Mic preamp homebrew, or "How did I get talked into this??"

[CurtZHP]

After building several mic preamps from kits and plans, I decided to take the plunge and attempt to build one completely from scratch.  I figured, "I know just enough about electronics to be dangerous.  What could possibly go wrong??"

I happened upon a few old UTC transformers at the radio station where I work, and figured I'd incorporate one or two in the design.  I also wanted to keep it simple by using just one dual op amp (for two gain stages).  I started breadboarding the thing around a 5532, but I'd like to use an OPA2604 (I happen to have a few).

Anyway, I started out with the UTC.  The one I have has a 50 Ohm primary and a 500 ohm secondary.  I take one side of the secondary to ground; the other side goes to the non-inverting input on the op amp.  Based on an old audio circuit book I found, I put a 20K resistor across the output and the inverting input.  Also on the inverting input is a 1k resistor in series with a .1 uF cap to ground.  This feedback network is supposed to give me a gain of 20 if I read it right.

I connected a scope to the output and fed the input with my Neutrik Minirator, using a sine wave at -40dBu.

The scope shows a clean sine on the input pin, but when I move to the output pin, the wave is the same amplitude (size), just fuzzier; like there's no gain, just noise being added.  I tried another 5532 with the same result.

Putting in a 10K resistor instead of 20K cleaned up the waveform, but there's still no additional gain.

Running that output to the second input on the 5532 and using a larger feedback resistor gives more gain, but it looks terrible as far as noise goes.  I haven't actually listened to the output.  I'm afraid to!

What am I doing wrong??

:-[

 

[abechap024]

Try replacing the .1uf with something larger say at least 10uf...

If I'm thinking correctly your op-amp should be amplifying but its only amplifying the frequency that can squeeze through the .1uf cap...hence the fuzzy sine wave...

check this out
http://www.waltzingbear.com/Schematics/API/API_312.htm

 

[CurtZHP]

That was it!  Not only was I looking at the old schematic in the book, I was comparing it to a few others and got my schematics mixed up.  The one in the book called for a 15uF cap there.  Another design was calling for something else, but it was an entirely different concept.  I found a 10uF and that made a huge difference (obviously!).  I found something way bigger, but that didn't seem to make any improvements, so 15uF it is!

Thanks for winching me out of the ditch!

;D

 

[PRR]

> a 1k resistor in series with a .1 uF cap

1K resistor needs 10uFd cap to reach -3dB at 17Hz, an audio benchmark.

Just a mark on the bench. The system may need other values. Usually you want much-much less than -1dB at 20Hz, even with multiple stages, so each stage should be closer to 2Hz. 1K and 100uFd. OTOH when tirmming the flab from a guitar amp's voice you may want to shave bass. 1K and 2uFd, 1uFd, whatever.

Likewise for other resistors:
100 ohms - 100uFd
100K - 0.1uFd

But not 0.1uFd. Against 1K, that cuts everything under 1,700Hz, most of the audio.

> 20K ... 1k resistor .... supposed to give me a gain of 20 if I read it right.

Actualy 21. But I applaud simple rounding.

> sine on the input
> the wave is the same amplitude

What frequency ? ? ?

At 1KHz, you have resistor-ratio says gain of 20, but you are about 2:1 below the 1K+0.1u cutoff at 1.7KHz, so gain is half of that: nearer 10.

 

[CurtZHP]

Putting a 10uF cap in there definitely made a big difference, but I've got some larger ones around looking for a home.  I ran the tone generator from 20Hz to 20kHz and the waveform on the scope stayed pretty solid, shrinking ever so slightly below 40Hz.  Flat enough for me!

Now I need to figure out how to control gain.  My first thought was to make the feedback resistor on the second stage a pot, but I've read that varying the resistance on the feedback loop can sometimes mess with the frequency response.  Any truth to that?  The other idea I saw was simply putting a gain pot between the two stages, possibly with a decent coupling capacitor to prevent any DC silliness.

Right now the thing just has two gain stages with no control.  The first stage presently has a gain of about 10 (not counting what the input transformer is supplying), and the second is set for a gain of 100 (100K feedback resistor). 

I plugged a mic in and sent the output to a mixer.  Needless to say, it had buckets of gain; so much so that anything louder than a stage whisper clipped the snot out of it.

Here's another wrinkle....
Not sure if this is right, but the output transformer (UTC A-21) is presently connected as a 200 Ohm primary and 200 Ohm secondary.  Seems to do the job of providing a balanced output, but I'm wondering if that's going to be the correct impedance for things down the line.

I thought about just doing a simple cap coupled output, but then I'd need to put in another op amp to invert the signal for the "-" output.  Don't really want to add any more active parts.

 

[abbey road d enfer]

Putting a 10uF cap in there definitely made a big difference, but I've got some larger ones around looking for a home.  I ran the tone generator from 20Hz to 20kHz and the waveform on the scope stayed pretty solid, shrinking ever so slightly below 40Hz.  Flat enough for me!

Now I need to figure out how to control gain.  My first thought was to make the feedback resistor on the second stage a pot, but I've read that varying the resistance on the feedback loop can sometimes mess with the frequency response.  Any truth to that? 

Somewhat true... Make it too high and the opamp will run out of juice at HF so you'll have plenty of midrange gain and bad treble; make it too small and the opamp may enter into oscillation if it is not stable enough. But in many cases, it will work, within limits. At least, that's how it works in thousands of Neve, SSL, MCI,Harrison...

The other idea I saw was simply putting a gain pot between the two stages, possibly with a decent coupling capacitor to prevent any DC silliness.

That's an other option. The problem is there'll be situations where you experience clipping or noise problems.

Right now the thing just has two gain stages with no control.  The first stage presently has a gain of about 10 (not counting what the input transformer is supplying), and the second is set for a gain of 100 (100K feedback resistor). 

That's about 75-80dB overall gain; way too much for most common needs.

Here's another wrinkle....
Not sure if this is right, but the output transformer (UTC A-21) is presently connected as a 200 Ohm primary and 200 Ohm secondary.  Seems to do the job of providing a balanced output, but I'm wondering if that's going to be the correct impedance for things down the line.

I would be much more concerned about how your second gain stage feels about driving a low-Z xfmr. A 5534 is generally capable of doing it, but it depends on the real parameters (inductance) of the xfmr. Regarding the actual load at the secondary, a typical lowZ xfmr does not require to be matched (loadZ = ratedZ) or Zobel'd (RC load for taming HF resonance).

I thought about just doing a simple cap coupled output, but then I'd need to put in another op amp to invert the signal for the "-" output. 

No you don't. You can run the output impedance balanced. Takes just an additional R and C.

 

[CurtZHP]

Well, I backed off on the 100K feedback resistor on the second gain stage.  It's down to a 40K now. 

Just for kicks I tried a 10K pot between gain stages for level control.  Not really digging it, so out it goes.  At this point I'm considering making the second stage feedback resistor a pot for that purpose.  Not sure why putting it in between stages didn't work for me.  I have another preamp I built from a set of plans that does the same thing (transformer coupled input too), and that works great.  (It's also a much more complicated design.)  Got some clipping without what seemed to me to be a lot of level.

I'm also starting to kick around the idea of having a single stage.  Not sure if I can pull that off successfully with such limited knowledge, though. 

The only other major problem at this point is hum.  Probably a ground loop somewhere.  Also, with nothing plugged into the input, the thing is noisy.  Like buzzing, humming, crackling noisy!

I've got a pencil drawn schematic of this mess.  I think I'll scan it and post it for your amusement.

 

[CurtZHP]

Scary, ain't it!

 

[abbey road d enfer]

Running that output to the second input on the 5532 and using a larger feedback resistor gives more gain, but it looks terrible as far as noise goes.  I haven't actually listened to the output.  I'm afraid to!

Moving from 100k to 40k doesn't make such a big difference. It's only 7dB less. So you still have too much gain. I reckon you should be doing fine with another 10k there.
Have you used a linear or log pot? The problem with linear is that the gain jumps at you in the first 1/8th of rotation.

 

[Samuel Groner]

I suggest you have a look here: http://www.groupdiy.com/index.php?topic=16537.0

IMO the best way to control gain with a single pot.

Samuel

 

[Speedskater]

Your schematic needs two different ground symbols. XLR pin #1 connects to chassis ground. The transformers, P.S. and pot connect to audio ground. Then one wire connects audio ground to the chassis.

 

[MagnetoSound]

(... also you seem to have an extraneous resistor in parallel with the LEVEL pot.)

 

[CurtZHP]

Moving from 100k to 40k doesn't make such a big difference. It's only 7dB less. So you still have too much gain. I reckon you should be doing fine with another 10k there.
Have you used a linear or log pot? The problem with linear is that the gain jumps at you in the first 1/8th of rotation.

I've been trying several different values, but that makes sense.  I believe it's an audio pot.  Listening to the output, it sounds pretty smooth over its entire travel.  (I think it came out of an old console.)

Your schematic needs two different ground symbols. XLR pin #1 connects to chassis ground. The transformers, P.S. and pot connect to audio ground. Then one wire connects audio ground to the chassis.

That's probably why the thing hums like crazy.  While breadboarding it, I was just hooking up grounds wherever I needed one.  Speaking of grounds, there is an error in the schematic regarding the input transformer.  It does not actually have a separate ground from the secondary winding.  When I sketched the input stage, the transformers were reversed.  The output transformer does have a separate ground lug.

(... also you seem to have an extraneous resistor in parallel with the LEVEL pot.)

That was a holdover from the old textbook schematic I was borrowing from for the input stage.  It was a 560 Ohm resistor, but it didn't seem to make any difference whether it was there or not.  I even put in something much larger at one point, just to see what would happen, and that didn't seem to matter either.

 

[CurtZHP]

I tried putting 10K feedback resistors on both stages, and even at maximum the gain was inadequate, so I put a 20K on the second stage.  That looks good.  Right now I'm not bothering with any sort of level control until I solve a few other problems.

I also swapped out the input transformer (UTC O-12) with another A-21, same as the output, just tapped for a 50Ohm primary and 500Ohm secondary.  I was getting some really nasty noise with the o-12 in there.  Putting in the A-21 cleaned up a lot of that.

I also completely dismantled what I had breadboarded and redid it, cleaning up the ground scheme.  That got rid of the loudest hum, but there's still a bit of background noise.  I'm also faintly picking up the radio station where I work (not a big surprise considering the transmitter is just 5 miles away with nothing in the way).  It also still generates some nasty hash when there's nothing plugged in.

I noticed that the noise changed somewhat when I tried different caps in the second stage feedback network (the schematic has a 10uF there).  I guess I just need to do some filtering here and there.

I also did some checking into these transformers.  The UTC A-21 is described as a line matching transformer.  I'm now wondering if this is at all suitable for my needs.

 

[CurtZHP]

I'm reposting the schematic to reflect some changes:

1) Input transformer now the same type as output
2) Audio ground and chassis ground now differentiated
3) Got rid of cap in second stage feedback network.  It didn't seem to matter.
4) Added ferrite beads to secondary of input and primary of output transformers.  Otherwise the thing made one heck of an FM receiver.

What you can't see on the schematic is that I've connected the level pot to the circuit using Belden 8451 2-conductor shielded cable with the drain wire connected to ground.  I tried a 47K pot instead of the 10K, and it seemed to get noisier around the middle of it's travel.  (Could have been the pot itself.  It wasn't as good a quality as the 10K I have.)

Overall, the thing seems to sound pretty good so far.  I've still got the tiniest bit of hum listening over headphones.  It's definitely coming from the preamp, because when I drop that fader on the mixer it's plugged into the hum goes away (Mackie 1402).  I'm thinking it's just because the thing is on a breadboard and not installed in a properly shielded metal case.  For a power supply, I'm using the 15V +/- supply built into the breadboard, but I intend to build a proper supply for it, including phantom power.  Right now I'm only able to test it with a dynamic mic.

The other thing that I've noticed, comparing this to the other schematics I've been studying, is that there's only one capacitor in the whole thing.  Is that normal?  I'm kicking around putting some filter caps across the supply rails to ground just to keep junk off the rails.  Should I consider putting them elsewhere?

 

[Speedskater]

Still have a problem on the output XLR pin 1.  The wire should only go to the upside down Christmas tree ground.

 

[Speedskater]

Something like this.
Didn't have time to look for better symbols.

 

[CurtZHP]

Much prettier than my chicken scratch! 

I attached XLR pin 1 on the output directly to the "chassis" ground, and that didn't really seem to make any difference.  Then when I accidentally held the mic too close to the side of the powered breadboard, this tiny hum got noticeably louder.  I guess what I'm hearing is coming from the power transformer inside the box.  Unfortunately, I don't have a nice toroid laying around to test it with; but I do intend to build a nicer power supply to go with this thing (with phantom power). 

So I guess I've got the basic circuit down.  Now time to add some refinements:

1) +/- power supply with phantom power.
2) Bass roll off.  Not sure where to include this.  I was thinking between stages, just south of the pot.
3) Phase reverse.  One schematic I saw had this after the secondary on the output transformer.  I figure that's better than running tiny mic level signals through a toggle switch.
4) Maybe a 20dB pad?  Not sure where to put that either.  Maybe I shouldn't even bother.
5) Find a decent case to put it in.