Torsen housing axial loads

[ebashevkin]

I'm Brown University's drivetrain engineer for 2010. In the past we've always run a Salisbury-type differential, but I'm looking to switch to the Torsen University Special this year. I'm currently in the process of designing a housing to make sure it will fit on our car.

One of the things I'm particularly worried about is the axial loads the housing sees when a torque is applied and the element gears push the side gears axially due to the helix angle between the two. However, since we do not have a Torsen differential on hand (we're waiting to buy one until we can be sure it'll fit) I have no way of measuring this helix angle and by extension calculating the axial forces on the housing. I've tried contacting Torsen but they were unable to provide me with the information. Has anyone measured what the axial loading on the housing is as a function of applied torque, or could anyone with a University Special on hand let me know what the helix angles are so I can calculate it myself?

Thanks a lot in advance!
-Eli

---
Brown University FSAE
Drivetrain Engineer

[ebashevkin]

I'm Brown University's drivetrain engineer for 2010. In the past we've always run a Salisbury-type differential, but I'm looking to switch to the Torsen University Special this year. I'm currently in the process of designing a housing to make sure it will fit on our car.

One of the things I'm particularly worried about is the axial loads the housing sees when a torque is applied and the element gears push the side gears axially due to the helix angle between the two. However, since we do not have a Torsen differential on hand (we're waiting to buy one until we can be sure it'll fit) I have no way of measuring this helix angle and by extension calculating the axial forces on the housing. I've tried contacting Torsen but they were unable to provide me with the information. Has anyone measured what the axial loading on the housing is as a function of applied torque, or could anyone with a University Special on hand let me know what the helix angles are so I can calculate it myself?

Thanks a lot in advance!
-Eli

---
Brown University FSAE
Drivetrain Engineer

[Hector]

SAE technical paper #2002-01-1046 "An Evaluation of Torque Bias and Efficiency of Torsen Differential" describes how to model the loads of a Torsen differential. At the end of the paper it's got a table of inputs and a table of expected outputs (to test your spreadsheet/MATLAB file/whatever you use to solve for the loads). SAE provides each team with up to 30 free papers per year. Not sure where the link is, but you can search the FSAE forums for the link.

As far as I know all Torsens use 45 degree right-hand helical angles.

[Tim.Wright]

Try this document
http://www.torsen.com/files/Tr..._Control_Article.pdf (http://www.torsen.com/files/Traction_Control_Article.pdf)

The axial force would be the component of F1 acting in the axial direction. I had a MATLAB file which solved all those equations but I can't find it.

You should be able to set it up though. Its just solving for about 12 or so unknowns so you will get all the forces not just the axial force.

Actually, I just thought of another way. Take the torque produced by the diff, divide by the radius of the element gears and multiply by the cosine (or sine depending on the convention) of the helix angle. That should work??

-Tim

[MalcolmG]

On a slightly unrelated note, why are you moving away from the Salisbury diff? I personally think the Salisbury is one of the more ideal diffs out there for FSAE - with the complexity of the design/tuning being one of the biggest downfalls - but if you already have done them before then why stop?

[TimR]

While you are designing the housing make sure you use some centre-ing design (not just 'Cocentric' in solidworks or bolts)so that the two parts of the housing don't run eccentric. The guy who did last years one didn't do this, and the diff underwent some automachining. Not good to drain the diff oil and find glitter in it...

[Hector]

Who said anything about this being a two-piece housing? There's a lot of successful one-piece units out there.

/Is it a two piece housing?

[ebashevkin]

Hector: Yeah, I'm trying for a one-piece housing. It'll be a pain to machine, but I think it's worth it. Also thanks for the info on the paper, I'll be looking at that.

Timo: What I was planning on doing was just using the helix angle to calculate axial force, as you suggested. I originally didn't have the helix angles though (since we don't have a unit on hand) but since Hector said it's 45 degrees that should be easy to do.

MalcomG: First off, I'm not entirely sure if we're switching yet, I'm just looking into it and making sure a Torsen would fit in our packaging. But it seems to me that the Torsen may, over the course of a full race, transfer a larger percentage of the power to the wheels than a Salisbury. On the other hand, the more I've been looking the more similar the two differentials look. Both lock more under power, it just seems that the Salisbury has a non-zero intercept (when graphing torque applied vs. locking power) whereas the Torsen crosses the origin.

---
Brown University FSAE
Drivetrain Engineer

[billywight]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">But it seems to me that the Torsen may, over the course of a full race, transfer a larger percentage of the power to the wheels than a Salisbury. --- Both lock more under power, it just seems that the Salisbury has a non-zero intercept (when graphing torque applied vs. locking power) whereas the Torsen crosses the origin. </div></BLOCKQUOTE>

Keep researching...

[ebashevkin]

After further researching, this is what I get:

Both the Torsen and the Salisbury allow your wheels to independently but try to prevent the wheels from doing so. Both accomplish this by designing interfaces into the system that generate friction when the wheels are moving at different speeds. In the Salisbury, this is from the clutch plates. In the Torsen, it's from the gear meshes and the friction between washers. In both, the friction generated is proportional to applied torque. In the Salisbury, this is because a higher torque means more force pushing the clam shells apart (due to the ramp angles) which means more force pushing the clutch plates together. In the Torsen, this is because more torque means more gear-gear friction and because the helix angles generate side forces, more force pushing the washers together.

Thus they act pretty much the same - the main difference would be that on the Salisbury you can have a force resisting differentiation even when no torque is applied thanks to preload while this isn't possible (easily) on the Torsen. However, the Salisbury's characteristics change over time thanks to the wearing of the clutch plates whereas the Torsen stays (mostly) constant after the break in period.

---
Brown University FSAE
Drivetrain Engineer

[Thrainer]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by ebashevkin:
... However, the Salisbury's characteristics change over time thanks to the wearing of the clutch plates ... </div></BLOCKQUOTE>

I can see how wear would change the preload, but not how the lockup under acceleration or braking is affected. Can you please explain?

[ebashevkin]

Well, the characteristics change over time primarily due to the change in preload. This also reduces (slightly) the locking under braking / acceleration. In addition at some point the wear starts to change the coefficient of friction of the clutch plates. But most of the effect is due to the loss of preload. And if you can't have preload why run a Salisbury?

[billywight]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">And if you can't have preload why run a Salisbury? </div></BLOCKQUOTE>

Why would you want to run preload in a Salisbury??? What are the advantages?