Lishui Controller Modification - Firmware Flash Project

[saneagle]

I think in the spirit of fairness I should reinstall a similar sized KT controller and get a direct comparison.

That's the point. There are a whole load of people riding around on their Woosh bikes and other bikes with speed control controllers saying how happy they are with them, likewise all those guys with crank-motors, but they never tried anything better. It's only when you try everything that you know what's good and bad.

 

[Koner]

that was great fun read, I'm planning to buy lishui controller, its very compact compared to KT, 30a ls controller is smaller than 25a KT

 

[Woosh]

That's the point. There are a whole load of people riding around on their Woosh bikes and other bikes with speed control controllers saying how happy they are with them, likewise all those guys with crank-motors, but they never tried anything better. It's only when you try everything that you know what's good and bad.

For crying out loud, the constrast is very clear: you don't have to adjust anything with a Lishui controller, it just works and does that reliably. No post about it.
To contrast against it, the number of posts to help out someone who has just bought a KT on Aliexpress and run into settings problem. It's like 50-100 to zero. Does the KT go faster? No. Does it accelerate faster? No. Does it go farther? No. So what does it actually and measurably do better? Nothing.
It's just like asking someone to trust your liking. Not very scientific.

 

[Koner]

For crying out loud, the constrast is very clear: you don't have to adjust anything with a Lishui controller, it just works and does that reliably. No post about it.
To contrast against it, the number of posts to help out someone who has just bought a KT on Aliexpress and run into settings problem. It's like 50-100 to zero. Does the KT go faster? No. Does it accelerate faster? No. Does it go farther? No. So what does it actually and measurably do better? Nothing.
It's just like asking someone to trust your liking. Not very scientific.

I think KT controllers are cheaper and more accessible than LS(telling from experience of trying to find one)

 

[Nealh]

KT vs Lishui is simply availability, one is third party and one is 99% OEM.
It is no reason that D8veh and I recommend KT as are a couple other members now they have seen the light.
The KT current control imv is a nice ride then the Lishui I have used and I know if a scumbag tries to nick the lcd or damage it I can get a compatible replacement very easily.

For Woosh it is business and no customer being able to change any settings makes sense and life easier for the business, spares form Woosh are available. However the user can't simply buy a third party Lishui component and hopes it will work as soft /programming may still differ.

 

[Nealh]

If Lishui were that good then there would be no need for re-writing the the programmes for them via OSF, obviously there are alot of users not happy with the speed control element.

 

[Woosh]

If Lishui were that good then there would be no need for re-writing the the programmes for them via OSF, obviously there are alot of users not happy with the speed control element.

It's just an experiment. The OP didn't even try to compare the two firmware. He's happy with the results and generously explained how it was done. If I ever fall out with lishui, I will know how to help my customers!
Lishui supply also current control firmware. The comparison is like which version of Linux is better.

 

[saneagle]

For crying out loud, the constrast is very clear: you don't have to adjust anything with a Lishui controller, it just works and does that reliably. No post about it.
To contrast against it, the number of posts to help out someone who has just bought a KT on Aliexpress and run into settings problem. It's like 50-100 to zero. Does the KT go faster? No. Does it accelerate faster? No. Does it go farther? No. So what does it actually and measurably do better? Nothing.
It's just like asking someone to trust your liking. Not very scientific.

the one in Mikel's bike didn't just work and do it reliably. The MOSFETs blew. Now I have to replace it.

The KT controllers are measurably better. It's extremely rare, if it ever happened, that a KT controller packed up. Most of the queries are about setting up because people want to optimise the settings. The rest of the queries are about throttles and pedal sensors that don't work because these guys don't understand the importance of the correct wiring sequence. It would be the same if somebody with a MiRider bike wanted to swap his controller for a Lishui one based on your recommendation. In that case the Lishui wouldn't work and he'd be on here trying to figure out why.

The sinewave KT controllers are measurably better than the average Lishui in average bikes because the average Lishui isn't sinewave. Sinewave makes the motor nice ans smoothe and quiet. That's the first reason to change. The second reason is that KT use power control, while as the average Lishui uses speed control. We've been through that many times before. I don't understand why you can't see the advantage of current control. The third point is the way the throttle can be set up for more convenient rider control. I would swap for that one reason alone. Fourth reason is that it's virtually impossible to get Lishui controllers except from expensive OEM spares. Why would anybody want to spend a lot of money when they can get a better KT one cheaper. Fifth reason is that most KT controllers are dual voltage, which is very convenient when people want to replace their underpowered 36v battery for a 48v one when they're ready for it.

As you can see, I swapped out all the Lishui controllers for KT ones in my bikes. sometimes, it's quite a lot of work. Why would I do that? You've known me personally for about 12 years. Do you think I'm nuts or something? Remember, I was the guy that told you to chuck your 14A controllers and get 17A ones. I was the guy that told you to get 48v systems. I was the guy that told you that you can get motors rated at 250w, if you know what I mean. Were any of those decisions bad?

 

[Sparksandbangs]

When I started I don't think I realised what a potentially contentious issue this could be.

I'll run through my thinking. I've always been interested in flashing(sorry couldn't resist). In the past I've added increased functionality to routers and other gadgets. The thought of doing this to e-bike controllers has been floating around for quite a few months. What pushed it to the forefront of things to do was a couple bugs that I found in the KT controller I was using.

The first was the overspeed issue where after travelling down a hill at above the speed limit when going up the other side there would be a significant delay before the power came back on. It was definitely a bug (or maybe a feature that I can't fathom) because if the speed limit was removed it didn't happen. If the speed limit it was lowered then it happened earlier. Also another issue was if parameter C14 was set to 1 the power stayed the same in levels 1 to 4.

One of the things that I wanted to be able to change was the power delivered by the assist levels. Another was being able to set the wheel size exactly.

Let's be honest whatever controller you select is in someway a compromise. The only way you can get exactly what you want is to do it yourself.

My first thought was to buy another KT and flash that but the FOC Lishui controllers have better hardware and processing power. If you are going to wipe something and start again then might as well be the better option.

I have no experience with Lishui current control and couldn't comment on how good or bad it is. I don't know what the original software in the controller that I adapted was like because it was for use with a torque sensor.

As far as I am aware there are several types of controller supplied by Lishui. The one supplied by Yose Power uses square wave and won't run EBiCS although there is an alternative firmware. Hochsitzcola the guy that wrote both of these firmwares isn't interested in displays and concentrated on torque sensors. The torque simulation side worked but it wasn't perfected because it wasn't his thing. I think I've massively improved it in EBiCS, I would guess that splicing some of my code in to the block commutation version would make that better as well. It would be interesting to see how it works.

Anyway I was riding today and thinking how good it was. Whisper it quietly, it's better than the KT. Although I will admit that there might be three reasons why I think that. The first is confirmation bias. I've spent a bit of time on this and maybe I want it to be better, so I think it is. The second is that when you set something up for how you ride then it is probably going to feel better. Maybe someone else with different riding conditions would feel differently. Third one is maybe it is just better.

I will put the KT controller back on to do a comparison when I get time.

 

[Jodel]

I have no wish / intention to offend or irritate any forum member, but will add my own (limited) observations to the Lishui / KT debate!

Can I preface my comments by saying that I am rider a very long way past my ‘best before’ date, with a lot less pedalling power than a younger person. I have arthritic digits and a distinctly dodgy right ankle which can make starting from rest a bit challenging at times. A throttle to help starting off and a simple cadence sensing PAS suits me too.

I have two Woosh DWG22C hub kits, one on a tandem, one on a solo. The kits came with Lishui controllers and have worked faultlessly since they were installed several years ago. More recently, I bought KT controllers to try out current control rather than speed control. I know many users really like these controllers so I thought it worth giving them a try.

To cut to the chase, I can definitely see the attraction of both systems but if I had to pick only one, it would be the Woosh speed control implementation of the Lishui.

On my solo, I tried a KT controller (with the LCD3 display) which fitted to the Hailong HL1 battery case, just like the Lishui. I started off trying out current control, but I found I had to make a lot more gear changes that I did with the Lishui ‘set and forget’ speed control. I then set up the KT controller for speed control and much preferred that configuration. Although the KT controller can be modified extensively, I found that it wasn’t possible to have a linear step between power settings 1-5. 1-4 was OK, but then there is a big jump from 50% to 100% power when going up to level 5. What I did like about the KT speed control was the sensitivity of the system to minor variations in speed. The KT seems to react to +/- 0.5 mph changes, where the Lishui seems to need more like a +/- 1 mph change to respond.

I have a HL1 type 6 MOSFET KT on my solo (with the DWG22C motor) which has been reliable. On the same bike I tried a separate 6 MOSFET KT and also a 9 MOSFET KT. I intended to try those KT’s on my tandem, but wanted to test them out on the solo first. The HL1 type KT controller has been fine. Both of the separate KT controllers were less satisfactory. They both overheated and gave up after a few miles. All three were all configured exactly the same way so I don’t know why the separate KT controllers gave up the ghost.

Based on my experience, the reliability scores clearly favour Lishui, with 2 out of 2 working perfectly and only 1 out of 3 KT’s working properly. The Lishui / DWG22C system on my tandem has now done about 6K miles and has never faltered.

I have no experience of riding with Lishui on other bikes, but whatever Woosh do to ‘tune’ their controllers suits my preferences more than KT.

 

[Nealh]

For the end user (self build) the KT fits the bill, easily accessible to buy and reasonable price to buy.
Features such as current variability, dual voltage and the mention of a legal speed control device makes it a firm favourite . Lishui's are not available for the self builder.

 

[saneagle]

When I started I don't think I realised what a potentially contentious issue this could be.

I'll run through my thinking. I've always been interested in flashing(sorry couldn't resist). In the past I've added increased functionality to routers and other gadgets. The thought of doing this to e-bike controllers has been floating around for quite a few months. What pushed it to the forefront of things to do was a couple bugs that I found in the KT controller I was using.

The first was the overspeed issue where after travelling down a hill at above the speed limit when going up the other side there would be a significant delay before the power came back on. It was definitely a bug (or maybe a feature that I can't fathom) because if the speed limit was removed it didn't happen. If the speed limit it was lowered then it happened earlier. Also another issue was if parameter C14 was set to 1 the power stayed the same in levels 1 to 4.

One of the things that I wanted to be able to change was the power delivered by the assist levels. Another was being able to set the wheel size exactly.

Let's be honest whatever controller you select is in someway a compromise. The only way you can get exactly what you want is to do it yourself.

My first thought was to buy another KT and flash that but the FOC Lishui controllers have better hardware and processing power. If you are going to wipe something and start again then might as well be the better option.

I have no experience with Lishui current control and couldn't comment on how good or bad it is. I don't know what the original software in the controller that I adapted was like because it was for use with a torque sensor.

As far as I am aware there are several types of controller supplied by Lishui. The one supplied by Yose Power uses square wave and won't run EBiCS although there is an alternative firmware. Hochsitzcola the guy that wrote both of these firmwares isn't interested in displays and concentrated on torque sensors. The torque simulation side worked but it wasn't perfected because it wasn't his thing. I think I've massively improved it in EBiCS, I would guess that splicing some of my code in to the block commutation version would make that better as well. It would be interesting to see how it works.

Anyway I was riding today and thinking how good it was. Whisper it quietly, it's better than the KT. Although I will admit that there might be three reasons why I think that. The first is confirmation bias. I've spent a bit of time on this and maybe I want it to be better, so I think it is. The second is that when you set something up for how you ride then it is probably going to feel better. Maybe someone else with different riding conditions would feel differently. Third one is maybe it is just better.

I will put the KT controller back on to do a comparison when I get time.

The late kicking in is not a bug in the controller. It's caused by the motor freewheeling when you're pedalling faster that what it's spinning. It will be spinning at 15 mph, when you're travelling anything faster than that. As you slow down to below 15 mph, the freewheel cannot re-engage until the motor has slowed down a lot more. It's nothing to do with the controller. It's a mechanical issue.

 

[Sparksandbangs]

The late kicking in is not a bug in the controller. It's caused by the motor freewheeling when you're pedalling faster that what it's spinning. It will be spinning at 15 mph, when you're travelling anything faster than that. As you slow down to below 15 mph, the freewheel cannot re-engage until the motor has slowed down a lot more. It's nothing to do with the controller. It's a mechanical issue.

Except it isn't, or it would do it with every controller. Changing the controller removes the problem. It's not mechanical.
Pound to a pinch of it will start again when I reinstall the KT.

 

[saneagle]

Except it isn't, or it would do it with every controller. Changing the controller removes the problem. It's not mechanical.
Pound to a pinch of it will start again when I reinstall the KT.

When you go down a steep hill and turn the pedals, the motor will spin at 15 mph because it's the motor speed used for the cut-off, not the bike speed. It'll be doing that all the time you're above 15 mph and after you've slowed down until the clutch re-engages. When the clutch re-engages, it will slow the motor down to the bike speed. The controller can't do anything other than see the motor speed from the hall pulses and control it accordingly. There's no way the software can do what you suggested.

The reason that some other controllers don't behave in the same way is because they use the speed sensor to regulate the cut-off, not the hall sensors. For a KT controller using the hall sensors to regulate the cut-off speed, when going down a hill faster than 15 mphwithout pedalling, the motor is cut and will only be rotating at slow speed due to some friction. When the motor is turning slower than the bike and you demand power, the clutch can easily engage, but when the bike is travelling faster than the motor down the hill, the motor will be spinning at 15 mph, and the clutch can't easily engage when the bike slows down.

The reason some bikes exhibit the behaviour that you indicated and others don't is because they have different types and sizes of clutches in them. Some have springs to help the rollers engage and others don't. Some have large rollers with large mass, others have smaller ones.

In other words, if you have a KT controller set to current control or speed control at maximum, you will get the behaviour in some bikes, but not all. If you have a controller that uses the speed sensor for the cut-off, you won't get the issue.

Finally, I just analysed the Speedict chart that I showed ealier. It plots motor current and speed. It clearly indicates that when I slowed down to start a climb after free-wheeling down a hill, the power came back on at exactly 15 mph. That bike had a KT controller and Xiongda motor. The event is at 8488 and thereafter. 15 mph cut-off speed is the flat part of the mauve line. You can just about make it out again at 8352. Those two times were the only times I was freewheeling down a hill during that part of the record. Most of it is uphill:

 

[Sparksandbangs]

Thank you for taking the time for the detailed explanation.

To summarise. The behaviour is down to a combination of two things. The particular motor installed on my bike and the way in which the KT controller determines the speed cut off using the hall sensors. As the modified Lishui uses the motor speed sensor for cut off it nullifies this behaviour.

Real world experiment. I could modify the code so that the speed limit is activated by the hall sensors and the display information only comes from the speed sensor and then run for a couple of days and see what happens. Problem is that the code is designed to operate from either the speed sensor or the hall sensors. Start to combine the code and there is a good chance of introducing an error or two needing correcting.

So here's what I propose. I will switch the firmware over to run speed measurement from the hall sensors. I will also install an external cycle computer to accurately measure the wheel speed. The behaviour you have explained should occur on the uphill as no other components of the build have changed. Is this acceptable?

 

[saneagle]

Thank you for taking the time for the detailed explanation.

To summarise. The behaviour is down to a combination of two things. The particular motor installed on my bike and the way in which the KT controller determines the speed cut off using the hall sensors. As the modified Lishui uses the motor speed sensor for cut off it nullifies this behaviour.

Real world experiment. I could modify the code so that the speed limit is activated by the hall sensors and the display information only comes from the speed sensor and then run for a couple of days and see what happens. Problem is that the code is designed to operate from either the speed sensor or the hall sensors. Start to combine the code and there is a good chance of introducing an error or two needing correcting.

So here's what I propose. I will switch the firmware over to run speed measurement from the hall sensors. I will also install an external cycle computer to accurately measure the wheel speed. The behaviour you have explained should occur on the uphill as no other components of the build have changed. Is this acceptable?

That's right. Your test should confirm if it's correct. Thanks for doing it. I still have the Speedict somewhere, but the software has long gone in an old phone. If I can find anything to download, I'll do some specific test with it. That old chart was to monitor what happened with the 2-speed drive.

 

[saneagle]

Incidentally, I was just riding my folding bike with KT controller down a hill. It doesn't have a speed sensor, so it picks up the speed signal from the halls. Each time I pedalled, the speed indicated exactly 15 mph, and when I stopped pedalling, it went to zero. This is pretty obvious really.