Banzai's KM84 DIY Body & PCB kit build thread

[kingkorg]

So the 3U capsules are vented…since the body is vented also, does that have a noticeable effect on the pattern or frequency response?

Response of these capsules will heavily depend on what's behind them, how large the chamber is, if it's vented or not. Km84 has shallower chamber and no additional vents that MXL bodies have.

That is true, but not necessarily better. The dome which difuses the sound will be placed considerably lower in the tube of the MXL compared to km84. Once again, these things are way more important and will introduce more audible deviation from the original than any change in the circuit of this particular mic. Even if you compare mxl stock circuit versus the km84.

Slot size on km84 capsule is carefuly tuned and is the part of the delay network. Using more slots is in a way the same as widening the holes on the rear backplate of the capsule.

As far as i can see, this capsule is more of a km184 based. It has tighter slots, so potentially even more important.

I suggest using some foam to make the chamber behind shallower, and listen. Record white noise with and without the foam at same exact position and look at spectrum analyzer for differences. 90°, 180° as well. Big part of km84 is off axis sound.

 

[Ricardus]

So the 3U capsules are vented…since the body is vented also, does that have a noticeable effect on the pattern or frequency response?

Yes. I am interested in knowing this too.

If need be I can close the vents on my MXL bodies with bondo, just like I filled in the embossed MXL logo on the bodies.

 

[oongbac]

Yes. I am interested in knowing this too.

If need be I can close the vents on my MXL bodies with bondo, just like I filled in the embossed MXL logo on the bodies.

Description on product page says "Please do NOT close the vents on the mic bodies." But it worth to do some sciencing...

I'm intersted too!

 

[TwentyTrees]

Thanks for the test @oongbac, really helpful. I agree with the comments others have made - to my ears the Maiku capsule sounds recessed in the upper mids, almost like it's sat a little further back from the mic. The 3U sounds more balanced across the frequency range to me, though I can see that the Maiku would work on a brighter source such as mandolin or tambourine, possibly even fiddle.

I'd like to try the 3U capsule myself, from what I've read elsewhere it sounds like the capsules are plug-and-play and no modification is needed (please shout if I've misunderstood!). I think it's the Flat Cardioid capsule from 3U's "mm5" range, correct?

- - -
Update: having tried both 3U and Maiku on my own builds, the Maiku is indeed a little darker and the 3U flatter. I've decided to keep the 3U, but the Maiku would absolutely work in different contexts. They are both plug-and-play.

 

[oongbac]

Thanks for the test @oongbac, really helpful. I agree with the comments others have made - to my ears the Maiku capsule sounds recessed in the upper mids, almost like it's sat a little further back from the mic, with a slightly disconnected HF transient. The 3U sounds better balanced to me.

I'd like to try the 3U capsule myself, from what I've read elsewhere it sounds like the capsules are plug-and-play and no modification is needed (please shout if I've misunderstood!). I think it's the Flat Cardioid capsule from 3U's "mm5" range, correct?

Yup! No modification is needed with banzai body if you choose mm5.

It's the new one they released at 14th December 2022 for slotted body.

You can find the informations here.

http://www.3uaudio.com/Show.asp?id=701

 

[TwentyTrees]

Yup! No modification is needed with banzai body if you choose mm5.

It's the new one they released at 14th December 2022 for slotted body.

You can find the informations here.

http://www.3uaudio.com/Show.asp?id=701

Brilliant, thank you!

 

[JW]

Kind of off topic, but you know how km84 is used on acoustic guitar, along with another mic on vocals. What kind of capsule have you all found best for mixing with a vocal mic? Omni, cardiod, or hypercardiod? In terms of phase? 3u has a hypercardiod.

 

[McIrish]

Personally, I always use the cardioid. KM84i capsule

 

[maarvold]

For comparison, I am attaching screen shots of Neumann's KM84's response (from their downloadable info pdf) and the cardioid capsule under discussion (from the www.3uaudio.com website). Probably worth noting the differences between 100-200, at 4k, at 9.5k and above 12k. BTW, I am not saying these differences are deal breakers, but they are areas of what should be reasonably audible deviation from what seems to be some peoples' expectations (or hopes).

 

[kingkorg]

For comparison, I am attaching screen shots of Neumann's KM84's response (from their downloadable info pdf) and the cardioid capsule under discussion (from the www.3uaudio.com website). Probably worth noting the differences between 100-200, at 4k, at 9.5k and above 12k. BTW, I am not saying these differences are deal breakers, but they are areas of deviation from what seems to be some peoples' expectations. View attachment 103129View attachment 103128

Sadly you can not compare any two measurements of this kind made on different equipment, different period, manufacturers. There is difference in reference mics used, smoothing, environment, standards...

Also no km84 i ever encountered had this flat response. Neumann's measurements are all awfully smoothed and pretty much unusable. Nothing more than descriptive.

Probably because of graphs like this people think they don't tell much about mic sound. Proper graphs actually do, but none of these two posted are any good.

Published Sony c800g graph is example of a good FR graph.

 

[maarvold]

Sadly you can not compare any two measurements of this kind made on different equipment, different period, manufacturers. There is difference in reference mics used, smoothing, environment, standards...

Also no km84 i ever encountered had this flat response. Neumann's measurements are all awfully smoothed and pretty much unusable. Nothing more than descriptive.

Probably because of graphs like this people think they don't tell much about mic sound. Proper graphs actually do, but none of these two posted are any good.

Published Sony c800g graph is example of a good FR graph.

When I bought my first AKG C451 (before there was phantom power)--in the mid 70s--included with the mic was a graph of that actual mic's frequency response that was printed on the skinny version of fan fold printer paper: with the perforations along both sides and the trace drawn by what looked like a fine point dark blue felt tip pen. And that graph mostly looked like it was drawn with a ruler: very much like the one in the Neumann info document. Please remember that I am giving you eyewitness testimony that this type of thing is possible before you start trying to discredit what I am saying. I'm pretty sure I remember there were similar graphs that came with my Sennheiser MD-421s during the same time period as well. A little over 10 years ago, I received similar documentation with my pair of Sennheiser MKH800s. Those did not look as perfect as the AKG 451, but were not far away.
IMO, with the help of the internet, these frequency response graphs, which were once tools, have now become weapons wielded by people who often don't understand exactly what to make of them. When I recently went looking for a response graph for my Sennheiser MKH40s--the design concept one in Sennheiser's published data sheets--I was unable to find one on their website and had to look around a lot to find one. That's because the 'like button' (with its built-in potential for confirmation bias) has become the more powerful tool. I find this unfortunate.

BTW, I would think, by far, the 2 most difficult things in trying to get a truly accurate frequency response measurement would be 1. designing, then paying for construction of a truly excellent anechoic chamber and 2. getting a 20-20k transducer to play back with nearly perfectly linear amplitude.

 

[kingkorg]

I have no reason to try discredit what you are saying. Only one remark, there's no tranducer as far as i'm aware capable of reproducing 20-20k linear response. It is not requred, that's why they use calibration mic to compensate for the error.

I have 5 different refference measurement microphones with certified factory calibration files one of which B&K. They all measure slightly different. Depending on which you use for compensation the DUT mic will yield slightly different response. That's why i said comparing measurements from different sources is comparing apples to oranges. 3U audio doesn't have anechoic chamber.

 

[maarvold]

I have no reason to try discredit what you are saying. Only one remark, there's no tranducer as far as i'm aware capable of reproducing 20-20k linear response. It is not requred, that's why they use calibration mic to compensate for the error.

I have 5 different refference measurement microphones with certified factory calibration files one of which B&K. They all measure slightly different. Depending on which you use for compensation the DUT mic will yield slightly different response. That's why i said comparing measurements from different sources is comparing apples to oranges. 3U audio doesn't have anechoic chamber.

The implication of what you are saying is that there is no possible way to obtain an extremely accurate frequency response plot of any microphone in any location? Because establishing a true standard is not possible? How are companies like B&K/DPA able to credibly measure the performance of their individual microphones?

Also, I can understand how the application of a convolution (measurement/calibration file) of a highly accurate measurement microphone could be used as the 'ruler' that other DUTs (microphones) can be measured against. And it seems like using this type of procedure could largely eliminate the non-linearities of the transducer as a factor in the measurement process--the same would probably be true for the acoustical properties of the testing space--because, when the DUT was placed in the same physical location as the measurement microphone was placed, one could calculate what would be required to make the DUT behave like the measurement mic performed under the same test... the frequency response plot could be derived from that comparison. Is this how it's typically done now? It seems obvious that the accuracy of the measurement would be compromised by a less-than-accurate convolution capture process as implemented for the measurement microphone.

In the 70's, when I bought my AKG 451, I can't imagine them doing anything besides frequency response measurement in an anechoic chamber, although I could understand how there might be some applied filtering/EQ to compensate for the unavailability of a truly linear reproducer. Comments? I'm trying to learn something here... at least hopefully.

BTW, there's a very well-known engineer--now a mastering engineer--that worked for Motown for many years that used to use KM84s as his measurement mics for monitor voicing.

 

[kingkorg]

How are companies like B&K/DPA able to credibly measure the performance of their individual microphones?

I believe we agree on most of things, but this is not exactly what i was talking about. I was talking about extremely smoothed Neumann's measurements published for their mics, and 3U audio's quasi anechoic measurements, yet again smoothed.

When i say quasi anechoic i don't mean that in pejorative way, there are modern ways of getting extremely reliable measurements. It's what i do. Anyone has anechoic chamber at their disposal, outdoors. You just need some luck and knowhow. However these are heavily dependent on reference mic to establish the flat line. Which is tricky, but i can't go into tiniest of details because of some proprietary stuff i use for the measurements.

When it comes to various measurement techniques this channel is gold.

https://m.youtube.com/@ListenInc/videos
I don't say there is no flat km84, i just say i never encountered one.

 

[molke]

I was talking about extremely smoothed Neumann's measurements published for their mics, and 3U audio's quasi anechoic measurements, yet again smoothed.

At least for the frequency response given above, I don’t think this is meant to be a ‘measurement’, but rather a ‘target response’ (see the allowed tolerance). Neumann is probably perfectly capable of taking smoothed and unsmoothed measurements of microphones. These are apparently not published in a marketing brochure. Comparing the polar plot on the other hand seems much more interesting.

 

[kingkorg]

At least for the frequency response given above, I don’t think this is meant to be a ‘measurement’, but rather a ‘target response’ (see the allowed tolerance). Neumann is probably perfectly capable of taking smoothed and unsmoothed measurements of microphones. These are apparently not published in a marketing brochure. Comparing the polar plot on the other hand seems much more interesting.

Couldn't agree more. Sure, Neumann can pull off anything they like.

Polar plot is where all the money is. Eq-ing the 0° angle response to get it flat is totally legit process, and no different to using originally flat mic. However you can't eq front independently of off axis response and that is why it matters so much.

Measuring polar plot and interpreting it is a whole nother can of worms. You really need anechoic chamber. But even then the plot will vary with distance. Two different mics, with same exact plot response at 1 meter probably measure (and sound) way different at 0.5m or 10m.

 

[maarvold]

...Polar plot is where all the money is. Eq-ing the 0° angle response to get it flat is totally legit process, and no different to using originally flat mic. However you can't eq front independently of off axis response and that is why it matters so much...

Very interesting--I hadn't considered that. Would you say a tightly drawn polar plot is an indicator of capsule quality? Sounds like it.

 

[kingkorg]

Very interesting--I hadn't considered that. Would you say a tightly drawn polar plot is an indicator of capsule quality? Sounds like it.

Sure, you would want best possible rejection over wide range of frequencies. And that off axis curve should be nice, smooth and gradual with no huge dips or peaks. In a lot of LDC cardioids there is substantial peak in high mids @180°.

Take a look at this LDC measurement i took. 180° peaks just 5db below the 0° in very sensitive area. Very audible. It is so strong that it causes a slight dip in 0° response at the same spot.

This is very typical for this particular kind of capsule, so not surprising. However idealy you wouldn't want that.

 

[maarvold]

Most of you folks don't know me, but I earn my living recording and mixing music for film & tv underscore, video games, theme parks, etc. The fairly recent postings in this topic kind of pushed a hot button for me because, in particular, the banzai/Maiku mic I built specifically for snare drum (which, among other things, incorporates George Massenburg's suggested KM84 mod--you can find it by doing a search), to me, sounds fantastic and has ended my 20 year search for my snare drum mic; obviously YMMV.

On the 4 previously-posted sound examples, I was interested to try corrective eq as an experiment because, based on what I heard, I would need to do that with both mics if I was mixing. My methodology was to attempt to only subtract problems (and I was pretty faithful to that), but to leave the basic personality of the capsule in place so I could get a deeper understanding of what each capsule was about sonically.

I attached audio files and EQ screenshots, for anyone interested, and used the same basic settings for both A versions and both B versions; none of them are 100% the same though. The main problems that were bugging me were too much on the bottom and a couple of body resonances (the same idea as wolf tones in orchestral celli). I really tried to touch the files with eq as minimally as possible, yet get to what I felt was a basic, professionally-acceptable result.

I know pretty much exactly what I think about the results for what I like* but I thought some here might like to have a clearer picture of the type of result they might achieve with either capsule in a mix situation so they can clarify how they feel about their personal choices and criticisms.

The EQ was done with a GML 8200 and the settings captured with Waves' Q-Clone; a Topping D90 was the DA and a BLA White Sparrow was the AD for the hardware insert... all very short cable runs used in the analog chain, probably 12' or less.

https://www.dropbox.com/s/aonjomdbtkx5df6/ Maiku vs 3UAudio In Banzai DIY84_24-48 Wave.zip?dl=0
To refresh you, A = 3UAudio, B = Maiku

*and there were things to like in both capsules for me

 

[Aleguitarpro]

Hi.
I'm among the crowd that ordered these hoping will sound close to km84. Not in my experience as others reported and it's not a matter of circuit component choice or transformers.
These capsules are far from an original.
I had a couple to make comparison and plotted an EQ curve that bring Maiku on the same territory as originals.
Attached you'll find Fabfilter Pro q preset.
(You have to change extension to .ffp)

Hope this helps to understand their difference: it's not a solving receipt but if you want to understand "static" differences in frequency response between Maiku and original.
Off axis response of original it's something unbeatable.