View Full Version : Linear power versus linear voltage

Crash & Burn
05-07-2006, 02:12 AM
I've come up with the following idea and I want to know what you technies think about it:)

All the transistor based controllers I've seen sofar all had the same stepsize per wiperband. Typically something like 33 ohm per band with a 10 band wiper board resulting in 330 ohm total resistance.
This will result obviously in a linear increase of the voltage supplied to the engine.
I've done some calculations to obtain the resistance values per band in order to achieve a linear power increase instead of voltage based on tha fact that voltage relates to power in square (how do you say that in English:):)

Below the results for 10 band wiperboard. First column is the resistance per band. Second column is the cumulative resistance of the wiperboard.
I've implemented this on 4 controllers
and the general perception is that the controller from a " feel" perspective is much more responsive due to the fact that every step you move on the wiperboard results in a linear progression of the power to the engine.
Inserted a link to the excel where I did the calculations.




05-08-2006, 01:06 AM

Me and my pall Steen, makes the CarSteen controllers that are so popular around here. It is a MOSFET construction, and one of the design goals has been to make a linear voltage regulator. All 25 steps are equal in voltage size. This is especially important for the "first step", since our home track has an "almost flatout curve".

Now lately we have been wondering if steps should be linear in voltage or in power (watts). We have therefore added 4 small options switches which makes it possible to modify the power responce according to 4 different curves apart from the linear model.

In my experience the voltage linear setup is in general the best.
Maybe because an electric motor is supposed to take RPM's in linear responce to the applied voltage.
However you might occationally come across cars that perform differently. Such cars with very uneven power bands, can sometimes be made more controllable by applying non linear control.
However I would in general not like to drive with a power linear controller. That feels somewhat like a controller with the wrong "Ohm" or "Attach" setting.

We can make the curve both concave and convecs. The one you would use for a punch and brake situation where you almost just needs an ON/OFF switch. The other for a wildly overpowered car like a eurosport F1, where you need an excessive lot of slow speed control.

Best regards,
Carsten Groennemann.

05-09-2006, 10:24 AM
Just as soon as I win the lottery, I'm going to send a check to Carsten to buy one of his controlers. I've always been impressed by it's elegant and minimalistic approach. I'm not sure yet that it's concept for choke control on an open GP7 car is the best approach for managing motor life in an open race. But nobody so far has proven or shown results for the concept.

I do agree that people that think that applying a linear voltage range to a motor is going to provide a linear response ARE JUST PLAIN WRONG. The output speed is probably closer to a logarithmic curve. And under load that curve would probably exagerate. One hundred bands of linear bliss is just voodoo science.

Send me one of them there inside outside round curve controllers. My 55 year old brain has a hand to eye kink just waiting for a log controller.

05-09-2006, 09:20 PM
One of the interesting options on the difalco design is the "plug 'n play " feature that was the standard on the older difalco. They usually came with a slight logrithmic slope to the response, and are therefore VERY tunable, resistor by resisor.

05-10-2006, 03:53 PM
You need to consider both the theoretical view and subjective view.

Basic engineering theory dictates that motor's speed is directly proportional to the voltage and that the torque it develops is directly proportional to the current flowing through the arm. That much isn't voodoo science.

Load has an effect..but that's not the issue at hand. The real issue is controller "feel".

Controller "feel" is highly subjective. What may feel mushy to one can be just right for another...or waaaay tooo twitchy for an old guy like me with slow reaction times.

I owned an electronic controller who's voltage curve changed from linear to exponential as it's sensitivity was adjusted. I (and several of my racing buddies) noticed that the controller grew mushier on the bottom end and twitchier on the top end as the sensitivity was reduced.

I did some investigation to find out why and plotted out the voltage curves. There seemed to be a strong correlation between the controller "feel" and the shape of the curve. The controller felt mushy where the voltage didn't vary much from band to band, and was twitchy where the voltage varied a great deal.

I theorized that if I was able to develop a transistor drive circuit that only changed the initial starting voltage on the first band w/o changing the shape of the voltage curve, I'd be able to get the sensitivity I wanted on the first band without sacrificing responsiveness on the rest of the lower bands.

It worked...and again, that's not voodoo science.

Zamzickles...you may prefer a controller with an exponential voltage curve and for you, that may be the right choice to go fast. However, I've found a large number of customers that prefer the "feel" of a controller that has a more linear voltage curve. If they can feel the difference between my Linear 100 controller and the electronic controller they already own...enough to shell out $200 to $300 of their hard earned money...than I strongly doubt voodoo science is at play.


05-10-2006, 04:46 PM
Let me offer up this freebie. I'm a carpenter, and one of my greatest faults is overthinking solutions. I race with a talented guy that designs chassis, and has tried his hand in building controllers. One of my ideas to improve controller response ( from slower to faster and vice versa.), is to change the shape of the bands themself. I think that if you used the pivot point of the trigger as the hub point, and shaped the bands like pie slices, all you would have to do is the find a way of altering the wiper button position of the trigger arm to change response feel. The further out on the arm, the bigger the distance that the button has to travel and the bands get wider. If you look at just about every controller out there, you have plenty of room to make the board bigger, especially in relation to the trigger pivot area. I think it is at least worth making a prototype and testing it.