Beesneez Lulu Fet - KM84 SMD mod

[trans4funks1]

I have 4 Lulus myself and enjoy using them.

I may not understand the scope of this project; are explaining that you duplicated the LULU circuit using a differing fabrication technology or did you modify the circuit in your quest for a lower noise floor?

It seemed like you describe starting with a Neumann K84 schematic and then using the LULU component values. How much differs between the KM84, The LULU, and your interpretation?

I think this project will be especially compelling if it can be adopted to upgrading some of the $50 SDC donor mics that are available into microphones comparable to the mics mentioned above.

Thank you.

 

[Khron]

If you had actually read through the entire thread, you'd have (at least some of) your answers already:

I noticed that while most of the components were identical the Lulu had a couple of value changes with capacitors. Only a couple, but I decided to stick to Ben's choices instead of the KM84 choices. (I actually don't really know what difference it should make. That's the next thing I need to learn.)

As well as https://groupdiy.com/index.php?topic=66037.msg836559#msg836559
Hint: open the photo, and read the notes on the bottom...

 

[GraemeWoller]

Yeah, they're great mics alright! It's cool to hear you have four of them.

Basically the scope of the project really is to learn, I think my initial post made it seem like I had a beef with the performance, but that's not the case at all. I've edited that first post now. I heard the difference in self noise between various mics, including the Lulus, and thought I'd look into it and learn a bit while I was at it.

The circuits in the KM84 and the Lulu FET are identical. The only difference is the Lulus have no -10db pad, the values of C3, C5, and C8 are different, and the transformer is a 4:1 Cinemag instead of 6.9:1 BV-107 in the KM84.

The KM84s use:

C3: 1.5µf 15v
C5: 5µf 25v
C8: 5µf 25v

The Lulus use:

C3: 4.7µf 25v
C5: 4.7µf 25v
C8: 4.7µf 25v

My version uses exactly the same values as the Lulu circuit, but I'm trying a Cinemag CM-5722 which is based on the BV-107 from the KM84 to see how that goes. AND, I thought it would be fun to try the whole thing as SMDs instead of through-hole, just to see if I could.

I agree, I think it could be interesting to folks who want to give it a go for other mics. I've provided the Eagle files in a previous comment incase anyone is interested in what I did with mine. So far I've learnt that I can lay things out on a PCB fairly neat and tidy! Yay me!

How much difference is there in capsule polarisation voltage between different capsules out there?

Rock On!

G

I have 4 Lulus myself and enjoy using them.

I may not understand the scope of this project; are explaining that you duplicated the LULU circuit using a differing fabrication technology or did you modify the circuit in your quest for a lower noise floor?

It seemed like you describe starting with a Neumann K84 schematic and then using the LULU component values. How much differs between the KM84, The LULU, and your interpretation?

I think this project will be especially compelling if it can be adopted to upgrading some of the $50 SDC donor mics that are available into microphones comparable to the mics mentioned above.

Thank you.

 

[e.oelberg]

Hi GreameWoller

I completely understand your approach. You will learn a lot. Anyway SMD is fun, easy to solder and good for a mic with so little real estate. Also HiZ parts are much cheaper in smd. It could be also usable to upgrade gefell mv690. I never liked there amplifier but the have  good capsule options

 

[Khron]

Added to that, for DIY prototyping, it allows one to skip the hole-drilling step That being said, hand-matching parts (where needed) can be a slight bit more tedious.

That is, unless you want to spring for more expensive, tighter-tolerance parts - although in some cases, that's not an option because there are none available. Resistors yes; capacitors not so much (you'd be hard pressed to find sub-5% parts above the pF-range); semiconductors - not gonna happen

 

[RuudNL]

That's interesting, feel free to get technical. Is it due to the smaller diaphragm naturally having a lower output and the amp section having to make up for that?

Small diaphragm capsules have a lower overall capacity than large diaphragm capsules.
LDC's are usually in the order of 80 pF, SDC's are in general something like 35 pf.
And a lower capacity means a lower variation in capacity due to to movement of the diaphragm.
Q = C * U.
The equation shows that with a fixed charge (Q) on the capsule, the voltage (U) generated by the capsule has a direct relation with the change in capacity (C).
So smaller capsule capacity means lower capacity change (as a result of diaphragm movement) means lower output voltage!
And since the noise of the impedance converter/head amplifier stays the same, SDC's have a worse signal to noise ratio than LDC's,

 

[GraemeWoller]

Small diaphragm capsules have a lower overall capacity than large diaphragm capsules.
LDC's are usually in the order of 80 pF, SDC's are in general something like 35 pf.
And a lower capacity means a lower variation in capacity due to to movement of the diaphragm.
Q = C * U.
The equation shows that with a fixed charge (Q) on the capsule, the voltage (U) generated by the capsule has a direct relation with the change in capacity (C).
So smaller capsule capacity means lower capacity change (as a result of diaphragm movement) means lower output voltage!
And since the noise of the impedance converter/head amplifier stays the same, SDC's have a worse signal to noise ratio than LDC's,

That's a great breakdown, and makes perfect sense. Thanks for that! That's along the lines of what I was thinking, but your description makes it so wonderfully clear. 

 

[GraemeWoller]

Hi GreameWoller

I completely understand your approach. You will learn a lot. Anyway SMD is fun, easy to solder and good for a mic with so little real estate. Also HiZ parts are much cheaper in smd. It could be also usable to upgrade gefell mv690. I never liked there amplifier but the have  good capsule options

Thanks! Yeah, the learning is something we all enjoy, I'd imagine. It's certainly true to say that I'm digging having a go with the SMD parts. Even using the "moderately large" 3216 packages I've been amazed at how much more space there is on the board!

I'd had another thought for some later revision/test that it would be interesting to do a SMD/Through-Hole hybrid where all the signal path components are top-notch thru-hole but the support parts are SMD.

The only reason is the test the argument, (or just have fun fiddling), that SMDs are not as good for the signal path... I'd like to see some figures on that though, to be fair.

 

[GraemeWoller]

Part Four - Getting The Boards Back

The first thing I noticed when I got the boards back was "the mistake".

It certainly wasn't Seeeds fault. I could tell right away it was mine. I'd wanted to create a solder pad part that was the right dimensions, but by the looks of things I'd not organised the soldermask properly. There's some great tutorials around explaining how to make Devices in Eagle, but I'd obviously missed something that I'll need to fix in a V2 board.

So, I whipped out a Stanley Knife to clean it up. It didn't take too long to sort it out, just a little patience.

I'd laid out the dimensions for the transformer slot based on the Cinemag spec sheet, but when it came time to check the sizing IRL I found that the spec sheet didn't allow for the wire wrapping overrun and the outerwrap, so it was off to Mitre 10 to pick up a milling tool… And, that leads me to an important point…

Don't by a Medalist rotary tool.

Ever.

I ended up using the less bent grinding bits in my cordless drill and pretended it was supposed to work like that.

With the benefit of experience I can plan for a larger sized cutout in the next versions. The same goes for the notches at the XLR end to give a little more clearance for the body end. I also fount I had to carefully drill out the hole width for the capsule connection turret. I didn't need to use a turret, but I'd read that the impeadence is so hi before the FET that flux can cause a bridge and add noise to the circuit, so I wanted to keep as much flux away from the hi-impedance part before the FET as I could. I thought it might be worth the experiment.

I think one of the things that SEEMS obvious, but is worth reminding yourself about, is to be patient. Take your time to check, and recheck. If there are any alterations after the fact take the time to do it carefully and not use the "bull at a gate" approach. You can always mill off more, not less. You can always clean off more soldermask, not less.

It sucks to make a mistake, but it sucks more when it's irreversible.

 

[GraemeWoller]

Shit tool.

 

[GraemeWoller]

Make-shift Dremel.

 

[GraemeWoller]

Cleaning Solder Mask.

 

[GraemeWoller]

Trimming down the XLR connector.

 

[trans4funks1]

Hi Graeme,
Thank you for explaining.

Thank you.

 

[GraemeWoller]

Part Five - Paste and Chips

When everything was tidied up I was ready to start smearing past and placing components on the boards. I'd chosen 1gigΩ resistors that had silver connectors and noticed on the spec sheet that tin based solder was fine as long as it had a reasonable amount of silver in it but lead based was out of the question. That's really good to know if you're building anything that's hopefully going to last a reasonable amount of time. The last thing you need is to have your connections corrode away!!

This is one area where my impatience reared it's head. I could've ordered an appropriate syringe to apply the paste, I could only get it in a pot, but I didn't. I wanted to get on with it when it arrived so I pulled out a trusty bamboo skewer and started carefully "blobbing" solder paste onto the pads of my first board. It wasn't as sticky as I'd imagined, but I finally had a technique, of sorts, down and was left with a fully pasted up board!

I can totally see why people would have a stencil made.

If I were to have a larger run of these made I would be ordering a stencil. I'd probably aim to do 10 to a sheet and do the lot in one go, BUT this isn't one of those jobs so the skewer it was.

The very first board was my test case.

I knew I'd be fine to apply the paste, give or take a bit of fiddling here and there, but the tricky bit was the reflow process.

Reflow was a completely unknown quantity, as such. I've soldered with a soldering iron before, but I could tell I wasn't going to be having a hell of a lot of fun trying to solder even big 3216s by hand with a 40w iron. I was betting on reflow being the business, but I was pretty sure I wasn't going to mod a toaster oven to do it. I was going to have to try the "other", slightly less dignified, option.

I used a fairly flat pancake pan.

I pulled out my digital kitchen thermometer and started warming up the pan. The reflow temp on the solder paste indicated a reflow point of around 220-ish, so I did my best to test the heating of the pan with a soak period then ramping up to the reflow point without anything in there. Once that seemed to measure ok I did the same again with the pasted first board. But, strangely, while it worked fine and the paste reflowed with no issue I noticed it did everything measuring a temp on the thermometer of around 80° lower than expected. Now I know kitchen thermometers tend to be a bit off in general, but I thought that result was probably a bit ridiculous so if anyone has any thought on why that might have happened I'd love to hear.

With one successful test under my belt I repeated the paste-fun-times and proceeded to manually place TINY LITTLE COMPONENTS very, very carefully on my second board. Well, they seemed tiny to me. Especially the 2012 cap at C1 that I'd accidentally brought instead of the 3216 size that I'd intended to stick to.

Once the components were on I ran the whole hillbilly-reflow process again and, voila!, I had a mic PCB with all of the SMD components on!!

N.B. ("Noob Bollocks"): When you are ready to send off your layouts to your chosen PCB company make sure you have markings for all of your components. I had most of them, including orientations for the tantalums, but some of the resistors didn't have package outlines on the silkscreen layer and that lead me to place the 2.2k resistors horizontally instead of vertically on the board. I'd gone through and tested all the components for shorts before going any further and found that one of those resistors was showing a short and found the mistake that way. I should've double checked.

 

[GraemeWoller]

SMDs on, ready for reflow.

 

[GraemeWoller]

Hillbilly-Reflow!

 

[GraemeWoller]

Reflowed SMD PCB having a natter with its antecedent.

 

[Khron]

That soldering technique reminded me of AvE ;D https://www.youtube.com/watch?v=-vhHYo8WOFk

I'm intrigued about the solder specs for those 1G resistors. Got a link to the datasheet, or wherever that's mentioned?

 

[e.oelberg]

I even solder the really tiny sms parts with a small pincher an my Weller iron. It's fast after some practice, and much easier if you need to experiment with values, take parts in and out.