Behringer TD-3 bugs and thoughts
This article was originally called "Behringer TD-3 bugs list",
because that’s basically what it was.
It has been slightly updated and a conclusion has been added.
I have sold my TD-3 not long after writing the initial article.
As someone who has spent a lot of time analysing and replicating every single detail of the original 303 sequencer (along with Anton Savov for the n0nx0x project), I feel I’ve become somewhat an authority on the matter.
It came to me as a big surprise that Behringer decided to clone the original sequencer, after all, each of it’s predecessors decided against that. Honestly they didn’t do a terrible job at it too, although it’s obvious they chose a completely different implementation under the hood, which inevitably leads to bugs and annoyances, the latter especially evident when you are used to working with the original.
10/3/2020: Initial version based on firmware version 1.2.5 (21/2/2020).
12/3/2020: Some minor changes and publication
15/3/2020: Added release notes for 1.2.6. Indicate that midi slide bug is fixed there.
23/1/2024: Slightly updated the corpus
24/1/2024: Added conclusion
Behringer TD-3 bugs list
Transpose: In midi slave mode, transposing a pattern when the sequencer is not playing will not apply the transpose to the pattern when playback starts.
Transpose: In midi slave mode, stopping a transposed pattern and restarting will keep the transpose for 1 loop and then play it normal. When pressing transpose it will show the pattern as non-transposed.
Impossible to press both Pattern Section (A/B) and a pattern slot at the same time, which makes switching patterns between sections very hard and error-prone.
Note attributes (e.g. UP) are not always stored. They appear stored and then when switching to another pattern and back they are gone. Still need to find the exact steps to reproduce this, from the 30 somewhat patterns I entered manually from pattern sheets, it happened twice, after using the copy feature.
You can only enter 16 notes in Pitch Mode. Not a "bug" per se, rather a wrong implementation choice which also causes you not to be able to write multiple patterns in one go (which also blocks tap entry across patterns).
Tap write: on the 303 you can hold any of the pattern slot (1..8) keys to sustain tapped notes. This really helps with inputting patterns with tied notes. This is completely not implemented on the TD-3.
Tap write does not work in midi slave mode.
Transpose: In midi slave mode, if the first pattern in the track is transposed, it will not play transposed the first time.
Transpose: in some situations the transpose is not saved. I’m still trying to understand when/how this happens. Sometimes it is saved and then next time I check the track is wrong again.
On a 303 you can hold Transpose during track play to see the pitch changes on the LEDs. Here it does nothing.
When holding the Bar button to see which bar you are on, the speed of the blinking is wrong (should be faster).
No issues noticed. Requires more testing.
|The bug below has been fixed!
As expected (everyone gets this wrong!), a combination of slide, accent and identical notes does not work the way it should. Too bad, as those are usually the source of many interesting patterns. I tested based on an article by Anton Savov which investigates the problem across 303 clones, and provides possible solutions. The result is as described in the article:
The pattern becomes fragmented, and/or some notes get cut-off short, and/or some slides are gone.— Anton Savov
The 303 and MIDI
That is exactly what happens here (first is n0nx0x2, second is TD-3): https://soundcloud.com/rv0/n0nx0x2-td3-midi
Update: This issue is fixed in 1.2.6: https://soundcloud.com/rv0/n0nx0x2-vs-td3-midi-round-2
This are the copied release notes in case the Behringer links change/go dead:
- Version 1.2.6 (12 Mar 2020) link
Fixed: MIDI slide doesn’t work when the overlapping note is same with the previous one
- Version 1.3.7
- Version 2.0.1
Saw this on my friends TD-3 MO. No idea if this is the same.
Did they nail it?
Not really.. They tried to mimic the original sequencer without understanding how it really works under the hood. As a result it’s very similar but there’s all sorts of quirks/weirdness and incomplete things.
Patterns are stored in a very memory-wasteful way, as the engineers did not have to deal with early 80ies hardware restrictions. But in doing so, they made things a lot more difficult (for themselves) than it should be. There’s a lot of elegance in the simplicity of the original software.
Other than that, the switches on the machine are bad, causing double triggers, making programming the thing a frustrating hit and miss.. That’s the main reason I got rid of mine.
While I’m quite happy that they chose to clone the original (superior) sequencer, I’m afraid the quirks will just confirm the stereotype about the 303 being hard to program. Most people will just use MIDI, which is a shame. Although it can’t be denied that the inexpensive TD-3 changed the landscape of 303 clones forever, it could have been so much more.