Interesting! I've never seen a gearbox encoder like that. The disc has another cool feature: it uses Gray code instead of plain binary to avoid multiple simultaneous transitions in the disc tracks. In Gray code, only one bit will ever transition at a time for reliable reading. I like the new logo.
Thanks, the gear box was a big surprise for me as well but I can think of a few places I would want to use one now that I know about them. I figured it was time to get a logo, still mucking around with it, thank god for SVG drawing.
I can't think of a way to make a tube but it might be possible to make a rectangular section with one side flat and the other side curved. It would be usable as a mono-directional level if used as one side of a container.
At Brooks Automation we used a 14 bit absolute encoder, worked a little differently than this one. I have a 25' tape measure that has an absolute encoder that can determine position to 1/16", have yet to decode the 'binary' pattern but know it only has marks and spaces of length 1, 2 and 3
with the allen bradley absoulte encoder there's a link 1 and link 2, is it the same or should use it as input for link 1 and output for link 2 as always? if many encoders are in series. or i can use both link 1 or 2 and it doesn;t matter?
Actually you can make one for the type used on SINAMIC motors .... specifically how the interpolated incremental plus absolute 12-13 bits increases overall resolution to much more than 12-13 bits.....
A great video. I would like to know which encoder is used? that gives 4096 points means its resolution must be 360/4096= 0.0878 degrees right? please let me know weather there is any encoder with less than 0.01 degrees in modern invention. thank you
The drawing that you have made on paper and on timing diagram is for binary code - where as the disc seems to be coded in gray code. Some of your explanation would really change to address this. However, seeing the disassembly was very informative.... including the multi turn aspect. Never opened a multi turn absolute encoder myself. Clear communication ....tks....
Excellent video! I was a bit confused though. 3:00 with only outer pattern you said that direction can be determined? How? I thought only speed could be determined?
hi, i have a pulse generator whose glass disc have bad secretes on some portion. Will it work or not? if not what specification required to get new one? on glass disc "HUBNER 2000" is written. What does it mean? Waiting for response. Thanks IRfan
This is so cool; I wasn't aware how rotary positions could be digitally implemented, such as automotive throttle position sensors. I suspect the two different systems are used to prevent overrun errors, at high RPM operation, where it wouldn't be feasible to scan every bit, but to sample at regular intervals to calculate the rate of revolution, as well as smooth out sampling errors which, with this technique, will inevitably accrue. In the gear train, how many sensors were there? Each sensor is effectively a binary bit, so 4 sensors would give 16 combinations, 5 -> 32, 6 -> 64, 7 - 128. If the position sample rate was 60 Hz, the disc would appear still at just 3600 RPM, but with the extra bits on the gear train to help, non-extrapolated positions could be calculated for rates of revolutions that would likely be in excess of the equipment's physical capabilities. We used to have difficulty monitoring high traffic gigabit and faster interfaces, using SNMP v1 polling, which used 32 bit counters: any interface that passed over 4 gigabytes of traffic between 5 minute poll intervals would result in useless graph plots. Thanks again for demystifying another electro-mechanical utility!
There are 6 slotted gears in the multi-turn section so that's 2^6 + 2^5 + 2^4 ... = 127 (right?) Either way the geared section is one of the coolest and cheapest bits of electronic-voodoo I have seen in a while.
Close... the last bit is 2^0, which gives you the +1 to make odd numbers, so it'll keep track of 64 revolutions. I think in order to reliably gauge roll-overs, there has to be at least two readings (I'm gonna ask someone who'd have that info about counter-rollover readily at hand) I made a simulation of such a disc in (heh) Logo... for unbounded rotation, (as in, not a stepper motor), the frequency of nonsense readings is pretty high. I must try stepping the rotation to see if that cuts down on it; i'll post the simulation soon.
how does the encoder know where it's at if it has been turned ie 5 full turns when the power is off. it knows the position it's at when power is applied to it. How does it know it's now 5 turns ahead and not at the same position it was at before power off?
If you took the pattern on that glass disk and unrolled it into a straight line it would work as a linear encoder. the longer the distance the more tracks it would need with the first one being black for half the distance and white for the other half. Just like the inner track of the disk.
One I'd like a couple dozen to make NC equipment. true it would be a only a small part of a much larger system. Even a two axis machine still would need ball screw servos and programmable control unit.
Hello Nick! How can I prove that the encoder of Heidenhein type: ECN 1313 2048 , ENDat01 type is IN DEFECTIVE?! IT is connected to a sinchronous electric motor (N.C. 41 A, from Montanari Gulio) and the encoder interacts with a VVVF drive (FRC-F7).
EnDat is a serial digital interface so it can be a bit tricky to check the absolute encoder part. The best way to check the function of the encoder is to use Heidenhain's PWM 20 tool. It's expensive, though, like 2-3000 bucks. EnDat01 has an additional incremental track (1 Volt peak-peak), which could possibly be tapped and looked at with an oscilloscope. But it's separate from the absolute part of the encoder.
Fantastic vid, really interesting to see how one of these works. I guess this combination of a gear train encoder and the main disc encoder was the cheapest, however gears could never be as reliable in the long term. If the gear train was used to aid with high speed readout, then the better solution would be a second disc encoder with a lower resolution driven by the same shaft. You could also use a spiral pattern and phase modulate the encoding which would work well at high speed.. a quick google search revealed a patent for that filed in 2013 haha, patent US8546747. Your t-shirt is awesome haha! Took me a few seconds to get it lol ;)
Glad you enjoyed the video, I was surprised they had the gears at all as the way this encoder was being used restricted its movement to 180 degrees back and forth. Thanks for the patent info, that's a really interesting design, too bad someone beat you to it. If I can get my hands on any other NFG industrial gear I'll do some more "how do they work" tear downs. The principals are always simple enough but I love all the different ways they are implemented. I made the shirt with a printing service that allowed the alpha layers in uploaded designs. I've got another in the works of a stylized circuit and I might do the camera logo in the future.
Nick Moore That's pretty interesting.. I guess the 180 degree restriction was just because of that specific application. Volume discounts often mean that you use what you have even if it's a bit wasteful. :> Since you're in Canada as well, GC Surplus can be a great source of random industrial things, though it's a blind bidding system and lately they've been raising the minimum prices haha. I was told that the safest method is just to bid half of what the item would go for used and you'd have a decent chance of winning it. Someday I'll print some nerdy shirts for myself too haha! XD Thanks for responding!
i can't understand the reason for using both patterned disc and gear train , and how the combination of the two affect the accuracy of the readings. thanks for the video
+Morkvonork No - the reason of the gearbox is different. The big central disc is absolute for a single turn. The gearbox allows to detect positions in different turns (as the signal image of the big center disc repeats every turn).
What Felix said. This could be done electronically but this method allows the number of turns to be saved when powered off. A better method is using Weigand Wire method, but this method is pretty neat also.
Hello Cody. That is your personal opinion: Weigand is the "better" method. Probably it uses less mechanical parts than a multiturn with gears, but that does not make it the "better" method. Knowing from "inside", gear failure is a very rare reason for encoder failure. Much more importance: bad voltage (overvoltage, undervoltage), wrong connection, mechanical impact on the shaft or vibrations outside the specifications kill encoders. And you find the affected parts as well in Weigand Sensors as in those that use gears. Currently, I do not know from any Weigand-Sensor available on the to fulfill SIL3 approval (machine safety, EU "Maschinenrichtline). But there are certified ones with gears.
Everything is expensive when someone else can make money with it. Nature of industrial gear. On top of that, this has to be a robust enough design and manufacture to withstand a pretty harsh environment with liquids, heat, abrasive dust, vibrations and shock in order to provide reliable and safe operation.
Your penciled diagram is wrong. On these encoders there is never two transitions occurring at the same time. The edge of the following segment is always 50% overlapped of the previous. This is why binary coding is not used in absolute positioning encoders.
Thanks for adding this correction. I initially didn't understand what you meant by binary coding not being used, but after a bit of research it started to make sense. I found a paper called *Gray Code Encoder Output vs Binary Output.* It helped me understand your comment.
