Cycle Analyst: increasing proportional assist

Get all your technical information about electric bikes here.
Post Reply
User avatar
judson   10 mW

10 mW
Posts: 34
Joined: Jun 29 2015 11:51am
Location: Seattle, WA

Cycle Analyst: increasing proportional assist

Post by judson » Mar 06 2016 1:28am

Hey All,

I hope this is the right place to post this. I bought an eZee rear hub motor from GRIN and Justin has been phenomenally patient in helping me work through some unique obstacles posed by my particular bike. In any case I was hoping to give him a break and appeal to the ES community for this likely straight forward question.

So I have a TDCM torque sensing bottom bracket running through a Cycle Analyst v3 on a cargo bike with an eZee controller and motor. I installed the bottom bracket and programmed the CA with the recommended settings form the user guide. When I hop on the bike it runs really well, spiking up into the 600 to 700W range when I am really grinding up a hill but usually hovering in the 200 to 400 range. When my wife tries it, she does not get the same umpf. She reports that even when climbing a hill she is only seeing 200 to 300W at most. I assume this is because she is not pushing as much torque through the pedals. This bike will mostly be ridden by her not me so I was trying to figure out the best way to ramp up the assist level with her in mind.

Reading the user manual my best guess would be to change the AsstFactr, but was hoping someone here could confirm this. Also, how much should I change it?

Also, tomorrow I intend to set the TrqScale for real. Up until now I have been using the suggested default of 50Nm. My other thought for uping the assist level was to tweak this number, but that seems a little hacky.

Thanks yall!!
Cargo Bike: eZee rear hub, 25A controller, 48V 11.6Ah LiMn battery, TDCM BB torque sensor atttached to Cycle Analyst v3

Daily Driver: BBS02 750W mid-drive, 48V 13.5Ah battery

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst: increasing proportional assist

Post by teklektik » Mar 06 2016 9:46am

judson wrote:When I hop on the bike it runs really well, spiking up into the 600 to 700W range when I am really grinding up a hill but usually hovering in the 200 to 400 range. When my wife tries it, she does not get the same umpf. She reports that even when climbing a hill she is only seeing 200 to 300W at most.

I assume this is because she is not pushing as much torque through the pedals. This bike will mostly be ridden by her not me so I was trying to figure out the best way to ramp up the assist level with her in mind.

Reading the user manual my best guess would be to change the AsstFactr, but was hoping someone here could confirm this. Also, how much should I change it?
The CA is designed to accurately display rider statistics for folks who are more of the 'cyclist' rather than 'ebiker' persuasion. These are displays on the 'Human Power' status screen. Of interest in this case is the 'human Watts' field which shows the power the rider is putting into the pedals. When pedaling faster than 55rpm, 'Human Watts' is multiplied by Trq->AsstFactr to develop the Assist power (a more detailed explanation of the calculation is presented in the Guide, but this is close enough for this discussion). This calculated value is what is shown in the 'Watts' field of the 'Electrical Only' screen.
  • CaV3_AssistPowerCalc.png
    CaV3_AssistPowerCalc.png (28.7 KiB) Viewed 1532 times
    Detail: Although 'human Watts' are always calculated and displayed accurately, Power is (rpm x torque) and the actual assist power is based on the maximum of (true rider cadence) or (55rpm) to give some extra assist on the getaway where cadence is zero or very low. 'Human Watts' is a separate calculation that does not involve this '55rpm' fudge and so is always accurate. Above 55rpm, the two calculations have the same result. Also, for simplicity, we are ignoring Trq->AsstStart in this discussion.
So - two things:
  1. the wife likely weighs less and so the bike needs less power to climb the hill at a given speed, and
  2. the wife is not generating the same human Watts as you - easily seen in the human Watts field.
To ballpark a 'fix' for her, we ignore #1 and just focus on the human Watts. Just increase Trq->AsstFactr by the ratio of your human Watts to hers (Set to that screen - watch the max value climbing the same hill):
  • (Wife Assist Factor) = (Hubby Human Watts) / (Wife Human Watts) x (Hubby Assist Factor)
In short, if she produces half the power, you will want to double the AssistFactor.

This is a hack calculation, but will get you in the ballpark.

Frankly, for PAS installations, I strongly recommend that you install a pot (CA3_AuxP or equivalent - see Guide) on the AuxPot input and set it for 'Limiting' and 'PAS Level'. This will let you configure Trq->AsstFactr larger than necessary and then dial it down by an arbitrary 0-100% 'AuxPot' factor on the fly (Assist % rolls on the AuxPot pop-up screen when you dial the knob so reproducing settings is very easy). This lets you tune the assist while riding for exercise, cruising, different terrain, or in this case - different riders. This feature is particularly nice to allow tuning effort/cadence when matching speeds with other riders. This is a simple and intuitive adjustment that even the most non-technical folks can easily master.

In your case, a second strategy is to set up two presets with different Assist Factors so you can quickly 'hot swap' presets depending on the rider. This would work with or without the pot feature.
judson wrote:I intend to set the TrqScale for real. Up until now I have been using the suggested default of 50Nm. My other thought for uping the assist level was to tweak this number, but that seems a little hacky.
Ya - it is, but -- setting this accurately is necessary only to calculate an accurate 'human Watts' value for display. If it is inaccurate, things generally work fine and exactly the same power assist and operation can be achieved -- just with slightly altered Setup values. As shown in the image posted above, offsetting the true torque by some factor can be compensated by a similar inverse adjustment in the 'assist factor'. So - getting the torque scaling accurate is good, simply to make the human power display more meaningful, but not strictly necessary if you don't care and just want the right assist. Again, the Guide details the assist power and human Watts calculations and you can see the affects on the calculations quite plainly.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
judson   10 mW

10 mW
Posts: 34
Joined: Jun 29 2015 11:51am
Location: Seattle, WA

Re: Cycle Analyst: increasing proportional assist

Post by judson » Mar 06 2016 9:52pm

Wow! Thanks for all the info.

So I calibrated the TrqScale by hanging a 10lb weight from the crank and doing the math suggested in the manual (ended up changing it to 69.7Nm/V for what it's worth). I also upped the Assist Factor and now the wife is zinging along without breaking a sweat.

Also thanks for the suggestion about adding an Aux Pot. I already have it and will be installing that soon. Before I add that layer of complication though I wanted to get all the kinks worked out of the system.

For instance the motor seems to be pulsing intead of providing a more constant push. Just as I'm typing this it occurred to me that I should check the electric only statistics to see if the pulsing is registering there. I read in the torque sensing tuning hints from the user guide that a possible fix to this is to increase the Trq->AsstAvg but to be totally honest I can't find that setting. I scrolled completely thought the SETUP TRQ SENSOR options and did not see it. I get Sensor Type, Trq Scale, Trq Offset, Asst Factr, and Asst Start but no Asst Avg.

Any thoughts on where to find that? Or even what else may be causing the pulsing? When I got home from the ride I noticed that my rear fender was rubbing a bit, but this pulsing has been around since before the this latest ride, and in fact since before I made the aforementioned tweaks (though it was a little less pronounced when the motor was not working as hard before upping the Asst Factor).

As always, thanks for any help you can provide!
Cargo Bike: eZee rear hub, 25A controller, 48V 11.6Ah LiMn battery, TDCM BB torque sensor atttached to Cycle Analyst v3

Daily Driver: BBS02 750W mid-drive, 48V 13.5Ah battery

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst: increasing proportional assist

Post by teklektik » Mar 07 2016 8:32am

judson wrote:Any thoughts on where to find that? Or even what else may be causing the pulsing?
Descriptions of the various parameters can be found on the Grin tech V3 page in the 'Setup Menu' section.
This parameter should be the last one in the TRQ setup menu:
Grin Tech V3 Page wrote: SETUP TRQ SENSOR
...
Asst Avg:
This value lets you control over how much of the pedal rotation to average the torque signal, as measured by the number of PAS pulses. Because the pedal torque pulsates heavily with each turn of the cranks, it is essential to average the signal to prevent corresponding pulses of motor torque too. For a sensor like the THUN which only measures the left side pedal torque, this should be a multiple full rotations, so either 8, 16, or 24. A higher value leads to smoother power output but also a slower response to changes in pedal effort.

For the 12-pole TDCM sensor which senses both left and right torque, it can be set in multiples of half pedal rotations, so 6, 12, 18 etc.
If the pulsing is in sync with cadence, then the averaging parameter should address the matter.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

Post Reply