Member postings for Andy Ash

A couple of weeks back there was a thread here all about electronics. In particular motor control. Various schemes were discussed but a common suggestion was that something like an Arduino could be used to generate PWM to control a motor. Understandably, many model engineers are not very keen to enter the electronics realm, let alone software. Vive la difference!

Over the years I have seen many threads talking about inverters and reversing switches for single and three phase motors. Sometimes I have been frightened by some of the things that I have read. This is especially because high power switchgear is not a good place to do learning. Often, fault finding and problems are posted. You know that if you were there in person it would be a simple fix, but without your own eyes and ears it is just too hard to diagnose. Often it seems that posters want to know which colour wires to connect, as if there is some universal standard. I wish there were.

I wanted to put up a really simple speed controller, that anyone could have a go at, in complete safety. One which could still be useful if needed. What I'm going to post might create a fire if you get it wrong, but it cant shock you so badly that you will be unable to put the fire out! Look on the bright side, I say!!

Actually this controller can be quite powerful. It runs from a 12V supply and is rated for at least 50amps. That gives you 600watts, nearly 1hp. You might find a motor to do that from 12V, but you're less likely to find a power supply which will deliver 50amps. My recommendation is to use a car battery as an accumulator, and set it up with a trickle charger. In the workshop environment you can drape your 12V power cable across a machine tool and not worry too much about electric shocks. Perhaps it could be good for gear hobbing or tool post grinding. It doesn't even have to be a motor, it could be a heater or anything really. That's up to you. Do use cables which are up to the job! If you don't they'll get hot and melt.

The reason I did this thing was that I remembered my demister fan in the car had not been working properly for ages. The two slow speeds didn't work, but the fastest and the next fastest did. I had a good idea that it was the dropper resistors that had failed. I also had the idea that I could make a speed controller to replace the resistors. The picture below shows the faulty resistance wire.

The module has three resistors chained and tapped. The diagram below shows the electrical arrangement.

The resistors project into the airflow from the fan. It's a great idea, the fan cools the resistors that control its speed..... The problem is that the resistors make more heat when the fan goes slowly, and so cools less. They could have rated them properly in design, but evidently they didn't.

The great advantage of a PWM speed controller is that it is much more efficient than resistors. With the resistor scheme, perhaps the motor runs at half speed. The effective resistance of the running motor is the same as the dropper resistor. Half the energy goes to the motor, half to the resistor. I measured the running current on the next to fastest speed and found 9 amps. That's 40 watts of power wasted to go at half speed. The heat has to go somewhere and that is why the resistors burn out, as so many people find. Hot in the airflow, the resistor wire oxidises readily and becomes brittle. Cars vibrate and demister fans fail regularly.

Iv'e had my tea now too!

The new data on the scales, shows that they are kind of "SSI" esque.

It might be possible to develop an interface, but as I said before it would never be ideal.

How good or otherwise an interface would be depends strongly on more data that would be needed about the scale, and on the demands of the application.

One would need to know the maximum traverse rate of the slide. Also the maximum data rate from the scale.

The DRO expects infinite time resolution, and much respect goes to its designer for that decision. A home hobby lathe probably isn't that demanding. The thing one learns about these systems is that once the time information is gone it can't be got back. Never to throw it away is a good thing.

Absolute encoders can be more precise than incremental ones, for a stationary axis.

Sometimes there is no substitute for absolute information. One may not have the scope to "zero" an axis.

The exquisite acceleration information given by an infinite time incremental signal, means that an end effector can follow a precise path no matter how fast it travels. With the absolute system, the faster the rate of travel, the poorer the quality of the path.

High end encoders are usually available with all the different systems in a single unit, but the price is physical size and usually money too. If you're building a really good robot you need all the information anyhow.

It is clear now that Yuri is doing more than he apparently claims, but not that much more. He's got quite a nice idea though. The technical difficulty isn't high, but it would be great to have on a massive milling machine. I can just imagine standing on the table of a vertical borer or something. You'd be able to reach into your pocket and your mobile phone would tell you where the tool point is. You might even be able to jog the machine through the user interface.

Aha!

New information....

Working. Standby for analysis.

I looked at the iGauging website and they had all sorts of different encoders.

I don't know which ones you have, but there's no data there anyway.

If I had your scale in my hand I could stick a scope on it and work out what is being issued. As it is I have no idea.

At least there is some information on the DRO. It talks about TTL quadrature, which is enough to know. I've always used high end kit like SICK, Heidenhain, Renishaw. If you pay proper money, you get a proper datasheet.

It sounds like Yuri's Android wotsit is reading ASCII strings and blatting it out to a graphical display. Converting ASCII to Quadrature TTL is what you need, and it is not an ideal thing to do.

Quadrature TTL is fast, it's like the slide giving your DRO an update for every micron of movement.

The ASCII scheme will update you periodically in time. If the slide moves an inch between updates, then first you'll think it in one place then another. It's like the slide dwells at A for ages, then suddenly moved between A and B in zero time.  The TTL quadrature scheme tells you about all the points in between, with infinite time resolution.

It's not ideal to convert between a high level scheme like the ASCII one and a low level one like the quadrature one. Normally you'd go the other way.

iGuaging have loads of different scales, so some might work.

Broadly though it looks like Yuri's Android software is only a display, and possibly an interpreter. Your DRO is smart enough to read a barebones (proper) encoder. The encoders that Yuri reads are smart in their own right so he has to do less in his software.

I'd say you need to inject the signals from your scales into the middle of the electronics of the DRO somewhere. Either that or pull signals from the middle of the scale electronics. Neither is very practical.

My recommendation would be to get some proper encoders;

http://www.industrialencodersdirect.co.uk/

You still would need to think about scaling and calibration. I didn't read the manual of your DRO enough to establish if it could be calibrated to different quadrature encoders. Obviously on some a pulse might represent a micron, others it might represent a meter.

Broadly you will want an "Incremental Encoder" to get the Quadrature TTL signals that your DRO datasheet requests. I think all those encoders are rotary rather than linear. You will find linear incremental scales if you look, but you might end up with Mitutoyo, Renishaw or some other expensive alternative.

Try some of the more generic Chinese scales. You might come up lucky!!

Edited By Andy Ash on 27/03/2014 19:13:04

Surely someone should just think about a bit of basic physics before wielding the H&S stick?

I don't really know why some people are trying to apply full size railway rules to miniature railways. They're just not the same, and physics says so.

Kinetic Energy = 0.5mv^2

If an accident occurs then the energy dissipated to achieve rest (and that which will cause harm) is expressed by this function. If a 5" miniature locomotive achieves 6mph it's doing well. If it's full size brother gets to 60mph, it's a bit slow. Even then, the speed is 10 times greater. A full size locomotive travelling at 6mph would have 1/100th of the energy.

Then, in the full size system, a carriage might be 400 times heavier than an individual passenger. On the 5" railway the carriage weighs something less than the weight of one passenger.

If something goes wrong and a collision occurs in full size, the fear is that a carriage would damage a passenger through deformation, confinement, and fire. At 5" gauge that is unlikely due to relative lack of energy, lack of mass, lack of combustibles and open carriages. Actually the worst that is likely to happen in a collision, is that the driver ends up sitting on top of a hot boiler.

Braking on a 5" railway is almost irrelevant, save that a train can be made to stop in order that models do not become damaged. In the worst case you can probably use your feet to stop.

As the sizes become larger, more consideration is needed.

If speeds increase, much more consideration is needed.

Actually, the big hazard with miniature railways is *loading gauge*. Ideally track furniture would be arranged such that no body part of the largest person can reach any stationary part of the railway from a moving carriage. If, god forbid, someone were to lose part of a limb it is very unlikely that a better braking system would avoid it.

That parents often choose to carry their kids in their arms, and even hold them out whilst on the move, never ceases to amaze me. I also find it incomprehensible that they often resent corrective instruction.

Edited By Andy Ash on 23/03/2014 15:45:38

You could flycut the end square, then drill and ream a hole.

If you then made a separate barrel with a peg on the end you could bolt, weld (or both) the end on.

I don't thank I'd have the patience to keep stopping the lathe to keep adjusting the tool.

I have a feeling you can get an automatic boring head, but I've never understood how those work.

Does anyone acutally use a unijunction transistor outside of those demo kits? I'm not sure myself.

Understanding how to use transistors in number is still extremely important. Even if you can get a part to replace them, it is often the case that the part isn't actually that complicated. There is still a high premium on ideas.

So, if you're designing a product which needs an analogue function which is a single source supplier, and your product is intended to have any kind of volume, you don't want to use the single source unless you have absolutely no choice.

The transistors on semiconductors can be smaller and closer together, electrically they can achieve performance attributes that could not be available on a circuit board. For a low volume product this can be a good reason to use a single source part. It makes a product more valuable.

Using an exotic amplifier in a battery charger is problematic. You want to make thousands of battery chargers, but you can only get hundreds of the exotic amplifier. Even so you still need the exotic amplifier because it makes charge termination more accurate and prolongs battery life. That's quite a realistic scenario.

The only way through is to understand how the performance enhancement can be had, and to develop something out of transistors which will never go out of fashion.

Until the Chinese copy what you're up to, it's the kind of edge which can build business advantage.

If you look at the mobile phone of yesteryear, it doesn't have the volume exotics that it does today. A modern mobile has a hugely exotic, nearly single chip solution inside. Equally they make it in massive quantities. The only reason an exotic chip like that can be manufactured so cheaply is because of the volume. That volume of sales had to be built.

If you can get an advantage out of a mobile phone part, it might be good. The trouble is that if you use one of those parts, the mobile phone company can lose interest overnight. If they don't need it for their product, the manufacturer just stops making it. Suddenly your product can't be made any more. That doesn't happen with transistors.

For the model engineer it makes sense only to use big powerful chips, because you'll only make one of whatever it is. On the other hand those big chips don't give you quite what you want. You end up having to learn about transistors anyway.

My only worry with the GCSE type training packs, is that they focus quite strongly on common emitter configurations of transistors. Obviously digital radio is quite complicated, but there is no reason not to build a low frequency superhet out of discretes. It's quite rare to see such a thing.

Edited By Andy Ash on 06/03/2014 18:59:14

I've never done any vacuum tube stuff directly myself, but I'd quite like to have a go.

I did work on a control board a few years ago which went with a travelling wave tube assembly. The TWT was about a foot long with a separate high voltage supply. That's why it needed the circuit board that I worked on. Isolation is really important with modern low voltage systems connected to powerful old tube systems.

Transistors are cheap as "chips". Get a bucketful for a couple of quid. You can do anything if you have enough of them. "3904" is NPN "3906" is PNP. expect to pay about 2p each for a hundred. Diodes are similar "4148". Although individually resistors are cheaper, you will have to buy a minimum quantity of a range of values. Getting started with resistors is more expensive.

Start off with things like Colpitts Oscillator, maybe a TTL inverter, build up to complete Op-Amp. Once you can understand and build an Op-Amp from transistors, I'd say you're pretty much there. The next thing after that is to understand the physics of the semiconductor materials.

The only reason to need that is if you want to go off in the direction of chip design. There is some in this country but it's quite rarefied. You'd probably only need or want to go that way for a job. Some people have tried making their own thin film transistors at home. You can succeed, but it is very difficult. You certainly can't make as many as you would need for a simple chip with reliable results.

You just have to have infrastructure which is impractical at home, if you want to make semiconductor chips.

The same is not true for vacuum tubes. These can be made at home with success. The only tricky part is obtaining barium for a "getter". Obviously you still need a vacuum pump, gauge and a glass lathe.

I'm thinking about trying to make some tubes at home myself.

I have a Chambers Science and technology dictionary for that. I was given it for my 21st birthday I think.

I don't use it that much, but sometimes someone will say something about biochemistry and it will get me going in the right direction to understand.

Mainly it has to be wikipedia. I'll be honest and say that it rules my life. Type the jargon into Google and follow it up with the word "wiki". So for example the word "commutator", it actually brings up a whole section on maths and conjugation. Don't be put off though because at the top of the page it says "Commutator (electrical) disambiguation", click that and you're there.

The other thing I noticed was that you were already on track with the 555 timer.

You knew you probably needed one, but wanted to know more about what it is. This is a really normal situation.

If you're close enough to an electronic part to know its number, get the data-sheet. Parts without data-sheets are rare. Some of the more mechanical components may not have part data, but all the electronic ones will. Mechanical parts include inductors, relays and connectors. Semiconductors always have them.

Some semiconductor parts, usually with hundreds or even thousands of pins will attract licence fees for use. In that small group of parts a detailed data-sheet might not be available. Usually a product summary would still be available instead.

The devilish group are parts that *used* to have a data-sheet. They're old. Everyone has forgotten. You might not be able to figure it out. Reverse engineering is your only hope.

For data-sheets, type the part number into Google and follow up with the word "datasheet".

Sometimes it can be hard to find the data. Be confident, know that it exists. This is when knowing the manufacturer begins to help.

Once upon a time parts were big enough that a manufacturer mark could be stamped on. These days manufacturers have to choose between a microdot, which can only be read by a machine, or maybe two characters that could be read by eye. Larger parts you can still work it out, but with surface mount the only time you know for sure is when the part comes out of the tape. Even then you are dependent on your supplier to put the right sticker on the tape.

It is not uncommon for the suppliers to get it wrong, and the first you know is when the product doesn't work!!!

Edited By Andy Ash on 03/03/2014 13:08:41

You can use a 555 timer as a triangle wave generator.

If you compare the triangle wave with the variable output voltage from a potentiometer, using a comparator, then you have PWM without needing C or assembler.

I have to say that I don't know that I would recommend Arduino for learning motor control.

If you want to learn about Arduino, go for it.

If you want to get a small motor control project working quickly, and you know Arduino (or any other embedded micro), then perhaps Arduino is the way to go.

The trouble is that unless you delve into the motor control details you will probably end up with a stepper motor solution. There is nothing wrong with stepper motors. Actually a stepper is a very complicated piece of design. It's just that you can do more different things with other types of motor.

You can still drive a simple AC permanent magnet motor with a very coarse square wave (an Arduino). The AC motor will always be more efficient. Driven correctly it will give just as much torque for a given winding current, and will do it for longer without overheat. Steppers are speed limited. A simple motor driven from a variable frequency source will achieve rotational speeds limited only by the physical materials that they are made from.

The only advantage of a stepper is that you can use it open loop, achieve angular precision, and avoid the stability problems which are inherent in closed loop servo control. These are not a huge hurdle, and I would argue that understanding closed loop servo control is understanding real motor control. At the very least, it is understanding the mechanical world in electronic terms; gain margin, phase margin.

If you use a stepper you avoid understanding the detail of motor control. Obviously, from a design perspective, you may not want the additional electronics for closed loop control. The stepper is still a valid solution.

More than just this (as others have suggested) if you don't know Arduino then you have that to learn before getting close to motor control. Just because it is electronic, does not mean it has to use an Arduino.

The only real down side to a simple permanent magnet AC Motor is that you might have to make it yourself, unless you want a really big one! I'm pretty sure that the RC plane types have already realised the efficiency benefits. I think you might get small motors from the states and china on e-bay.

Clearly the efficiency benefits are so great that they overcome the weight penalty of a chemical battery even in an aeroplane!!!

Edited By Andy Ash on 28/02/2014 12:14:48

I think if you were researching a decision to buy Horrowitz and Hill you might want to do a Google for the free online PDF file!!!

Edited By Andy Ash on 27/02/2014 20:12:31

The Art of Electronics is an excellent book.

If you want to go bonkers I could recommend books which would be very expensive, and numb your soul.

You would not need them.

I have learned more from semiconductor application notes than ever I did in my studies.

The 555 timer is simply a flip-flop (one bit memory cell) and a window comparator. You can use it in any way you could imagine, but it has a high current output ideal for driving a capacitor. This makes it a good timer. Normally a proper LSTTL non-retriggerable monostable is considered to be a better formal design candidate, but it depends on what you are doing. For the sake of your stock shelf a 555 timer can do more different things.

I've not got one, but I'm pretty sure you could buy a book on just the 555 timer if you wanted.

Speed control on motors is interesting.

It depends very much on the particulars of the motor you are using. If you are using a synchronous motor, then dual speed control from a single control input is easy. If you are using an asynchronous motor then it is not.

If you are using a typical brushless fan motor with an electronic commutator, then you have a means of synchronising the two motors. The commutator requires you to detect the position of the rotor for commutation. Since you have to change the polarity of the magnetic field yourself, when you do you know the rotor angle.

By counting revolutions in time you know the speed of the motor.

You can use potentiometer to specify a speed command signal.

Your commutator design would take an input which controls the current in the winding. The commutator would automatically switch the current polarity every half revolution of the rotor. The commutator would integrate (count) reversals in time. This signal would represent the speed of the motor.

You would then design an error circuit. The error circuit would compute the difference between the potentiometer speed demand, and the reported speed from the commutator. The result of the sum would be amplified greatly and fed into the current command of the commutator.

Obviously if you had two motors, you would have two electronic commutators and two error circuits.

The motors would maintain the same command speed irrespective of the load placed on them individually.

It is important to realise that it would be a "speed match" and not an "angle match".

For an angle match; control over, and feedback from the motor is required at a better precision.

Another, older way of doing this is with a synchro/servo pair. Actually with this scheme a pair of three phase synchronous servo motors can be directly controlled for both speed and angle using a synchro transmitter. All you need is wiring and a three phase power source. The transmitter and the servos, are purely wound rotating components.

These days servos are driven from inverter drives. A typical lathe motor is asynchronous (squirrel cage), but a permanent magnet three phase motor connected to an inverter can be controlled with with speed and angular precision.

Hope that all made sense!

>>>>Quoted from Russ

Sorry if I came across as bragging if that's what your pointing at, I just wanted to heat my workshop free of charge...... and thought, this is probably something someone might want to replicate for their model engineering .......

In terms of technology the first electron microscope was built in 1931.... as skilled as he is (and I certainly couldn't achieve that), it's not exactly cutting edge - whereas, I don't imaging the world networking together their mechanical calculators to join forces and battle cancer in 1931,

Not sure what you mean by the PC's, iPhones and eChuff but since Valve Corperation who employ Ben Krasnow are at the bleeding edge of "PC's, iPhones and eChuff" and it's all American I don't know anyone who confuses it as UK tech.

>>>>>

Not at all, you came across as someone who would heat their shed with their PC, and claim the money back by using it to generate virtual money.... Or something.

Maybe you should have a look at at Ben's vids.

I think Valve, the US game software company, have employed Ben to carry on building stuff in his shed.

Unlike here in the UK, there is a recognition of the benefit that brains in sheds can practically realise.

Valve have a demonstrable history of doing this. I can name other people that they "look after" too.

Comparable in the UK is Sage. Renown for accounting software. It's a corporate stitch up. I certainly don't think it would be worthwhile to approach them for an engineering development bursary, on the basis of my youtube vids!!!! (Not that I have any)

You get a company like Valve, and they recognise that all the software in the world isn't going to get you to the moon. Most importantly, they have something in them, which makes them look outward. Looking towards that which might be lost, that which could be gained.

It's not that you are bragging. It is that if we look like we are bragging, then I think we look stupid.

Maybe all model engineers have a computer big enough to heat their workshop, but I think not; I know I don't.

Edited By Andy Ash on 25/02/2014 23:00:18

Real electronic technology isn't specifically about personal computers or even ASICs.

Check out Ben Krasnow on "the tube".

He's built an electron microscope, on his Bridgeport, in his shed.

I think he got a really good job (and a respectable, open, future - with possibilities) with Valve through that project.

Valve..... now that's a commercial idea we could use in the U.K.

I think it's sad that electronic technology in the UK is assumed to be PC's, iPhones and eChuff.

I don't think it amusing, and it's definitely not worth bragging about.

As an aside, it is my opinion that someone like Burt Rutan, simply could not exist in the UK.

It is exactly this which I am attempting (but probably failing) to pinpoint.

Edited By Andy Ash on 25/02/2014 21:18:30

A really good reason to ditch those old MEM DOL starters is that they are full of asbestos.

Obviously when you wire them up the first place you end up putting your nose is straight in the asbestos cloth, so you can see what wire you're following.

I won't say you shouldn't.

Just be mindful. If it's not you it'll be someone younger than you in future.

They were great bits of kit. Much better than the modern plastic stuff, but they're still a health hazard today.

Broadly all motors should have both over temperature, and NVR.

NVR for power failure.

Thermal trip is important, because a simple fuse will not detect a winding burnout until after the event.

If you've never seen the smoke from a burning motor or a transformer, you wont understand that you'll leave the area before ever turning the power off.

If you have a choice between asbestos and no electrical protection, then asbestos it must be.

All IMO of course.

The cheap far eastern one is £30 on ebay, and it's "no worries".

Edited By Andy Ash on 07/02/2014 00:39:21

Ferrite is a nightmare for almost every ordinary manufacturing process except sintering, which is how ferrite cores are generally made. It's too hard to turn. If you get a tool point to bite, the ferrite will shatter. If you grind ferrite it will quickly clog your grinding wheel.

The shape of an inductor is often an important consideration, especially if you want to focus magnetic flux to increase flux density. For this reason, complex shaped sintered ferrite coil formers are very popular. There are some frequency response reasons too, but I think the ease of manufacture for complex shapes by sintering is the commercial driver for ferrite use.

The frequency response stuff is important because you might need to control the relative permeability of the material at specific frequencies, to improve the Q factor of inductors for filter circuits. Because sintering is based on powder metallurgy, it is easy to accurately introduce impurities which support frequency selectivity.

For ordinary turning, filing, and general working, soft iron (Fe - 0% carbon) is an excellent magnetic material. It is often better than ferrite, in the sense that it often has a higher relative permeability.

Don't use steel, or cast iron. Both are poor magnetic materials, unless you want a permanent magnet. The carbon impurities and other alloying elements lock the magnetic domains down. You can magnetise steel and cast iron electrically, but it has remnance and a non-linear B/H curve. You either have to heat treat it to release the remnance, or electrically demagnetise it.

If you actually want a permanent magnet, then you can use ordinary carbon steels (like silver steel). All you have to do is to figure out how to harden it (as hard as you can get) whilst it is under the influence of an external magnetic field. As the material falls below the curie point the field is frozen into the object. You can use this scheme to produce odd shaped magnets, but odd shaped magnetic fields.

Generally the harder the material the better the magnet. Obviously modern rare earth magnets use exotic materials.

Edited By Andy Ash on 27/12/2013 15:01:21

It is just not right to present Bullied as a second class engineer, unable to see the limitation of technology.

The poor bugger made an incredible job out of a very difficult deal, and he never complained.

Bullied not only understood the limitations of technology. He worked with the limitations and found new and better ways to solve problems. He had courage, and everyone knew he was inspired.

You have to view bullied from the perspective of the Brighton locomotive works and the constraints that it had always posed since the time of Craven. It was not poised to be easily converted to the manufacture of diesel or electric traction. No-one wanted to lose the skills or lay off the workers. Least of all bullied.

The writing was on the wall early in the 20th century. Brighton wasn't big enough for the manufacture of the engines it was still designing at the end.

The history, the knowledge, the lineage was present right up to the end.

Far from not understanding the technology, Bullied designed an engine that would ideally use all of the capabilities at Brighton, and ensure continuity of the site.

As it happened, days of steam were numbered, and no-one would sanction switch-over of the site (it's skills, facilities and capabilities) from steam to diesel or electric. The site at Brighton wasn't big enough, and had nowhere to expand. Logically it could never have competed with larger locomotive works.

There is no way that Bullied could have gone off on a tangent on his own. He worked within the scope of his financial masters. They allowed it to happen because they knew they were on the line too. They needed his inspiration, and he delivered something that no-one else could.

In my view, when Brighton finally died, it was inevitable that Pendolinos would be manufactured in Italy. Just as Isetta micro-cars would be made on the site of the old railway works. From the birth of the modern age, to the recent turn of century, the story is complete. It is now almost impossible (except for the observant) to know that the engineering works was there.

The thing is that although the closure of Brighton railway works was bitterly lamented for a long time locally, it's brains were always best. It always did more with less. It's just that in the end land costs were too high.

Edited By Andy Ash on 12/12/2013 14:28:47

You could use that switch for reversing a single phase motor, but not the way you are thinking, I fear.

The switch is of a type intended to go inside a cabinet with other control gear. It's not rated for motor switching (inductive load) even if it looks like it might be by the numbers.

It would be desirable to have sequenced contactors/switchgear mounted up on a DIN rail. Your rotary switch would sequence the switchgear. The switchgear would switch the start and run windings, provide no volt release and thermal overload for the motor windings.

A no volt release unit is nothing more than a limited version of what could be done with your reversing switch.

In the end, unless you get someone to do it for you, you're going to have to be confident about what you are doing. If you insist on doing it yourself but are not confident, just buy a cheap Chinese motor starter on e-bay. The instruction sheet is clear about how to connect it.

Single phase motor reversing isn't complicated but there is no hard and fast rule about how to implement it.

All you have to do is arrange for the start winding to be electrically reversed. The run winding will power the motor equally well in either direction no matter which way around it is connected. If you can reverse connection of the start winding relative to it, then you will "poke" the motor one way or another from a standstill. Which ever way you poke it, it will continue to run.

Normally you will have five fully rated conductors passing between a single phase motor and the reversing switch. 1 earth, 1 fixed run winding pair, 1 reversible start winding pair.