OPA Alice in a split config for upright bass mic?

[ccaudle]

@joulupukki did you notice the earlier note about pins 2 and 3 of U1.1 connected in reverse? It is still incorrect in that latest drawing.

 

[joulupukki]

I'm assuming the electrolytic is C3? ..... in which case yes, it can be rated at 25v.

I see you've corrected the schematic, to connect R1 to the half rail (which you call 'VGND' ).
The op amp connections are still wrong.... As Voyager10 pointed out in post #6:
•The capsule /R1 junction should be connected to pin 3, and not pin 2.
• Pins 1 and 2 are connected together.... Not 1 and 3 as you have at the moment.

Thanks. I'll get the OpAmp connections updated soon!

 

[Voyager10]

Just looking again, pin 4 of the op-amp, the negative supply, has ended up connected to 'VGND', not the GND input.

 

[joulupukki]

Speaking of GND and VGND, should the ring around the capsule connect to GND or VGND?

 

[joulupukki]

Ok. Fixed pins 2 and 3 and also the GND connection on the opamp. Thanks for catching that! This also has the capsules ground connected to circuit ground and not VGND. Hopefully this is at least the right circuit. These would be teeny boards!

 

[ccaudle]

These would be teeny boards!

If you take voyager's advice then C3 could be a surface mount ceramic cap, and the capsule board would be even smaller.

 

[joulupukki]

Ok, so I renamed C3 to C2. C3 is now on the body board.

R4 and R5 are now 1M and C2 is a 1uF MLCC. Does it matter that C2 is not polarized in this case (since it replaced a 47uF polarized electrolytic cap)?

The capsule board is 14.5 x 14.5 mm. The body board is 20 x 44 mm.

 

[jp8]

Speaking of GND and VGND, should the ring around the capsule connect to GND or VGND?

See my post #7. I would propose to connect it to Vgnd.

The idea behind this is that the OPA circuit amplifies the signal between its input node (pin 3) and its AC reference node, which is Vgnd. Ideally, C2 shorts Vgnd to GND for AC signals and there should be no difference, but as C2 has to be X7R at this value, capacitor microphonics and piezo voltages would add up to the capsule signal. To avoid that, connect capsule chassis (= capsule ground) to Vgnd. Any voltage on Vgnd caused by said microphonics or piezo effect will appear as common mode voltage on your preamp input and will for the most part be rejected.

Jan

 

[joulupukki]

Lastly, depending on the length of the cable between mic and power board, the capacitive load on the OPA output may exceed 100pF, which could cause instability or overshoot issues. Consider splitting up the 47R resistors into 2 x 24R, with one 24R on the mic board.

It wouldn’t matter that 1 set of the 24R resistors would be on the other side of the 47uF caps (C4 and C5)? I assume you’re referring to R9 and R10?

See my post #7. I would propose to connect it to Vgnd.

The idea behind this is that the OPA circuit amplifies the signal between its input node (pin 3) and its AC reference node, which is Vgnd. Ideally, C2 shorts Vgnd to GND for AC signals and there should be no difference, but as C2 has to be X7R at this value, capacitor microphonics and piezo voltages would add up to the capsule signal. To avoid that, connect capsule chassis (= capsule ground) to Vgnd. Any voltage on Vgnd caused by said microphonics or piezo effect will appear as common mode voltage on your preamp input and will for the most part be rejected.

If I take a look at this board: https://www.jlielectronics.com/diy-accessories/opa-alice-single-channel/

… it looks like that board is connecting GND from the capsule to common board GND. VGND from the op amp runs to the side of the 1G resistor not connected to the capsule signal wire but not over to the other wire coming from the capsule. That’s what made me wonder. The original schematic and board on this page also show capsule running to board GND: https://www.instructables.com/OPA-Based-Alice-Microphones-a-Cardioid-and-a-Figur/

I guess one thing I could do is put two different pads on the board. One for VGND and one for GND and I could try both and see which one seems to work best.

BTW, what do you mean by “X7R” ?

You also mentioned earlier, “It will also deprive you from the possibility to add a capacitor shunting R3 to add some high-end roll-off.”

To achieve a low pass filter, would I insert a capacitor on the OPA side of R3 to VGND or GND?

Or could I put a band pass filter on the body PCB, something like this?

 

[jp8]

It wouldn’t matter that 1 set of the 24R resistors would be on the other side of the 47uF caps (C4 and C5)? I assume you’re referring to R9 and R10?

Nope, it wouldn't matter.

I guess one thing I could do is put two different pads on the board. One for VGND and one for GND and I could try both and see which one seems to work best.

If you have the space, you could do that, indeed. If the capacitor were an ideal zero impedance path for AC signals between Vgnd and GND, then it wouldn't matter whether the capsule ground were tied to Vgnd or GND. In the original design, the 47 uF elcap comes closer to a zero Ohm impedance than the 1uF ceramic cap, so you might as well tie the capsule ground node to GND. And if the capsule touches the metal mic housing, it's the only choice you have.

BTW, what do you mean by “X7R” ?

X7R refers to the dielectric material of the ceramic capacitor. You'd typically only want to use C0G (aka NP0) in audio circuits because they are about the most stable capacitors around and do not add distortion and don't have piezo-electric properties. But they only come in small capacities. X7R packs more capacity in the same or smaller ackage, but its capacitance changes with temperature, DC voltage, frequency, mechanical stress (microphonics!) and about any other physical quantity you can think of. And it's piezo-electric. If you have space for a small 1uF 25V elcap or a 1206 size MLCC capacitor, then you could opt for that, instead of a 1uF X7R type MLCC. The largest capacitance with C0G dielectric in 1206 package is 470nF. If you further increase the voltage divider resistors to 2.2Meg, then you maintain the same cut-off frequency.

To achieve a low pass filter, would I insert a capacitor on the OPA side of R3 to VGND or GND?

No, you would shunt R3 with a capacitor, creating an integrating low-pass function. You could try and simulate that in LTspice. It is not the same as in the U87 circuit, because you only include the low-pass function in the inverting "leg" of the circuit. But you can achive a few dBs high-end roll-off.

Or could I put a band pass filter on the body PCB, something like this?

I wouldn't do that. R11 and R13 would add noise and attenuate the signal significantly. Suppose your preamp has an input impedance of 2k, which is not uncommon, you would lose 6dB. And SNR would degrade proportionally. The only resistors you want in your mic output are in 22...75 Ohm range, preventing oscillations of the mic output stage due to the capacitive load of cables and RFI filters.

Jan

 

[Voyager10]

If you needed to apply some EQ, the best low-noise option would be a capacitor across R3. It would give you a -6dB shelving EQ effect, as half the output voltage (on OUT2) is rolled off towards zero at high frequencies.

(Unconventional but not actually wrong - as long as the impedances on pins 2 and 3 are the same, the output is still 'balanced').