Gear train subject to fluctuating loads 3

[FeX32]

Hi guys,

I am developing a system that partly involves a system of gears connected together. One of the gears is connected to a system that will exhibit a fluctuating torque which will react into the gear train (3-5 gears in series).
The amount of torque is about 5-18Nm in its fluctuating component from 25-300Hz.

Another part of my system within this gear train is measuring strain. This measurement must be accurate for the system to work. I am concerned with this fluctuating torque causing backlash in the gear train and thus impacts in the train that will effect measurements. I have a few options. But, I have not dealt much with gear trains before so I would like to ask these simple questions:
1) With those torque values and nominal pitch diameters of about 150mm, is it possible to spec gears that will not produce backlash?
2) What companies would you recommend attaining the gears? I have used a very long time ago these guys:

http://sdp-si.com/eStore/Catalog#

They don't have specs, I was hoping to see max torque or power ratings of the gears.

Thanks,
Fe

 

[tbuelna]

All gear meshes require some amount of backlash to operate safely. With 3-5 gears operating in series, the total backlash from input to output can be significant. If the torque oscillation and frequency you describe is not fully reversing then it should not present too much of a problem. But if the torque oscillation is fully reversing at that frequency it will likely create some dynamic loading conditions at your gear mesh contacts that you need to watch out for.

One thing you might consider to minimize dynamic tooth contact loads is using helical gears rather than spur gears. Or also designing your gears to maximize contact ratio, by using a lower pressure angle and/or more numbers of smaller teeth.

 

[Strong]

Fe,

If you believe that the AC torque starts at 25-Hz, then why not low-pass filter the measurement to eliminate the AC component? Numerical averaging of the DC torque value can also reduce fluctuations and improve accuracy.

Walt

 

[gearcutter]

There are gear trains that exist that can operate without backlash. Known as zero backlash systems; they are generally found in low-load applications requiring precise movement, such as rotating turrets, telescope mounts, dish antennas, etc. I have also seen them on production lines.

 

[tbuelna]

....Another part of my system within this gear train is measuring strain. This measurement must be accurate for the system to work. I am concerned with this fluctuating torque causing backlash in the gear train and thus impacts in the train that will effect measurements....

The torque oscillations don't cause backlash in the geartrain. The backlash is the result of how the gear drive components are manufactured and assembled. As I noted, a simple gear mesh requires some amount of backlash to prevent teeth from jamming tight into the mating tooth spaces. The minimum amount of backlash required for a simple gear mesh is mostly a function of things like gear profile/index error & runnout, CTE mismatch of the materials used in system, structural displacements in the system at operating loads/speeds, etc. However, as gearcutter points out, there are some methods that can be used to mitigate the effects of backlash in a gear mesh. The most common approach is using an additional parallel gear at the mesh with some mechanism to preload it in the opposite direction.

A scissor gear is one such design. In reality, each individual gear is still manufactured with some backlash, but any freeplay at the assembled gear mesh is eliminated. The drawback with using devices like a scissor gear is that they are limited in the level of reversing load they can accommodate efficiently. There tends to be lots of contact sliding in the trailing mesh providing preload, and with high preload levels this can result in significant friction losses.

 

[gearcutter]

tbuelna - while the use of a scissor gear arrangement is one way of creating a backlash- free mesh; it wasn't what I had in mind.

The most common systems that I've seen use pairs of beveloid gears (tapered-teeth spur or helical gears). Opposing tapers are meshed together on a fixed centre. Take-up of backlash is done by an axial movement of either one or both members. The systems that I've seen - use a spring to maintain an axial force, usually backed-up by adjustment nuts.

Backlash-free worm gear drives are another example - in this case a 'dual lead' worm shaft is used. Similar to the beveloid set, an axial movement of the worm is used to eliminate backlash. Several of our gear hobbing machines use this system to eliminate backlash in the systems that drive the main work table.

Below is an example of a beveloid set. The taper is exaggerated to 5 deg, usually 2 - 3.5 deg is used.

 

[tbuelna]

gearcutter- The arrangement you show will force the teeth at the mesh into contact on both the drive and coast flanks if an axial preload is applied. A tooth flank taper of only 2-3deg does not seem to be sufficient to permit axial back-driving of the gears when the direction of shaft load reverses. Even with oil lubrication, I think you would need an included angle of more than 8deg or so for the tooth flank taper to overcome the frictions at the tooth contact and create sufficient axial force to adjust the position of the gear.

The arrangement you show would likely work fine for a single gear mesh driving in only one direction, assuming the axial preload from the springs was minimal. But I don't think it would work properly if there were more than two gears in series.

 

[FeX32]

Thanks for the responses guys.
I've seen the scissor gear type and even tried to spec some. They are far too compliant for my application.

I cannot filter out the AC component, because I am after the AC component that will be reacted into this gear train (by what I will attached it too). I can't say what it is I will attach it to...

I did spec out a motor I can program to act like a break - as long as it controls the system to have more nominal break torque than the AC component, I should be good.

Does anyone have a good supplier or gear selection and design references? I want to make sure I get these speced right.

 

[FeX32]

And are gears typically speed rated? Thanks,

Fe

 

[tbuelna]

Can you provide more details of your gear train? How are the gears arranged? How is the torque applied/reacted? The max torque oscillation of 18N-m you described does not sound like something a scissor gear with a 150mm PD couldn't handle.

 

[FeX32]

The gears are in series. And the torque is applied to the gears through the shaft of one of the gears.
I have not found a good scissor gear reference.
Thanks!

 

[tbuelna]

And are gears typically speed rated?

There are several factors to consider with any gear drive design. It is not clear what you mean by "speed", but one important consideration with gears is pitch line velocity (PLV). High PLV can increase dynamic tooth contact forces, increase losses due to windage, and increase stress levels in the gear body from dynamic forces. Another consideration is "relative contact sliding velocity" at the gear mesh. This relative contact sliding velocity increases in relation to PLV and can result in significant heat generation within the lube oil film, which can quickly produce scuffing failure.

Gear drives operating at low shaft speed and/or low PLV are usually limited more by tooth contact or bending stress levels, which are mostly a function of torque.

 

[FeX32]

Thanks tbuelna!