Hint for better Diode at MIDI in

[JSx]

Hi,

in my own old serial MIDI interfaces for my Macs I could solve receiving problems with some devices or rather long MIDI cables (more than 5 m) by using a Schottky-diode rather than the normal 1N4148 at the input of the optocoupler.

One problematic device is my AKAI MX73 master keyboard. OK with short good cables, but not with longer or low quality cables. For me it seems this is a principal design flaw in the MIDI specification.

I can't remember the ones I tested on my heavily used MIDI interface (it's a bit delicate to open it up), but I'm sure it was a BAT-type.

The last ones I solered in have been BAT47 (I also have ordered BAT46 which I might have used in my first interface), but I have not checked them with long cables or problematic devices.

Since I'm mostly a software guy, some more electronic oriented people may want to comment my suggestion.

Maybe you want to order them instead of the 1N4148 or together to try it out. It's just a few cents from Reichelt.

I hope this helps,

JSx

[stryd_one]

Welcome aboard JSx :)

It's always a good idea to think outside the box and come up with new ideas.... but I have to say, I've never seen any problem like this with any of the hundreds of midi devices I've used. That said, I never used serial-midi on a mac. I think it's more likely an issue with your setup than the MIDI spec ;) Has anyone else seen anything like this which could not be attributed to a particular device's fault?

[lylehaze]

The classic Amigas could also use a serial-MIDI adapter. There were many circuits online, some were a bit less developed than others (did I say that right?)

I have not seen problems like you describe, but if I had a flaky MIDI link, I would look first to the optocoupler on the affected MIDI IN. Optos have a history of being flaky, as well as a documented loss of signal peformance over the device lifetime.

On the other hand, they do perform "magic".. carrying a digital signal across an electrical isolation barrier. If you think about it, that's pretty cool.

LyleHaze

[JSx]

Hi folks,

I have not seen problems like you describe, but if I had a flaky MIDI link, I would look first to the optocoupler on the affected MIDI IN. Optos have a history of being flaky, as well as a documented loss of signal peformance over the device lifetime.

If you look at the schematics of a MIDI Interface, you see that the diode is in parallel with the phototransistor of the receiving optocoupler.

More than 10 years ago I tried to build a MIDI adappter for a soundblaster with spare components in my little lab (Using a 1N... optocoupler, because it was a bit difficult at that time to get the PC900 from my preferred distributor). At that time I also thought that the optocoupler was problematic, but it was not the culprit. After changing the 1N4148-diode to a schottky-type, it worked very well.

I know that my AKAI MX73 Masterkeyboard has a badly designed MIDI out circuit, but sometimes you don't want to change the electronics inside an industry-manufactured MIDI device. (Probably not to loose guarantee if it is a new one or not to damage it by accident.)

But even with a PC900 (on my Mac interface) I got better results (with problematic devices or very long MIDI cables) with the schottky-diode than with the 1N4148. It seemed to me that the signal was degraded by the 1N4148, because when I desolderd the diode, all problems went away. (The diode is present only for security purposes and not needed if the cabling is correct if I read the schematic correctly.)

BTW, the old serial mac interface is basically a RS422 to 5mA current loop conversion interface which also feeds one pin on the RS422 with a 1 MHz clock signal (independent of the voltage-current signal conversion).

Why not make a schematic better if it doesn't hurt?

BTW, I used a different type of Schottky-diode in my original tests than the one I mentioned in my last post, but I need to get home in a few days to check which type it was. (But I'm almost sure, that the type of Schottky-diode is not so much important.)

I don't wanted to provoke a discussion if you like the idea or not or somebody thinks that there might be no need for changing a standard schematic, but would be happy to get feedback if somebody with more knowledge than me thinks it would be a bad idea (probably because the schottky diode may be killed (?) if the MIDI input lines are reversed by accident or something similar) to replace the 1N4148 with a schottky-diode. I'm not familiar with the drawbacks when using schottky-diodes instead of normal ones.

Build your interface as you like. I just thought some of you might want to order the schottkys and the 1N4148 and then check them out instead of reorder them afterwards.

BTW, I still know a event on the last day of International Computer Music Conference 1988 at Cologne where we needed MIDI-cables more than 15m long to network several studios who wanted to jam with their software through MIDI and we had problems with some devices, but not all of them.

JSx

[stryd_one]

Why not make a schematic better if it doesn't hurt?

if... it would be a bad idea

Bad idea. If it ain't broke, don't fix it.

You're the only person I've ever seen who's had these problems, so I don't think it's 'broken' by design, rather I think it's broken by your implementation...

[JSx]

Bad idea. If it ain't broke, don't fix it.

Unfortunately you missed the point. MIDI devices are part of a larger system and this whole system of

MIDI devices is made to be able to change them. This might happen more or less often. In my case very often. I want EVERY device to be able to communicate at least with ANY of my MIDI Interfaces.

In your citation I asked for electrotechnical advice, not for your general personal opinion.

And even this missed the point, because it was not MY implementation that was bad.

Sometimes you have to live with equipment with bad implementations. It is not always possible to throw any device with bad technical design in the trashcan.

I know that I often work in unusual environments. So I might have more experience in situations where the specifications MUST be violated (very long MIDI cables, more than 10m long) or where I need to use specific devices/interfaces. (In 1988 we didn't have MIDI repeaters or they would have been much too expensive to connect more than 10 studios in the presentation.)

I NEVER told anyone that the diode MUST be changed, just gave an idea and reasons why it might make sense in certain situations. AND IT DOESN'T HURT in normal situations if you use the schotty-type, so you can just have a more reliable MIDI In in critical cases with the Schottky-Diode.

After some reading and with my experiments in mind it seems that with a standard pn-diode the signal edges at the optocoupler are not as steep as without a diode at all (I checked this too) or with the Schotty-diode. After some reading about the function of schottky-diodes and pn-diodes the leakage of schottky-diodes seems to have a lot less impact on the signal than the discharging of the pn-area for this kind of application.

For those who read until here:

I'm not completely sure, because I couldn't find the rest of my diodes I used, but I'm rather sure that I used 1N 6263 schottky-diodes in the successful tests with my MIDI interfaces.

BTW, my masterkeyboard uses standard TTL-signals for MIDI out, not open collector as in the reference implementation and this causes trouble with MIDI cables longer than a few meters and if you make heavy use of continuous controllers. I'm rather sure this was done to save a chip and I'm almost sure this wouldn't be the only device where this engineers/marketing decided to hurt the standard to save a few cents.

Since this gets more and more off-topic and I didn't got any useful replies yet, I will stop now. You got the information. Use it or don't use it, it's up to you.

[nILS]

Why does this always happen in my threads?

Instead of writing this lengthy reply you coulda confirmed your order ;-)

edit: ... which I see just happened :D

[Twin-X]

Why does this always happen in my threads?

To test if you are paying attention and not make stupid mistakes like mirroring the midibusses on the pcb  :P

[nILS]

Well I did that to test the people checking the schem for me ;-)

[lylehaze]

Somehow I get the feeling I'll accomplish nothing with this post.

Regarding changing to a Schottky diode at the MIDI receiver circuit:

"The most evident limitations of Schottky diodes are the relatively low reverse voltage rating for silicon-metal schottky diodes, 50 V and below, and a relatively high reverse leakage current. The reverse leakage current, increasing with temperature, leads to a thermal instability issue." ref:Wikipedia

So it would appear to be a poor choice to replace a diode that is normally reverse-biased, as you described in your above post.

As for myself, I operate my equipment within the specifications, and it works reliably. I have no great desire to design outside the standards because it "seems to work better" with questionable equipment or practices.

You are welcome to hack your stuff any way you want to, but if you want the rest of us to adopt your changes as the "new standard", you might reach a wider audience by petitioning the MMA directly. I'm sure they will review your suggestions carefully.

Have Fun,

LyleHaze

[stryd_one]

Unfortunately you missed the point.

How so? Seems like you're doing that thing that people often do to me where they say that I don't understand, because I don't agree with them.

I want EVERY device to be able to communicate at least with ANY of my MIDI Interfaces.

ANY device that follows the MIDI specification WILL work correctly. What's up with the caps?

In your citation I asked for electrotechnical advice, not for your general personal opinion.

That was my technical advice. Logic is pivotal in technical discussion.

And even this missed the point, because it was not MY implementation that was bad.

How'd I miss what point? I disagree, I think that changing the diodes unnecessarily is bad. That doesn't mean I missed your point, it means I don't agree with it.

Sometimes you have to live with equipment with bad implementations. It is not always possible to throw any device with bad technical design in the trashcan.

It's not a bad technical implementation, you've got a bug somewhere - and it's not the diode, they seem to work fine for the entire planet except you. That's probably not an indication that the entire planet got it wrong....

I know that I often work in unusual environments. So I might have more experience in situations where the specifications MUST be violated (very long MIDI cables, more than 10m long)

No, that is not a situation where it must be violated, you should place a device in the cable to act as a repeater of the signal.

(In 1988 we didn't have MIDI repeaters or they would have been much too expensive to connect more than 10 studios in the presentation.)

Lack of funding is not equivalent to a technical flaw.  Going hack on something to make it behave like you need is great, but I don't think it constitutes reasoning for having the spec changed.

I NEVER told anyone that the diode MUST be changed, just gave an idea and reasons why it might make sense in certain situations. AND IT DOESN'T HURT in normal situations if you use the schotty-type, so you can just have a more reliable MIDI In in critical cases with the Schottky-Diode.

Seriously what's up with all the shouting. I never said, that you said it "MUST" be changed. I just said don't change it if it matches the spec, cause the spec works.

Maybe while you're changing diodes you could fix your attitude.

[JSx]

Hi LyleHaze,

I want to thank you for your hints, even if I think a bit different than you (both is absolutely OK, I think).

Regarding changing to a Schottky diode at the MIDI receiver circuit:

"The most evident limitations of Schottky diodes are the relatively low reverse voltage rating for silicon-metal schottky diodes, 50 V and below, and a relatively high reverse leakage current. The reverse leakage current, increasing with temperature, leads to a thermal instability issue." ref:Wikipedia

So it would appear to be a poor choice to replace a diode that is normally reverse-biased, as you described in your above post.

This is relevent technical information. Thanks.

It's not complete, because it just mentioned the drawbacks of Schotty-diodes, but it helps me.

I don't agree with your conclusion, because you only mention the drawbacks, not the advantages.

If there wouldn't be any advantage, I would totally agree with you.

50 V Limit is no problem for MIDI, I think. MIDI is 5 volts.

Maybe for this reason I overlooked that part in wikipedia.

If you put more than 50 volts through the optocouplers LED will be damaged anyway, won't it?

The leakage seems to be much smaller than the current through the Optocoupler-LED, so it doesn't affect operation in normal cases. (OK, no new discussion about going beyond specification, please!)

You are welcome to hack your stuff any way you want to, but if you want the rest of us to adopt your changes as the "new standard", you might reach a wider audience by petitioning the MMA directly. I'm sure they will review your suggestions carefully.

If I would have found out this kind of problems 20 years ago, I would have talked to the MMA about it.

After 20 years I doubt, that they would change anything. But indeed a good suggestion to contact the MMA.

I don't want you (all) to adopt my ideas, just give a hint that those who want for the reasons I mentioned or for their own reasons be able to choose which way they want to go.

to stryd_one:

Sorry, I won't discuss any further with you. This is simply stupid.

I know my attitude is OK, but yours seems rather strange to me.

I know some people have problems to recognize correctly what read, for cultural or educational reasons.

Of course this is even true for me, but probably to a different degree than in your case.

[nILS]

Don't make me delete the entire thread you guys.

[bugfight]

there's this thing called a split...

[stryd_one]

http://www.schickt.de/mbquote.php?id=107

to stryd_one:

Sorry, I won't discuss any further with you.

You just did.

This is simply stupid.

Damn right. Smart people know not to treat me like a fool.

I know my attitude is OK, but yours seems rather strange to me.

You just keep telling yourself that. You come here to tell us to reinvent the wheel because you've got a flat tire, then you come to the SCREAMING conclusion, complete with putting words in my mouth (grrr), that I've missed the point because, when you asked, I answered that I don't agree that it's a necessary measure...

Then you follow up by saying that the replies you received are 'useless', because, once again, they're not in line with yours. If you don't want honest opinions, don't ask. If you're looking for the yes-men forum, this ain't it! (In fact I think this might be the polar opposite) :D

...And because I actually make some sense, you say my attitude is strange. Actually, I guess that would be strange to you, if you always act like this.

I know some people have problems to recognize correctly what read, for cultural or educational reasons.

Of course this is even true for me, but probably to a different degree than in your case.

?Language barrier...

I'd suggest you go with your last try:

I will stop now. You got the information. Use it or don't use it, it's up to you.

Look, it was an interesting topic, but you acted like a troll. Go away or chill the fuck out and show some respect.

[lucem]

I don't even understand why one could blame a protecting diode eating up (possible) negative voltage swings for non-working MIDI connections.

Also, this "open collector" nonsense further up the list...

MIDI is a current-loop, meaning there is exactly 5mA or 0mA flowing through a wire. 10m of cable only pose a problem

under very weird circumstances, like inproper signal feed, or inproper impedance matching between cable, sender and receiver.

(Just for those who maybe read this stuff any day later upon a forum search wondering what the hell was going on here)

[lylehaze]

You stated:

"50 V Limit is no problem for MIDI, I think. MIDI is 5 volts."

But the Wikipedia reference stated "50 V and below"

Last time I checked, 5 volts is below 50 volts, at least for positive voltages.

The failure rate for optocouplers is relatively high, both for complete failure modes and for gradual degradation. Replacing the optocoupler is a good practice for a failed or temperamental MIDI Input.

I would change the opto with a new, fresh replacement before trying to re-design the circuit.

Have Fun,

LyleHaze

[philetaylor]

I've not been around here for very long but even I know that it isn't a good idea to attack s1 in your second ever post. JSx If you intend to stay around here then I suggest you cool it with the attitude. s1 is one of the most helpful and supportive people I have seen on the many forums that I use so I would follow his advice and chill out

[stryd_one]

It's all love yo. Just, be nice  ;)

[lylehaze]

"JSx If you intend to stay around here then I suggest you cool it with the attitude."

True Stryd is very helpful, but then users like JSx are very important too. Stryd gets hungry sometimes. We need a steady supply of newbies to feed the monster.. :-)

Just joking folks.. Smiles all around the thread. ;-)

LyleHaze

[stryd_one]

LOL

Yeh that's very true. Dudes who are new to midibox are very helpful with their advice, because they have the benefit of an outside perspective, which can make things much clearer to them (and consequently, us). Dudes who have skills in electronics but are new to midibox are especially helpful, as they're able to make similar advice from a similar perspective, but with the added advantage of their experience.

That's why everyone, including/especially newbies, are encouraged to politely discuss any issues they see. Everyone wins! :)

[nILS]

Do I win too?

[stryd_one]

No lollipop to you! Eat your cupcake and go straight to your room young man!

[madox]

Just on a technical note, the capacitance of a silicon diode could possibly degrade the signal of a poor current driver, with a long or poor quality cable. The cable introduces RLC impedance components (mostly R at MIDI frequencies). This is just an RLC response (with low L), and the output waveform is subject to a MIDI interface's drive capability.

However, as stated above, the problem is not in the silicon diode. The silicon diode is suitable for this application, as circuit protection.

Most likely, problems exist in the driver, the cable, or the optocoupler gain. Opto's vary in performance, as well as degradation.

Sometimes in electronics, one has to patch a problem in some way that does not truely address the cause, though the diode would definitely not be the first thing I would look at in thid case.

[madox]

The classic Amigas could also use a serial-MIDI adapter. There were many circuits online, some were a bit less developed than others (did I say that right?)

@lylehaze - Were you an Amigan? I grew up on the Music-X sequencer.