My latest thing

[mayfly]

So with the sessions for Cornflower Blue's third album coming up, I've been experimenting with recording techniques.  Especially those that capture a stereo image.  I'm thinking I'd like to record all the instruments in stereo, moving them around in acoustic space by literally moving them rather than using the pan pot.  the goal is a more natural sound than what we have achieved before.

The only issue with groovy stereo microphone techniques is that you need multi-pattern mics.  So the TLM102 that we used on the last record is not on for this one.  Buying similar quality mics from the same manufacturer is verrry expensive.  So, I decided to do what I always seem to do.  Build some!  If you hang out in GroupDIY enough, you get to know some characters there and what they do.  A lot of them make parts for a variety of mics.

I've managed to just now accumulate all the required parts to make a couple of U87 clones:

the capsules are from a guy named Eric, the boards from a guy named Dany, and the mic bodies and transformers are from a guy named Chunger.  Oh - and the box of parts is from digikey.

The plan is to transform all this into a couple of useful tools for recording.  Here we go!

 

[swarfrat]

I was all pumped about mid side recording when I realized the figure 8 gets only room sounds, not source spacing.  And my recording space sucks. Still a cool project though.

Right now I'm focused on developing ideas rather than demos.

 

[Cagey]

"Micraphone" parts? Somebody needs a spanking!  :icon_biggrin:

This should be interesting. I'll be looking forward to hearing your results. In my mind, it sounds sorta like a waste of time, But I'm certainly not a recording engineer and the only way to learn is to try things. The benefit I could see from it is to change the ambiance of different tunes, but I'm pretty sure you could do that successfully with panning. Of course, things bounce differently so you could end up with some subtle tonal enhancements that simply panning wouldn't produce.

 

[swarfrat]

Sources? I found a place just selling the capsules, and place selling kits with bodies, but they only had a handful that didn't really interest me. (Either tube or cardiod only).

 

[AndyG]

The only issue with groovy stereo microphone techniques is that you need multi-pattern mics.

Not necessarily .... both spaced pair and x-y techniques require 2 cardioid mics.  The figure 8/cardioid combo is what you would need for the Blumlein (mid-side) technique.

All 3 techniques have their merits, depending on your source and/or recording space.

But building a couple of U87 clones sounds like a cool project ... and I'm guessing will cost maybe a third of what one U87 would cost you new!

 

[Re-Pete]

Judging by what I've seen from the OP on this forum alone, I'm sure this will be a resounding success. If it was me attempting this, I'd probably put the wires on the wrong way around and blow up the house's fuses and/or my home recording system!

Will be interesting to watch these come to light, Mayfly. Please post some 'in progress' pics.  :icon_thumright:

 

[mayfly]

Hi Folks,

Things are moving right along in the mic department.  First thing I needed to do was to find some matched components, particularly a pair of resistors and some mosftets. 

The resistors are 2.2k ones for the phantom power in the mic and they must be matched to within 0.4%.  You can't go out and buy those, so instead I bought about 20 1% versions and set to match em!  I needed a matched pair for each mic.

Well, I have to tell you - my old fluke 73 was no 'match' for this task.    It's too old and not nearly sensitive enough.  So I whipped up a wheatstone bridge jig. 

The idea that because of the current flowing through all the legs is approximately equal, very small differences in resistance will show up as a voltage delta across the horizontal pins.  Put a 9V across the top pins, and start to match!  Here's a closer shot:

I first used the bridge to find three that were pretty close, as in within under 1%.  Then I used these 'calibrated' ones in the bridge to find other ones that were even closer. 

Bob's your uncle.

 

[Cagey]

How very clever of you, agent Mayfly!

 

[bagman67]

Trevor, I have no idea whether your designs are worth a damn (I've heard good things), but your habits of thought and approach to problem solving really make me want to work more like you do.

 

[Cagey]

He's an engineer, you're a lawyer. You basically have the same skillset: you're both problem solvers who respect rules. You're just focused in different areas. But, you can certainly infringe with no trouble. A little explanation here and there, and you're golden. People with your intellect can have a lotta fun. Look at what you're doing now with guitars just by hanging around here and thinking on your own. Not many years ago, you'd have never attempted some of the things you do pretty well now.

 

[mayfly]

Thank you gentlemen.  Now for my next trick, we will match transistors.

each mic uses a single mosfet for it's gain.  The tricky thing about any 'single' transistor circuit is that they all vary.  A lot.  This is one of the reasons why people use op-amps these days - all the matching crap is taken care of for you in the design and implementation.  But I digress.

The circuit for the U87 is designed to include a bias resistor to tweak things, but the transistor must have a particular source-drain current to fit in the design parameters (between 4 and 10mA depending on who you ask).  In addition, I want to use these mics as matched pairs - so I want the transistors to be as matched as possible. 

Here's what I did.  I made another jig that runs the 9V between the source and the drain, with a spot to insert an ammeter.  I got the transistor to turn on by shorting the gate and source together.

Unfortunately I again ran into the limitations of my old meter (does not measure anywhere near a mA), so I borrowed this nice little guy from work:

Using that I went through my selection of mosfets and got some values for Ids:

I picked a couple in the 7.something range. 

 

[mayfly]

BTW, I can take no credit for the mosfet matching method.  I got that from the u87 build thread over in Group DIY

 

[Cagey]

The circuit for the U87 is designed to include a bias resistor to tweak things, but the transistor must have a particular source-drain current to fit in the design parameters (between 4 and 10mA depending on who you ask).  In addition, I want to use these mics as matched pairs - so I want the transistors to be as matched as possible.

I hate the "depending who you ask" thing. You know they're all right, and they're all wrong. I wonder if you could make use of some trim pots to change characteristics? I know some hate to use them as they're rarely a practical thing from a manufacturing point of view - you find a value and run with it forever, so it's money wasted - but on a project like this I can't imagine the tiny bump in cost being a hardship.

 

[Patrick from Davis]

The threads over at the other board go into excessive detail on the minutia at times, but the have done most everything possible to recreate the proper sound characteristics of the mics.  Several compare the clones to rented originals.  The matching of specific parts is required at times, and the reasons are not always obvious.  However, buying several components and matching two turns out to be more cost effective/easier than re engineering things a lot of the time. 

I have found a very small soldering iron tip is quite helpful for things like this, it is a little larger than a quarter.

While Mayfly has gone into the mic building side of things, I got hooked by the API500 module bug.  I have two mic pre's, two subtle distortion modules (colours, endless options) and a filter module.  I built the rack enclosure to house the modules and the power supply for it as well.  Other than the filter, all of the stuff is DIY and has links over there.

I have two stereo compressors to build as well, one based on a SSL 4000 bus comp and the other one is based on the Pye Compressor like Cat Stevens used when he was Cat Stevens.

I am very interested in this build, because there is a pair of ELA M251 clones I want to build and a pair of EF800 U47s.  Money, sigh.  Well, keep powering forward, this is fun for me Mayfly.
Patrick

 

[Patrick from Davis]

Sources? I found a place just selling the capsules, and place selling kits with bodies, but they only had a handful that didn't really interest me. (Either tube or cardiod only).

If you go to the GroupDIY web Page, there is a Microphone forum, and a White Page forum that have ordering or links to ordering.  The Microphone forum will have lots of discussions on the relative merits of just about all available mic capsules.  Then in the White pages you can usually find someone to supply you, or if it is a commercial operation, whom to contact.

Just like this place, I warn you, it is very easy to spend a LOT of money if you start reading over there.
Patrick

 

[mayfly]

I have found a very small soldering iron tip is quite helpful for things like this, it is a little larger than a quarter.

That's quite the op-amp you have there!  I built a couple of the seventh circle audio Neve clones for myself.  Might build some compressors as well, but first I want to see what these mics sound like straight in.  One thing that I probably will build is a little mid-side mixing/summing dohicky, so I can hear it in stereo as I record...

Fun stuff!

 

[Patrick from Davis]

That one went into a Classic API VP28.  Which is a marvelous pre if you like the API thing.  Those GAR2520's are a real test of your eyes to be sure you are getting everything in proper.  But at 25-33% of the price of a pre built one, yeah, I'll suffer a bit.  I really want a JML LA500 comp.  Man o man for vocals they are just so nice, that LA vibe but without all of the sourcing oddball parts.  Also it is nice to have the HPF integrated into it.  Okay, enough of me blabbering about this stuff to you, you need to get back to work!!  I wanna see those things.  Sigh, I guess this behavior is not going to end anytime soon...
Patrick

 

[mayfly]

Some more stuff.  The boards from Dany went together very well.  The BOM has some variants, so I chose to use polystyrene caps for the low signal stuff (instead of film) and electrolytics for the power filtering (instead of tantalum).  The polystyrene caps sound great, but they are verrrrry sensitive to heat.  You gotta be fast with that iron!  Here's the boards:

Next I thought it would be fun to play with the capsules.  These were made by a guy named Eric, who did a wonderful job:

They were mounted to the mic body using the smallest screws known to mankind:

Once that was done, it was time to attach the lead wires for the backplates.  U87's use a capsule that is essentially two capsules in one.  Messing about with the bias on the back capsule is what causes the mic to have different pickup patterns.  This means that I needed to attach wires to two electrically insulated backplates.  One was easy - I re-used a mounting screw.  The other side was tricky - I didn't have an appropriate screw that small.  Sooooo - I gently unscrewed one of the screws around the perimeter then used that.  Hope I didn't kill it in the process 

 

[mayfly]

OK!  There was an adventure!  An adventure in annoying mic problems.  A story of epic crackle and fizz.  and a whole lot of new four letter words.

But first:  Here's the unit with the boards installed, FET bias set to a nominal value, and ready to apply a little phantom power:

So I went to my little rehearsal room, plugged it in, put on some headphones, flipped on the power, and pushed up the fader.  Holy Crap it WORKED!!  and it sounded great.  I thought I should try the various modes, so I flipped on the 10db pad, and that was fine.  Then I tried the low cut, and that worked well too.  I flipped it to omni, and that was OK - but strangely a little dark on the back side.  Then I tried figure 8 - the pattern that was the whole reason for building the thing in the first place.  It crackled.  It popped.  It sizzled.  It was annoying as HELL!

So - what's a guy to do?  start debugging:

I pulled it apart and checked the connections to the capsule.  Fine.  I tried flipping the back and front capsules.  No change, only the cardioid worked from the OTHER side now.  I started re-soldering connections.  No change.  I cleaned the pattern switch with contact cleaner.  No change.  Checked, checked, and re-checked the circuit.  Everything looked right.

Then I thought "hey, some modes work, and some modes don't.  Why don't we use the power of deductive reasoning along with the schematic to eliminate possible sources of the noise?".  Sounded like good advice, so I followed it.  Most of the circuitry is common to all patterns, especially everything on the loud side of the MOSFET.  so I eliminated all that.  Then I eliminated any components on the low signal board that were for the pad and the low cut filter.  Then I eliminated all of the circuitry for the front capsule.  There were only a handful of components left - some resistors and a couple of capacitors.

Now if you hang around components long enough, you get to know them.  You know that resistors are pretty bulletproof.  They don't typically go bad or are bad out of the box.  Caps though, those buggers are a different thing entirely.  There was one Cap in the circuit, C2.  On a board filled with caps in the pico farad range this one stood out because it was 0.01uf.  A cap that big had to be used to pass signal at full bandwidth, and from the schematic that's exactly what it looked like it did.  It was a fairly pedestrian film type, so I didn't feel bad about pulling it.  I also had a bunch of nice Mallory film caps from my tube amp days so I said "what the hell - let's swap the forker!"

The result?  A beautiful sounding figure 8 pattern, and a nice sounding omni to boot.  Heaven!

a tale of two caps:  Bad one on the top, good one on the bottom:

a pretty nice mic!

 

[Cagey]

Congrats! I'm glad that worked out for you. You must be very pleased - I'd never even try to build a mic. Wouldn't have thought it possible, really.