Following a firmware update to version 0.75, three of my listed caveats have been corrected. If you wish to correct the behaviour of the pitch bend wheel, the individual pitch bend out, and the arpeggiator, please go to this address and follow the firmware installation instructions.
https://support.waldorfmusic.com/products/kb37
Will strike out the three complaints in the text below.
Any further updates will be added to this section, eventually.
=========
Original post:
=========
So i've had a few weeks to actively sit down and play intensely on Waldorfs' KB37 keyboard. This is intended as a critical review complementary to others out there
(i felt the SOS review for example mostly parroted specifications not giving it enough time to test its features thoroughly).
In sum, it's been a bittersweet experience so far, but i've not regretted the purchase. Mostly, it is a solid, nice instrument that keeps me playing.
There are some things you should be aware of before buying though, and i feel this information is not readily available elsewhere, so here it goes...
Also: i'm going to go into quite some detail about some kinks, quirks, bugs and (from my perspective) design contradictions.
That's not not to say the instrument is bad. Used to ui/ux design meetings and bug testing, i'm just bent on being thorough with describing caveats and use case scenarios.
So the problems i describe will surely be out of proportion in contrast to the good stuff.
My usecase:
I think it's important to declare my standpoint and intents of use, since the review is informed and filtered through that light.
It is supposed to be a computer-free, standalone keyboard for playing, jamming, practicing and taking it between the bedroom, living room and the rehearsal on the other side of town.
It is sometimes connected to a bigger rack of modules, but most often on its own and trying to be focused; sometimes switching modules out. I intended to use it free of distractions of a DAW,
various studio gadgets or a MIDI rig. It's just me and my modular. And sometimes friends / band mates jamming with me.
The good stuff first:
First off, let me say that i've never been a good keyboardist but the kb37 have been the perfect keyboard for me to encourage more thorough practicing.
It has simply been enjoyable practicing scales and coming up with little riffs. They keybed, which i read somewhere was a unweighted fatar 37, is far more fun and smooth to play than my
80s yamaha workstation synth i otherwise mostly use. I feel i've gained a new surface of directness to modules i didn't have before, and above all a nice standalone modular instrument i
can pick "modules of the days' mood" and put into and focus on just those modules.
The potentiometers and the wheels are very decent, even if i prefer broader wheels with a little cutaway rather than the half-centimeter lever these have. Each to their own, not a biggie.
They're solidly playable!
The buttons are sort of wiggly and don't offer tactile feedback, but on the other hand they're not supposed to be "played". They're more of a setup interface, so they're passable for their intention, too.
The switches feel good and direct enough and their functions are well planned.
It's very sturdy. And heavy, if you like that. I wouldn't mind if the underside was made from a more lightweight material than the thick aluminum sheet, but i feel this was made for touring a long time,
provided you're using a motor vehicle. The fatar keybed should be easy to replace if an accident ever would happen outside of warranty. It might also be easy to replace the PSU,
which by the way seems silent enough. I'm not detecting any problems by ear, anyway.
Pitch 1 and gate is pre-routed to the gate/cv bus, which is nice.
Firmware is supposedly easily updatable via USB MIDI through any sysex tool (current firmware is 0.75. i don't know if they every updated it but manual the is v1.1).
There are some mysterious extension/expansion labeled headers, both internally and on the back. Not sure what's on those pins.
The manual even includes some easy-to-make DIY circuits to plug into the sensor jack on the back. That's nice!
It blends in pretty well in a home, which is more than you can say about most synthesizers out there. See picture.
Pricing vs features estimate.
I think a comparable alternative would be the moog mother if you don't have some good bread and butter modules already and don't mind someone deciding a set of functions
for you to expand outside the keyboard later, and save a pretty penny in the process. Some control features might even be more well polished; see caveats.
If you want a self-contained, structurally robust modular keyboard and fill in modules to your preference yourself; i'm not aware of an option that compares in price/features ratio.
Btw, thomann.de dropped the list price on KB37 recently by several
. It's a much more attractive buy since! The caveat emptor part - including some tips how to get around some of the problems:
Now for the things that i think are off the mark in varying degrees. Sometimes, it comes down to the firmware containing bugs or not being thoroughly use case tested. I also sort of sense that waldorf,
while being veterans in the keyboard and MIDI game, were relatively green in the modular scene when they designed this instrument and made some more design choices that feel a bit malplaced
in a modular context. You can judge for yourself.
Duo(Para-)phonic problems - three points:
1a) First off, let's get one thing clear. It is marketed as duophonic, but the way the firmware is set up, it is paraphonic. This was a selling point for me,
so i guess it's on me for not reading the manual thoroughly before buying.
But as i'll explain, this false advertising (maybe lost in translation?) didn't really need to be false if the firmware had been programmed a little bit differently.
On a surface level, it has two pitch cv:s (the 2nd is assignable to CV X) and one shared gate cv (with optional retrig).
1b) Not the most useful kind of paraphonic.
Scratching the surface - it is a specific kind of paraphonic. If you release the second key, the pitch doesn't stay put. Instead it drops or raises to the value of pitch 1.
This behaviour is useful sometimes and perhaps even intended, if you have two near-identical voices and want them to stack for a beefier or detuned sound when playing one key.
Or if the 2nd voice is regressive in comparison to the 1st so that it can 'hide' inside the spectrum of the 1st.
You can also get a little more mileage out of this mode by switching patch cables for which pitch cv goes to which voice, which affects the keyboard note priority interaction in a sometimes useful inversion.
But with a modular setup, you're more likely to have two distinct voices.
So, problem B with the paraphonic: The equalling behaviour of pitch 2 ought, in my not very humble opinion, be optional since it simply just doesn't work for 9 out of 10 of my patch scenarios for 2+ voices.
For example, you could at first glance get away with not having a separate gate for voice 2 by using mutable instruments' "rings" with just the pitch patched in, which means rings will generate a gate based
on sudden differential pitch changes over time. This way you can actually get a 5-note polyphony - 1 on your 1st pitch CV and 4 on the 2nd through rings.
But due to the pitch normalling between the two pitch channels, you must make quite the acrobatic act on holding keys while playing not to create false triggers this way.
Compare this to the moog matriarch where pitch normalling between pitch cv channels is optional through a function press. The kb37 certainly has enough vacant function combination presses to conveniently
change this behaviour, but it seems noone thought of the need.
So ok. What about a second gate instead? Turns out, which is caveat 1C...
True duophonic opportunities missed by a thin margin
1c) There are two vacant CV outs that have no self-contained function whatsoever. These are cv out Y and cv out Z. Just like cv out X, their primary intended function is to route midi learn configurable cc data.
That means in normal mode you need an external midi device to make use of these three jacks. Such as a DAW or another keyboard or digital sequencer or something*.
But by pressing setup + x button, you turn cv x into the 2nd pitch.
Why not turn cv y into 2nd gate and cv z to 2nd velocity? If this is routed on the firmware level, it just seems unfinished to leave these outputs completely without a self-contained function of any sort
(be it even a static voltage source or modwheel inversion or anything but nothing). Especially one that is sorely needed to live up to the promise of being duophonic.
*As a small side point, i think this sort of reveals Waldorfs familiarity in a MIDI studio context vs relative unfamiliarity with a cv / modular one. For being a modular interface, the product sort of has a tendency
to prioritize midi-oriented design goals over cv/modular design goals. I think someone reading this will see what i mean later on as well.
1d)Unnecessarily distracting LED blinking vividly for the wrong reason.
This is sort of a petty point but it's mildly annoying still.
When entering self-contained mode for cv x (ie, it produces an internally generated pitch), the cv x button is constantly blinking in a stressing, bright pace. It is distracting and, honestly, semantically misleading.
I've known blinking to indicate that something needs urgent attention, to indicate tempo, or that a volatile mode has been introduced that will go away with the next function button press.
I've never known it to signify a persistent mode (even across power cycles). It just sits in the corner of the eye begging me to turn it off, but at the same time it is the only sensible mode to be in
if not using external midi devices. If they're making a firmware update, i advice just keeping it constantly lit when cv x out is in use internally.
Arpeggiator function bug (confusing at first but you can avoid the problem): FIXED in firmware 0.75
2a) The order of the arpeggiator functions differ between the keyboard label and the actual function. The manual v1.1 (and the unit) is in this order: down, up, down/up, random, ordered.
The label reads: down, down/up, up, random, ordered. Probably some communication error in development. Not a biggie once you know it, but if they make a firmware update,
i think they should update the arp functions to match what the keyboard labels actually say.
2b) Weird design choice about clock tempo.
If in tap tempo mode, turning the tempo knob will exit tap tempo and resume whatever tempo the potentiometer assigns. Apparently this is prone to false trigs, and the manual therefore recommends
you to put the potentiometer at either extreme before entering tap tempo mode, as to minimize that risk.
That's inconvenient in two ways.
-having to remember and perform this precautionary step to not risk causing any accidents with the clock.
-you cannot use the tempo knob to bend the tapped tempo.
At least the first, or potentially both caveats could be solved if you were to instead exit tap tempo mode by holding the function button (the usual, intuitive action modifier) and moving the tempo knob.
Pitch bend wheel hiccups (sadly unavoidable). FIXED in firmware 0.75
3a) The manual states that the pitch wheel has three dead zones - at the extremes, and at the middle, as to not generate false midi messages. That's all well i guess, but the actual function generated
from the pitch bend wheel is upfront faulty and nothing like the manual description.
Assume the pitch bend range is +- 12 seminotes (you can select between off, 2, 7, 12 and 24 semitones).
Now, do a slow pitch bend down. Once you get out of the plateau (center dead zone), you don't start progressing down in pitch - you drop a near semitone instantly! This makes subtle bend downs impossible
at this bend scale. Same goes in varying degree for +-7 or +-24. +-2 is tolerable, but still present to a hindering point when bending subtly/slowly.
Worse still - now bend upwards at a +-12. Once you get out of the center dead zone, you drop - yes, drop - a near semitone, before you can begin bending upwards.
This is not a problem solveable by setup calibration. All calibration does is defining what the extremes of the full range is in regards to MIDI pitch bend messages.
Pardon my drastic phrasing here, but whoever did the testing in this department that day must've had a sudden errand out of office. I feel it's impossible to miss unless you always bend and
release very quickly and/or to the extreme of the travel of the wheel at low pitch bend settings. Again, and barring an individual unit error which seems very unlikely, i feel Waldorf R&D needs more time working on
having the two worlds of CV vs MIDI work more seamlessly together.
The only way for the user to mitigate this problem is to set pitch bend range to +- 2 which makes this big bad bug less noticeable, but it's definitely there, and all the same each bend up will include a small
unintuitive/unmusical wobble that's clearly noticeable even at this level of attenuation. What's more, it makes subtle pitch nudging physically impossible through this wheel, as it stands.
I experimented with turning the pitch bend to main pitch cv off alltogether and use the separe bend out jack, but it is affected in the same way. So i believe it is most likely a bug in the firmware programming.
Which lead me to discover another little kink.
Nonneutral pitchbend bias FIXED in firmware 0.75
3b) when using the separate bend out to influence the pitch of my oscillator, i noticed a slight pitch drop. About a quarter note or so. At first i thought it might be a fault on my part or the patching or the mod
input of the oscillator, but i got the multimeter out and measured the bend output. With the bend wheel in the resting "dead zone" position, bend out measures a bias of -0.049v which confirmed my suspicions.
This *may* be individually variant from unit to unit so i'm not saying every unit is like this. It's not a big problem in any case but something you should be aware of when patching it to pitch sensitive inputs.
If there's a trimpot anywhere, it's hard to see without taking apart the chassi/voiding any warranty. The manual leaves no clues.
Too wide middle dead zone this seems a lot better as of firmware 0.75
3c) The dead zone in the middle is a too wide for my taste either way. It makes pitchbending up and down in a pitch-neutral fashion uncomfortable (if it were possible to begin with, see the above point),
and honestly, actual cv directly affecting the pitch of your modules should be prioritized before worrying about perfect midi out accuracy. I'm not going to use this thing inside a moving truck or next to a rock drill
and am not vorried about false pitch bend messages of minute nature. And even if it would trig a midi pitch bend value of a few sudden increments/decrements occasionally, who's going to notice? By comparison,
i notice the both the dead center and bug with the bend wheel *every* time i use it for pitch or bend CV. Just use a midi filter in your daw or workstation keyboard if false midi bends are a problem, or better
yet find another way to program around it. It just doesn't feel natural that there's so much travel in the middle before something happens. This is not a calibration problem, either. I feel a firmware update could
safely shave down this center zone margin by maybe at least half. As it is now, and combined with the pitch drop when stepping out of the dead zone problem, it is very hard to play the pitch wheel subtly.
You're better off using an external controller which kind of defeats some of the purpose of the unit.
Some minor first-time setup woes:
4a) The manual and package listed a mains cable, but no such was included.
4b) The manual instructs you to check the 110/220v toggle before powering on the first time, but there's no easy way to do that without disassembling the PSU or the chassi since it is in a really uncomfortable spot.
The manual doesn't detail where the toggle is. There's a vague label inside pointing in the direction where it is hidden. I decided to just turn it on rather than pulling it apart.
4c)This struck me as a bit odd. The reference note at startup was set to C2, meaning the two lower octaves on the octave switch all emit a 0v. The first thing to do if you're actually going to voltage control eurorack
modules and not just use it as a midi interface - set reference note to C0. Do it like this: Set octave toggle to -2. Hold setup. Push the key with the "reference note" label. Release both. Push the lowest possible key.
Now all octaves work across the whole range, as intended. Not sure why this isn't the factory setting.
So, there's that. One last wish is that the clock knob could have a bit wider range, even beyond the clock division settings/tap tempo. Especially upwards for use as a "ping" in audio rate. I could nitpick about certain
functions not being quite "there" in regards to modular design philosophy, but most of those nitpicks would be displacing this post into the speculative "what if" territory which would only be sensible in discussing a
mk2 of the hardware itself.
I feel all of the other problems could be more or less approached through the firmware if it only were open to modification or in active development. Again, i want to repeat that it is a great instrument in almost
every other way than the cons listed above. The firmware (mostly, hopefully) just doesn't feel quite final and has some pronounced problems, especially in a modular context.
Hopefully this is useful to someone (apologies about the bad structuring and a few repeats). And here's to hoping for either a firmware update ironing out some of these quirks, or alternatively an open release of the
firmware source.
Edit: fixed some typos and unclarities.
I put a bit of white tape on the tap-tempo key yesterday to make it quickly obvious in varying stage lighting. I often use an E-RM MIDIClock box as a master clock for my devices too, to dial in a precise tempo.
there seems to be some expansion headers inside "south" on the busboard too, did you notice?
