Mechanical vs Optical rotary encoders

Cwejman, Livewire, TipTop Audio, Doepfer etc... Get your euro on!

Moderators: Kent, Joe., analogdigital, infradead, lisa, parasitk, plord

Post Reply
User avatar
Sync
Common Wiggler
Posts: 155
Joined: Sat Jul 29, 2017 5:40 am
Location: California

Mechanical vs Optical rotary encoders

Post by Sync » Sun Jan 12, 2020 12:48 pm

This topic was inspired by some comments on a particular module where odd behavior of encoders were being discussed.

There's a classic flaky mechanical encoder problem that will often come up after something like 6-months to a year of use.

Here's the issue-- the encoders used today are often mechanical encoders, because they're the cheapest. The output they produce is a pulse-train derived from contact closures that are subject to "bounce" which produces noise at the transition. The software in the gear can perform de-bouncing to an extent, but it's not a complete solution to the problem for a couple of reasons. And the problem gets worse over time because the root cause is oxidation of the contacts.

The reason it gets worse is the fact that, even if the encoder's internal contacts are gold plated, the plating will largely wear off soon because it's very thin due to the cost of gold. Then what you get is likely copper making the contact, which oxidizes very easily. Once it oxidizes, it produces more noise to the point where the software de-bounce just isn't enough. The encoders work by producing pulse trains that the software has to interpret into both movement and movement direction, and noise causes spurious transitions that can confuse the de-bouncing algorithms. The symptom tends to be that when you turn it, it's misread as going in the opposite direction, or sometimes going in one direction more often than the other so it's a real pain to dial in things. You try to move a parameter up, rotating to the right, and it goes down instead. Sometimes it goes down regardless of whether you turn it up or down. And it's intermittent-- sometimes turning it faster makes it worse, if you baby it a bit by turning more slowly the problem can be reduced.

The fix generally requires replacing the encoder. But you can expect it to happen again before long. A far better fix would be if the manufacturers used optical encoders instead. That's one reason I bought the Synthesis Technology E352 wavetable oscillator-- they advertised that an optical encoder is used for the main dial and everything else uses pots. They clearly "get it," though it looks like they may be using a more expensive optical encoder than is necessary (can't fault them for that), there are ones available that are less expensive that might work well enough. In any event, this was a "stand-up-and-take-notice" point about the E352 that, besides it's a very interesting module, impressed me quite a bit so I chose that one over some of the alternative wavetable oscillators.

Korg also has this encoder problem in most of their synthesizers, and despite the fact they have some great designs in their equipment, I learned this lesson the hard way as I've purchased three of their synth machines over the years, (two electribes and a keyboard synth), and all three of them developed the problem within about 6 months time. They clearly either "don't get it" or don't care, but they should, they've been in business long enough. And I live in a dry desert climate where I'd think oxidation should be less of a problem. Decent optical encoders can be had cheap enough, especially for larger manufacturers, but most of the manufacturers just don't understand the problem well enough or don't care.

Traditional pots do NOT have this problem-- one reason is, even if they become more noisy over time, they will rest at a certain point on the internal resistor and don't depend on direction-detection via a mechanically generated pulse trains. Pots essentially, are far more resistant to noise and don't require de-bouncing. They'll generally settle to pretty much where they should be even if they're noisy.

And unfortunately, it's not so easy to replace a mechanical encoder with an optical one after the fact-- there are extra pin connections because the internal LED of an optical encoder requires power-- it's not pin-for-pin compatible and needs to have the right power connected. Plus, the output is not a contact switch closure but a pulse train. It may be possible to use one as a mechanical replacement, but it's non-trivial to say the least. Also, even the inexpensive optical encoders can be a little larger than the mechanical ones.

Better de-bounce techniques can extend the life a little, but it's limited as to what it can do-- enough noise will stymie pretty much any de-bounce algorithm. Still, there are sites out there if you google it where people have figured out better algorithms and it would behoove manufacturers to study the problem if they insist on staying with mechanical encoders and do the best they can, and help to minimize the problem via software updates when their module supports updates.

Based on my experience with this problem, I avoid anything made by Korg or Electron, even though I like their designs quite a bit so it's disappointing-- they use way too many of these encoders and if you check the forums on their gear, many users have seen this problem with their equipment-- it's pretty clear it's way too common.

I DO have several Eurorack modules with mechanical encoders-- it's hard to avoid without giving up some great modules. It concerns me but I figure at least I can probably do a replacement myself if necessary, as Eurorack is easier to deal with in that regard. I noticed the problem in my Pamela's New Workout's encoder just the other day, so there's one I'll probably have to operate on before long. And, I've had it about 6 months, so it's right on schedule.

One thing we should all do is 1) be aware of the problem and 2) make sure when we see a savvy manufacturer who gets it and advertises they use optical encoders like Synthesis Technology, we should make sure that we call attention to it. If we can get the reviewers to be a bit more savvy, that'd help too. Yeah, it'll add to the cost of making a module somewhat, but your users ultimately will appreciate it.

I'm not affiliated with Synthesis Technology in any way, and there may be others who are using optical encoders as well. If you know of any, it might be worth posting it to this topic. I'm not so inclined to call out specific Eurorack manufacturers who are using mechanical encoders though, they are often smaller manufacturers where I don't entirely blame them as it's not a subject on which everyone is that familiar-- and in time they may switch to optical, while threads can persist essentially, forever. People can find out for themselves, as any manufacturer that goes to the trouble of using an optical encoder in their product really ought to advertise that fact like ST does. The point of this topic is education, not to attack small manufacturers. The bigger ones though, like Korg and Electron, I'm less concerned about, and in fact, they have enough experience and charge enough for their gear that they really ought to be aware of and address the problem by now. But use your own judgement about that, I'll say no more about particular manufacturers except to celebrate those who've gone the extra mile.

For those wanting to know more about optical vs mechanical encoders, here's a couple of resources:

https://www.cuidevices.com/product-spot ... acz-series

The C14 series is one option out there-- though from US suppliers seem to be just under $17 ea. even in quantity, which is a little pricey. There are cheaper ones out there from China I think, but they seem to come and go- I've got a couple here I got from eBay at one point, not seeing any there at the moment though. Synthesis Technology mentions theirs is a Grayhill part, which I've no doubt is a good one and is one of those in that $17 range from the US suppliers at least.

A quick search on Alibaba nets these linked below at just cents apiece but I'm skeptical, the specs don't seem to mention "optical" though the title does. They do seem to have 5 pins if the picture is accurate, where mechanical ones can get by with 3. If I were a manufacturer, I'd first order the smallest quantity and tear one apart to make sure it is in fact not a mechanical encoder inside being misrepresented:

https://www.alibaba.com/product-detail/ ... 88275.html

So it likely takes some shopping around to shave the cost as much as possible, but to me, this would be a necessity if I were to build a product for sale that needed encoders.
--

Sync

autopoiesis
Ultra Wiggler
Posts: 797
Joined: Tue Mar 10, 2015 6:00 pm

Re: Mechanical vs Optical rotary encoders

Post by autopoiesis » Mon Jan 13, 2020 1:29 pm

thanks for starting this thread.

I would like to see more manufacturers explicitly calling out what kinds of encoders they use, as this absolutely factors into how many repairs we'll need to send our modules in for. I'd also rather pay more up front for a digital module that uses optical encoders than end up having to send one in for multiple replacements within 1-2 years of owning it; postage + post-warranty repair costs add up, and trustworthiness on the second hand market takes a hit.

as of now I don't know any other manufacturers who are using optical encoders.

User avatar
Fog Door
Wiggling with Experience
Posts: 366
Joined: Sun Feb 24, 2019 1:04 pm

Re: Mechanical vs Optical rotary encoders

Post by Fog Door » Mon Jan 13, 2020 5:48 pm

Great post Sync, both fascinating and horrifying in equal measure, as I had been living in blissful ignorance of this encoder issue :woah:

User avatar
peripatitis
Super Deluxe Wiggler
Posts: 2880
Joined: Mon Aug 09, 2010 2:53 am
Location: Greece
Contact:

Re: Mechanical vs Optical rotary encoders

Post by peripatitis » Mon Jan 13, 2020 6:14 pm

A very interesting read, thank you!
Henry Fronda:
"this is where they put you, when they don't like what you say"
viewtopic.php?f=7&t=226390&start=75

https://soundcloud.com/peripatitis

User avatar
justin3am
Let's go do some crimes
Posts: 1502
Joined: Mon Aug 25, 2008 8:26 pm
Location: Southern California

Re: Mechanical vs Optical rotary encoders

Post by justin3am » Mon Jan 13, 2020 6:45 pm

It's also becoming more common to see endless potentiometers (sometimes referred to as 360 pots or continuous pots) on some devices, particularly where it's important to get through a wide range of values in a single turn. Endless pots have their own issues but they tend to be less expensive than nicer encoders.

I believe 1010 use endless pots on their modules. Companies who make MIDI controllers have been using them for a while.
"Thank god for the drugs and drums." El-P
3amNoise | Synth Videos

User avatar
kay_k
Veteran Wiggler
Posts: 706
Joined: Thu Mar 30, 2017 7:33 am
Location: Berlin
Contact:

Re: Mechanical vs Optical rotary encoders

Post by kay_k » Tue Jan 14, 2020 11:31 am

I want to add that it does in fact make a difference what brand of mechanical encoders you use!

Post Reply

Return to “Eurorack Modules”