Best MIDI connection? Old-school cables versus USB.

[ScoobyDoo555]

MIDI is "hot-swappable". And has been from day one. Otherwise, as a safeguard, the connectors would have locking screws of sort (such as SCSI)

Whoever started this rumour doesn't know their a$$ from their elbow. Absolute BS.

All that happens when midi connectors are pulled..... is that the data flow stops.

SCSI, on the other hand, deffo "no-no" - I've seen sparks out the back of an Akai S3000 due to some dullard doing this. Quickest way to blow the SCSI board.

BUt midi? lol.

Edited to add reasoning.

[SeedyLee]

Whilst it's true that USB provides substantially higher bandwidth than MIDI, that doesn't mean it is faster overall. MIDI was always designed to have low bandwidth requirements, instead placing an emphasis on the requirement for accurate timing. Other than a few specific cases, such as bulk sysex dumps, MIDI over USB really doesn't benefit greatly from USB's increased bandwidth.

MIDI over USB operates in an asynchronous transfer mode, rather than the isochronous mode used for USB audio. Timing of asynchronous transfers is not guaranteed and can easily be affected by other data on the bus. MIDI doesn't suffer the same issues - Timing remains as accurate as the clock in the sending device and isn't affected by unrelated transfers.

MIDI connections are hot swappable, as the connections are opto isolated. Pluggind and unplugging USB devices causes jitter in the timings as handshaking takes place. The simple protocol of MII doesn't suffer such problems.

So despite the higher bandwidth of USB, surprisingly the 30 year old MIDI protocol still has many advantage.

[StephenKay]

Whilst it's true that USB provides substantially higher bandwidth than MIDI, that doesn't mean it is faster overall. MIDI was always designed to have low bandwidth requirements, instead placing an emphasis on the requirement for accurate timing. Other than a few specific cases, such as bulk sysex dumps, MIDI over USB really doesn't benefit greatly from USB's increased bandwidth.

MIDI over USB operates in an asynchronous transfer mode, rather than the isochronous mode used for USB audio. Timing of asynchronous transfers is not guaranteed and can easily be affected by other data on the bus. MIDI doesn't suffer the same issues - Timing remains as accurate as the clock in the sending device and isn't affected by unrelated transfers.

MIDI connections are hot swappable, as the connections are opto isolated. Pluggind and unplugging USB devices causes jitter in the timings as handshaking takes place. The simple protocol of MII doesn't suffer such problems.

So despite the higher bandwidth of USB, surprisingly the 30 year old MIDI protocol still has many advantage.

USB isn't hot-swappable?

I've been happily connecting my USB keyboards and disconnecting them from multiple computers during software testing dozens of times a day for the last 10 years or so, both windows and mac - never experienced an issue.

And - I could be wrong - but if you stack up a whole bunch of note-ons on a single tick, there would potentially be a skew between each note-on with a conventional MIDI connection, whereas there wouldn't be (or much less) with USB. My understanding is that the 31.25kB rate is not how fast the keyboard can respond to MIDI messages, but a limit on the 5 pin MIDI connection. Therefore, the timing of dense passages of MIDI notes should be better with USB. Although I could be wrong...

[SeedyLee]

Both good points Stephen. Naturally USB was always designed to be hot-swappable, but USB's more complex nature also means hot swapping is more complex, with device enumeration and driver instantiation being something that's unnecessary with MIDI. For example, I can merrily swap MIDI keyboards over using a single port whilst recording uninterrupted in my DAW, which simply isn't really feasible with a USB connection. Combined with MIDI's opto-isolation I think it gives it a slight edge in terms of it's ability to hotswap.

I'm curious to know how the packet-based nature of USB affects simultaneous notes vs regular MIDI. Might have to loo into that one a bit more

[Scott]

Both good points Stephen. Naturally USB was always designed to be hot-swappable, but USB's more complex nature also means hot swapping is more complex, with device enumeration and driver instantiation being something that's unnecessary with MIDI. For example, I can merrily swap MIDI keyboards over using a single port whilst recording uninterrupted in my DAW, which simply isn't really feasible with a USB connection. Combined with MIDI's opto-isolation I think it gives it a slight edge in terms of it's ability to hotswap.

True, I would not want to hot-swap a USB device in the middle of recording/playback. USB devices are not so immediately available as they have to identify themselves, etc. But also, I have had some odd things happen on some keyboards when their MIDI connections were made after they were powered on.

(Though really, does anyone need to hot-swap whilst recording?)

[Scott]

And - I could be wrong - but if you stack up a whole bunch of note-ons on a single tick, there would potentially be a skew between each note-on with a conventional MIDI connection, whereas there wouldn't be (or much less) with USB. My understanding is that the 31.25kB rate is not how fast the keyboard can respond to MIDI messages, but a limit on the 5 pin MIDI connection. Therefore, the timing of dense passages of MIDI notes should be better with USB. Although I could be wrong...

I think you're right. According to

http://www.usb.org/developers/devclass_docs/midi10.pdf

"USB can easily handle heavy loads of MIDI data while preserving the timing integrity of the data. Hundreds of MIDI note messages can be sent all at the same time."

[StephenKay]

And - I could be wrong - but if you stack up a whole bunch of note-ons on a single tick, there would potentially be a skew between each note-on with a conventional MIDI connection, whereas there wouldn't be (or much less) with USB. My understanding is that the 31.25kB rate is not how fast the keyboard can respond to MIDI messages, but a limit on the 5 pin MIDI connection. Therefore, the timing of dense passages of MIDI notes should be better with USB. Although I could be wrong...

I think you're right. According to

http://www.usb.org/developers/devclass_docs/midi10.pdf

"USB can easily handle heavy loads of MIDI data while preserving the timing integrity of the data. Hundreds of MIDI note messages can be sent all at the same time."

Thank you for that!

So are we done with this discussion?

USB = Faster, better timing

[Scott]

Thank you for that!

So are we done with this discussion?

USB = Faster, better timing

I'm no expert, but at least according to the spec, yup, USB should have that advantage!

There is still an issue of being more prone to hum, but there are solutions for that.

See http://forums.musicplayer.com/ubbthread ... cs/2488410 and scroll down to message #2488467 and read the page from that point to the bottom. Also
http://forums.musicplayer.com/ubbthread ... cs/2473835