imsolidstate

Spring tester / weight scale

This is another one I finished a couple months ago but haven’t posted. I wanted to test the rate of the springs in the coilover kit on my car, and manufacturers treat this like it’s some kind of trade secret. Except for a few; my kit was from Bilstein who gave me the rates but they weren’t very believable. I knew they were higher than what I was told.

So I built a spring tester. It is basically a weigh scale that can go up to 1000 pounds or so, with a way to safely compress the spring and measure displacement. For the scale I used four load cells with one at each corner of a 1/4″ aluminum plate and another 1/4″ plate to distribute the load to the four load cells and allow fixtures for various springs. The electronics for the load cells are INA103s and some more op-amps for gain. Since load cells are bridge devices a TDK DC-DC converter drives the in-amp rails with +/- 12V. The signal from the in-amps is offset and fed to an ATmega8 which does ADC and puts the values on an LCD. I am displaying the values for each load cell as well as the sum so I can see if any load cells are not being loaded equally which could result in an overload. There is also a button on an interrupt that allows zeroing the summed weight output.

To compress the spring I used a Harbor Freight hydraulic press. The spring is completely captured at the top and bottom, if you try this at home be very careful as compressed springs can be dangerous. I made a few different fixtures so that the springs could not move in the horizontal plane. I take no responsibility for anyone trying this, if you compress a spring without constraining it properly bad things can/will happen.

For displacement measurement I used a Mitutoyo linear scale I had but for this application it’s overkill, you could just tape a ruler to the side of the press.

So let’s get some results! Below is a plot of the stock spring, the garbage Bilstein spring, and the very nice Vogtland spring. Note that I’m not doing the typical 2″displacement minus 1″ displacement equals rate. I think that’s misleading especially since “progressive” springs are more common these days. I’ll explain why after the chart:

So Bilstein told me that the “working rate” for the B14 spring is 215 lbs/in. I’ve gathered that “working rate” means rate at ride height, which if you look at the graph is actually 366 lbs/in. This is the “secondary rate” after the dead coils have all collapsed. The rate Bilstein gave me is off by 70%. You would think they might be concerned about that but I couldn’t get anywhere with their tech people.

That’s the first problem with progressive springs, the meaningless measurements that manufacturers will rarely tell you anyway. The second problem I have with progressive springs is this: to get the secondary rate you have to put in extra coils that will flatten at a certain load/height. This severely limits the travel of the spring. The Bilstein spring at ride height only had about 1.25″ of available compression before full coil bind, which is exactly what it does over bigger bumps. The bumpstops would not engage before the spring would just slam flat. There are marks on the coils where the paint has worn off to demonstrate this. And seriously, only 1″ of compression travel? Fire that engineer. I’ll bet an F1 car has about an inch of compression travel.

In contrast consider the mostly linear Vogtland spring whose picture is at the top of this post. That picture is at ride height and there is still plenty of available travel, which means the bumpstops can do their thing and keep the car from feeling like riding in a wagon. Between the B14’s spring and ridiculous damping I can’t say anything good about it. It’s probably overdamped even for a race car, but I haven’t built my shock dyno so I can’t quantify that statement.

And as usual here is a schematic. Please note though that I messed up the output buffer stages in this diagram, guess I hadn’t had enough caffeine that night. It’s just supposed to be a unity gain buffer but it was shown here wired totally wrong.

And finally source code for the project.

Tags: ATMega, AVR, C, Electronics, LCD, PCB