Lessons from ShaftLab - 1

What ShaftLab is


Let's start by repeating the brief history of ShaftLab given earlier.

In the early 1990s, the R&D lab for TrueTemper shafts came up with a measurement tool they called ShaftLab. Its purpose was to find out more about how golfers bend the shaft during the downswing. It did that job admirably. In fact, it created such a stir that TrueTemper decided to market it to clubfitters as a high-end shaft-fitting tool. They have indeed sold some, though it is hardly common to find one in a custom club establishment. It is still too expensive. Also, it is limited in function for the price; it is closer to cost-effective as a research tool, which is how it started its life.

Shortly after this article was first released (2007), TrueTemper stopped making ShaftLab available. They discontinued it, both as a product and for in-house fitting use. (I don't know if they still use it for research.)

What ShaftLab is

ShaftLab is an instrument that allows measuring and recording the shaft bend during the swing. In order to do that, TrueTemper engineers rigged a golf club to measure its own bend and send the reading to a computer. The computer records and displays the shaft bend in a variety of ways. Once TrueTemper "product-ized" the ShaftLab, it added software to interpret the recorded bend and prescribe a shaft of a particular flex for the golfer.

That was very brief. Let's go into some more detail.

How it works

First let's look at the instrumented golf club.

The TT engineers attached four strain gauges to the shaft at 90º intervals. A strain gauge is a device whose electrical properties change when it is stretched or compressed. When a shaft bends, the outside of the bend gets longer and the inside gets shorter. So a strain gauge on the outside of the bend will experience stretching, and one on the inside of the bend will experience compression.

The strain gauges on a ShaftLab club are arranged so that two measure lead-lag bend (the red ones in the picture) and the other two measure toe-heel bend (the green ones; only one is visible, the other being hidden behind the shaft). For instance, if the shaft experiences "lead" bend, the red strain gauge on the left is compressed and the one on the right is stretched. This produces an electrical differential that is sent to the computer.

How is the signal sent to the computer. So far (as of 2007), the club is tethered to the computer by a flexible cable. There have been many user requests to use some sort of wireless technology, but so far TT has not seen fit to do the modification.

So the computer gets two sets of signals from the golf club: the amount of lead-lag bend, and the amount of toe-heel bend.

It "sees" the bend as a point on two-dimensional graph. For instance, the red triangle in the X-Y graph at the left represents a bend that is considerably toe-up and somewhat lagging.

The computer's job is to save these points often enough to give a smooth curve of the bend during the swing. I think they sample every millisecond, but it may be every 2 milliseconds. The result is that the computer can plot the shaft bend variation with time for as much of the swing as you'd like. As the software comes from TT, the plot is designed to cover the downswing and looks like this....
The blue trace is the heel-toe bend, and the red trace the lead-lag bend. The swing is that of tour pro Peter Jacobsen. TrueTemper includes with the package the ShaftLab plots for a variety or tour pro swings. When I had an opportunity to play with ShaftLab, it was the 1999 edition -- so I have the pro swings from that product. I'm sure it is a different set of pros today.

We can see from the graph that Jacobsen has a fairly long downswing: about 570 milliseconds from start to impact. (Later we will learn this isn't exactly true.) If we didn't read this from the graph, we could see it at the top of the page, where the software tabulates a few important parameters of the swing.

While this form of plot is certainly interesting, there is another plot that is at least as informative, that doesn't come with the ShaftLab software. (I used the Excel spreadsheet's plotting capability to generate it from the ShaftLab graph for Jacobsen.)

The graph at the right is a time-lapse picture of the X-Y plot we saw earlier on this page. The blue numbers next to the X-Y points are the milliseconds before impact. So we are tracing the two bends -- toe-heel and lead-lag -- for the 570 milliseconds of Jacobsen's downswing. A few points worth noting, which are typical characteristics of a pro swing:
  1. The first 400msec of the swing (more than 2/3 of the downswing), all the bend is toe-up, with very little lead-lag bend at all.
  2. The swing then turns into the [Toe-up, Lag] quadrant of the graph.
  3. With a little less than 100msec left to impact, the total bend peaks, a combination of toe-up and lag bend.
  4. For only the last 50msec do we see the bend get to toe-down, then show some lead. The actual direction of bend here is very chaotic. We will go into the reasons below, but you can also see this from the canonical ShaftLab output graph above.

What's in the ShaftLab product?

There are two interesting things to say about the product that TrueTemper ships:

1. The clubs: The instrument comes with four clubs: a driver and a 5-iron each in right-handed and left-handed versions. All the swings for fitting are made with these clubs. The shafts are all TrueTemper S-flex steel shafts, so they are very predictable from sample to sample. Therefore every ShaftLab trace you will see should be comparable, even if taken on different ShaftLabs.

2. Fitting software: ShaftLab is sold as a high-tech fitting tool. It would be nice if it lived up to that billing. Unfortunately it falls short in several ways:
I do some consulting for a ShaftLab owner. His approach is to use the same parameters the ShaftLab software uses for fitting, but he converts it into a frequency rather than a TrueTemper shaft model. That way, he can select from any shaft he wants. It would be nice if ShaftLab did this. But an obvious conflict of interest prevents TT from adding this capability. So you are basically paying a high price (currently at $6000, down from several tens of thousands a few years ago) for a device that tells you to spend more money with TrueTemper.

What ShaftLab can and can't tell us

Let's start this by discussing the assumptions underlying ShaftLab:
  1. The shape and size of the shaft bend trace is sufficient to tell you what you need to fit the swing to a shaft.
  2. The shape and size of the shaft bend trace does not vary with club length. (The instruction manual does have graphs to scale the recommended frequency with club length. But this just covers the standard frequency-length slope, not the possibility that the golfer would actually swing the club differently if it is longer or shorter.)
  3. The shape of the shaft bend trace will not materially change if a different shaft is used. The only thing that will change is the size. In other words, the trace magnitude scales with shaft stiffness.
Do I agree with these assumptions? Some yes, some no.
  1. This assumption says that shaft profile does not matter. But it does! These days, driver fitting is done with profile and weight as much as overall flex. So ShaftLab at best is really only doing part of a fitting: the overall stiffness of the shaft.
  2. This assumption is contrary to my observations. I have seen golfers who have a definite ideal club length, and their swing mechanics change for longer clubs. At the other end of the spectrum, I know some very tall golfers whose swing mechanics break down if the club is too short; it hurts their posture. I have never tested this assumption on a ShaftLab directly, but I suspect the swing changes due to seriously wrong length will affect the parameters ShaftLab uses to recommend a shaft.
  3. I believe this assumption. We'll discuss it in a lot more detail as we get into the lessons ShaftLab teaches us.
One final limitation. It's not a flaw but, given all the things that ShaftLab can do, it's worth mentioning it can't do this. And, while it is not important for fitting, it is important for ShaftLab's role as a research tool. ShaftLab separates the bending into toe-heel and lead-lag, but it can't relate that to in-plane vs out-of-plane. A few pictures (from Jack Nicklaus' book "Golf My Way") to show what I mean...

At the start of the downswing, the club's heel-toe plane (green line) is aligned in the same direction as the swing plane (blue line).

At the moment of impact, the clubface has squared and the lead-lag direction is aligned with the swing plane.

So... At the start of the ShaftLab trace, heel-toe bend is the same as in-plane bend. This continues for a while into the downswing. But as soon as interesting stuff starts happening to the shaft bend, this relationship ends. Then, at impact, lead-lag bend is the same as in-plane bend. But this is not true even a few milliseconds either side of impact. In order to relate shaft bend to any analytical model, it is important to know the in-plane and out-of-plane bends, rather than heel-toe and lead-lag bends. ShaftLab cannot tell us this.

Bottom line: I think ShaftLab is one of the best fitting technologies out there. But it is still far from perfect.

Last modified  11/07//2017