somewhat OT: DIY condenser mic

[mdh]

I'm thinking about making some DIY condenser microphones, based on the Alice Microphone, and I'm wondering about the JFET preamp that was used in that mic (schematic here). In particular, is the 1 gigohm gate resistor really necessary (mic capsule spec here)? I ask because that's a rare and exceedingly expensive value (a 1/4W 1G resistor from Mouser is about $7), and I'm inclined to guess that it was chosen in an effort to make the input impedance as high as possible, without regard to practicality or necessity.

Any thoughts? Is a $7 1G resistor just mojo, or does it matter?

[Mark Hammer]

I'll freely confess to not being an expert on microphone matters, but I find it hard to imagine how much benefit there might be to having a 1G resistor there, as opposed to three or four 22M resistors in series.  I mean, whatever 1G might add to something like top end or noise specs, can be easily taken away simply by the quality of one's construction of the chassis, right?

Certainly sounds nice for a $7.50 microphone.

[mdh]

Indeed, it does sound nice.  And it looks nice, too, though the looks, not to mention the 1G resistor, jack up the construction costs a bit beyond the $7.50 for the capsule.  But if you can resist making things too pretty (I usually can), you could build yourself a nice arsenal of mics for recording.

[Mark Hammer]

Some folks have used cigar tubes for this purpose.  Although lord knows how much the sort of cigar that comes with a nice screw-top aluminum case of appropriate dimensions might set you back these days.

Too many folks are acquainted with the little electret capsule from their inclusion in various things like cassette recorders, answering machines, or other devices with "built-in recording".  My own experience is that usually the physical placement of the capsule is simply poorly thought-out, with the result that all sorts of horrible resonances and bandwidth limitations are introduced that make the capsule sound worse than it really is.  Also note that the full bandwidth of an electret capsule will not necessarily show up on a cheap cassette recorder or digital answering machine.  I'm sure a Neumann U47 would sound pretty crappy in those circumstances too.  Those same cheap capsules, with favourable placement, suitable electronics, and a non-challenging playback medium, can actually sound pretty good.  Not as good as the recordings you linked to, but certainly better than you'd think, given their cost.

[earthtonesaudio]

High impedance?  1Gohm?  Sounds like a good candidate for this month's FX-X topic (bootstrapping).

[ACS]

Wow, you know I'd never even considered building a mic... that looks eminently do-able though...  Trouble for me would be getting parts here in Aus!!

I feel my next project coming on 

[sean k]

It's been a long time since I was into this stuff , and it wasn't electrets but the big diaphrapm stuff, but if I remember rightly the big, big resistors on the cap were about getting voltage on the cap without current while also being about the caps seeing a high enough impedance so the lower frequencies didn't get lost in the roll-off.

So why not just a get a few big resistors, say a few 10M and see how the response is. And if you don't intend to go really low anyways then why bother with a super wide bandwidth you ain't gonna use.

Wow, those builders have a busload of dynamic mics 

[Gus]

google DIY microphones etc

I build condenser microphone tube and solid state for fun.  Every part matters and you need to think about what choices you make when you trying to build the best.

look for stuff at the neumann site

an106

other forums

There are reasons for the 1 gig and there are some old tube CF microphones that have bootstrapping

100meg to 3 gig is common in modern microphones grid/gate to ground etc.

There is a classic old model that did have 8 meg and 30 meg grid to ground but that was for other reasons as best I can tell

The Alice is a variation of a very clever Schoeps design.  A lot of transformerless microphones from China use a variation of this Schoeps circuit.

All kinds of details that matter at the capsule to circuit interface with condenser microphones.

What are you going to use it for?

[sean k]

I had a quick look at what prices are like for high Meg resistors down in this bottom end of the world and your 7 bucks for 1G is a bargain. I could get a 100M, .8W, for 5 bucks, which is 2.50 in US dollars, but the 1G is 24 dollars NZ and 4W 

[Mark Hammer]

So what is the the highest "common value" resistor typically available?  I know I've got plenty of 22M 1/4W in my parts bin, but how much higher than that do they go at pennies-per-resistor?

[earthtonesaudio]

Here's a DIY idea, don't know how practical...
Find some of that "conductive foam" like they use to ship static-sensitive devices (the pink stuff).
Figure out exactly how conductive it is (gigohms per inch or something).
Cut a strip to the right length for the resistance you want, and attach it to your circuit somehow.

[Paul Perry (Frostwave)]

What I'd like to know, is WHY the resistor needs to be so high.
I can see why you need it at all - no happiness can come from having a fet with a floating gate - But what is the difference between say 10 M and 1,000M?
Suppose the capsule has a capacitance of only 0.5pf, then at 20KHz that's an impedance of more than 100M. So I suppose that would be it.

I see in the notes to the Alice mic, that the ultra high impedance leads to problems with humidity causing leakage paths, so the critical parts have to be coated.
I think the quality of the resistor is going to matter. I don't see using stray bits of conducting foam, or even a pencil line (back in the old days, in an emergency, people used to DRAW grid leak resistors on a valve base! True.)

Anyway, there is a thread here http://www.diyaudio.com/forums/showthread/t-28870.html that shows some light.

[sean k]

Kinda off topic, but I bought my first temperature compensated resistor today. A 1k .1 percent with a rating of +/- 3ppm/C where an average 1/2 W metal film has a ppm of +/- 100/C. It was 14 bucks NZ. Don't ask me what ppm is though.

[Paul Perry (Frostwave)]

Parts per million.
I can't imagine what you are going to do with it, though

[George Giblet]

> Suppose the capsule has a capacitance of only 0.5pf, then at 20KHz that's an impedance of more than 100M. So I suppose that would be it.

I thought the condensor capacitance formed a *high-pass* filter so we are interested in sub 20Hz.

This article implies the JFET input impedance is typically a few hundred Mega Ohm and the resulting response is a few hundred Hz (implying a capsule capacitance of say 3pF.)

https://www.national.com/appinfo/amps/files/aes2003.pdf

So I'm thinking:
- 1G Ohm might be overkill from a practical perspective.   
- a JFET alone might work - I seem to remember electret capsules operating this way.
- bootstrapping might be the go

[Gus]

Go to the neumann site.  Find the PDFs look for "Microphones" very good write up.

A capsule holds a charge and if the capsule does not have the charge leaked away, the voltage at the "skin"  has to change because the "skin" moves closer and farther away from the back plate
charge voltage X (change in capsule cap/total capsule cap)   is the capsule output into a very high load resistor value

Then there are charge amps

Also look at were the noise is at a min and gate resistor value in AN106

IIRC old AKG c60s had bootstrapped CFs.  The way to go it the way that gives the sound you want.

I have read were on the web that some old tube microphones had grid resistors made from a line of India ink on a substrate.   Look for the yahoo group micbuilders? and the lab at prodigy pro(go to the meta meta maybe one of the best places for info on microphones for the DIYer) and homerecording  sometimes people do group buys to get a good price on parts.

[Paul Perry (Frostwave)]

- a JFET alone might work - I seem to remember electret capsules operating this way.

Yes, the Panasonic works this way - but there is a trick to it, apparently the fet has to be specially designed to self bias (at least, that is my reading of it). That is according to a comment in the thread that I linked to above.

[George Giblet]

> apparently the fet has to be specially designed to self bias (at least, that is my reading of it)

> Those little Panasonic capsules actually let the gate float until the drain-souce leakage just balances the gate-source forward bias (just barely in enhancement), clever trick but lousy for dynamic range.

It seems to me the balance of leakage effectively puts Vgs = 0 (actually slightly positive).

Paul, I tried to understand the unconnected gate JFET some years ago.   One view I had was the channel-gate diode sits at near zero bias voltage and that would mean the gate sits at potential (in the true sense) which is effectively a voltage tap alone the channel.  That would make the gate quite positive and would not help dynamic range.  In fact it wouldn't take much signal to forward bias the gate-channel diode and that would drop the impedance (at least for transients).

It's case where less is complicated!

[StephenGiles]

You could ask over at prodigy pro.
http://www.prodigy-pro.com/forum/viewforum.php?f=2

[mdh]

Wow, this generated a whole lot of replies while I was cooking dinner and overnight Thanks, everyone. And thanks for the references, Gus. I knew I was in for a lot of reading, which I don't mind, but I was having a hard time finding where to look, aside from the general JFET preamp stuff that's out there.  I haven't had time to look at it yet, but I suspect that Neumann write up will be very valuable.  I'm basically just interested in seeing what I can produce in the way of a set of mics for home recording.  They could be used for anything -- acoustic guitar, micing cabs, saxophone, vocals, and I'm not at all averse to tweaking for different situations.  Drums are quite a bit less likely since I don't have a kit and I don't play them.  The thing is, I intend to do this in my particular take on the DIY spirit, which is to make some stuff, have some fun, and see what it can do.  I'm not going to freak out if I don't wind up with results comparable to a $1000 mic for $20.