Castle HV160 Controller vs Sensored (Hall vs Optical)

[MitchJi]

Hi,

My intention is that this thread can be a discussion of all the issues of Castle Sensorless vs Sensored Controllers. Two obvious limitations of the Castle HV series is the 15s A123, 12s Lipo Voltage limitation and the high price so those issues probably don't require discussion.

The questions I have are the reliability of Hall Sensors and the advisability of resolving this issue with optical sensors, the low speed performance of Castle Sensorless Controllers and the relatively large size of Sensored Controllers. Feel free to raise any other questions or issues.

Some quotes and sources of information:
http://endless-sphere.com/forums/viewtopic.php?f=28&t=16952#p247709

The bike controllers are typically sensored, and also amperage controlled. RC controllers are voltage controlled and typically non sensored.

The Castle HV160 is the best RC type controller I have used so far. It even has data logging onboard, and the sync is hands down the best sensorless in the industry. There are issues with blowing caps however, as the typical bike setup has a lot of voltage/amp ripple going on and it is hard on low resistance equipment.

Not an issue with additional caps.
http://endless-sphere.com/forums/viewtopic.php?f=28&t=12672&p=188900

The new HV140 is a freakin BEAST! That may be the best thing running right now. I have two and a friend of mine is running a 3220 and a HV140. He is pulling 200 amps off that controller without problems!

Matt

Richard's comment on the HV110:
http://endless-sphere.com/forums/viewtopic.php?f=28&t=8160&hilit=hv110&start=225#p140186

I'll have to repeat that I am very impressed with the startup torque from a dead stop. Sensorless is definitely the way of the future and this thing proves that it can work well with the right algorithm.

http://endless-sphere.com/forums/viewtopic.php?f=28&t=16414#p241062

I have a 10 pound drive unit sitting here (two Astro 3220s) that runs 28hp total (2,000 ounce inch torque at 14,000 rpm). RC motors are unparralelled in output power. But, the issue of startup is a frustrating one. On a light weight bicycle, this is not usually an issue. On something heavier (even a moped) it becomes problematic.

Matt

http://endless-sphere.com/forums/viewtopic.php?f=28&t=18681&start=30#p273140

You gotta give a sensored controller a shot yourself Matt. Those 2t motors you've got on your trike would be the perfect motors to find the RPM limits of the infinion. When you feel the buttery smooth low RPM operation with mega low-rpm smooth torque, you will forget you ever saw a sensorless controller.

D's Kona requires excellent low speed performance and I haven't heard him complain. OTOH Ypedal (whose bike should have more reduction) isn't happy with his Castle.

So how much worse are sensorless controllers on an Ebike? Is Luke's opinion partially due to not using a Castle esc?

Justin's Controller and comments (I'm quoting his complete list and putting the comments I are relevant to my questions in bold):
http://endless-sphere.com/forums/viewtopic.php?f=29&t=7134#p107193

The Controller

There are a few things this controller does which have been IMO long overdue in the ebike scene:
1) Programmable limits: All of the controller parameters such as maximum battery current, maximum motor current, low voltage cutoff, max regen voltage etc. can be set to whatever value you like within the limits of the controller. With this 6 mosfet version, the battery current limit can be set from 0 - 30 amps, the motor current limit can be set from 0 - 45 amps.

2) Completely Waterproof: The entire controller is potted in epoxy so there is simply no possibility even of water damage. It can be run entirely submersed, in fact the controller would like that for the great heatsinking.

3) Small Size: This thing is about 1/10th the size of the Crystalyte 35A motor controllers so it can actually mount discretely on the bicycle.

4) Proportional Regen: When the regen switch is activated, then the throttle controls the intensity of the regen braking. The regenerative braking is motor current (hence torque) regulated, and you can set the maximum regen motor current in software.

5) Pulse by Pulse Current Limiting: You can short the output motor leads and run the controller full throttle and not blow a mosfet. Nice feature to have when the axle spins out and severs all the motor leads.

6) Current Throttle: The throttle input directly controls the battery current and hence the vehicle power. If you have the battery current limit programmed to say 20A, and ride half throttle, the controller will deliver 10A of power regardless of your speed.

7) Sensored and Sensorless: If there is a hall cable connected it will run as a sensored motor controller, if there are no hall signals present it operates sensorlessly. If the hall senor was initially connected, and then fails in the middle of your ride, the controller will detect this and switch over to sensorless mode without loosing a beat. I don't really need to say much more for people to know what this means for reliability!

8.) Overtemperature Protection: Rather than shutting down when the controller gets hot and leave you stranded on a hill, the controller automatically scales back the motor current limit value as the heatsink warms up, so it will protect itself from getting too hot while at the same time never cutting out completely.

What is the reliability of Hall sensors. Are Justin's reliability issues due to moisture exposure in hub motors or cheap hall sensors in Chinese hubs? Are optical sensors a better and more reliable solution?

What is the availability of relatively compact 60v/120a or 80v/150a sensorless controllers?

Maybe we need to lock Justin in his shop until he produces a perfect ebike controller .

 

[gwhy!]

Hall sensor reliability I think comes down to hub motors, Most of the pictures that I have seen posted on here of opened up hub motors do not look like the best environment for hall sensors with the thin wires that are used and also the heat inside a hub motor can be a issue. The heat/vibration issue was my concern about fitting halls inside the can of my out runners thats why I opted for external fitting ( that and can be swapped out in seconds should a hall fail ). Hall sensors should be as robust as the any other electronic component as long as its kept to within its parameters and I wont mind betting that most of hall sensor failures on hub motors are of a mechanical nature ( i.e broken wire or hall sensor leg ).

I have not tried any of the CC controllers due to cost, but I have played around with a few cheaper versions and they do work very well ( I also have not had one give up on me ) what really puts me off using these sensorless controllers is the dead stop start ability and the fact that so many people on here have managed to blow rc esc's and sometimes for no real apparent reason ( even the CC seem to suffer from this ) and this will frustrate the hell out of me.

By the time you add the additional stuff to a rc esc to make it work for e-bike use the space taken up is more or less the same as a 6fet controller so I think size is not a issue.

sensored controllers are so cheap and far more easily upgradeable/repairable ( should they go wrong ) compared to sensorless rc esc's so sensored controllers win again.

TBH I don't think there is no competition, sensored controllers win hands down on every front.

 

[recumpence]

Apples and oranges..........

From a pound per pound and size comparison, NOTHING comes close to a Castle controller. However, these issues are rarely a problem on a bicycle like they are on a RC model. RC controllers are fine up to a certain point of power. Beyond that, they become fiddley and less than reliable. For a HV160, 8kw is fine (with a decent battery). Above that, things get strange. However, that is ALOT of power for a bike. That being said, even I am moving to sensored for my highest power setups.

I cannot see myself moving to a huge sensored controller for my very light weight lower power builds, though.

Matt

 

[etard]

I'm glad you brought this subject up, I have been contemplating the sensored controller design lately. I think the problem is not the size, but the SHAPE. Other than a trike, or using large cargo bags, there is no logical placement for a huge brick on a bike. Batteries used to have this problem, but now you get packs and arrange them according to your application.

Now, I am clueless about the layout of a sensored controller, but is it possible to "stack" or "turn" components so that a triangular or square shape can be designed? How about round or even as a ring that is especially designed for a certain size motor to be inserted into the middle for one complete, complimentary shape and design with integrated cooling fins? Like a sleeve, possibly cool the whole thing with watercooling and fins. That would be a feat, but somehow I think as a community we could come together and make this happen. Just base it off a 3210 for smaller builds and Matt's new Mega motor for the criminally insane or motorcycles. Take a 12 FET controller and spread it into a ring around a motor and you will have quite a small, compact package with short motor wires, easily monitor temp, maybe put some LED's on the thing to easily monitor everything. Is this possible? or is there something to the Shape of a controller?

 

[gwhy!]

I'm glad you brought this subject up, I have been contemplating the sensored controller design lately. I think the problem is not the size, but the SHAPE. Other than a trike, or using large cargo bags, there is no logical placement for a huge brick on a bike. Batteries used to have this problem, but now you get packs and arrange them according to your application.

Now, I am clueless about the layout of a sensored controller, but is it possible to "stack" or "turn" components so that a triangular or square shape can be designed? How about round or even as a ring that is especially designed for a certain size motor to be inserted into the middle for one complete, complimentary shape and design with integrated cooling fins? Like a sleeve, possibly cool the whole thing with watercooling and fins. That would be a feat, but somehow I think as a community we could come together and make this happen. Just base it off a 3210 for smaller builds and Matt's new Mega motor for the criminally insane or motorcycles. Take a 12 FET controller and spread it into a ring around a motor and you will have quite a small, compact package with short motor wires, easily monitor temp, maybe put some LED's on the thing to easily monitor everything. Is this possible? or is there something to the Shape of a controller?

I think the 12fet + controllers are way to big also for e-bike use and a little over kill for "normal" use( but still very good value for money and still not as big that it can not be mounted onto a bike somewhere ) I think the 6fet is a very nice size and if the fets were on a power fet/slave board it can be made into one very compact controller I am looking at a different package fet at the moment ( I think there are a few people on this forum also considering and working on this option ). The 6fet controllers that I have modded have been ok in burst upto 100A but are currently limited to 70A but will run 70A all day long. The fets that I am looking at along with a better means of cooling ( water cooled) Im hoping to get 200-300A constant and upto 85v, but this in its self will make it physically bigger due to the wiring and cooling area. It would be nice to make a 100v version but this will double the cost. But to be honest I think a bottom level 6fet modded controller would suite most peoples needs.

 

[Jeremy Harris]

There's no doubt in my mind that sensored controllers massively improve low speed torque and starting. I know that the Castle Creations controllers are great, but I'm certain that if you invested the same amount of cash in a sensored controller you'd get way more power, with the added advantage that the torque from a standing start would be significantly better.

There's no reason why a sensored controller should be any bigger than a sensorless one. All that's happened is that we're taking parts that are intended for two very different markets and adapting them for our purposes. It so happens that the sensorless controllers, which are a good match for the cube law rpm/power curve of a propeller, were developed for model aircraft, so are designed to be small and light. On the other hand, the sensored ebike controller designers haven't seen a need to reduce size and weight, so have built chunky boxes.

As gwhy says, the 6 FET controllers can easily handle around 70 amps, 100V and are little bigger than the RC sensorless ones. There's no technical reason why an RC controller couldn't be designed to take sensor inputs and have current limiting. If someone did this for a good price then I think we'd have a real winner.

In the meantime, I'll just keep plugging away (when I get the time) at making a mega amp small controller using the tiny 6 FET boards as a basis.

Jeremy

 

[MitchJi]

Hi,

Thanks for the replies.

It doesn't seem like anyone is concerned with the reliability of Hall sensors so that might be a non-issue?

Opinions vary on the necessity of sensors though:
http://www.endless-sphere.com/forums/viewtopic.php?f=28&t=18817&p=274457#p274322

…I bought a hv100 turnigy esc when I first started playing around with the higher power setups ( the controller is still going but not used anymore ) I was not happy with the startup sync so I fitted hall sensors to my motor and now using a cheap sensored 6fet controller the whole thing works really well.…

…the v4 is what I have yes it works ok but did struggle from a dead stop ( you would need to get the bike moving slightly before winding up the throttle) I think this also applies to the CC esc…

OTOH (even on trails which is probably a good test for startup and low speed torque):

I still have the Turnigy 100 Amp ESC on my bike. I also have a Castle 110HV, but did want to risk blowing it, so I figured I would run the Turnigy for a while. So far it has worked awesome. I have run 8000 watts through it in very short bursts with no issues at all. It does get warm when I am pushing it pretty hard on the off road trails, but cools off quickly once I am back on the streets. I have put a few hundred miles on it so far and am quite happy with it. I have no issues with starting from a dead stop, however I am running an Astro 3220 6 turn, not an outrunner...

The sensored 6 fet controllers sound good. Even if an HV160 has "good enough" start up it sounds like sensored controllers are better. Reasonably compact and inexpensive. Two things would help:
1. The (soon to be released) hall sensor kit for the Astro's from Matt.

2. An affordable turnkey compact high power 6 fet controller.

the controller is a 6 fet keywin controller modified slightly with higher rated fets and better cooling than standard (approx £20.00 uk pound addition of the cost of the controller ) its currently limited to 70A max but will run this all day ( have ran it at 100A max for a little while )

I wanted to know the source for sensored 6fet controller's that can handle 100a. It sounds like that would require modifying a controller which I don't think I'd tackle.

Here Ya go MitchJi:
http://endless-sphere.com/forums/viewtopic.php?f=31&t=18744

6x4110 Mini Monster Programmable Controller LYEN Edition $79

20A is the current rate at 72v or 25A at 48v

* For hackers, you may double and possibly triple the current output to make it a mighty wonder by adding braids wires,add solders to beef up the circuit board traces and shunt. It may also be needed to upgrade to larger capacitors, gauge wires for both the battery and the phase wires to the motor.

Heavy duty hacking to get to 75A. If the startup is ok an HV160 that is bullet proof with extra caps at 160A is a much more user friendly option.

 

[gwhy!]

Hi Mitch,
Please let me clear something up as regards the startup thing, When I say it struggled from a dead stop I am talking about feetup deadstop ( trackstand ) and maybe pointing up a 10%+ grade hill and not using the peddles for any forward momentum from the start, This is what I am referring to when Im talking about dead start startup, for trail riding it would prob be ok for most people as if you stop on a trial 9 times out of 10 you will put your feet down and when you pull away again its instinctive to push off with your foot and this will defo help getting a sensorless controller running if the bike is geared right.

The additional mods to a lynen controller would really not be that bad to carry out as most of the really touchy stuff will aready be done i.e the replacement fets.

 

[swbluto]

So how much worse are sensorless controllers on an Ebike? Is Luke's opinion partially due to not using a Castle esc?

Yeah, it doesn't sound like he's used a castle ESC before. From my experiences with the controller and LFP's experience with the turnigy ESC, it sounds like the castle ESC has better heatsinking of the fets and the firmware is probably better in most respects(It likely doesn't use the board). However, I still think the ESC would probably still blow with LFP's usage. I had to add a powered heatsink to keep the temperature below 120 with my 1 kw application. Otherwise, it'd meander to 220+ F.

 

[liveforphysics]

The HV160 wasn't released before I gave up on RC controllers.

The CC HV110 was the biggest they offered when I got my pair. Strapped as much cap right to the input as I could fit (like I did with all my RC controllers), and they ran for about 2minutes before burning up.

Amazingly, every RC controller lasted about the same for me.

200amp Jeti-spin. About 1-2minutes
Turnigy HV100. About 1-2minutes
Turnigy version 4 HV100 About 1-2minutes (with nic starting/sync)
Extremely heatsinked Turnigy version 4 HV100. About 2-3minutes, agai with nice sync.
Castle HV110. About 1-2 minutes, stuttering starts, but good sync after.

Add them all up, its a pile of a bit over $2,000usd in RC controllers. Some performed well, not a single one was reliable or lasted for any practical/useful amount of time.

The first infinion controllers I got were 12fet units. Upgraded to IRF4110s. Never could blow one, but I was only pulling 80amps through them. I absolutely hated them with a passion because of the "soft-start" curse. I would like to punch the guy in the face who thought of "soft-start". AKA, super dangerous non-linear, dangerously delayed throttle response.

The next set was the 18fet Methys. Can't complain a bit. Drive the motors past 10,000rpm, response is instant, controllers never get too warm. Response is awesome, and controllers are totally tuneable.


Things I never got happy with? Hall placement. Its extremly fussy, and has extreme effects on performance. Advance makes it run like a monster at high RPMs, feels like having an extra 20v in the pack or something. However, then the low RPM operation uses about twice the current just for putting around, and the range is crappy.

What we need is a controller that retards timing at low RPMs for efficiency, then cranks up the timing on topend and WOT when you wana rip.

 

[johnrobholmes]

Agreed, variable timing is really key. Or most optimal, zero timing advance sensored startup and sensorless running in the rest of the range with a programmable timing advance for best efficiency or power (as the user sees fit).

In the mean time, it looks like the china ebike controllers may be the most reliable option for now.

 

[johnrobholmes]

On the note of hall sensor reliability- There is nothing inherently unreliable about them. They are the best detector for the job, IMO. A properly designed motor will keep them far enough from the heat and have them shielded from the elements. Common failures that we see in hub motors can be contributed to either heat or crappy quality control in most cases.

 

[recumpence]

So, how about a servo that changes the hall timing based on throttle position? You could even use a servo delay board to make sure the timing does not advance untill the motor reaches high RPM.

Thoughts?

Matt

 

[johnrobholmes]

That would work for a standard sensored controller. I hate to see how fine the servo control has to be :lol: Maybe a servo arm about 5mm long will do.

 

[amberwolf]

As far as I understand it, usually we'd want to be advancing the timing of the motor, as opposed to retarding it, right? If so, then that means pushing the phase signals ahead of the halls, or delaying the hall signals behind the phases, correct?

If so, an alternative to physically changing the timing by moving the sensors, which adds vulnerable mechanical hardware inside the motor (or on an external shell bracket for RC motors with external halls):

Add a variable delay to all three hall signal lines. In controller firmware this would be easy, if one has access to the source to do it.

Assuming no firmware changes are possible, a hardware addon board could be made simply enough. There probably are already dedicated chips that can add the same delay to a number of digital input signals at the same time. But lacking that, one could use 555s or similar one-shots to delay the pulse, and control the pulse delay time via digital pots linked together. Then use one single input control (whatever would have controlled the servo) to control those digipots.

Alternately, should retarding be necessary, the same kind of delay could be introduced to the MOSFET gate signals, in the same way. If they need to remain analog due to pulse shaping, an analog delay would be necessary, but I expect that in most cases this is not needed.


It would be a more complex solution in design, but at least for the hall-only delay would be very simple to install for the end user and would be the same for all controllers that use the same hall signal input voltage to them.

To install a servo and other hardware to move a hall bracket around would be different for every motor.

 

[mwkeefer]

So, how about a servo that changes the hall timing based on throttle position? You could even use a servo delay board to make sure the timing does not advance untill the motor reaches high RPM.

Thoughts?

Matt

I know we could call it an Actuated HEI Distributor, oh wait.... that's right, they have been on GMs (vaccume advanced which increases with RPM) for years = )_
Great idea though, perhaps even going old school and mechanically linking the throttle would eliminate the need for servo but size could be an issue... I like it though and for motors like Astro's and other RC which seem best with a hall PCB mounted, this could be engineered right into the PCB and physical mount = )_

As always, great ideas Matt!
-Mike

 

[heathyoung]

You couldn't advance electronically (for obvious reasons) but you can run it mechanically fully advanced and electrically retard (delay) at lower RPM's. A PIC could manage this relatively easily. Determine your frequency (only need one hall input to do this) and work out what 'curve' you want based on a lookup table in EEPROM memory space. 3 inputs, 3 outputs. Power from hall sensors. EEPROM memory can be written to (once) when programmed, or be written to/read by a serial interface.

No I'm not going to write it at the moment, before anyone asks, I've got too much on my plate as it is.

 

[liveforphysics]

So, how about a servo that changes the hall timing based on throttle position? You could even use a servo delay board to make sure the timing does not advance untill the motor reaches high RPM.

Thoughts?

Matt

Since I don't have the skills to do it through uController programming (like it should be done lol), I do think this idea would be a pretty damn easy band-aid type solution to give the best of both worlds! Good thinking Matt!

 

[etard]

Hey, while you're adding servos, how about a servo that adjusts belt tensioning based on amp load? :lol:

 

[Tiberius]

You couldn't advance electronically (for obvious reasons) .....

Its quite true that you can't make an output appear before the input with an electronic circuit, but this is slightly different.

In this case we know what the output sequence is, so it is possible to create an advance electronically. It can be done with a phase locked loop quite easily; the trick is to set up the loop to track the input and then to introduce an offset. The offset can be positive or negative, and can be speed dependent, and it appears as a phase shift in the output.

I've done this many times in analogue circuits and I'm sure it can be done digitally too.

Nick

 

[heathyoung]

It needs to be very tight or very low variance (slow acceleration) for critical timing applications - this is why its a technique that isn't used in interceptors etc for fuel injection - its god awful to code well and implement successfully - whenever a value changes, you get mistiming.

The trick usually is to get full advance and retard from there. Much nicer.

 

[MitchJi]

Hi,

I think I finally figured out the answer posted here also:
http://endless-sphere.com/forums/viewtopic.php?f=30&t=14210&p=288124#p288124

That must be a matter of personal preference because I have never, EVER had anything but buttery smooth low RPM operation from my RC ESCs. That being said, they blow easily at low RPM. So, in that regard, low RPM sucks.

Matt

With low throttle demands, it stutters a lot, from a stop. If I just crank on the throttle from a stop, it stutters and then blows up. This happened twice, and that's when I stopped using this setup.

Could one or both of you explain what each of you are doing differently to get opposite results??? :?

D's Kona requires excellent low speed performance and I haven't heard him complain. OTOH Ypedal (whose bike should have more reduction) isn't happy with his Castle.

I think I finally figured it out (I'll post this on the other thread also so hopefully further discussion won't drag Burties thread any further OT).

Matt uses high kv motors with 3 stage reductions (probably at least 30:1 total?). Gary is using the lowest kv and the least reduction (about 9:1) he can to make a single stage reduction possible. And Ypedal's build doesn't have sufficient reduction.

So hopefully my planed build with a kv of 113 (close to Gary's) and a reduction of about 25:1 will be ok (the total reduction is probably the important issue?).

So, how about a servo that changes the hall timing based on throttle position?

The most elegant solution would be to have that as a programmable option in the controller based on RPM.

The next most elegant solutions in my opinion would be how its done on older car distributors, centrifugal or vacuum advance.

For most motors centrifugal would probably be simpler but for the HXT motors using the existing vacuum (assuming it increases with higher RPM) would probably be the easiest solution.

 

[TSSR22]

Hi!
First post on this forum but tons of post reading.

What about the features of this ESC?
http://www.teamtekin.com/rx8.html

I'm in RC car racing for couple of years now and its the best by far IMO for cars.

You can adjust via you PC the timing of the motor (sensored) in a RPM range
No timing at low speed and increase of timing when you get up in speed.
Timing advance,boost from x RPM to y RPM, turbo boost, current limiting, lipo cut-off etc... 
There's tons of features in this ESC but maybe the problem is its limited to 6s.

Maybe its the reason why there's not a lot of people using car's ESC in their projects.

I would like to do a project with this one but the lack of feedback about RC car ESC let me think its maybe not a good idea.

Very nice forum!

Remy

 

[liveforphysics]

A lot of nice control features.

Being limited to 24v makes it a bit of a toy.


I gotta give them full comedian points for putting this in the spec sheet though. Had trouble not choking when I read that. LOL

"Continuous Current: 210 Amps*

*Per FET manufacturer spec sheet"

 

[Hillhater]

Hi!
First post on this forum but tons of post reading.

What about the features of this ESC?
http://www.teamtekin.com/rx8.html

Hi Remy, welcome to the forum and i am glad you brought up the car ESC subject as i was about to do the same siting this unit..
http://www.nitrorcx.com/ezrun-150a-sd-brushless-esc-for-1-8-and-1-5-rc-cars.html
Have you come across it before ?

Luke, i take your point about the 6s limitations etc, but for some of us working down in that voltage range on friction drives , this type of ESC could be an option .....especially if it will truely work as a SENSORED motor controller to give us that low speed functionality and avoid the "sync " problems that seem to plague some of these systems.
so, anyone care to speculate if these ESC's will work a sensored Ebike RC motor ( a halled, Turnigy 63 -54, 250kv)..
.. It would only need to feed 50-70 A reliably !