Castle HV Controllers - Use Extra Caps!

[gwhy!]

What voltage can that esc run Gwhy?
great price
i'm still a little fuzzy on the caps thing tbh, i thought with Richards board we didn't need them? i better know whats what or i'll pop it lol!!

Cheers,

D

I liked the price deecanio
Its advertised as a 8c, but the input caps are only rated at 35v, when I have done all the testing that I want to do at 24v I will have a look at what the max voltage the components could handle, Then maybe test to distruction

 

[recumpence]

Hey Guys,

I did some real world testing today (15 miles of riding and experimenting). You EE guys will be able to fill me in as to the technical side of this, I'm sure. Anyway, I found that the ESC builds up WAAAYYYY more heat just starting off than it does cranking the bike up a hill or accellerating in general as long as it does not have to start the bike from a dead stop.

Basically, I went for a number of rides and did some temp testing. My ESC on my recumbent is in a no airflow spot. So, speed made no difference in that regard. Anyway, without a temp probe (I merely used my finger on the heatsink) I found the ESC gets HOT starting off the line from a dead stop even if I am not pushing it hard. My Eagle Tree showed less than 1000 watts when I started off the line, but the controller was too hot to hold my finger on for more than a few seconds and that was a couple seconds of ESC only startup (no pedalling). However, if I pedal first up to maybe a few mph (just a couple pedal strokes), I could pretty much hammer the throttle with only a luke warm temp build up on the ESC. So, assuming heat in the FETs is the main failure issue, asside from inductance (which the caps remedy), then the secret to making the HV110 survive high output use is to merely pedal first. Heck, I prefer pedalling first anyway. It is smoother than trying to find that perfect startup point of throttle movement.

So, here is my experience with the HV110 in one sentence;

Tripple the input capacitance to compensate for the ripple (inductance) and pedal first to get the bike rolling, then hit the throttle. If you do those two things, it looks like the HV110 is quite a cool running, reliable controller.

Matt

 

[etard]

That makes me wonder if Luke was doing just that before he blew his controller. Anybody else have a blown ESC story that might relate?

I also wonder if maybe adding a clutch that engages and "burns" through that first 5 mph might not solve this problem. Matt, can you do a test with your slipper clutch a little looser so that it slips more?

 

[swbluto]

That makes me wonder if Luke was doing just that before he blew his controller. Anybody else have a blown ESC story that might relate?

I also wonder if maybe adding a clutch that engages and "burns" through that first 5 mph might not solve this problem. Matt, can you do a test with your slipper clutch a little looser so that it slips more?

Luke attests to not having functioning pedals on his bike, so I doubt he "pedaled first".

Anyways, the amount of current the motor draws depends directly on the RPM of the motor. The greater the RPM, the lower the motor current. The lower the motor current, the less the controller heats. Thus, you can see, that at 0 RPM, the motor would be consuming the greatest amount of current and thus the controller would heat the greatest. If you have any "current limiting" or any other duty-cycle based limiting(Like partial throttle), then that greatly increases the amount heating as then the mosfets are switching on and off which entails heat generation from switching. The technical reason for this has to do with the fact that there's current flowing during the time it's switching from the on to the off state and the resistance of the mosfet becomes somewhat significant, so the heating suddenly becomes significant (As P = I^2*R - When the mofsets are fully on, R(resistance) is reaaaalllly low so the heating is very low). Again, the greater the motor current, the more the mosfets heat.

 

[recumpence]

That makes me wonder if Luke was doing just that before he blew his controller. Anybody else have a blown ESC story that might relate?

I also wonder if maybe adding a clutch that engages and "burns" through that first 5 mph might not solve this problem. Matt, can you do a test with your slipper clutch a little looser so that it slips more?

Yes, there is a noticeable reduction in ESC heat when the clutch slips more.

The beauty of this clutch is, the lower the RPM of the entire drive, the easier it slips. At or near stall, the clutch is set to slip at 1000 watts of motor current. At 30mph, that jumps to over 3000 watts. It seems to be perfect for this application.

Hmm, maybe I should put it into production.

Matt

 

[swbluto]

The beauty of this clutch is, the lower the RPM of the entire drive, the easier it slips. At or near stall, the clutch is set to slip at 1000 watts of motor current. At 30mph, that jumps to over 3000 watts. It seems to be perfect for this application.

Hmm, maybe I should put it into production.

Matt

So it's a constant torque type of application? That is, as the RPM goes up, so does the power at a given amount of torque as P = torque*radial_velocity.

Also, at or near 0 RPM, you theoretically have 0 output watts(or very low) but you do have mega input watts, which is why the efficiency is so low at low RPM.

 

[recumpence]

Man, you are a brain!

Thanks for the input! I constantly marvel at all the electronic info and technical explanations. I always seem to just "Know" what is going on and can give a pretty accurate guess as to how something will perform. But, I am rarely able to give the math to demonstrate WHY it is happening. Cool! 8)

Also, yes, the clutch is a constant torque design. It is not RPM affected as far as tension is concerned. It does work fantastically well, though.

Anyway, I agree we need a sensored solution. But, for now and for those of use who LOVE the HV110 programming features and other items related to RC sensorless controllers, it looks like we are compiling a pretty good understanding of what is going on and how to extract the most grin out of these little gems.

To a degree, I feel responsible to figure out how to get the most out of these controllers because I have been the one pushing people (or at least guiding them) in this direction with RC stuff and these controllers in particular. So, I am very glad we are seeing more data and getting a handle on this.

Matt

 

[fechter]

When I was testing the throttle circuit, I was mostly doing full throttle accelerations from a dead stop. I did notice the thing heats up quick, but I'm glad to hear it runs cooler at speed.
When you are pulling high loads at low rpm, the current going to the motor (and therefore flowing through the FETs) may be several times the battery current due to the buck converter effect. I measured somewhere between 3-4 times higher under worst conditions with a Crystalyte analog controller and my BMC motor. When the motor is at or near full speed, the PWM is at 100%, so the motor current equals the battery current.

If the power is reduced, you can operate at lower rpm and keep the heat down, but with reduced torque. This is essentially what the current limiter circuit is supposed to do. Not only does it limit controller heating, it limits motor heating as well.

Using a friction slip clutch seems like a real waste of energy. If the limiter keeps the motor torque down to the same level that the clutch slips at, the motor current draw should be the same, but the input power would be less, since the motor is going slower. Torque is a direct function of motor current for all practical purposes. Net result will much less energy loss, and decreased w-hr per mile.

 

[erth64net]

Using a friction slip clutch seems like a real waste of energy.

Personally, I'd rather burn up clutch pads, than a $300-$500 controller

If the limiter keeps the motor torque down to the same level that the clutch slips at, the motor current draw should be the same, but the input power would be less, since the motor is going slower. Torque is a direct function of motor current for all practical purposes. Net result will much less energy loss, and decreased w-hr per mile.

To me, it sounds like the controller simply needs the function to ramp-up power at a lower rate...not necessarly current-limiting, more, "demand" limiting

 

[northernmike]

Let me see if I'm understanding this correctly. I'm probably not, because I'm feeling pretty confused!

OK.

Scenario = The HV 110 is put into full-throttle condition from a dead stop. Motor is stopped. Bike is stopped. No slipper clutch exists in this example. We're talking direct drive. So full load, twist-and-go, like you're riding an automatic scooter, or a ski-doo.

The controller has to figure out armature position before it begins applying current, and then it applies power and begins trying to increase RPM. Do I have that right?

Doing so is creating a TON of heat, and by the sounds of it is able to destroy unmodified / unregulated controllers - and not necessarily in an instant - is that right?

Besides adding caps and regulating battery-side current, what can be done so that these controllers are able to provide the punch needed to take off from a dead stop?

Alternately, from the "adding hall sensors to brushless R/C motors" thread, is there a sensored brushless controller out there that will provide off-the-line punch?

5kw sounds impressive but if I have to pedal to get it moving at a stoplight, I'm still disappointed.

I'm honestly entertaining the idea of a Mars, and the option of brushes.

Please guys, tell me there's hope that I won't have to go that route?

 

[recumpence]

OK, here is my take on it. Bear in mind, I have no electronic training. However, what I have seen is;

A small amount of RPM helps the controller live rather than dead stop accelleration. I have hit the throttle hard from a stop and this seems to help in that the motor begins spinning quickly and the heat does not build up as much as dead stop starting at moderate throttle.

Startup is hard on these controllers. However, if you startup at very low throttle the controllers lives fine or get it past that transient point quickly it seems OK.

I do not know if I am making any sense here or not.

Matt

 

[northernmike]

Matt, that makes perfect sense.

You mean applying full throttle instead of partial throttle generates less of a differential between the battery-to-controller current and the controller-to-motor current - ? - I think this is what Richard means by the "buck" effect.

If that's true, though, without armature-side measurement to compare with the battery side shunt, the limiter circuit may be severely... limited...

A 'ramp' that is calibrated to balance armature and battery currents and keep the FET loading down would probably be the best solution - but this would have to be tuned, to the bike + rider weight, and the slope... wouldn't it?

 

[swbluto]

Yes, full throttle generates more heating than partial throttle at a given RPM, as partial throttle causes the mosfets to switch on and off which creates less heating. At *true* full throttle, there is NO switching.

So, basically, WOT makes for less heating due to the lack of switching losses.

In my experience, though, I found that heating can be controlled by adding an additional heat sink(And making sure it's adequate by checking the temperature, of course!). For most cases, this means problem solved.

 

[oofnik]

In my experience, though, I found that heating can be controlled by adding an additional heat sink(And making sure it's adequate by checking the temperature, of course!). For most cases, this means problem solved.

Don't mean to stray off topic, but how is that heat sink working out? Are you still using the folded foil method? Are all the FETs staying cool?

I did something similar but with a slightly smaller heatsink that fit in to the groove on the ESC. I may have to upgrade.

 

[swbluto]

I'm leaving the heatsink unpowered, so I just basically increased the controller's cooling surface area. After a relatively "moderate" run(Some slight hills, but nothing big), I've found the controller's hottest part to be at 165 degrees fahrenheit and after a "heavy run"(A big hill according to small scooter standards! It's probably a 200-300 feet rise over a 6% grade), the hottest part was at 185 degrees Fahrenheit and this was all inside an enclosed back pack. Before the heat sink and folded foil method, it would easily top 230 after a moderate run and my thermal protection circuit kicked in and shutted the controller off. BTW, is "shutted" a standard english word? I could've swore it is but Firefox's built in dictionary thinks otherwise.)

Edit: I just checked mapmyride and they seem to think the big hill was 312 ft. tall. and that it was between 7% and 11% in terms of slope. I was going on average of 10 mph, and the current was limited to 15 amps from a 48V volt battery. If you don't have any current limiting, your controller shouldn't heat up as much as mine did.

 

[northernmike]

Bluto, it would be correct to say "shut the controller off", not "shutted the controller off". I know there's a technical linguistic explanation for this, but don't have it!

What controller / motor / batteries are you using?

 

[swbluto]

Bluto, it would be correct to say "shut the controller off", not "shutted the controller off". I know there's a technical linguistic explanation for this, but don't have it!

What controller / motor / batteries are you using?

It seems queer that a past tense form would be the same as the present tense, but I'm sure there are similar exceptions in English like there seems to be everywhere. Most of these exceptions that I blithely use elude me until a fairly intelligent foreigner starts learning the language and starts asking and questioning "casual usage".

Anyways...

CC Phoenix HV-85 / HXT 200 Kv 30 mOhm motor(Half the weight of the larger HXT motor used for conversions) / Ping 1.0 48V 10 AH 220-250 mOhm.

 

[gwhy!]

In my experience, though, I found that heating can be controlled by adding an additional heat sink(And making sure it's adequate by checking the temperature, of course!). For most cases, this means problem solved.

This is what I am finding, I have increased the heatsink and also mounted it a lot better onto the fets rather than it just being held onto the fets with heatshrink on my cheapo esc and this has decreased the max temp whilst I have been testing it on the bench anyway.

To me, it sounds like the controller simply needs the function to ramp-up power at a lower rate...not necessarly current-limiting, more, "demand" limiting

Simply ramping up the drive to the motor at a slower rate is not really going to solve the problem if the motor is still in the stalled state ( not with the controller i am playing with anyway ). The circuit that fetcher has built is a very good solution to this problem as long as it reacts fast enough, which it sounds like it will do.

 

[northernmike]

I may be completely out to lunch, but this is worrying me.

Simply ramping up the drive to the motor at a slower rate is not really going to solve the problem if the motor is still in the stalled state ( not with the controller i am playing with anyway ). The circuit that fetcher has built is a very good solution to this problem as long as it reacts fast enough, which it sounds like it will do.

When you are pulling high loads at low rpm, the current going to the motor (and therefore flowing through the FETs) may be several times the battery current due to the buck converter effect. I measured somewhere between 3-4 times higher under worst conditions with a Crystalyte analog controller and my BMC motor. When the motor is at or near full speed, the PWM is at 100%, so the motor current equals the battery current.

If the power is reduced, you can operate at lower rpm and keep the heat down, but with reduced torque. This is essentially what the current limiter circuit is supposed to do. Not only does it limit controller heating, it limits motor heating as well.

But guys, doesn't this limiter circuit effectively CREATE a ramp?

My understanding of it is this:

- On take off, full throttle input generates full throttle PWM, creating an over-current situation. AT THE ARMATURE. Perhaps the battery side follows suit, and the shunt measures over-current there. But, do the FETs not still get hammered?

- The limiter circuit reacts by cutting down throttle, increasing PWM'd "off" time, INCREASING the "buck converter" effect.

- As battery side current drops, throttle (PWM) is increased (decreasing "buck") until over-current is hit again, at which point it loops again, right?

Anyways, the amount of current the motor draws depends directly on the RPM of the motor. The greater the RPM, the lower the motor current. The lower the motor current, the less the controller heats. Thus, you can see, that at 0 RPM, the motor would be consuming the greatest amount of current and thus the controller would heat the greatest. If you have any "current limiting" or any other duty-cycle based limiting(Like partial throttle), then that greatly increases the amount heating as then the mosfets are switching on and off which entails heat generation from switching. The technical reason for this has to do with the fact that there's current flowing during the time it's switching from the on to the off state and the resistance of the mosfet becomes somewhat significant, so the heating suddenly becomes significant (As P = I^2*R - When the mofsets are fully on, R(resistance) is reaaaalllly low so the heating is very low). Again, the greater the motor current, the more the mosfets heat.

Please let me know if my thinking is right or not?

I really need to know!

If PWM-based brushless control can't provide take-off torque, I'm out..

 

[gwhy!]

I have been playing with my cheap esc and have found that with the soft start enabled it tends to loose sync far to easy at the beginning of the throttle control with a small load on the motor, without a load on the motor it seems 50/50 to weather it would loose sync but even if sync is lost it still normally continues upto max throttle. with the softstart disabled I am unable to make it loose sync at all ( on the bench anyway) I was just wondering do you guys run your esc's with the soft start enabled, and was wondering if this can be the cause of some sync problems. Has anyone else played around with the softstart settings ... with what results..

 

[recumpence]

Mike,

The HV110 is very good at starting. I have never had a startup issue at all. It starts reliably every time.

I merely brought up a longevity finding related to hard starting the control versus pedalling (one pedal stroke) first. What I found is the controller heats up far less when a simple one pedal start was used versus starting from a dead stop. The controller was not overheating by any means. I merely wanted to share my experience with this to see what the electronic gurus had to say about it.

Again, the HV110 starts up reliably, every time from a dead stop. I just wanted to share the fact that a one pedal first start may add life to the controller as it operates cooler that way.

I hope that helps clear this up a bit.

Matt

 

[gwhy!]

I may be completely out to lunch, but this is worrying me........................................................

Now to be honest I am not the best person to try to explain anything but my understanding of this is : if the pwm is ramped and the motor is in the stalled state the you will be switching the fets for the time it takes to ramp up ( so this will make more heat ) and the motor may still not start moving until near full throttle anyway so I think the best solution is ( as long as the drive train can handle it ) zero to max throttle as quick as pos ( no ramping ( no softstart )) so less heat in the fets and more likely to get the motor moving but this can produce loads of current so there needs to be away of reducing the max current if the motor is still stalled ( but hopefully it wont be ) . I hope this makes sence.. please anybody feel free to pull this apart :lol:

edit:
As Matt has said any information that can shared to help stop blowing a controller up has got to be a good thing. It sounds like the CC's is a very good controller and as been said that most failures can be put down to over heating problems.

 

[northernmike]

Yes Gwhy, now we're getting into it.

This is why I think Matt has such "luck" with his clutch - and why I may abandon the limiter circuit idea.

I would like to know how the sensored brushless world of controllers deals with starting surge - how is it that the Sevcon Millipak is able to pull this off?

 

[gwhy!]

Yes Gwhy, now we're getting into it.

This is why I think Matt has such "luck" with his clutch - and why I may abandon the limiter circuit idea.

I would like to know how the sensored brushless world of controllers deals with starting surge - how is it that the Sevcon Millipak is able to pull this off?

It may have a current limiter circuit built into the controller . The esc's that we are using to be fair was not designed to have such a demanding life as we are using them for so I think under the circumstances they hold up very well and Im sure that with a few mods and additions they can be made to work very well on ev's and I think a circuit like fetcher's is a very good starting point.

 

[northernmike]

Starting point, perhaps, but I'm interested in the end point!

Why should things have to get hot, let alone hot enough to burn out? With enough paralleled FETs could this not be totally cured? If enough is 100, or 200, then so be it.

Can't current be accommodated rather than limited?