Care to review an aplication-specific preamp design for me?

Hi all.

I have to build a preamp for a specific application. It's a matched-gain 4-channel preamp to take an unbalanced, "higher-than mic-level-but-not-quite-line-level" signal from a soundfield mic (ST250) and make it go (balanced out) straight into a multi-channel recorder.

The output of the ST250 has +25V on the outputs (an anomaly of its design) and so there are blocking caps at the input, not shown.

I have built the last 2/3rds of this design (basically everything from R5 onwards) and it works well and tracks to a fraction of a dB, using a 4-channel pot which I found to do the job.

My problem is that the ST250 overloads rather easily when it gets close to a brass band if they happen to be playing at anything like full chatter, or in a 'loud' room. If I switch in the "attenuate" button, then the gain range of my currently built unit is not high enough (it currently ranges about ±11dB) to use the full bit range of the recorder, so I want to build a matched but variable gain stage ahead of the exicsting stage.

With the current unit (the last 2/3rds of the schematic pictured below) noise is not a problem (not since I de-mounted the regulated power supply) and I tried 5532s and 072s and measured THD and noise on both chips. In some gain settings, there was not much to call between them, but at one setting, the 072 showed almost 10dB lower noise, so I went with the 072s.

This is a 4-channel design, here is one channel shown:



Subtleties like local decoupling and inter-stage DC-blocking caps are not shown, for clarity.

The gain error between channels absolutely must be kept minimal at all times, and using a 4-gang pot, the inevitable mechanical resolution errors mean that the more gain you "put around" that section, the more inter-channel tracking error you can expect. I can't afford that, since this is for a "B-format" SoundField application, so a seperate, switched +10dB and +20dB section would be perfect, since 1% or better metal film resistors will match much better.

Can anyone see any great flaws with this approach? -I need it to be buildable quickly, easily and using what I have or can get promptly. -I have a few 4-gang double-throw switches, so if this is a reasonable idea I'll go right ahead with it.

I'm not a great designer, and I respect the ideas and noise theory knowledge (of which I have a woefully inadequate grasp) of folks like PRR and bcarso, so I though I'd post it to let everyone here treat it like a pinata for a while! :wink:

Thanks for lookin'

Keef

I'd a.c. couple into the second stage to reduce wiper noise in the pot. And then measure the exact end-to-end R of each pot. section and adjust your R5 and R6 propotionally---that will help your tracking among channels.

There will be a little wiper noise due to the d.c. offset of the second 072, but since you won't be adjusting this much in real time it shouldn't be a big problem. The bias current is negligible for this FET amp, which is helpful. You might also want a small C from output to inv input of the second op amp to compensate for the peaking due to the highish impedances in the input/feedback network.



Brad

Cheers, Brad.

Yes, as I mentioned, I'd omitted the AC coupling cap from the schematic.

The second stage works great, there is indeed a little wiper "scuff", but it peaks down at about 40dBFS and -as you correctly observe- it won't get the chance to happen during takes, since the gain will be set and left alone.

Do the numbers look okay for the front end? -By that I mean the resistor ladder around the two switches. -I don't care if it's exactly 10dB or 20dB, I just need it in the ballpark so that I can range up or down and use the variable second stage to take care of trimming the rest.

The output Z of the ST250 is 750Ω each channel, and the cables feeding the preamp will be very short (12" long).

The 25V DC on the ST250 outputs is due to it's various powering options: As well as AC mains, it can run internal from batteries or phantom power. the path that takes DC from the XLR mic outputs also allows DC to "bleed back" when the unit is under AC power however, so a 25V electrolytic and a good DC path to ground at the negative plate takes care of all that for me, right at the input XLR.

Nice thinking about scaling the 4 gangs of the pot padding resistors. -Without scaling it's well under a dB error, but you're almost certainly right, and I could probably reduce that still further; Thanks for the tip!

Keith

I'm wondering why you build up the signal and then knock it down.

The second stage has gain of 1/0.5/0.1 (up/mid/down). If you need 2V at the output to whack your bits into the recorder, you might need 20V out of the first stage.

> the inevitable mechanical resolution errors mean that the more gain you "put around" that section, the more inter-channel tracking error you can expect.

I don't believe this is necessarily true. You can have sloppy loss, too.

I'd be thinking one stage with a 6-position switch, 6dB steps, 30dB range. if you get within 6dB of full-scale you have all but one of your bits, and you probably have plenty anyway. A 4-pole 6-throw switch is not a standard part, but since you won't be gain-setting "live", you could use two 2-pole 6-throw switches, which used to be a Radio Shack standard. Then selected 5% resistors give gain-match to a fraction of a dB, without pot-gang tracking error.

[quote author="PRR"]I'm wondering why you build up the signal and then knock it down.[/quote]

I found that there is gain in the balanced output driver. The overall gain of the second and stages together is about +12/+2/-8 at top/mid/bottom gain settings.

I originally built the second stage as a +10/0/-10 then followed it with a balanced driver, but there was too much gain. Since I can't mess with the guts of the SSM2214 I opted to drop it a bit at stage 2. -I need some ability to drop gain a few dB for when the mic is switched to stereo output, then channels 1 and 3 are left and right outputs and have some more 'oomph' to them.

Keith

[quote author="PRR"]The second stage has gain of 1/0.5/0.1 (up/mid/down). If you need 2V at the output to whack your bits into the recorder, you might need 20V out of the first stage.[/quote]

Hopefully I wouldn't be of a mind to boost the extra 10dB then wind it all off again at the variable stage! :wink:

Keith