NEW - ESC vs ESC

[ub3r]

YGE 120 HV

The YGE 120HV is a dedicated heli ESC designed for up to 14S operation @ 120Amps. This ESC has the highest On-Resistance compared to any other 120Amp controller I've seen. Even the Turnigy 120Amp unit has lower IR. However, what makes this unit work so well is the Active free-wheeling under partial load.

What separates this ESC from other common ESC's is the German build, anti-spark circuit, and active free-wheeling.



SPECS:
14S @ 120Amps
114grams without connectors
Active-free wheeling
8-16Khz PWM
Controller: Atmel mega168 16KB @ 20Mhz
Drivers: National LM5101 http://www.digikey.com/product-detai...101M-ND/634651
FETs: Fairchild 045AN08 http://www.digikey.com/product-detai...A0TR-ND/965271

The MCU used in the YGE ESC's is an Atmel Mega processor running at 16 Mhz. This chip is probably the most widely used controller I've seen in ESCs. We see the same chip used in most Hobbywing, Turnigy, and even the Kontronik Jive ESCs. What makes this MCU so convenient is the on-board analog comparator (which is used for zero-cross detection) and the cheap pricing. YGE have decided to run with an external 16Mhz crystal, as opposed to just using the internal 8Mhz RC oscillator as seen in cheaper designs, giving us twice the processing power.


The LM5101 FET drivers used in the YGE are a proven reliable driver. These drivers are proven to work well, and are commonly found in battery chargers etc. They arent the fastest driver, but will do just fine driving average FETs.

The 120HV utilizes the same Active Antispark Circuit as we saw in the KOSMIK. It has a couple of FETs which conect the Capacitors to the Supply slightly after the unit is powered. If you look closely at the picture below, youll notice the SMD FETs sitting under the capacitors. YGE are one of the first manufacturers to give us the Active anti spark utilising FETS. The old passive anti-spark (as seen in turnigy ESCs) has been around for years.


The YGE power PCB is quite unique in comparison to other designs. They have decided to use large D2PAK FETs as opposed to the traditional microFETs we see everywhere. These FETs have a larger heat-capacity and power dissipation compared to the smaller FETs. As a result, they can get away without mounting them directly to heat-sinks. The Power PCB looks like a 4 layer utilizing 4oz copper tracks to reduce resistance. The ceramic capacitors are crucial for filtering high frequency noise during fast turn-on. These aren't normally seen across the input supplies of other ESCs.


The FETs used have 4.5milliOhms of resistance, giving us a ph-phase resistance of (4.5mR/3FETs)x2Phases = 3milliOhms. The benefit of cheaper FETs is a smaller die size, which gives us a reduced input capacitance and gate charge, resulting in faster switching times.. Therefore reduced switching losses. 3mR is quite high for a 120A controller, but they can get away with it due to the active free-wheeling and reduced switching losses.


Ive spent the past couple of days benching these ESCs, and the YGE does really well under partial load due to the AFW. The YGE 120HV (non heatsinked) at 60Amps runs cooler than the ICE HV80 (heatsinked) at 40Amps during partial load operation (the test was run for 3000mah).

One thing i didnt like about the ESC is the non-isolated governor output.

Ill describe exactly how active free-wheeling operates to lower temperatures in the coming days. Im hoping to get the footage organised soon so i can post it up.

[ub3r]

What is active free-wheeling?? And what are the benefits???

Some electrical basics first..
1) Every motor is an inductor.
2) The current in an inductor can never change instantly.
3) The voltage across an inductor can change instantly.
4) A FET is a like switch, and turns on/off with a given gate signal
5) A diode is like a valve, only allows current to flow in one direction.

Some BLDC Basics:
1) BLDC motors have three phases, which consist of coils (or inductors).
2) Commutation is achieved by energizing each phase in order, causing a rotating magnetic field, which causes the magnets to follow.
3) Motor power is controller by adjusting average motor voltage. Average voltage is varied by chopping the commutation signal with a PWM signal. 50% ON and 50% OFF means the phase is only seeing half the applied voltage.

Some ESC Basics:
1) Every ESC uses FETs to switch the voltages going to our motor phases.
2) The FETs used all have inbuilt diodes, called Body Diodes.
3) Every ESC requires some form of Free-wheeling. There are two types: Passive (Diode) Free-wheeling, or Active (FET Based) Free-wheeling.

But what is Free-wheeling??
Ok.. Its not what you think it is. Many of you think free-wheeling means: When you release the throttle, free-wheeling allows the motor to keep rotating freely.. I can understand why someone would think that.. Yes the name is a little deceptive. Freewheeling is actually much more complex than that, and to truly understand how it works, you need to understand the basics mentioned above, and have some general understanding with electronics circuits.

Passive FW
Ok, now the hard part.. Look at the picture below: You will see two arrows. A RED arrow showing us how current flows when the PWM is ON, and the Blue arrow showing us how the current flows when the PWM is OFF. When a Phase is energized, the current begins to flow from the Supply (U), through FET Q1, through the motor phases, then back out of FET Q6. Then FET Q1 Turns off (PWM-OFF at Partial load). Now as we learnt from the above basics, the current cannot cease flowing instantly. And it can no longer flow via FET Q1 because its turned OFF. So the residual current (stored in the winding inductance) finds a new path via D2 (the Body Diode in FET 2). Then FET Q1 begins to conduct again during our PWM-ON, and the cycle repeats with the ON-OFF-ON-OFF nature of PWM. The lower the PWM duty cycle, the longer the Body Diode D2 conducts, thus more heat. I hope this makes sense.

This is the in-efficient method of free-wheeling, or Passive FW. Why is it in-efficient?? Because diodes burn voltage off as heat. Yep, even FET Body Diodes burn voltage too.

The above free-wheeling method (Passive FW) is the type seen in Scorpion and Castle ESC's. These ESCs require heat-sinks to dissipate the free-wheeling losses.

So how do we make it efficient?? and not burn off voltage??
Well, we use Active Free-Wheeling.

Active FW
Ok, refer to this diagram.. As you can see our loop currents are the same. When the phase is energised, the current flows via FET M1, through the motor, and then out of FET M6. When the Phase is de-energized, the residual current flows via FET M2, instead of diode D2. . All we are doing is turning FET M2 ON when FET M1 turns OFF (to provide a current loop).

FETs dont burn voltage like diodes, thus providing us with a more effecint means of controlling phase voltage.


AFW is seen in Kontronik, YGE and now.. YEP.

Active free-wheeling is pretty complex, and difficult to understand. It also has other names like Synchronous rectification, or Active rectification. Whenever you hear these names, you can safely assume there is a power circuit somewhere, where an active component (like a FET) has replaced a in-efficient diode.

Video Explanations:

Helifreak Download Link

Youtube

[ame]http://www.youtube.com/watch?v=uqzOQGiwGnE&feature=player_embedded[/ame]

[ub3r]

Short time ago I was talking to John (Mercuriell) about reviewing some common ESCs, comparing their features, price, and performance at full and partial Throttle with varying loads. He happily offered his Kosmik 200 to include in the test, and I couldn't resist.

Over the next week or so I plan to:

1) Compare ESC cost and basic features
2) Do a complete teardown and review of the internals including an estimate HW cost.
3) Performance comparison at Full throttle (100% Duty cycle) @ 50% of the rated input current
4) Performance comparison at Partial throttle (50% Duty cycle) @ 50% of the rated input current
5) Transient load tests, from 0% rated input current to 100% of the loaded input current.

I'll also be getting into some technical explanations on how Active free-wheeling 'AFW' works, and how it affects ESC and motor performance. Also be discussing how motor inductance affects motor power, how to choose your ESC PWM frequency and timing and why low inductance motors, eg double-Wye (YY) connected, are key for high power applications.

I'll be including the ICE2 HV160 + HV80, YGE120HV, KOSMIK 200, PowerJive 120, Turnigy Dlux 80, and possibly a SJ Hawk 120.

1) Kontronik Kosmik 200

2) YGE 120HV

3) Active Freewheeling

4) HW review of the Castle ICE2 ESCs

5) Video - Full & Partial throttle tests of Kosmik and CC ICE2 HV 160






Ill be tearing them apart, posting up some photos, will briefly discuss the internal components, compare the HW including HW cost. Ill also be including oscilloscope snapshots of the phase voltages at full and partial throttle, show you how timing advance works, and compare motor RPM (Some say the YGE produces more RPM for the same input voltage and motor Kv).

Close up of the Kosmik. This has 10A/30A Cont/Burst BEC. Im going to stress test the BEC to see if itll handle the rated current.


Ill be using my Dyno as a load. The dyno consists of a custom made Scorpion HK4535, and a Hk4035-560Kv..


Close up of the dyno


The Dyno works by having one motor connected to the ESC, and the other motor (used as a generator) connected to a variable load bank. The output voltage gets rectified into DC, then dissipated into a MOSFET load bank. The load can be varied by varying the amount of current i draw from the generator. I used this dyno to test my 24S ESC and it worked like a treat.

Ill be uploading the results and videos to this thread as i progress through the work. stay tuned...

Feel free to throw in some ideas/comments...


Kontronik KOSMIK 200

Ok... I thought id start with a tear-down of the KOSMIK as many of you would be eagerly waiting to see the internals.

SHORT DESCRIPTION:
The KOSMIK 200 is probably the largest (and most expensive) RC ESC i have seen to date. Physically, Its quite large and heavy, however, this monster ESC is not designed for small machines. If your considering this ESC, you're probably building a 700 sized speed machine, or 800 sized hardcore 3D rig.

One thing you'll notice is that the wires are connected onto the ESC with tabs and M5 nuts and bolts, rather than solder. I do prefer nicely soldered connections, however soldering such large wires would be quite difficult..

SPECS:
14S @ 200Amps continuous.
200Grams without cables, 310grams with cables/connectors/screws.
Active free-wheeling
8-32Khz PWM
Hardware:
Main controller: STM32 32Bit CPU @ 120Mhz
FET: International rectifier IRF7759 http://www.irf.com/product-info/data...uirf7759l2.pdf
FET drivers: International rectifier IRS2186 http://www.irf.com/product-info/data...irs2186pbf.pdf
BEC Regulator: National Semiconductor LM5116 https://www.national.com/ds/LM/LM5116.pdf
Price:
$809 @ Readyheli



INTERNALS:
The KOSMIK consists of 3 x PCBS. We have a Logic PCB (the brains), a MOSFET driver and BEC PCB, and the Power PCB. They are all connected using 0.05' and 0.1' SMD pin headers.


STM32 main controller. Enough proccessing power to mine bitcoins here.


The BEC circuit consists of a National Semiconductor LM5116 synchronous regulator IC, a 10uH SMD high frequency inductor, 3 x 80Volt SMD FETs and a diode + capacitors. The synchronous design is required to keep temps cool at high input/output voltage ratios. FETs are used instead of free-wheeling diodes (which have a large voltage drop when conducting). I have yet to test it, but i have no doubt it'll hold up to the 10Amp continuous current rating.


Normally you'd only find 3 x FET drivers in a BL ESC. However, Kontronik have chosen to double that, and use 6 x FET drivers in the KOSMIK. When switching FETs at high Speed (high slew rates), and high frequency the FET gates do begin to draw a considerable amount of current, and begin to stress the drivers. It looks like they've added an extra driver, giving us 2 x drivers per phase, sharing the load. Good thinking.

The IR2186 is currently the most powerful Hi/LO FET driver in a SOIC8 SMD package. Having a robust driver is crucial in high current, high stress applications, because if the driver fails, there is the possibility of the upper and lower FETs conducting and causing shoot-through (or cremating ), as we saw in the previous ICE HV series...


I'm guessing the power PCB is at least a 4 layer @ 4oz copper.


We have current sensing shunts on each phase. This hints the possibility of future updates that include SINUSOIDAL or FOC commutation.


The FETs used are rated at a whopping 160Amps continuous each. With only 1.8mR resistance and 8 x FETs on each phase, we can see where the KOSMIK gets its 200Amp rating from.


There is an additional FET near the input terminals. This is the key element in the anti-spark circuit, connecting the capacitors to the main supply slightly after power up.


All the semiconductors are well known brands, and there is clearly no skimping on quality anywhere. Even the PCB is high grade. But does it justify the $800 asking price?? With roughly $200 in component cost (not including the CNC case, assembly and engineering) we are going to have to wait and see how it performs..

Next up the YGE 120 HV !!

[ub3r]

Castle ICE2 HV Series ESCs

The ICE2 series of ESCs come in a 40, 60, 80 120 and 160 amp versions. The only difference between all five models is some extra FETs. Ive been testing the 80 and 160, and they do really well at high throttle percentages. However due to the lack of AFW, they tend run hot at partial throttle + high current situations.

ICE2 160HV SPECS:
12S @ 160Amps
210g with cables
8-32KHz
Passive Free-wheeling with heatsink.
Datalogging.
FETs USED: 48 x Infineon 31n06 (http://www.digikey.com/product-detai...GTR-ND/2080621)
Drivers: IR IRS21867 http://www.digikey.com/product-detai...PBF-ND/3056635
MCU: SiLABs 8Bit 8051 Pipelined MCU @ 32MHz.
PRICE: 272USD

Castle ICE2 160 compared to the PJ 120.


Castle ICE2 80.


All these ESCs feature a nice chunky heatsink mounted directly onto the FETs to help keep temps low. The HV80 has 8 FETs per phase, and the HV160 has 16. The FETs are mounted on both sides of the PCB. The power PCB uses thick copper bus bars to help keep the track resistance to a minimum.
HV160.

HV80.


One thing i like about the Castle ICE ESCs is the datalogging feature. The logger requires current shunts to measure current. The shunts are mounted directly to the negative input terminal of the power PCB.


Castle uses the same Logic PCB in all their ICE2 ESCs. The Logic PCB consists of SMD components on both the top and bottom sides. Its a tiny PCB, and measures roughly 25 x 25mm. The FET drivers and MCU are mounted on the bottom side, and the power-supply and MCU are mounted to the underside. The Logic board connects to the power PCB via a PCB-PCB interconnect (http://www.digikey.com/scripts/DkSea...-15S-1&cur=USD). The ESC electronics are powered via a tiny LM5009 switching regulator.

Electrically, the drivers used in the ICE2 are exactly the same as those used in the KOSMIK. The only difference is the pinout. Castle asked International Rectifier produce the drivers in a custom pin-out for direct replacement with their previous ISL6700 driver. The previous ISL6700 drivers failed due to the negative transient voltages caused during fast switching. Spikes are caused due to the long trace lengths, from the Logic PCB to the Power PCB (adding extra inductance), and rapid change in currents.. Inductance + Current Change = Voltage. (Lenz's Law ). Faster change in Current = Higher change in voltage.

Although the IRS21867 can handle small spikes, Castle has still taken extra precaution and added ultra-fast diodes in reverse bias to the PH-pin of the driver, to block the negative transients. You can see the additional diodes circled below. (I described how diodes block transients in the beginning of my AFW video). The recent fires in the 160HV were due to a connector issue, and not related to electrical design.


The feedback circuit consists of an external comparator and a few voltage divider feedback circuits.


Just a few points to mention..

Castle ESCs do have 2 x opto-isolators, however this does not provide any extra protection against interference. The only reason why they went with dual isolators is because they need bi-directional communication as the servo lead is also used for programming.


The 160HV has the lowest internal resistance compared to any other ESC on the market, including the KOSMIK. Infact the 160HV has 20% less IR compared to the KOSMIK. Then why does the KOSMIK run cooler?? Well it depends on how its used. In my testing (Yes i know, i have yet to upload the vids). Ive found the 160HV runs cooler at full power, and the KOSMIK runs cooler at partial power. This is directly related to the AFW.

Also, why isn't Castle V4.1 compatible with external phase sensors??
Well in V3.8 a standard commutation routine was used which looked like this:

In this switching waveform, the upper FETs do the PWM switching and commutation, and lower FETs do commutation only. Because the upper FETs are switching, the lower FETs will absorb all the heat from the free-wheeling currents (as i described in my video). Free-wheeling heat is not an issue with AFW ESCs, and this method works well..

So.. To increase the performance of the heatsink, and to distribute the free-wheeling heat evenly around the heatsink, Castle decided to invert the switching from Lower->Upper FETs twice per commutation. Huhh?? Well look at the pic below and you might understand. This is a really cool method, and allows the heat to spread evenly both the upper and lower FETs, reducing hotspots.

As you can see, the switching is now shared between the upper and lower FETs. This method reduces the stress on the FETs, however, it brought forward some complications for our phase sensors. The older Phase sensors werent compatible with this routine.

Overall, i think the castle ESCs are really well built, and have awesome features. There were a couple of design flaws in the earlier versions, however the issues have been ironed out as the product matured. If Castle release a AFW software patch, which is possible, these will no-doubt be the best bang for buck ESC on the market.

[Elios000]

how about tossing the Hobby Wing 120HV in that line up too since it pretty popular now too

[Mercuriell]

Thats a nice dyno setup Tony, what sort of power is that generating, could be KWatts. !! Lot of power for that MOSFET load to sink

[Bob O]

Subscribed!!

[sarpilot]

Nice setup look forward to your testing as well...

Sent from my DROID3 using Tapatalk 2

[fr0stbyt3]

Looking forward to the results... Nice!

[ub3r]

Originally Posted by Elios000

how about tossing the Hobby Wing 120HV in that line up too since it pretty popular now too

Good idea. I include it as soon as i get my hands on one.

Originally Posted by Mercuriell

Thats a nice dyno setup Tony, what sort of power is that generating, could be KWatts. !! Lot of power for that MOSFET load to sink

Hi John. Yep definitely Kilowatts. I've had cases where both motors have come off @ 100°C or more. The FETs can dissipate up to 500Watts each under water, and i have 10 FETs connected to my small load (5KW), and 40 FETs (20KW) on my large load bank.. Ill be uploading a video demonstrating how it works..

[PlaneHazza]

Wow, there's like £1500 worth of ESC in that one photo. That's more than the worth of any one of my helis!

[kimmik]

I can send you my hw120, you can test till it burns. In fact I'd be mildly disappointed if it doesn't get burned lol.

Btw, for point 4:

50% throttle at 50% input current.

Would it be better to use 25% input current? 50% input current at 50% throttle is 100% motor current. Kind of risky to have 200A in the 4035. Also risky to the esc without afw.

Feel free to try it at 50% but i'm just concerned.

[ub3r]

Originally Posted by kimmik

I can send you my hw120, you can test till it burns. In fact I'd be mildly disappointed if it doesn't get burned lol.

Btw, for point 4:

50% throttle at 50% input current.

Would it be better to use 25% input current? 50% input current at 50% throttle is 100% motor current. Kind of risky to have 200A in the 4035. Also risky to the esc without afw.

Feel free to try it at 50% but i'm just concerned.

Hey Kim,
I have a dead Hawk 120 which i received off Luke (ranma103). Ill take up your offer i cant get it going.

Heres a pic of my 20KW load. It needs to sit in a water tank during operation.

[kimmik]

LOL OmFG! Thats win right there lolol

By hw120 i meant hobbywing 120. I dont have a hawk unfortunately. If you want the hobbywing just pm me your address

[gouki74]

i was expecting only performance was to be tested but it's really good to see what's actually inside these esc's! thanks for the review! can't wait to see the others!

[ub3r]

Originally Posted by gouki74

i was expecting only performance was to be tested but it's really good to see what's actually inside these esc's! thanks for the review! can't wait to see the others!

Yea i thought ill explain the internals while i run my tests/edit the footage. That way we can see how the HW affects the actual real world performance.

I'm not getting into reviewing features like soft-start or governors. Just focusing mainly on HW for now.

Originally Posted by kimmik

LOL OmFG! Thats win right there lolol

By hw120 i meant hobbywing 120. I dont have a hawk unfortunately. If you want the hobbywing just pm me your address

Ok cool. Ill grab it off you in the next couple of days as soon as I'm done with testing these.

[Mercuriell]

Tony's review of the YGE 120 HV hardware is up

[gadawg58]

Awesome thread and thanks so much for taking the time to review these ESC's in such depth!

[jamin_00]

Excellent thread.

Subscribed

[swalko]

Great work .