Request for help with extension-spring biasing problem (illustration drawing link provided)

Hello everyone,
I would like to ask if anyone can please help me with the following problem.
I have a small radial plate-cam and bearing housing (cam-housing)
that is oscillated manually by hand via a small lever. The cam-housing is oscillated both clockwise and counter-clockwise from it's centered neutral position at rest. The housing is centered by two opposing extension springs connected to the underside of the housing. The springs are basically connected in series with the housing itself being a connection link between one end of each of the two springs. The other spring ends are connected to anchor points, with one anchor point being adjustable.
I have created three small GIF drawings that illustrate the design layout, which I have zipped into a single small 44.8 KB folder. The file was zipped with the free 7-Zip but you should be able to open it with winzip or most anything. The drawings can be downloaded from the following mediafire link...
This is a dual roller conjugate radial plate-cam having two rollers that work on an inner and outer cam profile. The drawings show the two opposing springs connected to the cam-housing, but they do not show springs which are connected to the follower-roller swing arm, which counter the load that the outer roller exerts onto the outer cam profile or curve. At the centered neutral position shown in figure 1, the cam follower rollers are in contact with the center of a 3-degree dwell on the cam profile. After the lever is activated and released, the springs connected to the roller swing arm bring the cam back to it's home position, just to the point where the rollers start to contact the start of the 3-degree dwell. After this point, the opposing springs connected to the cam-housing bring the cam back to it's neutral position so that the rollers are in contact with the center of the 3-degree dwell on the cam profile, at the approximate mid-point of the cam profile. Another function of the opposing springs connected to the cam-housing is to hold the cam housing at it's centered neutral position, where the rollers are in contact with the dwell at the cam mid-point.
My main goal is to have the cam be returned as accurately as possible to it's centered neutral position after the lever is activated and released. As long as the cam is returned so that the rollers contact either edge of the 3-degree dwell, or anywhere in-between, the cam follower / rollers will be in their exact neutral position. So, I have a 3-degree dead zone or a 3-degree room for error when returning the cam to it's neutral position.
Even with no rollers in contact with the cam & nothing but the housing itself just oscillating on a shaft with no load, when I move the lever clockwise from it's neutral position and release it, it seems to return to the same exact position each time, but when I then move the lever counter-clockwise from it's neutral position, it is about 1.25- degrees off from where it was before. When the rollers are in contact with the cam and a load is applied, this seems to double the problem to 2.5 degrees, so the end of the lever is basically " from where it should be at neutral. Some of the problem is due to some mis-alignment I had between the cam & rollers which I can fix, and I can also reduce roller friction as much as possible, but this still does not explain why I am 1.25 degrees off even when there are no rollers or load in contact with the cam.
Adding some type of physical or mechanical stop at neutral would seem a logical way to bring the housing back to it's exact position each time, but there are some tactile requirements for this design so I don't want to put a physical or mechanical stop at neutral. I want to preserve the balanced & smooth feel in the lever around the neutral position as much as possible. The opposing springs I am using have a rate of 140 lb per inch, but I may be able to use springs having a rate ranging from 30 to 60 pounds per inch.
Here are some things that occurred to me...
1. Presently, I have an adjustment nut at one end of the two springs which are connected in series at the cam housing. This means that I can only lengthen or shorten both springs at once. I use this adjustment to initially align the cam so that the rollers are in contact with the center of the 3-degree dwell at neutral. However, If I could put an adjustment screw at the spring connection point on the cam, so that when an adjustment is made, one spring is lengthened and one spring is shortened at the same time, it seems this might help. Do you think an arrangement like this may solve my problem ?
2. The spring hooks may be repositioning themselves at the cam-housing as the cam-housing is oscillated. Perhaps I could use a flat-head screw through a loop-end instead of a hook, to connect the springs to the housing. This would prevent the spring ends from repositioning themselves on the housing as it's oscillated. I tried to screw the hooks down with a center screw, but it did not seem to help as I could still see them moving. Perhaps loops are necessary if a screw is used. Referring to the drawings, the spring hooks are connected vertically to the cam, but perhaps rotating the springs 90 degrees about their longitudinal axis and connecting the hooks horizontally would be better ? I have not tried this yet as I will have to redesign the spring anchors.
3. Perhaps the problem is coming from manufacturing variances in the spring rates and/or tensions. I could try to go with a precision made extension spring. Compression springs and torsion springs would be harder to implement, but still possible, if they could offer some advantage. I prefer to use extension springs if possible. Perhaps a urethane or non metallic spring would be better ? I've been thinking of something like a thin rectangular piece of rubber or urethane that can be screwed or connected to the underside of the housing. I'm hoping I can find a compact and cost effective solution (preferably a stock or easily made biasing element).
I don't need this thing to be perfect because the cam dwell gives me some room for error, but I do need to be able to reliably return the cam to it's neutral position so that the rollers are in contact with some portion of the 3-degree dwell at neutral. If I can get a repeatability of or even 1-degree in each direction from neutral, that should be OK.
Presently, the thing is darn close to where it needs to be but I just need to tweak it to get a higher degree of repeatability and accuracy when returning the cam to neutral.
I would appreciate any advice or suggestions anyone may have.
Thank you. John
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
Add image file
John2005 wrote:

Was too lazy to unzip your drawings, so maybe my tip is completely off. But the words "central position" and "two springs in series" together with "differences in springs" triggered:
You need a spring that looks like a hairpin; U-shaped. Maybe with several windings (and not just a half). The loop slips over some axle. The thing that has to be returned into neutral position has a pin that goes into the gap formed by the two ends of the hairpin. That pin has to be bigger than the gap (will be your returning force). Now there is a second pin that also goes into the gap. If you move that, the two pins get appart and tensions the spring more.
Maybe you understood what I wrote, if not, I can eMail you a sketch.
Here's a picture of such a spring: <
The lowcost-DRO:
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
Add image file
Upload is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.