CroBorg Super Commuter

Today was my first commute with the new configuration of the Magnetic eBrake Lever (changed it during my long weekend). The analog circuitry has not changed, it is still putting out about 0.8-2.3V into the Sabvoton with the eBrake gain set at 500. What has changed is the switch, last week I was using the front brake switch to enable the eBrake so I had to pull the front lever a little to allow the rear lever to work the eBrake. Pulling the rear brake lever alone did nothing, so coordinated lever operation was required, and this caused certain issues, such as pulling the rear lever first to no effect, then pulling the front lever a little and having the rear ebrake come on at a higher level than desired.

Now the setup is more conventional, both the switch and the analog voltage for the eBrake come from the same rear brake lever. The difference is subtle but important. The ergonomics is better and it feels great. Most of the stops and descents today didn't require that I touch the front brake lever at all. The eBrake, even though I don't have it set as high as I plan to still has enough braking to bring the bike to a very nearly stopped speed (perhaps 1 mph), and putting down a foot is easily enough to stop completely. I will raise the gain further, either directly or with the amplifier that I am building, but it is good enough now to control speed down a 15% grade and come to a stop at lights and signs.

So now it works and feels like a "power rear brake lever". There is just a light spring for resistance, so it is easy to pull this lever. One finger is more than enough. It pulls through until it hits the handgrip, so there is a lot of motion, but the physical pull resistance never gets very high, there is just the return spring on the lever to work against. So it feels much easier than a regular brake.

This is the way an ebrake should be.

Thanks again to zombiess for his customized Sabvoton controller that made this possible.
 
Most of the stops and descents today didn't require that I touch the front brake lever at all. The eBrake, even though I don't have it set as high as I plan to still has enough braking to bring the bike to a very nearly stopped speed (perhaps 1 mph), and putting down a foot is easily enough to stop completely.
Are you saying that virtually all the braking down Wildcat is done via eb rake only? If that is the case, this is a major factor in determining how much braking needs to be up front. Of course, speed, distance, and the road surface will influence how much braking is needed on both tires. But your ride today demonstrates how strong and smooth the ebrake is.
 
The eBrake of the rear hubmotor is adequate for all normal riding, descending and stopping.

The dual front discs are needed for emergency braking, or really aggressive late braking, holding stopped on a hill and parking. I don't use them much normally.

The eBrake is as effective as it is due to three factors:

1) Cromotor - motor that has plenty of torque capacity (both as a motor and as a generator)

2) Sabvoton controller - that has plenty of variable regen and eBraking power

3) Battery that can take the regen current
 
For comparison, what values do these components have to provide the level of ebraking? As i finalize sizing the specs on my build, it is very important to have similar level of ebraking due to nominal braking coming from front drum.
 
20150406_185028.jpg


Look what arrived in the mail today!

A very nice cable operated variable resistor. It only has a 2 wire output, but all three leads are on the pot, so the cable can be replaced if we want to use it that way.

Thanks Stevil!
 
windtrader said:
For comparison, what values do these components have to provide the level of ebraking? As i finalize sizing the specs on my build, it is very important to have similar level of ebraking due to nominal braking coming from front drum.

The two controllers I'm aware of that provide variable regen are the Adaptto and the Sabvoton. I don't know much about the Adaptto's variable regen. From my testing I know the Sabvoton has powerful variable eBraking. I have also used regen on the Infineon controllers, but it was not variable, and it was still quite useful, if a bit jerky. There were some settings in the programming for different levels like high/low, but once set it was just a switch closure to turn it on. On high it was enough to decelerate down the 15% gradient, but it was harder to use since you could not modulate it.

The Cromotor is the only one I've used regen with. It is a very capable motor so it is also a very capable brake. A weaker hubmotor will also be a weaker brake. My 9C is not as strong, it has some trouble climbing the gradient, so it might have some trouble ebraking on it as well. The Croborg easily climbs the 15% gradient, accelerating as you go. It may be double the torque of the 9C, plus the 10% smaller tire diameter advantage. The 9C is roughly at stall torque on this grade, so pedaling is required to accelerate beyond a low speed.

If you are going to try a Sabvoton make sure it has the variable eBrake feature, they do not all have it.

If you want to know about the Adaptto's regen/eBrake feature check with some owners, I don't have one.
 
Is there some way to estimate the amount of ebrake force that can be generated by the system? A formula that includes variables such as motor torque, motor output, max battery charging amps, controller regen rates, etc.
 
Alan - Do you have any numbers on regen amps you're seeing?

For public reference, the Adaptto is configurable from the display for phase and battery amps on regen, it can take in any variable input and scale appropriately - so for instance a 1.5-3v swing would be more than enough to accurately apply regen up to the defined limits. My mini-e will do 25 battery amps or ~2kw and will brake more effectively than my rear disc/caliper as it functions as a crude form of ABS rather than just locking. This is vastly superior to a conventional rear brake. The Max-e will brake even more effectively down to lower speeds due to high phase amps possible. I use a thumb throttle mounted on the left grip for braking and now rarely use physical brakes of any sort.

You could think of re-gen as a negative acceleration. If you're extracting 2kw from motion and putting into the battery, this acceleration is going to be directly comparable to extracting 2kw from the battery and putting it into motion. Generally due to weight shift under deceleration you're not going to be able to apply quite the same forces if you had no other limits (e.g battery max charge rates)

Note that while this does add some range (in my case long term average is ~5%) the primary use is as a maintenance free brake.
 
Thanks for the Adaptto data, that's a very useful comparison. It sounds fairly comparable.

The Sabvoton has a settable ebrake coefficient. The units show as Amps, but I've been told that it is off by a factor of 10. So I have it set to 500 which would be about 50 amps. That is probably in phase amps, because I see up to about 2-3kW of regen according to the CAV3 (which might be low as it reads a bit slowly and high regen drops off quickly as the speed comes off). I've had it even higher when I used a thumb throttle which has a higher voltage output. In that case, as the speed drops, the regen power turns from charging the battery to draining the battery and the strong eBraking continues on to a stopped condition. When I had the eBrake value set too high it actually turned the rear tire backwards in a hard braking situation as the tire slid on dry pavement, pushing the pedals into my feet. It was no longer regen, but reverse power into the motor, capable of breaking traction on dry pavement. A real eBrake, not just a regen.

I could raise the setting further, and I will experiment with that when the weather improves, we have a few days of storms to weather. By setting it higher I could get by without an amplifier. I would like to see if there is a difference in response between an amplifier and a higher gain setting with no amplifier.

When the settings are adjusted so that the rear wheel does not break traction, it is somewhat like ABS. Regen allows the wheel to turn, as ABS does.

I do notice a very slight delay between changing the ebrake lever position and sensing a change in the deceleration rate. The controller takes a moment to adjust to the new value. It is a small delay, but noticeable. At least that's what I think I'm feeling. It is not bad, it makes the braking smoother when you apply it suddenly.

I suspect any controller is a bit less efficient and less powerful in regen than it is during powering the motor, I think the max regen is at about 50% PWM while the max power is at 100% PWM.

But variable regen/eBrake makes a great rear brake. I want a real front brake, but this eBrake is good enough for the rear. I've run that way for approaching 7 or 8K miles.
 
Today we are having some badly needed rain, so no eBike commuting for a few days. The drought here in California is getting serious, they are talking about big fines for taking long showers. I gave up watering the yard and washing cars some time ago, it is getting hard to find ways to reduce water usage further.

Here is the main loop code for the eBrake Amplifier as it currently stands:

Code:
void loop(void) 
{
  int16_t adc0;
  int8_t i;
                     
  for (adc0 = i = 0; i < 8; ++i)         // average 8 readings
  {
    adc0 += ads.readADC_SingleEnded(0);  // read ADC 11 bits plus sign 2mV/bit
  }
  adc0 >>= 2;                            // scale to 12 bits, 1mv/bit

  adc0 -= INPUTOFFSET;                   // remove input offset
  if (adc0 < 0) adc0 = 0;                // wrap clamp
  
  adc0 *= 3;                             // gain
  if (adc0 > 4095) adc0 = 4095;          // clamp at max

  dac.setVoltage(adc0, false);           // output DAC, don't store
  
  digitalWrite(1,HIGH);                  // make the Trinket red LED just barely glow
  digitalWrite(1,LOW);                 
}

It doesn't take a lot of code to read the ADC 8 times, take the average, do the offset, scaling and clamping then send it to the DAC. An op amp or two could easily do this but we can do things here that cannot be done readily with op amps, and making changes is much easier with code. If we made a dedicated circuit board this would be a very small circuit.

I stripped out the library and initialization code for the example above. The main loop runs all the time. This loop executes about 60 times per second. It could be sped up as the I2C communications to the ADC and DAC is running at a low speed, but 60 times per second should be an adequate rate. There is already some delay in the eBrake processing in the controller, this will not add enough to be noticed.

If we want to change the rate the braking comes on with respect to the lever we can introduce different equations into the calculation, or make a map. This same hardware could make a torque throttle for a PWM controller by adding some motor current sensors. The ADC has four channels so it can read an input throttle and three other signals, or two other differential signals, such as two phases of the motor current.
 
After work this evening I hooked up the programming cable to the Sabvoton and cranked up the eBrake gain from 500 to 600. The subsequent testing around the neighborhood showed a nearly ideal result with skidding barely occurring on pavement undulations at maximum ebrake application at near full speed. On flat dry pavement it stops HARD, but doesn't skid until it hits a ripple in the pavement, then it skids just a tiny bit. It feels like ABS but smoother.

This is with no amplifier, just the magnetic sensor straight into the Sabvoton input. I'll test it during the next commutes, but this feels good, there may not be much need for an amplifier with this setup.
 
Alan B said:
20150406_185028.jpg


Look what arrived in the mail today!

A very nice cable operated variable resistor. It only has a 2 wire output, but all three leads are on the pot, so the cable can be replaced if we want to use it that way.

Thanks Stevil!

I made some measurements of the cable operated variable resistor. The pot looks like a slider for a sound mixing board. The full travel is about 29mm. I tested a Tektro cable operating lever with magnetic switch built in and there is about 17mm of cable pulling range (guessing where the lever might contact a grip). This is about 60% of the full range of the pot. The resistance ranges from about 2.3 ohms to 9.78k, depending on the position of the slider.
 
It was a cold commute this morning, but the surface dampness we had yesterday was not to be seen. It would have been a wonderful commute at 10 AM, however at 7 AM it was a bit on the cool side.

Now running with the simple magnetic eBrake lever (magnet and linear sensor, no amplifier) and the Sabvoton eBrake gain set to 600. It really works! Never used the dual front disc brakes except to hold at a stop. The eBrake releases below about 1 kmh, so to hold position the disc is superior. On a steep downhill the ebrake comes to a stop, releases, starts to roll, comes back on, etc. This is a straightforward and adequate solution.

I still want to try the cable operated resistor. I need to make up a cable for that.
 
That's a wonderful report. It is the most detailed experiment in maximizing and demonstrating the effectiveness of regen braking. It gives me greater confidence in initially going with the drum in the front on my pending build. Only those who know that section of road have an appreciation of how much braking is needed to keep from flying off the road.
 
Just to give folks some flavor, this is a 12-15% downslope section (part of the road from the Hall of Science to Berkeley Campus) that has a sharp turn, and two driveways that leave it and go into the Lab campus. The first driveway is also nearly as steep and departs to the right at 90 degrees, about 1/4 of the way down. The surface is reverse banked slightly at this turn, so the ABS on my 4Runner fires even at fairly low speed while descending and turning off. When I was using the dual 160mm front disc brakes and the top gate was closed, I had to navigate the whole descent including the sharp turn, and then stop to make the left hand turn into the lower gate driveway, the discs were so hot a wetted finger test sizzled, and they turned black from the hot stainless steel interacting with the moist cool morning air. They felt on the verge of fading at the bottom when I had no regen. With regen/eBraking the front brakes are not used, they are cold and ready for the emergency stopping situation that we hope we never need.

On this descent most bicycle brakes are generally marginal, but the Sabvoton/Cromotor regen alone is easily adequate. Afterward the motor is slightly warm, and the controller is cold, even with minimal airflow inside the covers, and a few tenths of an amp hour is back in the Lipos. There's no feeling of overheated brakes, or almost fading, just a firm controlled deceleration like a parachute, or a low gear on a manual transmission.
 
Made the first return commute run today with the higher Sabvoton eBrake setting of 600 and the simple magnetic eBrake lever described earlier (and no amplifier). The weather was beautiful with temperatures about 65-70 in the afternoon, clear skies and dry pavement. It was nice to not have it get so warm as wearing the protective gear becomes more difficult as the temperature rises.

I rode aggressively down the twisty road on Wildcat from Inspiration Point to the Dam road, in fact I had a car pull over to let me go by. I wasn't pushing them, but I came up on them fast and then slowed and backed off, so I was a bit surprised when they pulled over, but there are quite a few folks on this road who are very considerate and not in a hurry. Since it is downhill the speed is good, and some of the turns are tight, so there is a good test opportunity for brake modulation and control. The regen/eBrake system of the Sabvoton did well on this route, and there was no need for the front dual disc braking at all.

I would not ride without the front discs, but I like not having to use them. They won't wear out or get hot and fade if I don't use them often.

Excellent!
 
Blue sky and warming weather made the morning commute more comfortable (and not as cold) today. The traffic was Friday light, and one traffic signal I usually have to wait for was green when I arrived, so today's commute run established a new record for minimum time, just over 28 minutes. Normal is 29-38 minutes, but I don't recall ever coming in under 29 minutes before. The time comes from the GoPro recording time, so it should be quite accurate. Usually I have to wait at that light a minute or so, that alone can make the difference. I have four lights on my commute, and today they were all green. I pushed fairly hard the whole ride, and it shows in the energy used at just over 13 amp hours (at 66V nominal).

I'm still calibrating myself on the new eBrake setup, getting used to the feel of it, what it can do, and how far to pull it for the desired braking effect. It is working wonderfully. I should re-torque the wheel nuts to see if the NordLock washers are holding (I carry the wrench and check it periodically). Strong regen tends to loosen one of the nuts, but the NordLock lockwashers limit that to a small fraction of a turn, usually 1 or possibly 2 clicks of the ramped detents. After a wheel change they need to be checked daily for a week or so. There are no torque arms on this setup, so it is imperative that the axle stay fully forward in the Chrome Moly dropouts, and that the nuts stay fully torqued.
 
Another excellent commute, no events, no excitement. Just the way we like it.

I checked the hubmotor nuts and no loosening, even with all the hard regen/eBraking and acceleration. Perfect! Those NordLock washers are sure doing a good job. When you tighten them you can really feel the detents, and when you loosen them you can feel the tension increase before they release, which is how they avoid loosening under vibration. They create a local minimum in axle tension where turning the nut either way increases the tension, so vibration doesn't tend to move them, once they are properly tensioned and settled in. They are a good choice for this application - easy to tension, easy to remove and hold well under vibration. Nothing to replace like keys or wires, nothing to get stuck or fail like locktite. If you do decide to use NordLock washers, take care with how they are applied. They come in pairs, and must be used that way, and no other (free to rotate) washers can be used with them. One side bites into the frame, and the other side grabs the nut. In between the two washers are a series of wedge steps that create the increase in tension when loosening the nut. The nut and frame must be softer steel than the hardened washers for the proper bite. They seem to work even with the fairly hard nuts on my CroMotor, which has its own teeth that may help the stack find a good equilibrium in tension. I've used them on every one of my builds and have never had a loose hubmotor nut problem with them.
 
Some of you may recall that I installed a solar circuit breaker rated at 125VDC 63A. My original plan was to install two in parallel, as I have the current limit set for 80A, but I was assured that it would hold 80A for quite awhile.

Today I decided to take the whole Centennial gradient from the lower gate on up, usually I climb in the campus to the top gate and climb from there. This does add a little to the climb, but not much, however it does compress the climb into one long continuous 15% or so gradient. The CroMotor took it in stride, reaching a good speed at about 5kw. Near the top at Space Sciences it cut out and I momentarily wondered if I'd fried another controller. I quickly decided the circuit breaker probably tripped since the Cycle Analyst went out.

Sure enough the breaker was in the tripped position. A quick reset and I was back on the road.

So if I run over about 4kW for very long I run this risk. I'm not sure I like that, however it is quite rare to do that. Slowing a little would solve the problem as well.

At least things don't fail when the breaker trips. That is good to know.
 
striker54 said:

That would be nice. When I ordered the 63A breakers that was the largest Amazon had. The picture there on aliexpress shows 125A but the text says 100A. The price is low and shipping is free, so probably worth a try either way.

I could just lower the battery current limit to 63A, instead.

I also noted 30A of regen on hard eBraking.
 
When the power in these systems rise to this level, the chance of killing yourself becomes a very real possibility. It seems safety systems are woefully lacking in the majority of these systems. The fact experimentation on power breakers is a current topic implies this has not been fully sorted out by now.

When one reads that 100amp breakers are blowing, there must be enough power coursing through to kill one dead in a second.

Are there any threads addressing the safety features all high power systems should have in place?
 
windtrader said:
When the power in these systems rise to this level, the chance of killing yourself becomes a very real possibility. It seems safety systems are woefully lacking in the majority of these systems. The fact experimentation on power breakers is a current topic implies this has not been fully sorted out by now.

When one reads that 100amp breakers are blowing, there must be enough power coursing through to kill one dead in a second.

Are there any threads addressing the safety features all high power systems should have in place?

The Electrical Hazards of an ebike are primarily voltage (not the current), and under 100 volts DC is not considered a hazardous voltage. It can give you a nasty shock, but should not, under normal circumstances, stop your heart. It should still be treated with appropriate respect.

The high available fault current does create thermal risks - such as melting your wristwatch or ring, so conductive jewelry and tools must not be used in the vicinity of such exposed energized conductors. Wearing appropriate personal protective equipment such as safety glasses and gloves is a significant protection, leather gloves provide protection against arc debris and significant protection against shock current from voltage. I measured the pair I use at several million ohms.

The essence of electrical safety is to avoid contact with hazardous voltages by a combination of engineering and procedural techniques. This is generally implemented by insulation, barriers and isolating the voltages near the source in a dependable fashion, and testing before touching any of the circuits to be worked on. The testing must be done with a meter that is rated for the voltages that could be present, and the meter must be verified to be functioning both before and after the de-energization testing. During re-energization extra precautions such as safety glasses and leather gloves are prudent as faults often occur at that event.

In the CroBorg's case the cable to the battery can be unplugged, and fingers are protected from the high voltages by cable insulation, connectors, etc. The controller on this eBike has screw connectors that are not inherently finger-safe, and the lower Borg covers in that area are screwed on, so it takes a tool to come into proximity with the voltages there.
 
As I was gliding along in the bike with 50 mph traffic sliding by, I noticed a police car in the bunch. I didn't see it early enough to get a good look, but they didn't seem to be interested in my eBike. Later when traffic slowed I passed them, and a lot of other low speed and stopped cars.

It is great to not have to pedal, except when you want to.

My morning commute would be slower in a car when the traffic is thick, as it was today.
 
Back
Top