Just wondering... how to: digitally programmable gain on MIC inputs?

[Yansi]

So if I understand that correctly, it may be done like so:

input -> selectable 20dB pad --> fixed 20dB low noise amp -> selectable 20dB amp --> ADC driver

Otherwise I do not have enough imagination.

Considering a typical ADC full scale voltage is about 2Vrms, I can achieve these ranges, using the solution above:

overall gain 0dB (PAD is on, gain fixed 20dB)
overall gain 20dB (pad is off, gain fixed 20dB)
overall fain  40dB (pad is off, gain fixed 20 + another 20)

However I do see one major drawback of this, and that is that I am in fact reducing the dynamic range by 20dB with this, each time another gain step is switched in or out, the signal at the ADC will get 20dB higher/lower.

Or am I missing something? (probably am, this ain't workie workie)

//Just digging through the Qu16 manual, it has a 65dB range of analog gain control in 1dB steps: https://www.allen-heath.com/media/Qu-16-v1.8-Chrome-Technical-Datasheet.pdf
I smell a PGA2500 in there plus a selectable 15dB pad in front of it! But I might be wrong as well...

//EDIT2: Wow, they must smoke some good material at the Mackie plant. The manual is almost fun to read, some parts are funny as hell: https://mackie.com/sites/default/files/PRODUCT%20RESOURCES/MANUALS/Owners_Manuals/DL16S_DL32S_OM.pdf
The Mackie DL32S has analog gain range of 60dB in 3dB steps.

[Bassman59]

Hello,

I have randomly came across this very interesting component: http://www.thatcorp.com/datashts/THAT_5171_Datasheet.pdf
http://www.aes-media.org/sections/pnw/ppt/other/low_cost_programmable_microphone_preamp_gain_control.pdf

I have always wondered, how do they implement programmable gains on the MIC inputs in all those digital audio mixing desks and racks. So now I know... or don't I?

The 5171 (which is actually a two-chip solution as it needs the 1580 amplifier) and the TI PGA2500 are both good choices, easy to implement, and they use a small amount of board space.

If you look under the hood of a Behringer X-32/Midas M-32, you'll see basically the same discrete-transistor long-tailed pair in front of an op-amp preamp circuit that's been used since the 70s in pretty much every British mixing console and a lot of others. This is basically the same circuit that is in the TI INA163 and THAT1510 mic preamp chips. In all of these cases, the gain is set with a single resistor between the emitters of the long-tail pair of transistors.

That single-resistor for gain set is usually implemented with a 5k or 10k reverse-log potentiometer (in series with a big DC blocking cap).

What's done in the Behringer design, and I've done something similar, is to choose precision (or precise-enough) resistors in the correct binary steps to get the gain you want, and switch them in with a DG4xx-type of analog mux. You can change the op-amp's feedback resistor to extend your gain range. (For example, if you have a 10k feedback resistor, switch another 10k resistor in parallel to drop the feedback resistance to 5k.)

This requires a micro to control all of the mux controls. Some of the analog muxes have I2C for control, so you can easily daisy chain a few of them.

[langwadt]

I do not understand your comment, Mr PacketHead. Could you please elaborate?

what would you rather do: package 10 chips in an envelope ship it and make $50 or put 1M chips in a box ship it and make $1M ?

[Yansi]

How do you then explain with your kind of logic, that a pack of few 2N3904s still costs a few pennies? I don't think that packaging a few of those chips cost them a whole lot more, than packaging those transistors. Otherwise they would not sell transistors cheap, would they?

On the other hand, I understand well, that they might divide the packaging cost unevenly across different parts. However pricing of some parts seem insane anyway. It's just what I call "price of exclusivity", or you may want to look at that as cdev does.

Also interesting to note, you first started with a "customer support" and now we are talking about packaging cost.

[langwadt]

How do you then explain with your kind of logic, that a pack of few 2N3904s still costs a few pennies? I don't think that packaging a few of those chips cost them a whole lot more, than packaging those transistors. Otherwise they would not sell transistors cheap, would they?

On the other hand, I understand well, that they might divide the packaging cost unevenly across different parts. However pricing of some parts seem insane anyway. It's just what I call "price of exclusivity", or you may want to look at that as cdev does.

Also interesting to note, you first started with a "customer support" and now we are talking about packaging cost.

it comes down to the same, how much work need to be done per sale