Reference mic-pre

Hey gang,

A while back I enquired the group about techniques used for measuring the frequency response of speakers in a room. Thanks for everyone's input. (original thread) http://www.groupdiy.com/index.php?topic=29112

I have been busy reading everything I can get my eyes on and playing around with an extra set of room EQ's, a few software based RTA's, measurement solutions and a pair of Earthworks M30 measurement microphones, making adjustments to the EQ's, taking measurements and listing to the end results. So far so good. :grin:

What I want to do next is build a pair of simple ultra low distortion/ flat frequency response mic pre's to mate the the M30 microphones. I did a bit of datasheet surfing/ engineering and came up with this.

http://conwayrecording.gotdns.com/~benmook/personal/diy/ref-pre.jpg
{over-large image changed to link --PRR}

All IC's are the new LME49870, http://www.national.com/ds/LM/LME49870.pdf

supply is +/- 16V (probably will be in a API rack)

I haven't figured out the right resistor values for the stepped gain, but it will be equivalent to a 5K.

Phantom is always on. Do I need diodes on the input? Is there an easy way to get rid of the coupling caps?

Power connections are not shown for simplicity. Will probably be 10R resistors to some .1 decoupling caps.

Hoping for someone with much more experience to tell me what's fundamentally wrong with this.

From a quick glance R1 and R5 at 10k will be effectively in series with the mic and add a lot of noise.

They don't need to be such a large value.

JR

What do R1 R5 do? (EDIT: John beat me.)

Why do you need ultra low distortion to measure speaker response?


If not for the Phantom, I'd just wire a TL072 as a single-ended booster. If the room electrical field is too nasty for a short unbalanced line, you will probably have troubles before you get to the loudspeakers.

Or get a MIDIman AudioBuddy dual preamp, $99. Not the lowest noise, but lower than many rooms, and quite flat response (flatter than any loudspeaker).

> Is there an easy way to get rid of the coupling caps?

Transformer.

If you have researched many mike-amps, you will note that "all" transformerless phantom-power inputs use caps. I can think of an exception or two, very far-fringe stuff.

From a quick glance R1 and R5 at 10k will be effectively in series with the mic and add a lot of noise.

They don't need to be such a large value.

Click to expand...

Right. I thought that was high. I have no idea what needs to be there, if anything. 10R?

Why do you need ultra low distortion to measure speaker response?

Click to expand...

Your right. A TL072 is probably quieter than a power amp, but shouldn't I try to get the best signal to my converter?

The LME49870 datasheet claims .00003% THD into 600 ohms. I thought it would be interesting to use that chip, and @ $2.63 a pop it's not going to hurt my weekend plans.

Or get a MIDIman AudioBuddy dual preamp, $99. Not the lowest noise, but lower than many rooms, and quite flat response (flatter than any loudspeaker).

Click to expand...

Insanity is upon me. I have a problem where I make things harder than they should be so staying up all night listing to the radio while routing PCB's is extremely relaxing.

Hi Ben,

why not use one of the mic pres in the studio? I've used GMLs and Millennia for this in the past. Those pres are probably +/- 0.5 dB in the range that counts for these kinds of measurement. Even in a very very good room frequency response errors are of a much higher magnitude making such a deviation from perfect irrelevant. I'm not sure there is anything that a custom built preamp of any description can bring to this party.

Spent a long time doing this kind of stuff with a friend of mine who had an ugly room. We used DPA omni mics into a Millennia mostly into his PT interface. As we are on mac we used Fuzzmeasure pro and took a million response measurements as we changed speaker and listening positions and moved treatment. With these techniques and a lot of treatment he turned a pig into a really nice room. We experimented with eqing his monitor feeds but gave up on that as a course of action.

I'm pretty sure the rooms in your place of employment are not pigs!

Cheers,
Ruairi

As others already have at least partially pointed out, you should have a clear idea of what you actually need for acoustic measurements. I'd say it is:

* reasonably flat frequency response (say 0.1 dB, although 0.5 dB would likely do as well) from 20 Hz to 20 kHz, at all used gains
* precisely settable gain

Noise and distortion are only minor concerns, as for the later the mic will dominate anyway and the former does not really influence the measurements. What I would build for this: a THAT 1510, with gain settable to 20 dB, 30 dB and 40 dB (add 10 dB and 50 dB if you like). Use parallel/series connected resistors to make gain precise, that simplifies some calculations.

With respect to your design: it is actually worse than a simple one-chip solution (as suggested above) with respect to several aspects:
* noise is several dB higher
* CMRR as shown will be relatively poor as the feedback network of IC3 is untrimmed
* output voltage offset will be rather high

The LME49870 datasheet claims 0.00003% THD into 600 Ohms.

Click to expand...

Don't be fooled by this number. It's a long story, but your design as shown will never come anywhere close in achieving such low THD.

Samuel

[quote author="sonicmook56"]
The LME49870 datasheet claims .00003% THD into 600 ohms.
[/quote]

If you go past the first page marketing stunt and look at the curves provided you can only get that distortion figure under very specific operating conditions. Star alignment is one of them.

> staying up all night listing to the radio while routing PCB's is extremely relaxing.

It is your life. Squander it any way which pleases you.

But back on Tue Sep 30, 2008 you wanted to move-on with measuring speaker-room response. This is a problem with no true solution. And a "better approximation" takes a LOT of time.

To be blunt: Grab a preamp which don't suck and get on with the REAL problem.

It takes a lifetime to begin to know room acoustics. The best geeks are old, and experienced enough to do more with two rulers than some young kid can do with all the dream-gear on the market.

The gear is a learning tool, also cross-check on what experienced ears suspect; also good for dazzling clients with piles of print-outs.

Flattening the room response is useful, but often not essential. Your ears "learn" an acoustic space. If you take that radio into the room and move both it and you around, you will know the room. (IMHO, sitting with monitors and ears locked in place is the real problem.)

Room sound-bounce pattern is also important. Again, your ears will learn (more so if not locked in triangle with speakers) but certain patterns assist the ears' judgement.

Most rooms are just too damm small. And way too low. It is like looking at the Mona Lisa while locked in a broom closet. I'd generally prefer a large cheap untreated room to a too-small room loaded with $9K of foam and diffusers.

I'd just throw down a THAT 1512 and be done with it.

Your original circuit is pretty much an instrumentation amp, same as the THAT preamp IC but the 1512 has all the hard work done for you already.

[quote author="Svart"]I'd just throw down a THAT 1512 and be done with it.

Your original circuit is pretty much an instrumentation amp, same as the THAT preamp IC but the 1512 has all the hard work done for you already.[/quote]
hey svart

i have see some where the $5 pre amp... that use THAT 1512

but they use batteries....
i plan to get a cheap blue tube pre... and ripp off the guts...
now i have a box some knobs and the xlr jacks...

what PS you suggest for the that 1512...
the circuit is the one THAT publish as an example
Greets
Rolo.

http://www.forsselltech.com/gerbers/MikePrePS1.PDF

BUT, you simply can't beat Keith's design:

http://www.beatbazar.com/guests/ssltech/kps-1/index.htm