# Better support for Powercal like power meters

Following this discussion I’ve found a cool feature that could be implemented:

https://support.zwift.com/hc/en-us/community/posts/115018973403-Riding-with-Powertap-Powercal

Currently Powercal it’s an affordable Powermeter, with some caveats, but it’s commonly referred as one of the best entry-level power meters for people that it’s not sure if power training it’s going to be profitable long-term for them (therefore, impossible to access to this software)

Since users have to spend a lot of money to join the ride, Powercal permit users get the first taste of the product, and I’m sure that with a good experience in 3-4 months people will think on investing more money in their equipment, like direct power meters or trainers.

But I’ve found two main issues to use Powercal more appropriately:

1. Cadence sensors should be taken into consideration in order to not move the bike if the user it’s not running his pedals.

if cadence = 0 , then speed = 0.

If cadence = N/A then, not consider cadence for speed calculation

That simple. This is an easy fix that automatically gives Powercal a 100% more usability.

1. Also, another problem is the spikes in training. With this type of power meters, it’s impossible to grow more than 100 W in less than 5 seconds. So obviously it’s impossible to go from 200W to 600W for 10 seconds.

Solution? This is a little bit more difficult because it’s something specific of Powercal type power meters. In case it could be possible to identify the powercal as the

The only solution In case it could be possible to identify the powercal as the power meter, and in case the user has a cadence sensor, then, stablishing a multiplier between cadence variation and power output of the power meter, considering also the slope of the in-game session

Let’s see an example:

If the user it’s running with an average of 80RPM, and 250W and with the same slope % it increases from 80RPM to 120RPM (50% increment), then we could apply a multiplier of  1+(0,02 per % increment/decrement) to the Power Output (in this case 1+ (0,02*50) X = 2X multiplier, therefore 250W x 2X = 500W output. So in this case is the user wants to achieve that 600W target for 10 seconds as the workout may suggest, then the user should increase his cadence to 130 RPM (62,5% increments) to get a multiplier of 2,25X so 250W x 2,25 = 562,5 W (plus the increase of W per effort that may put up the power meter output to around 280W, therefore, 630W total output, accomplishing the target, with a more realistic estimation).

To adjust a little bit further the multiplier, we could take in consideration the in-game slope variations. For example by adjusting the slope % to the RPM variations. If the slope increases 5% and the RPM remains the same, then the software must understand that the RPM has increased a 5% variation.

And obviously, the variations should have a decay over the time with an inverse logarithmic function taking time as the input of the function

But to be sincere, I have oversimplified the equation, so in this case, the only feature that it’s short-time feasible it’s the first one I stated.

The one potential issue I see with the RPM power equation is, what if the user on a spinning bike has the resistance knob set on very hard and he produce 300W @ 40RPM, then turn the knob down to zero resistance but spin like craze @ 120 RPM. That does not mean he produce 600W he produce even less that 300W.

Yes Gerrie, but when you do a training session, that situation it’s impossible. After a high resistance workout at low RPM, you never switch inmediately to a fast cadence workout. That’s why I was commenting the logaritmic time decay (in order to eliminate the multiplier in a 10-20 second frame for example).

For example, let’s say you are simulating going up a slope at 40RPM, with a high resistance, then you get let’s say a 300W average with the power meter

In the top, you reduce the resistance, and you go for a couple of seconds at 60RPM with lower resistance. Your HR will decrease the power to 200W but the multipler will provoke an incidental 400 W (due to the 50% increase factor), but that 400W will decay in 10 seconds back to 200W. Then after 10 seconds, you start a really high cadence going down the slope. Your power will increase to 250W and from 60 RPM to 120RPM increasingly. With the logarithmic decay you will probably go up to 500W for 20 seconds or so, but afterward, you will go back to 250W

So essentially you will have clockwise power peaks: This does not reduce the error but permits power training with peaks of power when needed by playing with the cadence. Obviously, this scenario is only applicable to this type of power meters.

But obviously, you can always do that, and in that case, you will be cheating yourself (like any other cheating mechanism that could be designed). But I understand that the logic here it’s to introduce mechanism for improvement, not for cheating

I know this is pretty complex, and as a feature should be tested a lot further to fine tune the estimation. But the overall experience could be really great for this type of power-meter users.

I am using a PowerCal with a SuperMagneto Pro… and have to agree that cadience being brought in to the equation would help.  If I spin real fast with little/big ring combo then the power cal does not really jump much if I spin faster with a big/big  it knows I am putting out more power.  Not great but it works pretty well… Except if I freewheel it will drop to a moderate power and speed but I keep moving so I can cheat the game and myself.  If the cadience was part of the equation this wouldnt happen and I should be able to reach high watts.