OPEN SOURCE DIY Mic Project - ORS 87 - Stripped Down u87

[ubxf]

I'm in the good capsule and transformer camp and the rest being good normal parts (without hunting for special resistors or capacitors unless proven to make a substantial difference )

 

[Khron]

R16: 56k (for 24V zener, try 47k for 33V zener)

47k's gonna work just as well even with the 24V zener - the latter will just sink a little bit more current, so the (now) 47k's gonna drop a little more voltage, ending up at the same net result anyway.

R13: 330k

R15: 300k

Make both 330k, or even 1meg (like R8). All they do is filter the DC that biases the capsule; more is better; exact values are NOT critical whatsoever.

C8: 20u 6v

C10: 4u7 35V
C11: 4u7 63V
C12: 10u 63V

Why not make all four, 22u / 63v or 10u / 63v? They're virtually the same physical size anyway (5mm or 6.3mm diameter, 11mm height), and you'll be hard-pressed to hear any difference. If anything, the more filtering of the power rail, the better.

Then again, if you make R13/15 1meg, C11 at 4.7u or more will just lead to unnecessarily long charge-up time. Maybe make C11 470nF (like C9) or 1uF (like C7)?

C13: 10n

Why not the same value as whatever C5 ends up at? Once again a non-critical value.

I'm just "allergic" to seeing fifteen million different component values for no good reason...

 

[Wordsushi]

@Khron, wow, this is intense. Being able to winnow down the BOM into fewer separate parts is not something I had contemplated at all. If the net result is better filtering, resulting in a more consistent stability of the circuit, and potentially cleaner sonics, then what are the tradeoffs, if any, in doing so?

 

[Khron]

what are the tradeoffs, if any, in doing so?

Lack of "vintage accuracy"... whatever that's worth Or that's the best i can come up with, anyway.

But these are some points / ideas / criterias i've implemented in virtually all of my projects in the last few years... (some stuff "in the oven", hint-hint... )

 

[OneRoomStudio]

Yup, I agree with all of @Khron's points.

I think it's worth thinking about the direction you want to take though. What I originally proposed was a "simplified U87i," including all of the "vintage accurate" component values. It's what it says on the tin - a simplified U87i, but that doesn't mean it's necessarily the best cardioid-only mic using a K67/87 capsule. A lot of things about the circuit could be optimized or updated or changed to suit one's personal tastes. At the end of all that though, it won't be a "simplified U87i" anymore (but it would probably be a great sounding mic). It all depends on your goals.

An "open-source" modern K67/87-based mic is probably a more interesting project to a lot of people here who don't necessarily care about the vintage accuracy. There are a lot of ways to approach that goal though!

 

[kingkorg]

Some capacitor types are important depending on how they are used in the circuit. For example capsule coupling caps (typically 1nF) must not be "leaky" or microphonic. Nothing particularly fancy, but reliable brand is recommended. Doug Ford from Røde recommends polypropylene, polystyrene, taking care not to overheat polystyrene variant. C0G (NP0) are also great.

 

[Khron]

For example capsule coupling caps (typically 1nF) must not be "leaky" or microphonic. Nothing particularly fancy, but reliable brand is recommended.

... Unless it can be avoided altogether...

 

[Khron]

including all of the "vintage accurate" component values

What that implies is that extraneous power supply noise is somehow... essential(?) to the character of a mic like this? I surely hope that's not what you meant

You'll note that all the component value changes i proposed were strictly on the power filtering side, be it for the JFET stage itself, or for the capsule bias voltage - nothing more

 

[Wordsushi]

This is really great info. Seems like every day I learn more about mics thanks to this project. I just ordered a whole bunch of parts from the original BOM, but I'm going to build one with Khron's suggestions down the road to compare.

 

[OneRoomStudio]

Make both 330k, or even 1meg (like R8). All they do is filter the DC that biases the capsule; more is better; exact values are NOT critical whatsoever.

I would have assumed the same, but in simulating it, I realized something - there is an AC negative feedback path from the secondary of the transformer through the 2K2 phantom power resistors, and through those 300K/330K resistors. Making them 1M actually cuts bass (by about 1dB @ 40Hz).

In a perfect transformer, there should be AC phase cancellation, but no real world transformer is perfect (and neither are the resistors). Not a huge effect, but interesting none-the-less.

 

[Khron]

Making them 1M actually cuts bass (by about 1dB @ 40Hz).

"Woooow", big deal. Make the LF feedback cap 47n (or maybe even 39n?) instead of 33n and that's mitigated

 

[OneRoomStudio]

"Woooow", big deal. Make the LF feedback cap 47n (or maybe even 39n?) instead of 33n and that's mitigated

Definitely, just pointing out that you can't change these component values in a vacuum, you have to consider the whole circuit.

 

[OneRoomStudio]

Here's a "modern" version of the circuit that simulates well and uses far fewer different component values.
You'd still want to tune C5, C6, and the source resistor for the FET and capsule you're using, but the rest of the components would all be standard. This configuration would have slightly less output than the "vintage" version due to the FET gain, but it would have lower noise and higher headroom.

 

[OneRoomStudio]

By the way, all of these simulations are using inductance and DCR values for the UTM 87 transformer since they're the only manufacturer that publishes proper data about their transformers.

 

[Wordsushi]

Here's a "modern" version of the circuit that simulates well and uses far fewer different component values.
You'd still want to tune C5, C6, and the source resistor for the FET and capsule you're using, but the rest of the components would all be standard. This configuration would have slightly less output than the "vintage" version due to the FET gain, but it would have lower noise and higher headroom.

View attachment 126771

Thank you for doing this. I'm really curious to A-B this with the original. Next time I order from Mouser, I'll have to get a couple of LSK189's.
What would you speculate would be a target drain voltage with a 2n3819?

 

[Khron]

What would you speculate would be a target drain voltage with a 2n3819?

Around-about mid-supply? Ie. half-ish the voltage at the other end of the 47k(?) drain resistor. The minimum thd bias may be on either side of that - REW's signal gen & RTA will show you where (if you can be bothered to set that up).

 

[OneRoomStudio]

Another technique I use is to inject a 1kHz signal into the FD connection (through a capacitor), and look at the output using a spectral analyzer (a lot of EQ plugins these days have them). Start with the bias where @Khron mentioned (drain voltage roughly half of what you measure at the R12/R14 junction), and then adjust slightly up and down until you see the lowest harmonics on the spectral analyzer (you want the 1kHz spike to be the biggest by far, and the other frequencies to be MUCH smaller). Some people like the sound of some even-order harmonics though...so trim to taste.

 

[Papa Tango Charly]

I'll do you one better. Here's the layout:
View attachment 124217
"X1/P-" at the bottom means "XLR Pin 1 and transformer Primary -." You can figure out the rest. The blue line is just used to show that it doesn't intersect the gray lines it crosses. All touching gray lines do connect.

Here's the BOM. Part names somewhat match the original U87 schematic for reference. Tweak to taste:
R7: 1G
R8: 1M
R9: 7k3
R10: 25K Trimmer
R12: 47k
R13: 330k
R14: 10k
R15: 300k
R16: 56k (for 24V zener, try 47k for 33V zener)
R18 & R19: 2k2 (closely matched!)

C5: 33n (can go as big as 330n if you want more low-end)
C6: 220p (treble cap - smaller = more treble, bigger = darker)
C7: 1u (can tweak to taste. smaller = less bass)
C8: 20u 6v
C9: 470n
C10: 4u7 35V
C11: 4u7 63V
C12: 10u 63V
C13: 10n

Q1: 2N3819

GR1: 1N4749A (24V zener - can try 33V instead with smaller R16)

Tie the XLR Pin 1 to the ground tab on the XLR connector. Film caps should be at least 63V. Use teflon-insulated pins for the capsule connections and keep those junctions "in the air." The board needs to be about 38mmx80mm. Careful not to short to the rails (I used electrical tape to insulate behind the board). Any K67/K87 type capsule should work. Any 8:1 through 12:1 OT should work.

Experiment!

View attachment 124217
Hello, Could you tell me what program do you use to build this schematic please, thank you for this great discussion, very pedagogic!!! Peace out!

 

[OneRoomStudio]

View attachment 124217
Hello, Could you tell me what program do you use to build this schematic please, thank you for this great discussion, very pedagogic!!! Peace out!

It’s called “DIY Layout Creator” (DIYLC). You can download it here: http://bancika.github.io/diy-layout-creator/

 

[Thane Snipes]

Hello guys. I've been following this thread since before I created my account here with much excitement. I'm a total beginner (I prefer the term "talentless hack"), so needless to say, I have many questions, most of which I try to answer by learning a little bit more every day. I am dead set on building this project soon, and would appreciate your insights.

I own four older Behringer B2 Pros, and I intend to use the K67 style capsule from one of them for this project.

I have been putting together a component list based on OneRoomStudio's updated "modern" schematic:

Here's a "modern" version of the circuit that simulates well and uses far fewer different component values.
You'd still want to tune C5, C6, and the source resistor for the FET and capsule you're using, but the rest of the components would all be standard. This configuration would have slightly less output than the "vintage" version due to the FET gain, but it would have lower noise and higher headroom.

View attachment 126771

I have some questions regarding some of the components, mostly due to my own ignorance of the common terminology:

What is "K1" near the capsule?

What is the "R*" 4.7k resistor? On the OG schematic, there's a 25k trimpot on the source leg of the 2N3819 JFET. Is the "R*" just a standard 4.7k resistor?

The D1 Zener Diode is noted specifically as a CDZV36B. Will any decent 36V Zener Diode do the job in its place, or is there something else I should be aware of? (I wasn't able to source this specific Zener on RS, where I'll likely be buying most of the components from, more on that in a bit.)

What are "PR1" and "PR2" on this schematic?

And lastly, are pins 2 and 3 from the XLR connector joined here?

-

I have begun putting together a list of components on RS, but I'm in doubt of which cap types to use, so any input would be appreciated:



This list should include everything except K1, J1, PR1/PR2, R* and the transformer, capsule and XLR plug. Let me know if I have committed any grave errors. It seems I might have to source the LSK189 (J1) from Mouser or elsewhere, so I'm open to putting together a component list entirely from Mouser. I'll also be ordering the UTM0587 straight from UTM's website.

Another technique I use is to inject a 1kHz signal into the FD connection (through a capacitor), and look at the output using a spectral analyzer (a lot of EQ plugins these days have them). Start with the bias where @Khron mentioned (drain voltage roughly half of what you measure at the R12/R14 junction), and then adjust slightly up and down until you see the lowest harmonics on the spectral analyzer (you want the 1kHz spike to be the biggest by far, and the other frequencies to be MUCH smaller). Some people like the sound of some even-order harmonics though...so trim to taste.

What might this process look like if I wanted to bias a JFET using practically no specialized hardware? Would it be an idea to build a little bread board that could take a headphone out (for 1KHz sine wave) and P48 from my audio interface as well as feed the output back into the preamp and use a spectral analyzer of some variety in Reaper to do biasing? If so, could anybody chime in on how to build such a thing? Or should I just go ahead and build the ORS 87 Modern PCB and do the biasing straight on the board itself with a trimpot of some variety in place of R*?
-

Some of you might think I'm in over my head with this one, and you're probably right, but I look forward to proving you wrong haha

Any help is much appreciated. I look forward to learning from all of you beautiful eggheads