Treadmill hack - VFD for 3 phase AC Induction Motor

[rrriceu81]

Some background . . . I am a EE with general electronics, biomedical instrumentation, computer and embedded processing and some power management circuit design (switching supplies, buck, boost, maximum power point tracking). I have been interested in exploring computer control of 3 phase AC motors at some point.

A personal project I am taking on is custom control on a heavy duty treadmill. I want to design my own control of the tread speed and hence the motor speed, which is tied directly to the tread.

I opened the machine, expecting to see a DC motor with some variation on a PWM controller, which is apparently a commonly used design for smaller treadmills. I found instead a rather large, 5HP, 3 phase 230V AC motor. The machine runs off a 30 A, 120VAC power plug that I have adapted to a 20 A socket.

See photo / attachment. Hope google allows the photo reference, else it is attached below.

There are no large transformers, only a small one to supply logic level power, I presume to the controller board and the button and display panel. A control board receives the 120VAC, the output of the key pads and five identical black power wires from the motor plus output from a rotary encoder that may involve some sort of hall effect sensor as the sensor abuts the notched metal encoder wheel on one side only.

As per the label on the motor, the motor power appears to be supplied by three wires with some sort of thermostat switch provided by the other 2 of the five identical black wires. I presume the encoder is used only for speed feedback control and not really in the VFD control, but I don't know that for sure.

The controller board has six TO-247 packaged power semiconductors on a common heat sink.

So there are a few options I have considered to take control of the device by a computer . . .

1) emulate the keypad presses - the simplest idea and not a bad idea, but there are some limitations imposed by the software that I would like to override.
2) #1 plus reprogram the processor - not an option
3) buy a VFD with computer (pwm, serial, . . .) interface. An easy option but I won't be learning as much.
4) build my own VFD and program a dedicated controller (arduino, R-Pi, etc)
5) analyze and hack into (destroy) the controller boad and hijack the IGBT drive signals and build a microcontroller interface. Rather not destroy the board.

I did a little reading about VFD at the wikipedia level, but would appreciate a reference to a bit more info without getting into detail such as considerations for torque feedback, optimizing efficiency, that would not be applicable on this motor anyway.

If I build something, I would like to do my best to reduce power line distortion and there is apparently a bidirectional IGBT switching topology that does this.

I want to figure out how they are getting the DC drive voltage for the 230 V motor from the 120VAC line. Need to take a closer look at the board. Perhaps there is a switching supply there. There is a small transformer there. Maybe that is part of it.

What are your thoughts on tackling this project?

Thank you.

 

[squeeky]

Dear rrruceu,
I’m no expert but I will do my best to assist. As always I warn about taking care with the electrical side, ensure your earth leakage unit is functional. A little knowledge is dangerous and then there is rule 1. Rule 1. Trust no one. That said….

The five black wires are probably the run and start windings of a single phase motor with a single common return. To get a single phase motor to turn you need a phase displacement which you don’t get from a single phase so the make one more capacitive or inductive to give it an offset so when you start, there is a little kick which starts the rotation. After that, the start winding has no real effect unless the motor gets near stall speeds. Some designs take out that circuit with a small shaft mounted centrifugal switch. The other two are probably correctly surmised by yourself as the thermal safety device. When it gets hot it open circuits. Thus the instruction to connect it to your control or signal circuit.

I have a washing machine with a VFD in it so the prices are coming down and they will become more common. The reduction in price comes with disadvantages. Switching spikes for one. Noise generated back to the mains. This may effect other electrical equipment and you might see and hear it on RF devices. To counter this you need a choke. A large low resistance high impedance coil. If you plan to build one, watch out for resonance.

How to get 240v from 120v? (DC drive voltages) This is the principle of RMS voltages and high switching frequencies. If you switch something really fast into an inductive load, you can generate huge spikes. If you did this very quickly, you would end up with a series of high peaks very close together. Now if you take the RMS (Route Mean Square) voltage you will find that it is greater than that of the supply. RMS is the DC equivalent of AC. The faster you switch, the higher the peak, the greater the voltage within the limits of your circuit. The power output does not increase because of rule 2. Rule 2. Nothing for nothing.

Most modern electronics use this method of switch mode to make light weight power supplies. No transformers needed.
Be warned that these spikes can be many thousands of volts and if your motor insulation is only rated at 1000v it won’t be long before it fails. Shorts between turns and then to earth.

To get correct speed control you need to understand a bit more about motor dynamics. Most first year electrical course will cover this in some depth as motors are the most common form of energy transfer that I know.

 

[jraef]

No, it says it is a 3 phase motor. 5HP is surprisingly big for a treadmill though, it's damned near impossible for even a trained athlete to exert 5HP, but maybe that was the plan. One aspect of treadmill design is to take into account the possibility of regeneration, a common concern at about 3/4 HP and below, 2HP for trained athletes. So the strategy may have been that because a regenerative (4 quadrant) AC VFD is very expensive, using a 5HP motor all but guarantees that it can't happen.

So how that would work is that your 120VAC is put into what's called a "voltage doubler" circuit (do a search on that term), which rectifies it to DC, then feeds AC back into in through more diodes and caps to double the DC up to around 330VDC, the level needed for making 240VAC RMS with the transistors. Since this is not a COTS VFD, hacking into it may prove to be be difficult in terms of accessing control commands. It may simply not have what you might want, because custom built units tend to be very lean in terms of user interfaces and functions.

"Will work for (the memory of) salami"

 

[Skogsgurra]

No doubt about the motor. It is a three-phase motor.

What makes me wonder is that this must be a regen drive - you cannot take care of the thermal equivalent of 5 HP without either sending the energy back to the grid (regenerate) or heating a rather large resistor (which you don't seem to have).

Or, is there another way? The sheer size of the motor points to that. As Jeff said; 5 HP is not something a normal person can produce. Even 1 HP is real tough. Normal treadmill maximum absorbed power is probably more in the 1/4 - 1/2 HP range.

So, why a 5 HP motor? I think the answer lies in the way the brake function is achieved. Simply by adding a DC Component to the motor and letting the "athlet" work against the zero speed that the DC wants to keep the motor at. All braking power will then be developed in the rotor and it, therefore, needs to be ridiculously large not to overheat. Making the rotor oversize is a lot cheaper than using a full-blown regen drive.

I think that your #1 option is what you should stick to. Anything else will probably only make you confused. I say so because this does not seem to be a run-of-the-mill (oops) application - to learn more about VFDs, I recommend that you (carefully and thoughtfully) study and perhaps dissect a small standard VFD instead. Squeeky had a few things to say about personal safety that you should read and heed.



Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[squeeky]

Thanks for the correction guys. Of course they build single phase inverters with a 3 phase output. Didn't think the price had dropped that low. The pulse counter for feedback adds another dimension to the cost effectiveness and versatility of the VSD / VFD. Some smaller drives do have PLC logic in them with digital and analogue I/O as well. I would recommend to look at a cheap drive with PLC logic in it. The treadmill will have been built for replication and not sophistication.

 

[ScottyUK]

Interesting Gunnar, I hadn't thought of an injection brake for this application. Makes a lot of sense now you have said it.

 

[rrriceu81]

Thanks for the replies.

Squeeky, thanks for the safety tips! Yes, I was reading that simple VFD designs can induce plenty of power source distortion. I think this is because they probably employ a simple diode bridge or equivalent to get the high voltage DC. This technique has a poor power factor (PF), thus a PF corrected switching supply would be optimal here. And I did read somewhere about concern for motor insulation when using a lower price VFD with big switching spikes.

Jraef, the voltage doubler did cross my mind, but I suspect they aren't very good for something requiring considerable power. I've never done the calculations, but I suspect that the capacitors get pretty large as you go with a higher and higher order voltage multiplier with higher currents. I have usually just seen those used when lower currents are needed . . . but I'll take a look at it. Also bad PF on a voltage doubler but it would be the simplest solution to get 230V.

At most a two quadrant VFD might be needed since the thing should never need to go backwards. Being a fancier treadmill, I bet this one might has braking, perhaps needed in the situation of a large individual on a large incline with a new and slippery belt. I had a cheap treadmill years ago and I could get it to go into fault mode after I sprayed silicone lube under the belt and cranked up the incline. I think it depended on friction under the belt for braking and at low speed setting, high incline, I could get it to go faster than the selected speed and it did not like that.

I think I will consider #1 first, then maybe #2 and perhaps #4 somewhere down the line.

Any thoughts on

1) a good source for VFD's that would employ computer interface? Quite a few on ebay but I'm cautious about that stuff particularly anything Chinese as they lie about specs on that sort of stuff.
2) a general tutorial on VFD design without getting into PhD level dissertation? I am interested in 3 ph AC motor control with regeneration with respect to electric vehicles and want to learn a bit.

BTW, it does seem like a huge motor for a treadmill. A local gym was upgrading and selling about 10 of these for a few hundred $ each. These apparently went for $5K+ new! A couple of thoughts on the HP, BTW . . . The best a top athlete can crank out for maybe a minute would be 2HP or maybe 1hp for a a few minutes or about 250 watts continuously and that is from the cycling data that I recall. Cycling is the most efficient in terms of human caloric expenditure to ergometer measurement, approaching 25% at best. That is, to put 100W on a cycling ergometer, you are burning 400W or more calorically. Running presents a difficult situation for measurement of 'work' since when running on a flat surface at constant velocity you are not really doing any definable work on the treadmill. Paradoxically, while making you work harder, cranking up the elevation on the treadmill requires less work by the treadmill and
hypothetically could require a braking feature on the treadmill. Few treadmills offer a 'decline' option, but this is where the treadmill would have to do real work besides overcoming friction because it is pulling you uphill while you run downhill.

 

[jraef]

Just to clarify, in AC drives "2 quadrant" is usually referring to Fwd-Motoring / Rev-Motoring (Quadrants I and III), not Fwd-Motoring / Fwd-Braking (Quadrants I and II).

There are a couple of companies that use voltage doublers at larger HP, it's very doable, just not practical compared to using a transformer. But for a treadmill mfr, I can see them spending the extra for it in order to avoid the size, weight and heat of a transformer.

For braking, true Line Regen braking is very expensive, essentially twice the price. But if you have what is called Sensorless Vector Control (done well), it has the ability to accomplish a sort of hybrid form of braking, often referred to as "Flux Braking". In Vector Control the algorithm can separate flux producing current from torque producing current, that is how they improve performance. For FB, the drive takes regenerative energy off of the motor just like Dynamic Braking, but then instead of pumping it into resustors, it then uses that excess energy to increase the FLUX current only, which saturates the motor (needlessly from a motor standpoint) just as a way to get rid of the excess energy. It in essence just uses the motor itself as the braking resistor. This of course heats up the motor, hence the oversized motor for the task. Just a word of caution though, some of the low-end suppliers play very fast and loose with the term "Sensorless Vector Control" so make sure they offer Flux Braking (or some marketing variant) if you want to use that.

Yes, be very careful on the cheap FleaBay drives, especially the ones that appear to have no name (although you can often see the name in the photos). Many of those are Huanyang, very notorious for making false claims on specs, particularly that they run a 5 motor from a single phase source. That one is a lie, you don't find out until it arrives, then the FleaBay seller has disappeared.

"Will work for (the memory of) salami"

 

[Skogsgurra]

Flux Braking! That's news to me. Maybe because I don't work out much. Not more than needed to get around in the house and in some plants.

DC injection heats the rotor and Flux Braking heats the stator. In both cases, the motor needs to be bigger than normal. And the glass, of course, is twice as big as it needs to be.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[mikekilroy]

Does this thing run? If so plug it in and verify what is going to 3 motor leads....

Might be dc for injection braking if the treadmill is never meant to run by itself, just react to the athlete trying to move it himself; but if it needs to run by itself too, then it cannot be dc injection...

Might be voltage doubler input stage but also may not.

It could be so oversized to run SLOW without overheating; a 5hp motor running with only 1/2 load on it can run mighty near zero speed without overheating the motor.. So on that thought, it may simply be a 0-120vac rms vfd output; the ac can be rectified to 160vdc bus rather than doubled, then chopped back up and spit out to the motor as a 0-120vac at 0-30hz output so the motor effectively becomes 0-800rpm max speed. If geared for this speed for max treadmill speed, no need for voltage doubling.... maybe a KISS principle....

plug that bad boy in and MEASURE what goes to the motor. Or call the mfgr and get specs? Do some math to figure what treadmill speed is what motor speed to see what the range is to help decide if you even need more than 800rpm...

http://www.KilroyWasHere<dot>com

 

[Skogsgurra]

There are so many possibilities with induction motors...

You can have a standard scalar drive (to make the motor run) and then superimpose DC to have the braking function. All that is needed is to manipulate the PWM pattern so that there is a DC component that exerts a braking torque.

There you are; a running motor with braking ability. The constant braking torque means that the motor gets hotter than it would without the brake. And therefore is larger than it would normally be.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[rrriceu81]

mikekilroy - that is a very good point re the voltage to the motor - maybe it isn't 230V on each phase. That had crossed my mind. It is such a big motor, it could do pretty well with 160V drive. I'll check when it is running although I doubt I'll be able to fully load it to challenge the VFD to go full on.

Another note - I did notice a very large, 12" x 1.5" wire wound resistor in the motor compartment. Suspecting this is a dumping ground for braking EMF. I talked to tech support in CA about a few things hoping for some tips. They went into defensive mode about anyone trying to retrofit their machine and would not tell me much, but they did say that the resistor is used as I described above.

I will continue to pursue my interest in learning VFD. Meanwhile I have emulated the keypad presses with optocouplers closing the respective keypad lines.

I set up this page to document my work on it . . .

https://www.facebook.com/pages/Treadmill-Hack/552839608155278?ref=hl

 

[mikekilroy]

rrice, output voltage will be pretty independent of load; the motor requires a constant volts/hz so if you measure the output voltage (AND freq (lots of meters have this today) while you are at it) you will see, regardless of load, the volts going to the motor.

http://www.KilroyWasHere<dot>com

 

[Skogsgurra]

That resistor - sounds like a 250 W thing. Any resistor value (ohms) on it?



Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.