The reason for writing this article was a recent forum discussion (see the link below).
So, let us turn from words down to business. First, let us take a look on the input/output schematic recommended by the MIDI specifications.
As you can see, the only element that can introduce a noticeable delay is the optocoupler.
The specifications recommend using Sharp PC-900. We, however, use something simpler and widely available in sufficient quantities, namely EL3H7C. In our experiment we connect six MIDI Thru cascades in series.
According to the data-sheet, the CTR (Current Transfer Ratio) ranges from 200% to 400%. Taking the average results in a pull-up resistor value of 390 Ohm. To demonstrate the effect of incorrectly calculated values, one of the resistors is set to a much higher value (4,7kOm).
To make the experiment closer to the reality we add protection diodes, and connect every two cascades via a regular 5m MIDI cable.
|
View (download)
2098x1276
The rest is pretty simple. We connect a logic analyzer to the test points, power the circuit up, and forward a pulse with duration of 32μs to the input.
Observing the waveforms, note a leading edge pulse delay,
View (download)
1259x603
along with the falling edge delay
View (download)
1259x603
of about 5μs, as it is mentioned in the datasheet.
The presence of cables does not change anything.
View (download)
1259x603
When using a "wrong" resistor, the leading edge got slightly steeper, but the falling edge delay is stretched out and is about twice longer now. This is an inacceptable distortion of waveforms, and using such cascade at the specified speed rate is inappropriate.
View (download)
1259x603
The resulting total delay of six stages is 34μs.
I do not have any exact information concerning the ability of human’s ear to distinguish sounds. However, according to my own experience the threshold of sensitivity is above several tens of milliseconds. This is at least 3 orders of magnitude higher than the delay obtained in our experiment.
From the above experiment we deduce two conclusions:
| 1. | The "Pros" sometimes (really do not want to use the word "often") do not know what they are talking about. | ||||
| 2. |
Outtricking the laws of physics failed again. This, in turn, implies that if your MIDI Thru adds a noticeable delay into the sound then:
|
Needless to say that all written here refers to the case of using identical samplers/ton generators (or at least the ones that have similar playback delays).
Discuss on the forum
Feedback
Other articles
Home
Copyright © 2015-2021 by Dmitry Dubrovenko.