Home Hobby Hobber Part II

[@gyrogearloose1345]

Good work Cleff! Glad you made some progress, solved some problems and found new ones. Yes, the cumulative errors! Keep on bro'. On to Part III !

[@Oskupossu]

Good to see how gradually work is going on.

[@FCleff]

Welcome to my channel Oskupossu. Thank you for your interest and encouragement.

[@MegaChekov]

well done thanks for sharing all the best from John in Texas

[@marianodiaz461]

I have cummulative errors on commercial machines, when I try to do indexing that is not divisible by 360, I fixed that by writing the program to look at the n of teeth to be done , and putting the formula in the controller , IE , if I have to do 21 holes, tell it to keep count of how many holes it has done ,that every index is 360/21* space number that has to be machined ,

[@FCleff]

Hi mariano diaz, Thank you for sharing your experience with solving this kind of problem. I imagine that commercial machines have a much higher gear ratio/resolution (~50:1/ 0.0005 degree) on 4th axis devices (rotary tables, dividing heads) and very fine resolution (~0.0001") in X & Y-axis for circular interpolation. Your programming solution plus the resolution yields non-compounding, high resolution results with true positions better than 0.001 degrees. My controller can do the math with VERY high precision BUT the motion will never be better than +- 0.1125 degrees due to the open loop stepper motor combined with a 4:1 gear ratio. "Rounding" up/ down on position eliminates compounding errors BUT this precision is, in my humble opinion, not good enough for "precision" work. I've worked out a rather complicated differential gear drive that would permit both high precision dividing AND hobbing but life is short and other projects will most likely get priority. Cheers, F.C.

[@marianodiaz461]

F. Cleff , it really the precision does not matter,( within reason) as the error given is the error of one , and the cumulative error does not creep in.You are doing a great job! I would love to do this project , may be to old by the time they publish it...

[@RakeshKumar-kr5ck]

How to setup gear premium in the machine from easy way and having machine

[@Kaptainkneemow]

What a great setup. How do you know what rpm to rotate the spindle for a given hob? Sorry for the lame question.

[@FCleff]

Hi Kaptainkneemow, How are things on the Nautilus? ;-) Yours is not a lame question at all and like most speed/feed questions for the home workshop the answer is......"well, that all depends on the diameter of the tool, the size of the gear teeth being cut, the depth of cut, what material you're cutting, what type of material the hob is made out of, how rigid the machine and setup are and what type/volume of coolant you're using". Additionally, the Arduino Uno can only read the encoder up to a max RPM of about 400 so this setup is somewhat self-limiting. In these videos I'm cutting Delrin (soft and easy to machine) with a HSS Hob so 400 RPM works fine. Hope that helps? Thanks for your interest, F.C.

[@skidloaders1]

Nice job. in my opinion it would be wise to consider converting your mil to linuxcnc. It,s freeEven a partial convention is possible, like you have already done.It will open so many more possibilities and it's not harder to learn than arduino programming.Happy machining to you.

[@FCleff]

Hi Skid, Thanks for the heads-up on linuxcnc. IF I should ever consider converting my mill to CNC I'll certainly keep it in mind. However, for my goals on this project (learn about programming Arduino, learn about gear hobbing, keep the cost of the controller at ~$100 US, learn about Bluetooth/smart phone user interfaces, design/fabricate PC board via the internet, etc, etc) this was the best approach. Cheers, FC

[@andypughtube]

KZbin just suggested this to me, and then I saw the LinuxCNC mention, and thought I would chime in. Hobbing turns out to be pretty easy to add to a LinuxCNC machine. It was one of those ideas that popped up as I was falling asleep, and I made it work a few days later. kzbin.info/www/bejne/kJmsdKWYZbmVpJY It was initially just a few extra lines in the HAL file, driving the A axis a different way. Of course, I then ended up with feature-creep and converted a much better machine, and made a much better hobbing setup, but using the same principles. kzbin.info/www/bejne/hne3eYxrip2pY7c So I can confirm that LinuxCNC is a great choice for controlling a hobbing machine, or for making other machines into hobbers. It is probably worth noting that the first machine is using steppers open loop with an encoder on the spindle and the second is using servos with resolver feedback and another resolver on the spindle and the hobbing modifications to the HAL file are just the same in both cases.

[@andypughtube]

Oh, and reading further through the comments. In LinuxCNC the positions are double-precision floating point. You will get rounding errors eventually, but the sun will burn out first. (I exaggerate, it's actually only thousands of years)

[@aaronpadilla458]

Hi, are you sharing your code?

[@FCleff]

Hi Aaron Padilla, I'm not sharing the code YET but plan to some time in the future. Real life has been throwing me curves lately so, I can't say exactly the code (Arduino and Smartphone) will be ready for prime time. Thanks for your interest, F.C,

[@torquelimiter6776]

Good stuff, thank you for sharing

[@FCleff]

Thanks for your interest and encouragement.

[@asharahmadarrauf]

Is it a open source code that you may want to share ?

[@FCleff]

Hi Moto Wizard, The entire source code for this version of the Home Hobby Hobber is standard Arduino(eze). However, I no longer have access to it due to a computer crash. :-( Cheers, F.C.

[@thecorbies]

Hi. Presumably this is Arduino driven, but did you write the sketch yourself or did you find it somewhere? I'd love to make one of these systems at some point. Any advice/pointers would be greatly appreciated. Regards Mark in the UK

[@FCleff]

Hi thecorbies, Thank you for your interest in this project. Yes, I wrote the sketch myself however it uses a hardware Arduino library file to read the encoder. This is/was my first Arduino sketch and was intended as a useful learning exercise. IF you just want to hob spur/worm/helical gears, the code and mechanical hardware are very simple. I started with stepper motors/easy drivers but recently changed to closed loop steppers with integrated drivers to increase speeds and accuracy. Both designs work with the same code (more or less). The hob spindle and the work spindle must ALWAYS have the same resolution. On my system the work spindle and hob spindle now have 9000 pulses/rev. The code simply counts hob spindle/encoder pulses and fires a pulse when the tooth count is reached. This creates the 1/tooth count ratio. Please let me know if this was useful. Cheers, F.C.

[@thecorbies]

@@FCleff Yes, very helpful although having recently re-located what was my garden workshop to an industrial unit which I share with one of sons, I'm not yet at the starting stage of the project. I DO understand the principles, but have no 'hands on' experience yet. May I contact you again nearer the time that I am ready to start? Regards Mark

[@FCleff]

@@thecorbies Mark, absolutely, any time. IF you plan to do more than "hobby" work you might take a look at these guys: kzbin.info/door/sqnf9U8J4zu1THyuMVtGVQ in Germany. They have done a great job both in video quality and technical justification. They rely on a collection of actual gears in their design (even with digital control) but the design has advantages and, in my humble opinion, a lot more complexity. Cheers, F.C.

[@samrodian919]

Reading some of the comments, my, my don't some people try to complicate things! Especially as Mr Cleff has said his goal is a $ 100 US controller.

[@FCleff]

Hi SAMRODIAN, Thank you for your interest. I value the perspective of others and the opportunity to further explain aspects of this project that are not clear. Cheers, F.C.

[@rogercrier]

Here is a link to the Touchdro kit of parts in case you want to build one the easy way! I have just ordered one for my mill, to complement the one on my lathe. The lathe one is arduino based with a diy veroboard shield, and while it is perfectly fine, I cannot be bothered to faff about building another veroboard, shield, so have bitten the bullet and paid the £70 ( crap exchange rate plus postage) for properly made circuit board. Nice to see peeps are sending you info to help sort the non integer problems. www.yuriystoys.com/p/store.html

[@FCleff]

Thanks, Roger.

[@watahyahknow]

not quite sure but check the channel from cluogh 42 kzbin.info?search_query=clough+42 hes allso using digital ratio , hes building a powerfeed for the leadscrew on his lathe to tap treads without having to change the gear ratios on his small norton style lathe from wat i understood from his lesson on the subject he gotten the compounding errors fixed to at least verry little error with the way he has it calculated in the program dont ask me to explain it any better my eyes are still glazed over , though im not over in a corner with my arms around my knees teethering back and forth mumbling to make it stop anymore looking for some sort of digital ratio for a cilinder honing machine and i'm interested in homemade tools thats how i came across youre channel for the honing machine i can work with some compounding errors but need sumting calculated from 4 pieces of data inserted in the little computer (bore size of cilinder / dept of cilinder / speed of hone / cross hatc angle ) and have it set the fifth setting (distance per minute on stroke ) to get the crosshatch right , the rest like the stroke and maximum dept can be set and ajusted manually

[@reiniertl]

Stop using uSteps, that only adds error. Stepper motors are designed to make full steps, half steps is OK, u-steps is just messing with the motor and adding a lot of error. If you have a 400 step/rev motor your angular precision will be 0.9 degree at full step, 0.45 at half step and you add a pulley system, lets say 2:1 then it goes 0.225 steps/rev at half step. You don't have to use u-steps, pulleys will reduce a lot the precision factor. Wat you get with u-steps is worse angular resolution and repeatability and less torque. So go to full step and use pulleys for better resolution, repeatability and more torque, you loose speed, but don't think that u-step makes that much better. u-step tries to emulate a DC motor with stepper motors, in my opinion a really bad idea for most applications. Not just because it is available in most commercial controllers it means should be used without carefully thinking.

[@FCleff]

Hi Reinier, Thank you for you interest and comment and I agree, u-steps a neither accurate or precise. This is especially true of inexpensive stepper motors and drives. That is why I only use half steps for this HOBBER project. The motor resolution is 400 full steps per rev. The timing pulleys are 4:1 and the driver is stet at half steps. So........................................400x4x2=3200, the resolution of the system. It works surprisingly well in the HOBBER mode but, as you point out, is too imprecise and inaccurate for indexing. Cheers, F.C.

[@reiniertl]

I think you can overcome that problem by using an encoder on the index table. It is a cheap and very simple method and making your own encoder is not a hard task, given the fact that you have a fully functional workshop. I think the problem with stepper motors is the lack of feedback. If you miss a step you loose accurate position of your device, but if you have an encoder, then you take the position directly from the table or spindle and accurate indexing becomes a simpler task. It may not be easy to add an encoder to a rotary table, but it is very easy to a lathe spindle. The larger the disk, the more lines you can add and the better the angular resolution.

[@FCleff]

Hi Reinier, Thank you for the suggestion. Please understand that this is NOT a simple move-stop-move indexer. It is a hobber and, like all hobbing machines, it runs in a continuous process and it works quite well running open loop with crude error correction done on the fly through firmware. To date, I have successfully cut MANY gears with it. It is my opinion that adding parts and complexity will kill the design goal of $100 US for the controller. I have a move-stop-move indexing head that can easily work to 0.0001 degrees that uses a simple differential gear train and, again, works quite well. Cheers, F.C.