[sdiy] MIDI phantom power...over 5 pin MIDI connector ?

[rsdio at audiobanshee.com]

Hi JP,

I would strongly recommend against designing any product that pulls power from the MIDI cable. MIDI simply wasn't designed for that, and if you try to take advantage of the vague power that's there, you'll still run into many devices which don't work. The link that A provided below has long lists of incompatible MIDI devices (because MIDI Solutions is deviating from the standard by pulling power where it's not supposed to be used).

That said, the best clues (if you're going to do this anyway) come from the MIDI Specification.

You can see that pin 4 of all MIDI outputs is supposed to be tied to 5 V through a 220 ohm resistor. I believe that A is wrong in saying that you can get only 5 mA, because the 5 mA limit is the amount of current that is supposed to flow when a MIDI bit is being transmitted. That 5 mA flows out of pin 4 of the MIDI Out, in pin 4 of the MIDI In, through the optoisolator, out pin 5 of the MIDI In, and in pin 5 of the MIDI Out. That makes the total circuit.

However, if you make a different circuit path for current to flow, then you can potentially pull much more than 5 mA from the +5 V supply of the MIDI Out device. This current would flow out of pin 4 of the MIDI Out, in pin 4 of your MIDI In, through your charge pump, and then out pin 2 of your MIDI In, before finally flowing in pin 2 of the MIDI Out. Of course, the more current you pull, the lower the voltage - but that shouldn't be a problem for your charge pump. At 5 mA, the voltage you see will be 3.9 V, but if you double the current to 10 mA then you'll still have 2.8 V. I've used the MAX1595 charge pump (not for MIDI, but for USB-MIDI) in products, and it works with as little as 1.8 V on input. That would allow you to pull 14.5 mA, but there would only be 9.5 mA available when the MIDI signal is using its 5 mA.

Note that you might cause problems for the MIDI Out device by pulling more than 5 mA, because their 5 V supply may have been designed to only provide 5 mA and no more. If there isn't any headroom for extra current, then your device could overheat the MIDI Out circuits and cause them to fail. This is the biggest reason why I argue against designing anything like this.

MIDI Solutions suggests that customers who make their own power supplies connect ground to pin 2 and power to both pins 4 and 5. This suggests that MIDI Solutions is also pulling current from pin 5. Note that this will only work if the buffer circuits in the MIDI out are not open collector. MIDI Solutions is probably assuming a typical pair of inverting gates which drive both high and low logic levels. This would indeed provide some power to your charge pump. However, any MIDI device that uses an open collector output will only be driving logic low levels, and not logic high levels. This is perfectly valid for MIDI, but not so helpful when abusing MIDI for free power. I assume that most of the power is coming through the resistor at pin 4 because it's always available, whereas the Voltage on pin 5 depends upon the logic level of the data and the type of driver.

I think you'd need a couple of diodes to combine the power from both pins 4 and 5 to your charge pump, but those diodes would rob more voltage. You can select LDO (low-dropout) diodes to minimize the losses there. Or, you can just ignore the power that may or may not be coming in on pin 5.
 
Finally, you're depending upon the MIDI Out cable shield for ground. Personally, I think it's a bad choice to assume that the shield will have a low resistance to ground. There are many ways to shield a cable, especially when you are aiming for low noise. A good designer of MIDI audio gear might choose to put a 1 Mohm resistor in parallel with a 1 uF capacitor, and only connect the MIDI Out shield to ground through that filter network. This will make their MIDI audio device much less prone to picking up noise from the cable shield, and it's perfectly legal in terms of normal MIDI functionality. If you pay careful attention to the MIDI Specification, a different GND symbol is used for the opto-isolator versus the symbol for cable shield connections. This means that it's perfectly legal if they do not connect directly. By the way, this means that MIDI Solutions is wrong when they claim that MIDI devices that aren't compatible with their power supply "don't conform to the MIDI specification" - the truth of the matter is that MIDI Solutions has not studied the original MIDI Specifications carefully enough.

There you go. I think that's plenty of clues.

Brian Willoughby
Sound Consulting


On Sep 9, 2015, at 5:57 PM, eidorian at aladan.net wrote:
> 
> Also see http://www.midisolutions.com/faqs.htm
> 
> Maybe contact those guys and see if they're willing to send you a service manual?
> 
> Cheers,
> A.
> 
> 
> On 2015-09-10 10:05, eidorian at aladan.net wrote:
>> As I understand it, you get very little current (5mA) - just enough
>> to drive a small microchip but not much else.  If you need more
>> current than that then you'll need to use it sparingly (i.e. not draw
>> on it all the time, so it can charge up a capacitor for when you do
>> need it).  Having said that, I once read somewhere that some (most?)
>> MIDI outs can provide more than 5mA, but I'm not sure I'd rely on
>> that.
>> 
>> Not really all that helpful sorry.  Hopefully someone with some
>> actual practical experience in this area will respond :)
>> 
>> Cheers,
>> A.
>> 
>> 
>> On 2015-09-10 09:43, jpdesroc wrote:
>>> Does anybody have schematics of an easy way to send
>>> power using the 2 MIDI pins in a 5 pins cable ?
>>> I don't want to use any of the 3 remaining wires of the cable
>>> for universal use on ANY MIDI cables..
>>> I know this company uses that technology:
>>> 
>>> http://www.midisolutions.com/prodmrg.htm
>>> 
>>> but I can't find any circuit ideas that uses it..
>>> I know I'll need some kind of charge pump design to get something like 5 volts
>>> out of the MIDI (idle or in use) signal.
>>> 
>>> Any clues ?
>>> 
>>> JP