ure 2 is simple, but its output will need to be
ac- coupled to the next stage, particularly if RG is vari-
able to allow gain adjustment. 1510/1512 input stage.

Click to expand...

Here's figure 2 as referenced in the datasheet quote above:

Hi Kato,

Svart recently did up a pre using the 1512, home etch files and all:
http://www.groupdiy.com/index.php?topic=32025.0

cheers,
dave

Davo beat me to it..

;D

Ah cool. I see you did all the work for me already.

I was actually planning to do the whorehouse version just like in the schematic above.:
  •  fixed gain
  •  no phantom power
  •  no balanced output

Then if that worked, I was going to slowly add features one by one till it was a respectable mic pre.
I did a quick layout last night - five caps, three resistors, 1 chip - and the whole thing fit on a 2" x 1" board. Quick build I'd think.

Or I could just do it right and use your layout.   thanks.

[removed question - looked more closely at svart's schematic]

Kato

I see Svart chose 100uF as cap value for AC coupling that 1510 output to that 1646 input

Am I right thinking that this output capacitor forms a high-pass filter with the that 1646 input impedance ?
http://en.wikipedia.org/wiki/High-pass_filter

That 1646 input impedance seems to be 5k as seen in
http://www.thatcorp.com/datashts/1600data.pdf

So I found cutoff is -3db at 0.318309881481454Hz (from the formula described in wickipedia link),
does that seem correct ?

If yes, is there an optimized cutoff frequency range to let the low frequencies go through the DC blocking caps ?
I mean here, -3db at 0.31Hz seems low enough  But what values range could be considered as optimized for good bass response ?

uh, that isn't good enough??

;D

'lytic coupling caps are cheap enough so that you should just pick the biggest/best you can afford and fit on the board.  220uf works well with most BJT opamp input impedances.  So does 100uf and 470uf.  I don't see much use in going larger than that.

Thanks Svart, sure it is good enough 
I was thinking more about the upper frequency edge though,
like if it would be acceptable to have -3db cut at 10Hz for example,
but I see your point from the economic side, bigger caps are not so expensive
and I guess the better is to test the cap in circuit and hear the result

Even if the -3 point is .3hz and or the pole is at something like 2hz, there isn't even a remote chance of you hearing that.  I contend that nothing below 10hz is useful.

I'd go so far as to say 'signal below 10hz is bad!'

So all that's needed is 100uF for AC coupling? That's small enough I might as well just wedge it into my board. If I find it's not needed, I'll just jump that space. Thanks for all the input.

Layout per schematic in first post.
Let me know if anyone catches fatal errors.




The schem has one spot going to chassis ground, and all others electrical ground. Usually connect those, right?
I made a chassis ground pad that can be jumped to signal ground if needed.