dynamic mic preamps

[sean k]

A good friend, whos just starting out in electronics, wrote me yesterday about a simple preamp for dynamic mics and so I went searching and most of what I found were dual supply balanced input preamps which varied from fairly simple to kinda stupidly complex.

But I did come accross one or two which were intrigue-ing (don't know how to spell that!) and I've been drawing stuff up that looks kinda silly... but simple.

The basic premise is having the two inputs go separately into that configuration where the input is at the emitter and the output is at the collector which works as low/high impedance converter. Then the problem is how to merge the two out of phase signals into one combined signals and I started looking at AB1 amps where the input is split into two out of phase signals then recombined as a combination in the AB1 output and I'm thinking that that is what I want to recombine the out of phase signals from the impedance converter stage.


What confuses me is how to bias the bases on the PNP and the NPN and whether or not it will even work. Any ideas?

[tempus]

Read this:
http://sound.westhost.com/project66.htm

[JFX09]

Hello,

I have this on the breadboard:



and had to say it works lovingly, thought i'd let you know

got it from there: http://www.zen22142.zen.co.uk/Circuits/Audio/audio.html

Cheers and Merry Christmas !

jf

[Nasse]

http://www.diystompboxes.com/smfforum/index.php?topic=77912.0
http://www.audiomasterclass.com/arc.cfm?a=giant-killing-$5-mic-preamp-its-secrets-revealed

[JFX09]

But those are dual supply for condenser mics, no?

This one is extremely simple and works really well with my cheapie  Realistic mic, quiet and everything. I used a TLo72 because I didn't have a tlo71 at hand.

[PRR]

Short cables in electrically-quiet rooms: dynamics can very often be worked unbalanced. If you can get away with it, this is MUCH simpler. Any simple 2-transistor amplifier will do the job. For best noise from 200 ohm sources, run the first transistor hotter than you would on a phono preamp. Rod's #13 is an example.

Transformers are your friend. Radio Shed has an XLR-1/4" adaptor with a 1:7 transformer. This gives true floating input, some voltage gain, and a 1-transistor booster would make consumer line level.

> input is at the emitter ... which works as low/high impedance converter.

You do NOT need or want to have a low-Z input. Most dynamics are specified for >2K load. This gives maximum voltage transfer, and the mike is usually tuned to give flattest (or intended) response this way. (SM57 sometimes sounds "better" with a load; it is expedient to use a hi-Z preamp and add a 200 ohm resistor when desired.)

> I started looking at AB1 amps where the input is split

That's clever.... but it won't work as shown. Imagine the input makes Q1 collector rise and Q2 collector fall (as it should). Then -both- Q3 and Q4 turn off. Conversely the other way, -both- Q3 and Q4 turn on. While these stages need out-of-phase relative to each one's rail, because the rails go up and down, they really need "in phase".

And your bias probably won't work. Q3 and Q4 need quite different base voltages. You can do it with three resistors, rail-to-rail. But since the larger plan has issues, no point in doing that here.

Building a differential-input unbalanced-output amp with decent gain, usable CMRR, and OK linearity is not trivial. For a few pesos more, you can wire stuff like Rod's #66, which is rock solid. To run it single-supply, cap-couple the ins and outs, and tie his "GND" to half voltage.

[Eb7+9]

PRR once mentioned an idea for a discrete mic Preamp at Prodigy Pro ...
the circuit was Spiced for more headroom using typical device models and adjusted to work on +/- 9V

PRR-Allison Transformer-less MIC Preamp

to make the circuit run clean some degree of device matching is mandatory

Marshall Leach's BJT matching technique

[PRR]

> those are dual supply for condenser mics, no?

No. Different issues.

"Dual supply" comes from a concept of keeping signal centered around zero Volts DC, so you do not need coupling caps (or transformers) everywhere.

That actually conflicts with Phantom Power, where your mike-lines carry power, and sit at some positive (and uncertain) voltage. Preamps adapted for Phantom mikes almost invariably use input caps or transformer.

Many older preamps and systems then fed the amplifier +24V or +44V; later the fashion was to power with +/-15V or so. Rod's #13 is single 30V supply. The "$5 preamp" is from the dual-supply school of thought.

Aside from the possible coloration of coupling caps (note that Sean is not afraid to use them, they ARE useful parts), in "large" systems it is actually -cheaper- to throw in a dual-supply because you can avoid so many coupling caps. But Sean is probably not building a 24-input system (yet). For one/few inputs, single supply is nice.


> PRR once mentioned an idea

Huh? I didn't even recognize it. Honored that you(?) re-drew it so neat. But...

Ugh. Not my idea. Someone else had the front-end from some preamp (in fact Rod's #66) but didn't want to finish the job with chips. I just pointed out that it could be cascaded to get to useful gain, maybe sketched to know how bad it would be. Way too many transistors. Only made sense for the original guy who already had the layout, or spare boards, or something.

22K output resistors with 9V supply and 1uFd caps seems weak for modern 10K-22K inputs.... seems like there is better real-world grunt without those emitter followers?

> to make the circuit run clean some degree of device matching is mandatory

I don't see that. It should run clean at low gain and low level. It must get pretty sloppy at high gain or level. If SPICE says matching helps this plan a "lot", SPICE is probably unrealistic.

Whatever. I had thought (though not clearly stated) that Sean did not want to fool with "dual supply". Maybe that's me; I think they are overdone. However Sean did sketch-up a nice single-supply preamp (but it won't work as drawn).

JF's plan will work lovingly, as he says. Some situations may want the 22K changed higher. It isn't the perfect every-purpose preamp, but much good music has been captured well through lesser stuff than this.

Rod's #13 is simple, well-tested, has a good explanation, can be understood by somebody who is "starting out in electronics". Almost any modern Silicon NPN will work fine.

[Eb7+9]

I just pointed out ...

considering the characteristics of those pairs your off-hand comment struck me as a possible way of developing a fair amount of linear-ish gain while running open-loop ... of course, added degeneration is required in the second stage for this to work right

22K output resistors with 9V supply and 1uFd caps seems weak for modern 10K-22K inputs.... seems like there is better real-world grunt without those emitter followers?

recall that Darlington followers are converting the resistance seen at the base side to the emitter side divided down by Beta^2

assuming Beta=100, that's better than (2k2)/10,000=0.2 ohms AC of incremental source resistance each ...
or 0.4 ohms AC total

with higher gain devices that figure drops - already that's plenty good in a balanced environment ...
the idea was to avoid op-amps throughout while reaching for reasonably similar incremental drive levels

the 22k resistors where chosen to establish 1mA of bias current in the output devices
thus allowing for your customary 100uA (10%) pk load current limit

this corresponds to crossing a 1 volt pk signal with a 10k load - a modestly good target

I don't see that. It should run clean at low gain and low level. It must get pretty sloppy at high gain or level. If SPICE says matching helps this plan a "lot", SPICE is probably unrealistic.

there's a strong relation that exists between common-mode resistance and the linearity of the pair's DC-transfer curve - being passively biased it's more sensitive to device and component mismatch ... per se, Spice doesn't indicate that until you start playing with mismatched modeling ... the sensitivity connection is something I landed on by accident when I first started off-trailing with Spice years ago ... turns out the Skizlai pair is more advantageous than a common differential pair in this regard - that's what caught my attention

all I did in Spice this time was widen the linear range of the two stages relative to Allison's original values - the second stage receiving fixed degeneration ... I built several units using randomly picked devices and the circuit became unusable to varying degrees as you'd expect

here's an example with a Sennheiser e609 dynamic mic on a dobro using a fair bit of gain going into a 10k ADC :

PRR-AllisonPreamp demo mp3

the noise floor is nice and low - the response very natural
smokes my ADM 780's