Treadmill amperage spikes? 1

[43st]

We're having an issue with a small treadmill. We're using a 15amp DC motor and when someone is running on the treadmill we're getting current spikes that are faulting the system.

The treadmill is 4' wide, by maybe 5' long on the top, with 6" dia. rollers with a belt linkage to the motor gearbox.

What are your opinions on this? How do consumer treadmills deal with this? My inclination is that a mechanical absorber needs to be implemented on the belt stage to the rollers. Perhaps a flexible spider or helical beam style coupling?

 

[IRstuff]

And, how big are these spikes? Are they consistent with the load change?

TTFN

FAQ731-376

 

[ivymike]

I expect they deal with it electronically - you can usually hear the motor frequency changing with each step, and I would guess that current spikes are associated with the motor frequency changes. Perhaps a big cap and some diodes?

 

[43st]

We have a 220 lb. test subject running when the spikes occur. The spikes are all over the place. They don't occur at lower speeds, only at at high speed with the heavy foot fall. The spikes are over 30 amps.

In the past we've used much longer belts, and I guess the elasticity has acted as a torque limiter of sorts. (they didn't have this issue)

 

[Tmoose]

V belt or cogged poly chain type ?

 

[43st]

It's an Eagle Pd style belt.. we're removing it and putting in roller chain with a clutch. We'll see how it goes.

 

[zekeman]

Spike is the result of an exchange of linear momentum.
Guy drops on belt with impulsive force that tends to slow belt (and motor); motor speed control responds to increase the torque on the motor to accelerate it to recover the sudden loss of angular velocity;the torque-time impulse leads to a current- time pulse.
The stiffer the coupling, the more pronounced the spike.

 

[zekeman]

Mechanical solutions like absorbers or a clutch could help this but are expensive;I think an electrical solution like a current limiter in the electronics circuit would be much cheaper.

 

[electricpete]

For rotating equipment in general, providing increased rotating inertia (for example flywheel) is one means to reduce short-term load fluctuations seen at the motor.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[43st]

This is a one off, same is everything we do. So price really isn't a big concern. If we were manufacturing many of these then it would be more important. I have no idea about our automation drives, they are old DC servo based, if our automation guys could handle the spike they would've done so. But I will ask them tomorrow.

The torque limiter clutch is less than $100 from Martin (they have the same style in McMaster), and it's also adjustable. It's basically a sprocket compressed between two friction pads on a hub, pretty low tech.

 

[EngineerTex]

Which direction are the spikes coming from?

With DC motors, when they are back-driven, they act like generators. I would wonder if something is happening to cause the motor to be over-revved for a tiny amount of time, which would induce a huge voltage across the windings, which, in turn would try to find some type of load. When it doesn't find any load/resistance, it tries to put that load back to the power grid and there's pretty much no resistance there, so pop goes the fuse.

That could be completely out of left field, but I just don't think that an instantaneous stopping of the belt with a heavy footstep is suddenly going to send the resistance (reluctance) across the motor's windings to zero.

One way to check if my theory has any merit is to ask: Where is the motor? Is it at the front of the treadmill, or is it at the back? Depending on what's happening with each footstep, it should be able to tell you what's causing the overcurrent. I'm assuming that there's some amount of sag in the belt in the return area (below the running surface) If it's at the front, then the runner probably wouldn't be able to stop the motor instantaneously, because the sag would have to be taken up before the motor stops. If the motor is at the rear, then he probably wouldn't be able to over-rev it for the same reason.

Engineering is not the science behind building. It is the science behind not building.

 

[43st]

In this particular instance he was running toward the drive roller. There is no slack in the belt, we have an idler roller on the bottom to keep the tension tuned in. (we were unable to have adjustable end rollers due to space constraints)

The spikes were going up.. it's a 15 amp motor and we were getting 30-40 amp spikes I believe. When the guy walks it is within range with mild peaks. I'm pretty sure the drive is trying to maintain a constant velocity and when the foot hits it's trying to compensate.

 

[geesamand]

I suspect a torque limiter will be very dangerous for the athlete. I've seen knee joint instability studies are performed using a sliding floor plate, so I assume you're not trying to replicate that.

I've run on a few treadmills and low rotational inertia is typical of the cheaper treadmills. Larger inertia costs more in terms of material but it's also more realistic to the runner than something electronic. (If you think about it, the road is not made of anything elastic when you're running and pulling yourself forward, should the treadmill be?).

Another factor is the quality of the running stride. Good runners impact their foot below their CG. 220lb athletes typically don't have the cumulative mileage under their belts to have developed a highly efficient running stride, in fact they tend to be extremely powerful and not too smooth.

 

[43st]

Of course this will all be tested.. Safety is very very important when human subjects are involved.

The torque limiter will only slip in the upper 10% of the curve, it eases back in as the torque eases back. As I understand it this is precisely how home and gym treadmills work, and when they start to age they start 'sticking' when someone runs on it.

 

[EngineerTex]

I haven't yet been persuaded away from the idea that the runner is over-revving the motor for an extremely tiny amount of time.

Whether you are using a universal motor or a permanent magnet motor or any other type of DC motor, there is a matter of the reluctance in the winding that is going to have a resistive effect against the applied voltage. Your controller doesn't dump current into the motor. Current occurs because of an applied driving voltage. What type of voltage would your controller need to apply to give you 30 amps? Does your controller have some type of voltage limiter or maybe a current limiter to prevent itself from applying 30 amps to begin with?

However, if I'm wrong and it turns out that the motor is being stopped by the runner, then the answer would be much more low-tech than to add a mechanical clutch. All you would need to do is reduce the tension in the belt or use a more pliant spring on the tensioner. If the runner is stopping the belt with each footstep, the motor has about 90% of the belt to stretch, since it is trying to push the belt between the roller and the location the runner put down his foot (5 inches away) and at the same time, trying to stretch the amount of belt on the bottom between the two rollers and all of the belt between the back roller and the runner's newly-placed foot. Personally, I think that you have too much spring in the system already to support the idea that the motor is being stopped. But I've been wrong before.

Engineering is not the science behind building. It is the science behind not building.

 

[43st]

We will look at this EngineerTex, thanks for taking the time to explain what you think is going on. I don't know the details of our DC motor system.. it's all fairly old DC servo gear, on Allen Bradley drives.

Adding a spring to any of the tensioning means going back to the computer model and making something to fit into our mechanism. The idler roller is on an adjustable shaft and not sprung. Right now the fastest swap is the limiter assembly, off comes the drive belt and pulleys, on goes the #50 chain and sprockets.

There is a slight bit of flex on the drive roller shaft which translates to the pd drive belt bellying a bit when the system is being run, but it's really pretty minimal.

 

[jstewart]

Hi 43st,

A couple of thoughts (from an engineer at an exercise equipment manufacturer):

1. Most commercial and consumer-grade treadmills don't use synchronous drive belts. Ours use Micro-V, and others I've seen use similar. You're thus slip-limited in the torque that the motor can see. Also, the belt isn't quite as stiff; many manufacturers even use so-called "auto-tension" belts with low-modulus reinforcement, which reduce the stiffness still further. I don't think I'd care much for the feel of a chain-driven treadmill, but I've been surprised before.

2. The motor usually has a flywheel installed on the motor shaft. Even consumer treadmills usually have one.

3. We don't use high servo gain on the controller. I suspect the current spike (and high voltage) you're seeing is because your AB controller is trying to maintain a fairly constant speed on the motor. In fact, treadmills with somewhat spongy response (within reason) tend to feel subjectively better than those with tight feedback control.

- Jon

 

[jstewart]

One other thought - on over-revving the motor:

We don't usually see that happen except at steep incline angles, but even then, flywheel inertia tends to prevent "spikes." You tend to see reduction in load to the motor, and ultimately motor over-speed errors. Yes, you need to deal with back-EMF off the motor, but it's not a lot of voltage because the speed of over-run is low.

- Jon

 

[43st]

Great post, thanks jstewart!

This is good to know moving forward, if we need to do something similar in the future.

In our test our system was operating in local with no servo information being used. Right now the plan is the torque limiter and to loosen the belt idler tension. If that falls flat we will look into a v-belt, but we're a bit limited in diameter as well as depth so this will be more tricky.

I try to exhaust the simple fixes first, then we'll look at something more intrusive. So often we're just on the edge and a little tweak can get things moving perfectly.

 

[zekeman]

From all the comments I think there are 2 problems with the design, one electrical and one bio-mechanical

The spiking is a result of a tight
speed control system with limited belt friction and stiffness.Fixing the elecrical problem by increasing inertia or increasing the drive motor size will reduce the spiking but not the dynamic impact on the runner.Increasing the flexibility, reducing the servo bandwidth and increasing slippage will improve both, but the safety and performance of the system may be compromised.

The second problem has to be based on studies of the impact on the knees vs speed control and slippage design in the belts. The socalled "better" designs address this problem .

And BTW, increasing the inertia in the train is not a sign of a better treadmill but is based on the economies derived from a concomitant reduction of motor size and decreased electrical speed control.

Also, there is no easy comparison of a treadmill vs running
on a road. Did you ever see aa road move (outside of an earthquake)?