Questions about THAT1512 in mic pre

[hendrik]

Hi, I am new here. Before signing up, I read a number of threads on this forum and now I have a few question I hope the experts here can help answer.

I want to implement a 0 dB ... 60 dB microphone preamplifier using an anti-log 10 kΩ potentiometer, going from the output of the THAT1512 to an unbalanced TRS output via three different low-cut filters as shown in the diagram below. I've read through Wayne's thread about the improved servo and I prefer this one over the floating version that flies along with phantom power. However, I am not sure why I would need to servo the output offset, given that I can easily use high-pass filters at the output (and need to do anyways for the low cut filters). Given that I want to use a symmetric power supply, I don't think there'd be a power-on plop. Is this just a no-go for audio purists or are there other more fundamental reasons why I would want to servo the offset?



I've also seen Joe's implementation and the corresponding schematic. However, I couldn't find any description of the input section to the THAT1512. Is there any available somewhere? I understand the symmetric setup on RG and the 15 µH inductors to reduce RF susceptibility but the section between the relay switches and C1 (180 pF) is not clear to me.



In this implementation I wonder how symmetry is ensured, given that inductors have a lousy tolerance of the DC resistance. If I want to use a 5 Ω resistor to achieve a 60 dB gain, I have to subtract the DC resistances of the inductors but the most reasonably sized I could find for an SMD implementation of the preamp is a NLFV32T-150K-EF. This one has a typical resistance of 300 mΩ ±20 %. So that means for the actual resistor I have to choose two 2.5 Ω - 300 mΩ = 2.2 Ω resistors. At the given tolerance, the actual RG would be somewhere between 4.836 Ω and 5.164 Ω for a tolerance of ±1 % of the 2.5 Ω resistors. So the actual gain is lies somewhere between 59.7 dB and 60.3 dB which, I guess, isn't too bad.

Coming back to my question about whether the servos are needed: given that I use low-cut filters at the output of the THAT1512, is the feedback servo really needed? Would I experience some sort of clicking if I use a scratchy pot (as mentioned somewhere in one of Wayne's threads)?

Another question: in the schematic of the THAT1512 demo board (see here), there's C25 (220 pF) between IN+ and IN-. This appears to have been added after the fact because it shows up in the revision history of that document but with no explanation what this capacitor is inserted for. I noticed that Joe's design has this cap as well (though it's a 180 pF, so pretty similar). But as mentioned before, I can't find any description for his version of the mic preamp either.

BTW, I haven't decided yet whether to go with a massive 6800 µF or the servo'd version (if actual servoing is actually needed given my questions above). Any comments or recommendations would be nice.

[mediatechnology]

Hi hendrik thank you for joining us!

Joe's mic preamp was designed for use in Sam's remote truck where he worked with a lot of very long mic lines. For this reason there are a lot of inductors to reduce RF interference. You mention that you won't be requiring a balanced output so I conclude you won't be working with 500 foot mic lines.

To keep it simple I would eliminate the inductors both at the input and Rg lines and also use a capacitor in the Rgain line without any servos. Since you're using a pot for gain control having Cg keeps DC out of the wiper better than a servo.

C1 in Joe's preamp (180pF) or C25 on the THAT demo board are a good to have installed. I'm not sure why THAT added C25 as a revision rather than it being in the original. C14-C16 appear on both Joe's preamp and THAT's demo board also to provide RF protection. A Wye network is used to reduce the common mode capacitance. The 10Ω and diodes provide phantom power protection.

With regard to the input network you may have other questions that I haven't addressed.

If the outputs of the HP filter directly feed an output you'll want to add build-out resistor to isolate them from capacitive load.

[mediatechnology]

Another thing worth pointing out is that your gain pot will have a minimum "hop-off" resistance in the wiper.
On a new pot this is usually 2-3Ω.
At maximum gain the control may also be a bit twichy.
Since you're wanting to use a 1512 with a pot for Rg you might be better off putting some gain after the 1512, in the filters, and reduce the maximum gain in the 1512 to place less demands on the gain pot.

[hendrik]

Many thanks for your very valuable feedback. OK, now I know the background for Joe's design. Indeed, in my case the mic cable isn't expected to be more than a few feet long. So it appears best to go with the input stage as shown in Fig. 5 of the datasheet:



There's one thing that caught my eye though: both Joe's as well as the demo board design have a 1.2 kΩ (or 1.15 kΩ) resistor from the IN+/IN- lines directly to ground, whereas the datasheet has them tied to a single 22 kΩ resistor before going to ground (see R1, R2 and R7 above). What's the reason for that and which configuration should I choose?

OK, I get your point about the gain pot oddities and the idea behind the gain offset and the maximum gain. Thanks for that info.

What's still not clear to me, given that I follow with low-cut filters, why would I need the large CG?

I'm also not sure why I need a build-out resistor. If I understand it correctly, you are referring to a resistor at the output of the THAT1512 to prevent from oscillations caused by high capacitive load put directly at the output. Shouldn't that, however, be taken care of by the grounded resistors following the capacitors? Maybe I am missing something.

One thing I wanted to check: what's the maximum expected voltage difference between RG1 and RG2? Do I really need to go with the recommended 6800 µF at 10 V? A 6.3 V capacitor should also do, right?

[mediatechnology]

The 22KΩ resistor to ground raises the common mode impedance of the input which can improve LF common mode rejection by reducing the matching requirements of the input capacitors.

You can relax the value of Cg considerably owing to both the HP filters you intend to use and also by reducing the maximum gain in the 1512 to reduce adjustment sensitivity at high gain.

The build-out resistors are in series with the final output op amp(s) to provide resistive isolation between the op amp and cable capacitance. The resistors I believe you are referring to are shunts which discharge the output coupling caps.

In normal operation Cg has very little terminal voltage but during a phantom fault the peak voltage can be quite high. A 6.3V unit should be OK however.

[hendrik]

Once again, thanks a bunch for your reply. Now I understand your point about the build-out resistor.