I'm still wondering though - why is the LVC is set to 35V in a 36V system?...
The LVC is designed for lithium packs.
You said that one SLA had lost much capacity.
Ok
If you are using 3 SLA. 35v/3 = 11.6v. If one battery is not doing as good as another in series it could be down a volt. This pack discharged at 35v with a runt SLA would be very bad for EV applications.
Do a test. Hook up your multimeter to the runt SLA and fully discharge your pack on a ride to LVC. Now you will see that the failing SLA voltage isn't performing up to scratch, probably dropping lower than 8v and shunting everything from the other packs.
Your LVC is hitting in the blink of an eye.
When I charge a battery I can watch the volts go up slow till it is charged and almost see the highest volts on the meter up until float charge mode cuts the volts. When the battery is full, if I reinitialize bulk charge state by unplugging the charger and replugging the charger back in the volts will go up so fast I do not even see the highest volts on the meter. It only gets to 14.8v and stops instead of 15v. If I leave the charger on over night and do it again I am lucky to get enough time to see the volts go up to 14.5v This is really reaching 15v but my digital LCD display is not fast enough to catch it.
This could be happening to you when you twist the throttle and the volts go down so fast the LCD screen isnt able to show you you this.
To put a lot of energy and power into a hub motor with out overheating the controller or power cables the controller breaks up the power into little packets, When you throttle the controller expands and narrows the width of these packets to your throttle use, this is called pulse width modulation (PWM). PWM also makes speed control very efficient.
When the batteries obtain so much resistance, lack of current on a decent load delivered by PWM may start to pulse the the rail voltages down in the controller. So what you may see at the terminals maybe also different in the caps at the controller. You would need a good oscilloscope and probe in the controller to see this anomaly in action. The spikes and dips in voltages are very fast.
I also think there is an LCC and HCC in the controller too. Low current cut off and high current cut off. Not too much is documented with this but from memory, with the older controllers, there was a whisper of info about this some time ago.
Maybe try setting it at 24v to see if there is any difference but I feel you wont see much with the runt SLA's in the pack. Be carefull because 24v LVC will finish your SLA's and you could be heading towards controller failure if you push things too far..