• Howdy! we're looking for donations to finish custom knowledgebase software for this forum. Please see our Funding drive thread

Reduce long term power with CA and Kelly

EdTesla

1 mW
Joined
Jan 12, 2021
Messages
11
Hello again,
I am getting a Kelly Controller with Cycle Analyst the day after tomorrow. I would like to test something with the Cycle Analyst. It is about the e-bike legislation in Germany.

I would like to use the full power of the motor for 25 minutes. After that, the maximum power should be gradually reduced to 350W (electric power, approx. 250W mechanical power). After 30 minutes, 350W should come out constantly from the accumulator.

Can I change the profiles via the serial interface? I would build a small controller which sends a command after 25 minutes and sets the system to the reduced power profile.

Alternatively, it would be fine if I could vary the maximum power over speed and never get over 350W at the test point of about 25km/h.

Thanks a lot for your ideas.
 
EdTesla said:
I would like to use the full power of the motor for 25 minutes. After that, the maximum power should be gradually reduced to 350W (electric power, approx. 250W mechanical power). After 30 minutes, 350W should come out constantly from the accumulator.

Can I change the profiles via the serial interface? I would build a small controller which sends a command after 25 minutes and sets the system to the reduced power profile.

The serial port of the CA cant' be used to change settings except by the same method as used by the computer-based setup program, which reboots the CA when its' done. If it's ok for a reboot to happen at the behavior changeover points, then you can change any setting you like via this method.

But it is much easier if you simply use the Presets function of the CA, and either use the Aux Input to switch which preset is being used (you can see the CA setup info and manual on how to do this), or modify the CA to wire the front panel buttons out to your external control system.

Or just set a timer alarm and when it goes off, press the front panel button or Aux switch to change the preset with a finger, if that's allowable.

However, I don't know if you can use Presets to gradually change the power limit the way you want. When you switch from one preset to the other, the limit change is immediate.

You can, however, use the aux input as an analog control with a manually-controlled potentiometer (or a digipot controlled by your external "small controller") setup to control the power limit, in the CA's power limits menu.


Alternatively, it would be fine if I could vary the maximum power over speed and never get over 350W at the test point of about 25km/h.
If you mean, set a power limit to only 350w, regardless of speed, then you can do that in the CA settings in the Power Limits section. You can also set a speed limit in the speed limits section.

If you mean something else, you'll need to be precise and specific; there are a lot of things the CA can do--but not everything.
 
Many Thanks amberwolf,

I will use an additional controller and the analog aux.
 
Hello again,

amberwolf many thanks again. It works like a charm. But... :) if I plug in the cable for serial communication the display starts to flash a bit. I do have this effect sometimes when pulling the brake.

Should I be worried?
 
Do you mean the backlight, or the characters on the LCD?

The backlight is part of the series circuit that supplies power to the CA, so if power demands spike the brightness can increase. Depending on the actual connections you have, that may be normal. If it's excessive, it could damage the CA's power regulator.
 
It is the backlight. Well I have a 10V PAS and the throttle connected. Not more.
 
As long as the voltage into the CA to power it is low enough, you should be ok. The higher the voltage is, the less load you can place on the regulator, especially the 10v. There is a note in the manual and on the info page on Grin's site that discusses the limitations, but it is not exact.
 
I have now been able to isolate the problem. It only occurs when I connect the GND line of the audio cable to the serial interface.

It seems that I have a deviation between GND V_BAT and GND CA. Strange.
 
Normally that serial cable is not connected to something that is running from the same power source as the CA itself, so perhaps there is something in the internal CA wiring that can cause a ground lift or ground loop when it is connected that way.

If you use an isolated DC-DC to power whatever you are connecting to the serial port, it should fix that.
 
Thank you. It should buy a bag of food (that will last about a week and a half :lol: ) for my monsters (of which there are only three now, rather than four):
file.php
 
Nice Dogs. I have a Pyrenean Mountain Dog (70kg).

My configuration is now running legally according to German rules. The only thing that worries me now is the large currents during recuperation. We will use a Kelly with 90A continuous current and 270A peak. I hope that the controller can really handle these high currents in the long run.
 
EdTesla said:
Nice Dogs. I have a Pyrenean Mountain Dog (70kg).
Couple decades back I had a wolf-pyrenees-something mix, Lady; the least doglike of all the dogs I've had, and arguably the smartest. She wasn't very heavy (compared to say, Yogi, one of the St Bernards I have now) but she was very tall and lanky (though she didnt' look it in her winter coat, she did in summer).

My configuration is now running legally according to German rules. The only thing that worries me now is the large currents during recuperation. We will use a Kelly with 90A continuous current and 270A peak. I hope that the controller can really handle these high currents in the long run.
WIth the Kelly, you can change the settings in it directly for some things (which things and by how much depends on the specific controller version).

If I understand correctly, though, you won't be using those high currents much based on your CA limiting. Also, IIRC the Kellys use phase current rather than battery current for those numbers, and phase currents vary greatly depending on load. But even with battery currents, to get that kind of power usage, you'd either have to be going fairly fast, or you'd have to be going up hills or slopes taht are either steep or by pulling up a fair bit of weight.

The Grin Tech motor simulator at ebikes.ca can help you see how much power is needed for various circumstances, at least approximately, with differnet systems. While they may not have the specific motor or controller or battery that you are using, you can enter custom numbers to emulate what you have, or pick something similar to what you are using instead.



The one problem I've seen posted about with Kelly is that sometimes (rarely) they just stop working for unknown reasons, according to various posts around ES over the years. I have no direct usage experience with them yet, as I just got a used one recently.

I don't remember which specific models, except for the one I have here, a KLS7250D, that I haven't had time to do much with yet. The previous owner had one die not long after he got it, and it was replaced by Kelly under warranty then. Years later (couple months ago now) the replacement stopped working. He didn't troubleshoot it, just scrapped the Motorino it was on and sold all the bits cheap. All I can say about it so far is that it connects to the programming software fine, and I can read from it and write to it normally, and it does try to run the hall-angle-identify procedure, but something is wrong as it started to smoke the phase wires of the little 9C motor I used for the test. :shock: (the motor worked fine last time I used it, a few years ago). And measuring the resistance of the FETs gets normal readings, from each phase to B+ and B-. Opened it up and it all looks pristine inside, no blown anything, swollen caps, etc. So...dunno what's wrong yet. Gonna test with a much bigger motor as soon as I can.

Overall Kelly is probably a good enough controller; they're certainly better designed than the typical generic ebike controller, AFAICT (but I'm not an engineer).
 
Back
Top