Long spring goes opposite to the indented one, you can see the place for it. The reason it stopped working after contact cleaner is you removed the lubricant that provides drag between the shaft and the bump mechanism, so it would not move the switch. Probably would have worked with a drop of silicone oil down the shaft, to make a thin viscous film. The reason it failed was probably the low viscosity oil evaporating out of the switch.
I think the additional problem was due to me heavy handidly putting my knife down each side. The case bulged out. Most likely that's not good for the tight tolerances required. But desoldering/soldering could also be an issue.
@@EEVblog Glass filled nylon there, so likely robust enough to survive soldering. Think it just lost lube there, plus the contacts wearing the gold finish off on the spring brass wiper, so it stopped making good contact. Also probably not helped by too low a contact current, they have a minimum contact current on switches for a reason. This probably has 10k resistors, and thus not enough current flow to keep the contact surfaces clean long term. Probably needs 1k and a 1n ceramic capacitor across the switch contacts, to provide that little bit of spark to clean them.
My theory is that it failed due to wear on the cog and pin - on the assumption you didn't bend the left side gold plated strip during opening it, it looks visibly bent with a kink compared to the right side. To elaborate, as teeth wear down the white piece will gradually move higher (to the centre/pivot point). Once the wear reaches a point high enough, when the white piece moves sideways to push the gold plated strip to touch the outer pin, it will push it beyond the elasticity point and gets plastically deformed. I'd imagine it might begin to fail intermittently at first, like as it's rotated and clicks over say 24 times, electrically it may only contact the outer pin sufficiently to register 13 times. Also i wonder if one potential undesirable characteristic of this design would be that the strip may connect both pins at the same time (or neither pin) - with such tight tolerances i don't see how it wouldn't. Just a spit-balling theory though, i could be completely off base and wrong.. but super interesting how simplistic something so expensive is, oversimplified it's just a fancy bidirectional switch no? 😂
I have been using this kind of encoder (manufactured by Alps) in flight simulator hardware. It is used for selecting heading, speed, altitude, radio frequencies, etc. Extremely reliable, not a hiccup in 20+ years of everyday use.
@@EEVblog It's been 20+ years, so I can't say for sure, but I want to think it's the *SRBM1L0800.* I'm a bit thrown off by the "6 positions" title at the top of the page, but from the electrical diagram shown at the bottom of the page (two momentary contacts, each driven by CW or CCW rotation) it could not encode 6 discrete positions. Perhaps the "6 positions" title is carried over from the SRBM switch series, most of which are in fact simple 6-position switches, as seen by their diagrams. It does seem that SRBM1L0800 is the switch that I bought so many years ago.
LOVE that encoder. It's great for making rotary volume controls on devices with tactile vol controls built on the original boards. Just wish they made them with longer shafts and even a vertical mount.
I've got several "Alps" encoders and switches from various Film Projectors I gathered at my job throughout the years, there are some really fancy ones among them an they sure are reliable I've yet to find one that does not work.
At 12:29 rotate by about 75 degrees anticlockwise the part you are holding so that the remains of the plastic rivets line up in top and bottom halves. It is then obvious that you have the the spring is in the wrong location. Possibly the other long piece of metal then goes between that spring and the shell wall.
Long sprint pushes the white part against the cogwheel, the other shaped spring pushes against the cogwheel from the other side so the clickety ratchet action is balanced and smooth and probably to ensure a full step is made and the whilte part stops centered.
Many years ago, I wanted to convert an up-and-down buttons for a synthesizer to a shaft encoder. I wasn’t smart enough to design it so I put it in front of two other engineers. One used a PIC micro controller, and the other use some 74 LS something logic and both gentlemen able to do it. The same thing as your pulse encoder does in one little package. Very advantageous. I went with the logic and not the PIC…
LS logic would be a 74LS74 to do edge detection on the inputs, after running through 74LS132 with some RC filtering, to get a good sharp edge and debouncing, as slow edges are a bad thing with flip flops. One half provides the up pulses, and the other the down pulses, with only 2 packages and some resistors and capacitors.
@@SeanBZA Classic quadrature decoder is a D-type like the CD4013. Inputs to the Clock and Data, Q or /Q will give the direction. First used it around '72 so it's nothing new.
The quadrature encoders are trivially easy to decode in hardware - you only need a D flip-flop and you will get out a clock and direction. One of the encoder outputs is connected to the clock of the D FF, the other output is connected to the D input on the flip-flop. Now if you turn the rotary encoder, the Q output of the D FF will give you the direction, and the clock used to clock the flip-flop is a clock for pulse counting. It's not an efficient way as you cut your encoder pulse count down, but it's a cheap and easy way to get direction and clock decoding.
Certainly a cheap way to get speed and direction but you get unexpected results when changing direction. In one direction it will first count in the opposite direction when the clock signal changes before the direction signal - you can even get it to constantly count up by turning it back and forth - acceptable for a cheap input device but useless for accurate position tracking.
11:39 94€ per unit 🤨 It must be shipped in some really beautiful handmade wooden box decorated by japanese masters with a golden painture mixed with the tears of a unicorn.
I was trying to figure out how it would work and was expecting some kind of directional clutch but the actual mechanism is elegantly simple. I too wonder why it is not more popular.
If this mechanism turns out to be cheaper to produce but just as reliable as current implementations, someone will likely start producing them after becoming aware of it through this video, as any patents must have run out by now. Then again, looking at how cheap rotary encoders are nowadays, I can also imagine that no one will put in the effort.
I have encoders from ALPS that operate in a similar fashion, but which only have three terminals. The way they operate is when you turn the shaft clockwise, the common pin will first connect to the CW pin, and the CCW pin will be left open. Then for each step of the shaft rotation, the connection between common and CW will be broken briefly. Same goes to the counter-clockwise direction. The common will first connect to the CCW pin (and CW will be left floating), and for each step, the CCW pin will open for a brief moment. I bought these long ago from an electronics surplus store thinking they were quadrature encoders. Came with no model info, so can't point to a datasheet. Over the years, I've used them in a handful of projects, but they are a bit of a pain to decode reliably. They tend to occasionally give either 0 or 2 pulses for a single step of the shaft.
That's exactly why you want both the NO and NC contacts. By having two contacts for each direction, you can reliably debounce the signal. This is easy to do in either software or hardware and should completely eliminate the problem that you encountered.
@@gutschkeYou're probably right. There's a lot more redundancy for the decoding with the extra pins. I've attributed the additional pulses to the fact that even a tiny amount of back-rotation on the shaft will disconnect the direction pin from the common, and which occurs in time scales much longer than the contact bounce time. I experimented with many different lengths of dead time for the debouncing, but the double counts persist until the dead time is so long that it's otherwise unusable. You can get it to a point, where the double counts approximately equal the zero counts, and thus you're at least on correct on average. I've been using them with the common grounded and pull-ups on the direction pins, which to me feels like the right way to do it.
TH-cam serendipity! Earlier today, I was trying to locate a replacement Tektronix encoder made by Bourns. Its quadrature sectors are etched directly on the PCB board and mounted using plastic pins.
I guess it's not as popular especially in industrial use as quadrature encoding as there is less ability to detect a fault. With quadrature if one of the 2 outputs dies then you know the encoder is knackered.
But made inputs easier with a low code space, just debounce in hardware, and 2 inputs direct to a little code to increment or decrement a counter. No need to have fast code looking for edges using an interrupt, and then decide if it is valid, and update the counter. Just poll input, and look for a high on the enable pin, and then look for edges.
I saw your previous video, I think they only used the NO contacts, so it's like if you had two buttons, left and right. It would be very easy to hack in two regular tactile switches. This may be actually why they designed it this way, so that it could be retrofitted into existing designs that used up/down buttons without changing the code.
@@EEVblog Looks like it would be possible to bodge in a pair of SPDT microswitches and a 3D printed lever mechanism to activate one or the other when pressing it from either direction.
@@EEVblog Reading both the NO and NC contacts is probably done for debouncing. You could do the same, if you replaced the rotary pulse generator with two push buttons.
@@gutschke IMHO - exactly! These SPDT contacts allow you to use simple RS flip-flops for absolutely reliable debouncing (one for each channel). So basically this encoder + 2 flip-flops give you 2 logic output lines equivalent to UP and DOWN buttons, but with "built-in" debouncing. Very clever design, never saw it before! Thank you Dave for sharing this.
The problem with the design is that it is way too german to survive average users and our price competitive manufacturing. It just does not forgive any mistakes! It must be manufactured precisely to very tight specifications and it must be used in near perfect conditions or otherwise it just will not work. The approach is just not fit for the modern era of disposable everything.
Around 12:36, looking at the holes you had to 'drill out', you have the cover upside down. That detent spring would be on the side opposite the white piece, so it probably just presses against the black wheel's 'teeth', not the white 'pendulum' piece. That long thin piece of metal, I wonder if it fit into a spot in case to 'push' the white piece against the wheel?
looking inside the 'cover' (not the part with the contacts) I think I see two tiny slits for the straight piece to slip into. That would make it a sort of 'leaf spring' to hold the white piece inward toward the center and allow it to slide side-to-side as you turn the 'gear piece' each direction.
Did you turn it both ways when testing? I think I'll go back to the original video to check that out. I checked out the original video, and yes it was broken both ways. What we now know is the normally closed contact was permanently open.
This rotary pulse encoder can probably be emulated by normal quadrature pulse encoder and one dual JK type flip flop chip (e.g. 74L73): The two outputs A, B of standard encoder would have to be fed into 74LS73 like this: A-->J1+K1+CLK2, B-->CLK1+J2+K2. Then Q1 and /Q1 goes instead of the two pins on left side and Q2 and /Q2 goes instead of the two pins at right side. This is assuming TTL logic is used, otherwise 4027B can be used.
If you look at the locations of the formerly heat-welded posts that held it together, you'll see that the formed spring does not bear against the white pawl, but fits in the opposite side and will just provide detents by acting directly on the ratchet wheel teeth. I suspect that the flat strip is a spring that presses against the pawl.
Interesting- one thing that this design tradeoffs is that the backlash is not controlled. It depends on friction and how fast that little shuttle finds itself pushing on the contact. Quadrature encoders have well defined backlash important in positioning apps. This is better for a manual knob since its simple to decode. You can take some resistors and diodes and make a one terminal encoder- in one direction, the signal idles low and pulses high and in the other, idles high and pulses low- pretty simple to decode with just one pin.
I think you'll find the +1.25 and +1.75 numbers describing your glasses are not magnification. They are the reciprocal of the focal length in metres. So if your eyes can only focus on far distances, the first pair makes them focus to 1/1.25 = 0.8 m, and the second, 1/1.75 = 0.6 m. In practice your eyes can also focus closer than that, but the glasses enable them to focus even closer.
I would imagine this SPDT per direction would be easier to handle debounce, since you would only count the NO contact after switch had returned to NC contact, since the DT appears to be break before make.
I'm pretty sure it can be replaced by two push buttons. The micro only look for a pulse. There's nothing to decode here, it's juste two switches and the magic is purely mechanical.
If I remember correctly I had a similar thing in a kit for a model railroad transfer platform controller. I think it was from ALPS. Oh, sorry - that one was a quadrature encoder, and one of the best...
It seems kinda weird that the flat "contact bridge" isnt bound to one of the pins. It seems like the pressure against the "normally open" pin could pull the contact off of the "normally closed" pin
Yes, you have violated the relationship between the parts after disassembly. 6:53 You need to rewatch the video and it turns out that the curved spring is located on the opposite side. It is necessary to restore the sequence and there will be a place for a flat spring.
Very interesting. Maybe a interesting topic for another video: electronic torque measurement device, like from Amazon Aliexpress etc. Very curious of how they work and can be so acurate for that price.
Very elegant design, was wondering how on earth it worked inside, or at least how on earth they got it to work in such a low profile base. I do however wonder how fragile it is during the soldering process, the fact the pins are also the contacts and that the tolerances appear to be pretty tight makes me wonder if you reworked it a few too many times or dwelled a little too long or with a bit too much force if you'd break it or reduce the life expectancy.
@@Damien.D That is true for almost all solder in switches, where the contacts are going to melt the case a little, so they have a max temperature profile to handle that, so the pins do not move in the housing.
no, you won't need a micro to replace the encoder. just two buttons: one to go up, another to go down. It may be even an option to improve usability even now when you have encoders
I recently had to replace the quadrature rotary-encoder of my mouse because the wheel was getting flaky. I put in a cheap Chinesium one from eBay and it's back to being flaky (perhaps more so) after just a couple of months. I don't know if the encoder itself is the problem or the fact that it wasn't the right size like the listing said and maybe it's developed a cold-join from cracked solder from the stresses of middle-clicking. 🤔 I can't be arsed to reflow it right now and swapped mice.
I prefer the lookup table method. It does a good job software debouncing mechanical encoders. int8_t decode(uint8_t gray) { const int8_t states [] = {0,-1,1,0,1,0,0,-1,-1,0,0,1,0,1,-1,0}; static uint8_t old = 0; old
I've seen n something similar in climate control panel from a car, but based on a switch like ESE24MV1T from PANASONIC. protrusions end detents were built into knob and front panel. Using discrete switch could improve reliability and tolerances issue, but in this space this propably results in small number of impulses per rev
I wonder if they used this over the standard due to the application it was in. Like the static from the brushes could affect the instrument qualification with noise or something
German is not that hard for British: Drehimpulsgeber -> Dreh-impuls-geber Dreh --> drehen verb --> cognate with Old English throwen, meaning to twist, still in throw a pot from clay on a wheel. --> to turn, twist impuls --> impuls from Latin impulsus, the PPP of impello, to push geber --> noun of verb geben --> cognate to Old English giefan --> English to give --> noun: a giver --> turn-impuls-giver So easy.. :D
The data sheet lists some insanely large number of operations though. For most applications, you shouldn't see the part wear out over the lifetime of the product that it is used in.
Principle looks awesome to me. Sure I wonder how sensitive to dust it is because of the small force applied to contacts. Looks like little bit od dust would be able to act as an isolator. Anyhow something tells me if that encoder was scaled up just to like 1,5X of it's current size, it would be easy to make it reliable as well. As it is now it is like fine watch mechanic but using plastic parts as well. Sure isolating properties are needed but plastic is not most stable thing to use at that size. Nowadays there are better and reinforced plastics and Teflon's and Delrin's and such and possibly ceramics that would do way better job I think. Anyhow better plastic and bit bigger size, maybe even just 1,3 times bigger and that could be more reliable I think.
I was going to recommend some small buttons, if you couldn't find the replacement part. Not sure if they would have worked? It would have been slow dialing it in but, anything's better than bricking it!👍
Nice principle, so simple and clever, thanks for sharing. BTW, I never expected that there is a german expression for an electronics component and no proper english word. Usually, it is vice versa.
A long time ago I was working for the automotive industry R&D, blasting climate controls with an ESD gun very similar to yours Some batch of these climate controls had a problem with their encoders. They failed at 10kV I believe. I could not measure anything and wondered how on earth a simple switch contact could be damaged by an ESD pulse. I opened one up and found out they were optical with very complex light pipe action going on inside. Tiny little things with just about 10mm on each side. And sure enough after knowing what to measure I found out that one of the Phototransistors was open They had like a backplane with the active stuff and lightpipes leading to that backplane (probably to avoid ESD problems ... yikes) You could also take one functioning phototransistor out and replace it from another unit. I made a bunch of projects using these "refurbished" encoders
It's not even weird that the manufacturer does not market it internationally; they probably have their 20 German customers that you never heard of, that make stuff for another 20 German companies that you never heard of, that then make stuff for another 20 German companies that you might know if you are in a specific field, that then make stuff for companies you know very well. Typically all small/middle-sized, family run since 1887, industrial manufacturing companies, with HQ in a small town in the black forest or "somewhere near Stuttgart" or Cologne. The company would probably not even think of marketing this internationally, since they can't compete with the low cost quadrature encoders.
Did a project with a customer, where we needed a modification to one of EBEs sensors. The chief dev wanted to do it in software, but we were thin on memory (those were the days) so, we asked them and they delivered five handmade prototypes in two days.
Nice, I wish o could've found a replacement for my Yeasu 1500M encoder, poor design snapped the knob off then a couple years later they no longer had the replacement part.
@@EEVblog I got rid of it at a hamfest several years ago. Super glue worked but I wanted it to be original. It was just a poor design, if you saw the front of the radio you'd see a brick with a plastic knob ready to be struck by anything off to the side to snap the stem. Solid receive and my first Yaesu.
A simple up /down switch and some software would have provided a more reliable solution. Surely good design practice would require at least two independent sources for a critical component.
The question is why do it like this? Only advantage I can see is there's no need for logic to decode the direction as with a quadrature encoder, but that only needs a D-type and a use can usually be found for the other half of say a CD4013 if only as an inverter or buffer - how to do this is left as an exercise for the reader.
Those are just 2 microswitches with 2 outputs each in one housing? I don't know why it shouldn't work reliably as you get older. Ultimately, the well-known micro switches also work very reliably.
Yes.... I too, suffer the cursed age eye correlation thing..... I have my regular multi focal's..... Then a 1.8 and a 3 for table work..... So, I tote 3 pairs of of eye glasses around at all times, the 3 I only wear when doing PCB work as that is like 30cm max focal distance...... Getting old is not for sissies........😂🤣😂🤣 My daily carry tool... Leatherman, always have. 30 years now.
Similar situation here. Multifocals for everyday use, 4 diopters for minute work, plus an "arm's length" prescription, specific for the computer screen. This last one is highly recommended, much much better than mutifocals for large PC screens.
Hi Dave! Could you explain all the weird little phrases you use? Apparently these are cultural references, but people from other countries might be interested to know where they come from and what they mean.
It seems there is too much chance for failure with this design. Does the datasheet say how many operations this thing can handle? Asking a part to function by slamming its parts together over and over is super old school industrial revolution type stuff.
Rating is 187k revolutions or 6 million clicks. A normal PEC11 rotary encoder is only rated 30k revolutions. So likely either someone screwed up the design/implementation, or the spec is a bit of wishful thinking.
Detecting direction with a standard encoder is trivial with a processor or logic, not to mention pulse multiplying, above its standard pulse per rev rate. It's obsolete!
Whelp, totally having to eat my prior comment on this one, granted it's not a quadrature encoder, but an encoder is an encoder all the same, so you very much could have just replaced this with singular push-buttons for the up-down adjustments as you'd said. TIL!
Very cool video, I was curious how it worked! But, this is insane to me -- it feels like the market segment that this device fills must be created by incompetent and/or impossibly lazy firmware engineers. Given a choice between this clever-but-fragile mechanism, and a normal encoder plus a few lines of firmware, why on earth would you pick the former. Don't get me wrong, the device is a beautiful and elegant piece of engineering to solve the given problem. But the problem doesn't exist, and the solution introduces entirely unnecessary fragility. Or am I missing how crap microcontrollers were in the 90's?
Even today, many designs with microcontrollers are going to the edge of them, using all kind of tricks to get more functions out of the already available micro's. E.g. modern washing machines, or the cheap weather stations, show a fixed "amount" of functions, still different functions per model to choose from, but often the same amount. And I think that wasn't any different then. If you as a manufacturer could simply mechanize a part, it would be beneficial to a simple controller, and sell.
Long spring goes opposite to the indented one, you can see the place for it. The reason it stopped working after contact cleaner is you removed the lubricant that provides drag between the shaft and the bump mechanism, so it would not move the switch. Probably would have worked with a drop of silicone oil down the shaft, to make a thin viscous film. The reason it failed was probably the low viscosity oil evaporating out of the switch.
I think the additional problem was due to me heavy handidly putting my knife down each side. The case bulged out. Most likely that's not good for the tight tolerances required. But desoldering/soldering could also be an issue.
@@EEVblog Glass filled nylon there, so likely robust enough to survive soldering. Think it just lost lube there, plus the contacts wearing the gold finish off on the spring brass wiper, so it stopped making good contact. Also probably not helped by too low a contact current, they have a minimum contact current on switches for a reason. This probably has 10k resistors, and thus not enough current flow to keep the contact surfaces clean long term. Probably needs 1k and a 1n ceramic capacitor across the switch contacts, to provide that little bit of spark to clean them.
The white pawl is acetal aka Delrin which is self lubricating and will be happy to do its thing without lubrication in this type of mechanism.
@@theslimeylimey Yeah, shouldn't need lube I suspect. That's what she said.
My theory is that it failed due to wear on the cog and pin - on the assumption you didn't bend the left side gold plated strip during opening it, it looks visibly bent with a kink compared to the right side.
To elaborate, as teeth wear down the white piece will gradually move higher (to the centre/pivot point).
Once the wear reaches a point high enough, when the white piece moves sideways to push the gold plated strip to touch the outer pin, it will push it beyond the elasticity point and gets plastically deformed.
I'd imagine it might begin to fail intermittently at first, like as it's rotated and clicks over say 24 times, electrically it may only contact the outer pin sufficiently to register 13 times.
Also i wonder if one potential undesirable characteristic of this design would be that the strip may connect both pins at the same time (or neither pin) - with such tight tolerances i don't see how it wouldn't.
Just a spit-balling theory though, i could be completely off base and wrong.. but super interesting how simplistic something so expensive is, oversimplified it's just a fancy bidirectional switch no? 😂
im so glad you actually made a video for the teardown. very educational. thanks a lot
I have been using this kind of encoder (manufactured by Alps) in flight simulator hardware. It is used for selecting heading, speed, altitude, radio frequencies, etc. Extremely reliable, not a hiccup in 20+ years of everyday use.
Part number?
@@EEVblog It's been 20+ years, so I can't say for sure, but I want to think it's the *SRBM1L0800.* I'm a bit thrown off by the "6 positions" title at the top of the page, but from the electrical diagram shown at the bottom of the page (two momentary contacts, each driven by CW or CCW rotation) it could not encode 6 discrete positions. Perhaps the "6 positions" title is carried over from the SRBM switch series, most of which are in fact simple 6-position switches, as seen by their diagrams.
It does seem that SRBM1L0800 is the switch that I bought so many years ago.
LOVE that encoder. It's great for making rotary volume controls on devices with tactile vol controls built on the original boards. Just wish they made them with longer shafts and even a vertical mount.
I've got several "Alps" encoders and switches from various Film Projectors I gathered at my job throughout the years, there are some really fancy ones among them an they sure are reliable I've yet to find one that does not work.
At 12:29 rotate by about 75 degrees anticlockwise the part you are holding so that the remains of the plastic rivets line up in top and bottom halves. It is then obvious that you have the the spring is in the wrong location. Possibly the other long piece of metal then goes between that spring and the shell wall.
Long sprint pushes the white part against the cogwheel, the other shaped spring pushes against the cogwheel from the other side so the clickety ratchet action is balanced and smooth and probably to ensure a full step is made and the whilte part stops centered.
Many years ago, I wanted to convert an up-and-down buttons for a synthesizer to a shaft encoder. I wasn’t smart enough to design it so I put it in front of two other engineers. One used a PIC micro controller, and the other use some 74 LS something logic and both gentlemen able to do it. The same thing as your pulse encoder does in one little package. Very advantageous. I went with the logic and not the PIC…
LS logic would be a 74LS74 to do edge detection on the inputs, after running through 74LS132 with some RC filtering, to get a good sharp edge and debouncing, as slow edges are a bad thing with flip flops. One half provides the up pulses, and the other the down pulses, with only 2 packages and some resistors and capacitors.
@@SeanBZA Classic quadrature decoder is a D-type like the CD4013. Inputs to the Clock and Data, Q or /Q will give the direction. First used it around '72 so it's nothing new.
The quadrature encoders are trivially easy to decode in hardware - you only need a D flip-flop and you will get out a clock and direction. One of the encoder outputs is connected to the clock of the D FF, the other output is connected to the D input on the flip-flop. Now if you turn the rotary encoder, the Q output of the D FF will give you the direction, and the clock used to clock the flip-flop is a clock for pulse counting. It's not an efficient way as you cut your encoder pulse count down, but it's a cheap and easy way to get direction and clock decoding.
Certainly a cheap way to get speed and direction but you get unexpected results when changing direction. In one direction it will first count in the opposite direction when the clock signal changes before the direction signal - you can even get it to constantly count up by turning it back and forth - acceptable for a cheap input device but useless for accurate position tracking.
Nowadays obviously this would be done in software but the mechanical solutions always make me smile.
11:39 94€ per unit 🤨 It must be shipped in some really beautiful handmade wooden box decorated by japanese masters with a golden painture mixed with the tears of a unicorn.
If your reading glasses are marked 1.25, that's the diopter value, which is the reciprocal of the focal length in meters.
For reading glasses, you can actually specify several parameters. They can have diopters, astigmatism correction, and magnification.
@gutschke As far as I know prescriptions are expressed in diopters.
The spring was FLAPPING IN THE BREEZE dave
What an outside the box way to think about how rotary encoders work! Performs the same task in a really cool way.
I was trying to figure out how it would work and was expecting some kind of directional clutch but the actual mechanism is elegantly simple. I too wonder why it is not more popular.
Yeah, I was surprised by the simplicity. But the instant I saw it it was obvious.
If this mechanism turns out to be cheaper to produce but just as reliable as current implementations, someone will likely start producing them after becoming aware of it through this video, as any patents must have run out by now.
Then again, looking at how cheap rotary encoders are nowadays, I can also imagine that no one will put in the effort.
Awesome video, i love when you explore rare components
I have encoders from ALPS that operate in a similar fashion, but which only have three terminals. The way they operate is when you turn the shaft clockwise, the common pin will first connect to the CW pin, and the CCW pin will be left open. Then for each step of the shaft rotation, the connection between common and CW will be broken briefly. Same goes to the counter-clockwise direction. The common will first connect to the CCW pin (and CW will be left floating), and for each step, the CCW pin will open for a brief moment. I bought these long ago from an electronics surplus store thinking they were quadrature encoders. Came with no model info, so can't point to a datasheet. Over the years, I've used them in a handful of projects, but they are a bit of a pain to decode reliably. They tend to occasionally give either 0 or 2 pulses for a single step of the shaft.
That's exactly why you want both the NO and NC contacts. By having two contacts for each direction, you can reliably debounce the signal. This is easy to do in either software or hardware and should completely eliminate the problem that you encountered.
@@gutschkeYou're probably right. There's a lot more redundancy for the decoding with the extra pins. I've attributed the additional pulses to the fact that even a tiny amount of back-rotation on the shaft will disconnect the direction pin from the common, and which occurs in time scales much longer than the contact bounce time. I experimented with many different lengths of dead time for the debouncing, but the double counts persist until the dead time is so long that it's otherwise unusable. You can get it to a point, where the double counts approximately equal the zero counts, and thus you're at least on correct on average. I've been using them with the common grounded and pull-ups on the direction pins, which to me feels like the right way to do it.
TH-cam serendipity! Earlier today, I was trying to locate a replacement Tektronix encoder made by Bourns. Its quadrature sectors are etched directly on the PCB board and mounted using plastic pins.
I guess it's not as popular especially in industrial use as quadrature encoding as there is less ability to detect a fault. With quadrature if one of the 2 outputs dies then you know the encoder is knackered.
True.
But made inputs easier with a low code space, just debounce in hardware, and 2 inputs direct to a little code to increment or decrement a counter. No need to have fast code looking for edges using an interrupt, and then decide if it is valid, and update the counter. Just poll input, and look for a high on the enable pin, and then look for edges.
Very interesting component!
I saw your previous video, I think they only used the NO contacts, so it's like if you had two buttons, left and right. It would be very easy to hack in two regular tactile switches. This may be actually why they designed it this way, so that it could be retrofitted into existing designs that used up/down buttons without changing the code.
All 4 lines go to pullups and the micro.
@@EEVblog Looks like it would be possible to bodge in a pair of SPDT microswitches and a 3D printed lever mechanism to activate one or the other when pressing it from either direction.
@@NiHaoMike64 this guy here did exactly the same. th-cam.com/video/I41c4Bo2q7I/w-d-xo.htmlsi=IpoM_lLYuPleZbTM
@@EEVblog Reading both the NO and NC contacts is probably done for debouncing. You could do the same, if you replaced the rotary pulse generator with two push buttons.
@@gutschke IMHO - exactly! These SPDT contacts allow you to use simple RS flip-flops for absolutely reliable debouncing (one for each channel). So basically this encoder + 2 flip-flops give you 2 logic output lines equivalent to UP and DOWN buttons, but with "built-in" debouncing. Very clever design, never saw it before! Thank you Dave for sharing this.
The problem with the design is that it is way too german to survive average users and our price competitive manufacturing. It just does not forgive any mistakes! It must be manufactured precisely to very tight specifications and it must be used in near perfect conditions or otherwise it just will not work. The approach is just not fit for the modern era of disposable everything.
Well, for clumsy fat-fingered savages this clearly is unknown technology
Around 12:36, looking at the holes you had to 'drill out', you have the cover upside down. That detent spring would be on the side opposite the white piece, so it probably just presses against the black wheel's 'teeth', not the white 'pendulum' piece. That long thin piece of metal, I wonder if it fit into a spot in case to 'push' the white piece against the wheel?
looking inside the 'cover' (not the part with the contacts) I think I see two tiny slits for the straight piece to slip into. That would make it a sort of 'leaf spring' to hold the white piece inward toward the center and allow it to slide side-to-side as you turn the 'gear piece' each direction.
cheers dave, cunning bit of design :)
That is so cool and useful. You don't even need software to decode it.
Add velocity to your encoder interface so we don't have to turn the thing 100 times to change a value from 0 to 10...
Did you turn it both ways when testing? I think I'll go back to the original video to check that out.
I checked out the original video, and yes it was broken both ways. What we now know is the normally closed contact was permanently open.
I saw some like that in the USSR stuff before. Every one I saw was broken for what it's worth.
This rotary pulse encoder can probably be emulated by normal quadrature pulse encoder and one dual JK type flip flop chip (e.g. 74L73): The two outputs A, B of standard encoder would have to be fed into 74LS73 like this: A-->J1+K1+CLK2, B-->CLK1+J2+K2. Then Q1 and /Q1 goes instead of the two pins on left side and Q2 and /Q2 goes instead of the two pins at right side. This is assuming TTL logic is used, otherwise 4027B can be used.
If you look at the locations of the formerly heat-welded posts that held it together, you'll see that the formed spring does not bear against the white pawl, but fits in the opposite side and will just provide detents by acting directly on the ratchet wheel teeth. I suspect that the flat strip is a spring that presses against the pawl.
Interesting- one thing that this design tradeoffs is that the backlash is not controlled. It depends on friction and how fast that little shuttle finds itself pushing on the contact. Quadrature encoders have well defined backlash important in positioning apps. This is better for a manual knob since its simple to decode. You can take some resistors and diodes and make a one terminal encoder- in one direction, the signal idles low and pulses high and in the other, idles high and pulses low- pretty simple to decode with just one pin.
I think you'll find the +1.25 and +1.75 numbers describing your glasses are not magnification. They are the reciprocal of the focal length in metres. So if your eyes can only focus on far distances, the first pair makes them focus to 1/1.25 = 0.8 m, and the second, 1/1.75 = 0.6 m. In practice your eyes can also focus closer than that, but the glasses enable them to focus even closer.
I would imagine this SPDT per direction would be easier to handle debounce, since you would only count the NO contact after switch had returned to NC contact, since the DT appears to be break before make.
Just looked up Rotary angle pulse encoder, and the Wikipedia page uses a screenshot from this video 😂
😂😂
It's a little bit misleading without the toothed shaft.
Wow, imagine that!
@@Okurka. So it was a choice to not included a reference to the video or channel? 🤔
The way it pulses one pin or the other, it seems like it was intended to drive 74192/193 counters directly.
I'm pretty sure it can be replaced by two push buttons. The micro only look for a pulse. There's nothing to decode here, it's juste two switches and the magic is purely mechanical.
I use 1.5 for normal reading but bought a 3.0 for use as magnifier glasses. Now teardown a different version, please :-)
That is one of the standard way of sending control pulses to an industrial servo amplifier.
Excellent.
The bent one goes opposite the white piece to produce the detent. The straight one pushes the white piece against the teeth.
If I remember correctly I had a similar thing in a kit for a model railroad transfer platform controller. I think it was from ALPS.
Oh, sorry - that one was a quadrature encoder, and one of the best...
It seems kinda weird that the flat "contact bridge" isnt bound to one of the pins. It seems like the pressure against the "normally open" pin could pull the contact off of the "normally closed" pin
Yes, you have violated the relationship between the parts after disassembly. 6:53 You need to rewatch the video and it turns out that the curved spring is located on the opposite side. It is necessary to restore the sequence and there will be a place for a flat spring.
Very interesting. Maybe a interesting topic for another video: electronic torque measurement device, like from Amazon Aliexpress etc. Very curious of how they work and can be so acurate for that price.
Interesting Device,
Very elegant design, was wondering how on earth it worked inside, or at least how on earth they got it to work in such a low profile base.
I do however wonder how fragile it is during the soldering process, the fact the pins are also the contacts and that the tolerances appear to be pretty tight makes me wonder if you reworked it a few too many times or dwelled a little too long or with a bit too much force if you'd break it or reduce the life expectancy.
the datasheet is indeed recommending not to heat it too much.
@@Damien.D That is true for almost all solder in switches, where the contacts are going to melt the case a little, so they have a max temperature profile to handle that, so the pins do not move in the housing.
I just noticed the pixellation of the mechanism in the video thumbnail. That's some funny clickbait there. Well done.
You should try the bourns ACE encoders, they are also creative types of encoder
I think this is quite clever.
no, you won't need a micro to replace the encoder. just two buttons: one to go up, another to go down. It may be even an option to improve usability even now when you have encoders
Very clever. Not sure what pins are supposed to make or break as it turns.Looks a bit like a mini morse code key
I recently had to replace the quadrature rotary-encoder of my mouse because the wheel was getting flaky. I put in a cheap Chinesium one from eBay and it's back to being flaky (perhaps more so) after just a couple of months. I don't know if the encoder itself is the problem or the fact that it wasn't the right size like the listing said and maybe it's developed a cold-join from cracked solder from the stresses of middle-clicking. 🤔 I can't be arsed to reflow it right now and swapped mice.
Pretty interesting mechanism.
I think optical detection of rotation is easier/cheaper.
that's like ideal substitution for plus minus buttons
Two variations of a quadrature decoder:
int decode( unsigned gray )
{
static unsigned old = 0;
unsigned binary = gray ^ (gray >> 1);
int delta = (binary - old) & 3;
old = binary;
if ( delta == 2 )
delta = 0;
return delta - ((delta & 2) > 1);
int delta = binary - old;
old = binary;
switch ( delta )
{
case -1:
case +3:
return -1;
case +1:
case -3:
return +1;
}
return 0;
}
I prefer the lookup table method. It does a good job software debouncing mechanical encoders.
int8_t decode(uint8_t gray)
{
const int8_t states [] = {0,-1,1,0,1,0,0,-1,-1,0,0,1,0,1,-1,0};
static uint8_t old = 0;
old
Pretty cool. Hmm, coupling it with an up/down counter comes to my mind...
Debounce!
I've seen n something similar in climate control panel from a car, but based on a switch like ESE24MV1T from PANASONIC. protrusions end detents were built into knob and front panel. Using discrete switch could improve reliability and tolerances issue, but in this space this propably results in small number of impulses per rev
I wonder if they used this over the standard due to the application it was in. Like the static from the brushes could affect the instrument qualification with noise or something
German is not that hard for British: Drehimpulsgeber -> Dreh-impuls-geber
Dreh --> drehen verb --> cognate with Old English throwen, meaning to twist, still in throw a pot from clay on a wheel. --> to turn, twist
impuls --> impuls from Latin impulsus, the PPP of impello, to push
geber --> noun of verb geben --> cognate to Old English giefan --> English to give --> noun: a giver
--> turn-impuls-giver
So easy.. :D
So the double-pole arrangement is to avoid having to deal with switch bounce.
If I would guess The contact goes to the left 12:14
I’m kinda partial to optical encoders. That one seems like it would wear out too soon.
The data sheet lists some insanely large number of operations though. For most applications, you shouldn't see the part wear out over the lifetime of the product that it is used in.
You say it costs $6, but the Burklin page lists it for 93 Euro
@@Okurka. 6 Euro each, plus postage.
I guess the 93€ are mainly the cost for it sitting in a shelf since 1993 😂
doesn't a small spring with a tooth create a click on the gear?
Principle looks awesome to me. Sure I wonder how sensitive to dust it is because of the small force applied to contacts. Looks like little bit od dust would be able to act as an isolator. Anyhow something tells me if that encoder was scaled up just to like 1,5X of it's current size, it would be easy to make it reliable as well. As it is now it is like fine watch mechanic but using plastic parts as well. Sure isolating properties are needed but plastic is not most stable thing to use at that size. Nowadays there are better and reinforced plastics and Teflon's and Delrin's and such and possibly ceramics that would do way better job I think. Anyhow better plastic and bit bigger size, maybe even just 1,3 times bigger and that could be more reliable I think.
I was going to recommend some small buttons, if you couldn't find the replacement part.
Not sure if they would have worked?
It would have been slow dialing it in but, anything's better than bricking it!👍
Nice principle, so simple and clever, thanks for sharing. BTW, I never expected that there is a german expression for an electronics component and no proper english word. Usually, it is vice versa.
A long time ago I was working for the automotive industry R&D, blasting climate controls with an ESD gun very similar to yours
Some batch of these climate controls had a problem with their encoders. They failed at 10kV I believe.
I could not measure anything and wondered how on earth a simple switch contact could be damaged by an ESD pulse. I opened one up and found out they were optical with very complex light pipe action going on inside. Tiny little things with just about 10mm on each side. And sure enough after knowing what to measure I found out that one of the Phototransistors was open
They had like a backplane with the active stuff and lightpipes leading to that backplane (probably to avoid ESD problems ... yikes) You could also take one functioning phototransistor out and replace it from another unit. I made a bunch of projects using these "refurbished" encoders
It's not even weird that the manufacturer does not market it internationally; they probably have their 20 German customers that you never heard of, that make stuff for another 20 German companies that you never heard of, that then make stuff for another 20 German companies that you might know if you are in a specific field, that then make stuff for companies you know very well. Typically all small/middle-sized, family run since 1887, industrial manufacturing companies, with HQ in a small town in the black forest or "somewhere near Stuttgart" or Cologne. The company would probably not even think of marketing this internationally, since they can't compete with the low cost quadrature encoders.
Did a project with a customer, where we needed a modification to one of EBEs sensors. The chief dev wanted to do it in software, but we were thin on memory (those were the days) so, we asked them and they delivered five handmade prototypes in two days.
I feel like that functionality could be replicated with a miniaturized version of the rotary switch mechanism used on multi-meters.
Dave briefly mentioned expecting wiper contacts in this thing, just like a quadrature encoder would have.
Nice, I wish o could've found a replacement for my Yeasu 1500M encoder, poor design snapped the knob off then a couple years later they no longer had the replacement part.
Custom jobbie?
@@EEVblog I got rid of it at a hamfest several years ago. Super glue worked but I wanted it to be original. It was just a poor design, if you saw the front of the radio you'd see a brick with a plastic knob ready to be struck by anything off to the side to snap the stem. Solid receive and my first Yaesu.
Won't a SLA-Printed one do the job?
A simple up /down switch and some software would have provided a more reliable solution. Surely good design practice would require at least two independent sources for a critical component.
In the 90s you had Alps, EBE, Möller and Marquard.
@@falkmachtsachen there is still a less complex, more reliable solution using simple push to make switches.
Looks like the flipper on a ducks a$$. Great video!
I wouldn't say it's unreliabl:, it worked 20 years
That operation looks like fancy way of two buttons._
The question is why do it like this? Only advantage I can see is there's no need for logic to decode the direction as with a quadrature encoder, but that only needs a D-type and a use can usually be found for the other half of say a CD4013 if only as an inverter or buffer - how to do this is left as an exercise for the reader.
At least this is trivial to convert into two push buttons.
I guess it lost the precision over time due to metal fatigue.
Possible.
Those are just 2 microswitches with 2 outputs each in one housing? I don't know why it shouldn't work reliably as you get older. Ultimately, the well-known micro switches also work very reliably.
So just the 94 € then!
I wonder why no one uses them...
Yes.... I too, suffer the cursed age eye correlation thing..... I have my regular multi focal's..... Then a 1.8 and a 3 for table work..... So, I tote 3 pairs of of eye glasses around at all times, the 3 I only wear when doing PCB work as that is like 30cm max focal distance...... Getting old is not for sissies........😂🤣😂🤣 My daily carry tool... Leatherman, always have. 30 years now.
The Leatherman can't fit in your fob pocket.
Similar situation here. Multifocals for everyday use, 4 diopters for minute work, plus an "arm's length" prescription, specific for the computer screen. This last one is highly recommended, much much better than mutifocals for large PC screens.
Once he fixes it he's like, "how do I set the clock, ah whatever, I don't want it anymore, it's not broken."
Well done sir!
Hi Dave! Could you explain all the weird little phrases you use? Apparently these are cultural references, but people from other countries might be interested to know where they come from and what they mean.
And I'd have the eep explaining them over and over and over again, year in year out...
@@EEVblog So it's a kind of tradition)
Would you give me a link?
Brilliant and unreliable at the same time
EBE is known for the opposite. That gun is older than youtube.
First thing i liked about this video was the Bitcoin logo on the mic :D
Musisz to składać odwrotnie. Zobacz na kołki, które rozwierciłeś.
...but there are no kangaroos in Austria!
It seems there is too much chance for failure with this design. Does the datasheet say how many operations this thing can handle? Asking a part to function by slamming its parts together over and over is super old school industrial revolution type stuff.
Rating is 187k revolutions or 6 million clicks. A normal PEC11 rotary encoder is only rated 30k revolutions. So likely either someone screwed up the design/implementation, or the spec is a bit of wishful thinking.
@@jaro6985 As my Mercedes has a bunch of EBEs encoder switches: These are reliable but don't like being not used.
Detecting direction with a standard encoder is trivial with a processor or logic, not to mention pulse multiplying, above its standard pulse per rev rate.
It's obsolete!
Quadrature is THE way to go; the algorithm to decode the bits is super simple and there are way fewer moving parts.
This is TERRIBLE, Muriel!
Pfff mine glasses are 7.25, but with all the new technologies they aren't that thick.
WTF!
@@EEVblog without them it's all blurry after 10 cm ;)
"Everyday carry shootout" Lol
Whelp, totally having to eat my prior comment on this one, granted it's not a quadrature encoder, but an encoder is an encoder all the same, so you very much could have just replaced this with singular push-buttons for the up-down adjustments as you'd said. TIL!
Very cool video, I was curious how it worked! But, this is insane to me -- it feels like the market segment that this device fills must be created by incompetent and/or impossibly lazy firmware engineers. Given a choice between this clever-but-fragile mechanism, and a normal encoder plus a few lines of firmware, why on earth would you pick the former. Don't get me wrong, the device is a beautiful and elegant piece of engineering to solve the given problem. But the problem doesn't exist, and the solution introduces entirely unnecessary fragility. Or am I missing how crap microcontrollers were in the 90's?
Even today, many designs with microcontrollers are going to the edge of them, using all kind of tricks to get more functions out of the already available micro's. E.g. modern washing machines, or the cheap weather stations, show a fixed "amount" of functions, still different functions per model to choose from, but often the same amount. And I think that wasn't any different then. If you as a manufacturer could simply mechanize a part, it would be beneficial to a simple controller, and sell.
Normal encoders in a ESD gun...
TYRE TEARDOWN....🤣🤣 CAPS👍👍
Seems like a perfect encoder for a MEMS. But as we can see, at larger scales, its "dicky."
no need to censor anything man, i'd watch ur vids regardless and others would too (wrt the blurred insides of the encoder)
It ruins the surprise of tryign to guess how it works before I open it.
Drehimpulsgeber: er gibt drehimpulse
Hi! You get 2 pc. of encoder from German EBAY, not one for 6 Euro.....
Guess I'm getting 4 then.
"turny impulse-giver"
Und die Kandidaten raten