REVISED 19/12/09
Closed-loop control is a system that uses an electronic control unit in conjunction with feedback sensors to monitor and modify its own operation. In the context of semi-automatic gear selection, the closed-loop system monitors the gear position sensor to determine if and when a shift has been successfully completed. This information is fed back to the gearbox control unit (GCU) so that the pneumatic actuator, engine cut or throttle blip can be turned on and off as necessary to effect the fastest and most reliable gear shifts.
By contrast, an 'open-loop' system does not have any feedback mechanism and simply relies on fixed time delays for engine cut, throttle blip and shift actuator operation. At best, this is a significant compromise, which leads to increased shift times and numerous miss-shifts. At worst, the gearbox can suffer excessive dog wear or even catastrophic breakages.
To get a better understanding of why closed loop control is not only desirable but absolutely essential, we must first understand the basic mechanics of a gear shift using dog engagement. However, I'm not going to go into exact details here, as it will give the game away to our competitors! Only a very few of them appear to have worked it out so far ;)
Any gear shift, whether it's an up-shift or a down-shift, is a 2 stage event. First we must successfully disengage the current gear, then we must engage the next gear. Both these events take a varying length of time to execute depending upon various dynamic factors acting upon the vehicle at the time. Disengaging the current gear is the most challenging aspect of the shift. This is due to a phenomenon known as transmission wind-up. On up-shifts, this wind-up has a tendency to keep the dogs locked together for a period of time after the engine power has been cut. A similar thing happens (but in reverse) on down-shifts during the throttle blip period. The time it takes for the dogs to unlock depends upon several complex factors which are impossible to model or predict, suffice to say that the strategy to successfully get out of the current gear and engage the next one as quickly as possible is not just a case of cutting the engine or blipping the throttle for a few milliseconds and hoping for the best!
So, we now know that the time it takes for the shift to complete can vary, depending upon various factors. What you might not appreciate is how dramatic that variation can be. The results of our datalogging show that in the ideal situation of a clean engagement with no transmission wind-up, the shift time is basically as fast as the pneumatic actuator can move the gear lever. On most gearboxes, this is in the order of 25-30ms. This is the typical time that would be measured 'on the bench' with no load. However, in the real world with the gearbox transmitting torque, the shift time can increase dramatically. Typically we would see upshift times ranging from about 30ms to 200ms, or in some instances considerably longer.
Now, if a semi-auto shift system uses open-loop operation, the only way to avoid missed shifts is to cut the engine for sufficiently long enough to allow for the slowest possible shift. Using the example above, this would need to be in the order of 200ms. If the cut time is set to a shorter period, as is usually the case with most unintelligent aftermarket shifters, there will be an increase in missed shifts because the current gear has not been given sufficient time to disengage. Of greater concern is the fact there will also be some dog-damaging miss-shifts where the current gear has just been disengaged, but insufficient cut time allowed to engage the next gear. Some companies selling open-loop shifters are claiming typical cut times as low as 20ms - I don't know about you, but I would like to see the evidence before believing these claims. Strangely enough, no data-log evidence is presented on their websites. You can see some of our customer data logs here and here, which clearly demonstrate the necessity for variable cut times, and also put the 20ms claims into context!
While on the subject of shift times, we have read claims on one of our competitors websites (and rather amusingly, blatantly copied on another) of performance gains of 1/5 of a second per gear shift. Compared with what, exactly? 1/5 of a second is 200ms, which is actually longer than a decent driver will take to change gear using a stick! So where does the saving come from? Of course there are no such like for like savings. It's the same sort of nonsense argument that claims if you exit a corner 10kph faster, you will be 10kph faster all along the next straight. Utter rubbish. Even if the shift time was reduced from 200ms to zero, there would not be a corresponding track time saving of 200ms because the vehicle still has forward velocity during the shift. The only saving comes from the extra time that the vehicle would be accelerating, which isn't going to add up to much over a period of 200ms unless you have a hyper-performance car! That's not to say that there aren't worthwhile savings to be made over a race distance, but we don't sell our system on that basis. We tell you the truth, even if sometimes it's not necessarily what you want to hear, or have been led to believe by others.
Anyway, back to the plot. Next we must consider what happens if the shift completes before the end of the engine cut. In other words, the engine remains cut while the vehicle is in gear. When this happens, the forward momentum of the car drives the engine, and the gearbox is transmitting reverse torque. Remember that the throttle will be wide open at this point, and under these conditions the engine becomes nothing more than a compressor providing a very effective brake. So, instead of increasing performance, an unintelligent open-loop system will often reduce performance and potentially cause vehicle instability because of the large dip in the acceleration curve. The sellers of open-loop shift systems don't mention that on their websites do they? Furthermore, when the engine power is eventually resumed, the sudden torque reversal causes the transmission to shunt aggressively, leading to possible transmission breakages as well as the aforementioned chassis instability.If the shift strategy is not optimised by use of closed-loop control strategies, time-consuming or potentially damaging miss-shifts will occur as described above. Optimising the shift strategy is a very complicated affair and is governed by numerous factors. Even variations in track conditions can have a significant impact on the shift time. Hopefully you can appreciate that a fixed engine cut time rarely coincides with the actual shift time, and the cut will almost always be either too long or too short when using open-loop methods. So, what's the solution?
The only way to determine the required engine cut duration (and throttle blip duration on downshifts) is to monitor the gearbox barrel position sensor and alter the cut dynamically. When the Geartronics system makes an upshift, the engine cut is maintained until the GCU detects that the gear has been engaged. The engine power is then resumed without any time wasting excess cut period, and there is no transmission shunting that can cause instability in the chassis.
No doubt you will be told (usually by companies that are selling the open-loop stuff) that it's not necessary to use a gearbox sensor, even on motorcycle gearboxes. They will do their utmost to convince you that closed-loop control is complicated and unnecessary. Well, if that were the case, then why have we (along with every other professional paddleshift designer up to and including F1) invested so much time and money in developing such a system? In the case of Geartronics, it's certainly not so that we can charge a lot more for our system, because one well-known open-loop solenoid based system costs almost as much as ours.
We're not going to name the companies that are using open-loop controllers, that's up to you to do the research, but it doesn't take much investigation to find out. Simply ask the vendor if their system connects to the gear position sensor. If the answer is "yes", then the system is probably running closed-loop control. If the answer is "no" then we would strongly suggest that you either come to us (or one of our respected competitors who are using closed-loop control), or forget the idea of paddleshift completely and stick with a gear lever! Just to give you an idea of how few closed-loop systems there are out there, we know of only one electric solenoid system to use such technology - this is the highly respected Zytek 'EGS' as used in A1GP, Formula Nippon and several LMP sports cars. At the time of writing, we believe most commercially available pneumatic systems in the UK are using closed-loop control.
In summary: for any serious semi-auto shift system to work consistently & reliably, the GCU needs to measure as an absolute minimum the gearbox barrel position, throttle position and engine speed. In order to measure these three basic parameters, the GCU needs to be a fairly sophisticated microprocessor device running a configurable software programme. All of the current 'cheap' solenoid based systems that we are aware of use either very basic digital controllers that are adjusted by means of internal 'DIP' switches, or imprecise analogue controllers that are adusted by a set of screws on the side of the box! - we'll leave you to draw your own conclusions...
All text is protected by copyright, Neil Wallace & Geartronics Ltd. No unauthorised reproduction in part or in whole.By all means provide a link to this text, but under no circumstances whatsoever must you copy it to your website or otherwise publish it without the prior written consent of the author.