MMPro Observations

[BrianG]

With Patrick's permission, I can now talk about the results from beta testing.

First, some naked pictures of the early testing unit I got:







Some highlights:

• As you can see, the MMPro looks like a MM/MMM hybrid: the wiring layout is similar to the MM as far as the power/motor wires and status LEDs, but the throttle/switch wires come out one side of the case, the fan and sensor ports are on the other end. The PCB layout is similar to the MMM in how the controller PCB is mounted to the power PCB.
  
  
• Motor and power wires are 13GA, just like the MM.
  
  
• The MMPro physical dimensions are VERY VERY close to the MM, just a little taller (due to the fan).
  
  
• Castle Link has all the functions/features of the MM/MMM, and adds additional features for the sensored functionality. I never tested sensored mode though.
  
  
• Allows sensored and sensorless operation. Sensored mode has a cheat function, but since I never tested this ESC on a sensored motor, I can't comment on it.
  
  
• Adjustable switching BEC. Sweet.
  
  
• Link uses the standard Castle Link adaptor that every other Castle ESC uses, except the MM.
  
  
• Rated for 6s and around the same current as the MM (~80A), maybe a bit more.
  
  
• Three 220uF surface-mount style caps; 660uF total.
  
  
• Fan is replaceable without having to disassemble the case.

Now, let's go over the FETs. Some common choices:

4108 - Used in a lot of ESCs still.
4833 - Used in the HobbyWing 150A and the Tekin RX8 IIRC.
4933 - New OnSemi FET with better power than the 4108 and better switching characteristics than the 4833 (more efficient switching action=less heat). I can't find the datasheet for the 4933, so I can't go into the specs. If someone could find it and provide a link, I would be grateful. The OnSemi website doesn't have it listed at all, nor do any of my normal datasheet resources.

Anyway, the MMPro uses 18 FETs; 15 under the heatsink, and three on the other side of the power PCB just under the throttle/siwtch wires. By comparison, the MM has 30 FETs and the MMM has 36.

The first beta ESC I got used the 4108N FET. Very common FET used in a lot of ESCs. The second unit I got uses the new 4933 FET, and I believe this is the FET that will be in the final product. Patrick can comfirm this.

When my first beta unit came in, I installed Deans and new bullet plugs to match the CC motor (the ESC came prewired with 4mm short bullets), and downloaded the beta Link software and hooked it up to check/change my settings. Since it was nighttime, I then ran it in my e-Jato conversion (CC4600, 3s lipo, geared ~50mph) in the basement to get a feel for it. Seemed to run fine except for this one spot where it went forward a few feet, stopped (lost all power), and then kicked back in, but ran fine the rest of the time. The Eagletree showed a large 105A current spike at that time (normal spikes ~60A). Temps after a gentle 15 minute run were pretty high I though, but I was doing a LOT of partial throttle running. The next day (or the day after), I took it outside for a while and ran around the yard doing high speed runs and general bashing. Ran fine except a few spots where it would hesitate when slowing down and then get back into the throttle. About 5 minutes into the run, the ESC started running oddly and a second or two later, the dreaded smoke came out.

A couple weeks later, my second beta unit arrived (using the 4933 FETs). Put it back in the e-Jato and ran a LOT cooler. Then ran it in the CRT.5 on 3s and it ran cooler than the MM in that as well. Of course, I'm sure the fan helps. Still had that occasional hesitation in mid-acceleration I had with the old version, and when I looked at the ET graph, noticed current spikes over 130A. But this time, the ESC didn't smoke. The new FETs are definitely better.

Eventually, I got a notice from Castle saying there was a hardware and firmware fix to address the hesitation/cogging. I installed the new firmware, but still haven't sent it back for the hardware fix. Regardless, the hesitation seemed to be gone, but it never was a problem I could intentionally reproduce anyway so not sure if it was just a coincidence. I plan on visiting CC next week while on vacation, so I'll bring the ESC for the hardware upgrade.

Since then, I've run it without the fan and it compares to the MM as far as temps are concerned, maybe a tad cooler. I plan to run it with the fan (for warranty concerns), but it's nice to know the ESC can remain thermally stable if the fan unknowingly fails.

Aside from the e-jato and CRT.5, I've also run this ESC in the following setups:

- Slash truck; Medusa 36-50-2200 on 4s geared for ~40mph. Ran perfectly fine with cool temps, which is to be expected given the vehicle weight/size/gearing and motor kv used.

- MBX5/Corr truck; Neu 1512/2.5d on 5s geared for ~45mph. Worked perfectly fine on this too, but that little fan got a workout! I expected it to the MMPro to get hot on this setup though, but it still held up fine. To be clear, I do not recommend running a setup like this on the MMPro, but it was my job to try to kill it after all.

All in all, I am quite pleased with this ESC.

[brushlessboy16]

Awesome read brian!!

Thanks for the insight

[jhautz]

Nice review Brian. Any pics of the unit side by side with a MMM and/or MM? Just looking to get a better idea of how big this thing is.

[micke_b]

Thanks for great info.

Have you identified what MOSFET DRIVERS CC are using?

[BrianG]

Originally Posted by jhautz

...Any pics of the unit side by side with a MMM and/or MM? Just looking to get a better idea of how big this thing is.

No side by side pics yet (am at work right now), but like I said, the MMPro is VERY close in dimensions, just a bit taller because of the fan.

Originally Posted by micke_b

...Have you identified what MOSFET DRIVERS CC are using?

The 4933 FET is in the latest beta unit I have.

[micke_b]

Originally Posted by BrianG

No side by side pics yet (am at work right now), but like I said, the MMPro is VERY close in dimensions, just a bit taller because of the fan.



The 4933 FET is in the latest beta unit I have.

Yes, but what DRIVERS are they using? The three ic:s on the bottom of your first picture I think.

[BrianG]

lol, oh. Not sure. I'd have to scrape the conformal coating off to find out.

[micke_b]

Originally Posted by BrianG

lol, oh. Not sure. I'd have to scrape the conformal coating off to find out.

Could you? ;-)

Also did you find out what MCU they are using? I know that the Mamba Max I have is using: C8051F320. But I think this one looks different. Do you have the part number for it?

[Pdelcast]

Originally Posted by micke_b

Could you? ;-)

Also did you find out what MCU they are using? I know that the Mamba Max I have is using: C8051F320. But I think this one looks different. Do you have the part number for it?

Hmmm, why don't you ask me, since I designed the MM Pro?

And why are you so interested in the particulars?

Since you can find all this out yourself with a little research (and I really don't care if you know--)

FET drivers are commodity items now, everybody makes one -- and they all work very well (except for a few notable exceptions...)

The FET drivers are Intersil ISL6700IB. The processor is a Silabs C8051F367 at 50 mips.

The MOSFET is an NTMFS4933 (about 1.3 milliohm.)
(the NTMFS4933 isn't yet released to the public, and datasheets won't be available online until late January, early February. Castle gets the first run of wafers.)
NTMFS4108, BSC014N03s and BSC019N04s are suitable substitutes we may use if availability becomes a problem. (we use a half-million FETs a month, so sometimes we have problems with availability...) There isn't a significant difference in performance with any of these other FETs -- BUT, we do have to change the drive circuit values.

BTW BrianG: The production version shows almost no difference in performance between these different FETs -- the original Beta had an issue with drive speed on the 4108s -- the 4933 version you got had the issue corrected. Once the drive is correct, the different FETs have almost no performance difference.

Anything else you want to know? The switcher chip is an LTC3824 (Linear Tech), with an AO4421 (Alpha Omega Semi) external FET. The main inductor is from NIC. We are using a novel circuit to give variable programmable output voltage on the BEC switcher (I'm not going to give you details about how that works -- figure it out yourself!)
We are using a variable gain back-emf detection circuit. (Figure that one out yourself too!)
The fan and sensor connectors are from JST, and are not chinese clones of the JST part.

We use a TPS76133 (TI) regulator for main processor power (3.3V) -- And there are four total power supplies on the board (including the BEC supply.)

That about does it. Now if you want, you can run over to Novak, or Tekin, or LRP, or HobbyWing, or whoever and let them know. Good luck.

Patrick

[micke_b]

Originally Posted by Pdelcast

Hmmm, why don't you ask me, since I designed it?

And why are you so interested in the particulars?

Since you can find all this out yourself with a little research (and I really don't care if you know--)

FET drivers are commodity items now, everybody makes one -- and they all work very well (except for a few notable exceptions...)

The FET drivers are Intersil ISL6700IB. The processor is a Silabs C8051F367 at 50 mips.

The MOSFET is an NTMFS4933 (about 1.3 milliohm.)
(the NTMFS4933 isn't yet released to the public, and datasheets won't be available online until late January, early February. Castle gets the first run of wafers.)
NTMFS4108, BSC014N03s and BSC019N04s are suitable substitutes we may use if availability becomes a problem. (we use a half-million FETs a month, so sometimes we have problems with availability...) There isn't a significant difference in performance with any of these other FETs -- BUT, we do have to change the drive circuit values.

BTW BrianG: The production version shows almost no difference in performance between these different FETs -- the original Beta had an issue with drive speed on the 4108s -- the 4933 version you got had the issue corrected. Once the drive is correct, the different FETs have almost no performance difference.

Anything else you want to know? The switcher chip is an LTC3824 (Linear Tech), with an AO4421 (Alpha Omega Semi) external FET. The main inductor is from NIC. We are using a novel circuit to give variable programmable output voltage on the BEC switcher (I'm not going to give you details about how that works -- figure it out yourself!)
We are using a variable gain back-emf detection circuit. (Figure that one out yourself too!)
The fan and sensor connectors are from JST, and are not chinese clones of the JST part.

We use a TPS76133 (TI) regulator for main processor power (3.3V) -- And there are four total power supplies on the board (including the BEC supply.)

That about does it. Now if you want, you can run over to Novak, or Tekin, or LRP, or HobbyWing, or whoever and let them know. Good luck.

Patrick

Hi, thanks for the Answer Patrick. Of course I've should have asked you, just took the chanse of asking here since BrianG made me curious with all the pictures. ;-)

Im asking since Im trying to learn how these things works and I have a goal to try and build a esc of my own, but based on a AVR MCU. Sice I really like your products and it seems like you guys know what parts to use I thougth I save myself that trouble. I still have a lot to learn, but Im getting there.

Thanks for sharing.

[Pdelcast]

Originally Posted by micke_b

Hi, thanks for the Answer Patrick. Of course I've should have asked you, just took the chanse of asking here since BrianG made me curious with all the pictures. ;-)

Im asking since Im trying to learn how these things works and I have a goal to try and build a esc of my own, but based on a AVR MCU. Sice I really like your products and it seems like you guys know what parts to use I thougth I save myself that trouble. I still have a lot to learn, but Im getting there.

Thanks for sharing.

The AVR MCU is a decent platform.

Building a good quality ESC isn't an easy thing. We have years of experience with it (Castle has been in business for 12 years!), and are still learning new things every day. Before starting Castle I worked designing avionics for Honeywell and Garmin.

A few things I didn't point out -- the main FET board is eight layers, six ounce copper each layer. Just tooling for a circuit board like that costs several thousand. You don't want to know how much we pay for each board.

The control board is .005" track/track, 2 oz copper, four layers. (that's an expensive board to tool too...)

The heat sink is a custom made for us by Aavid.

If you are planning to build your own ESC for your own interest, great. If you are thinking about getting into the business -- let me warn you. The margins are terrible, the competition is fierce, the distributors make you jump through hoops so they can make more money on products than you do.

Most of the manufacturers in this business don't do anything new -- they just wait for new products to come from their competitors, then they clone them, steal the ideas and designs, and then undercut your (already very thin!) margins. And that's just the western companies -- the asian companies steal your ideas, cheapen up the designs by using cloned parts, recovered parts, sub-par materials (whatever they can do to make it cheaper to manufacture.) There are asian companies selling ESCs in the air market that have selling prices BELOW my BOM costs... Think about that... Their selling price is lower than my material costs ALONE.

It isn't a business you want to get into unless you are doing it simply because you love the hobby. You won't get rich doing this.

[lutach]

Originally Posted by Pdelcast

NTMFS4108, BSC014N03s and BSC019N04s are suitable substitutes we may use if availability becomes a problem.

Patrick,

Fairchild has a 30V MOSFET FDMS7650 with 0.99 milliohm. Can that also be a suitable substitute if availability becomes a problem?

[Pdelcast]

Originally Posted by lutach

Patrick,

Fairchild has a 30V MOSFET FDMS7650 with 0.99 milliohm. Can that also be a suitable substitute if availability becomes a problem?

Nope.

The BSC014N03 actually works well. The Fairchild part has serious issues.

[bdebde]

Sounds like another winner from Castle, and great review Brian. Maybe I should have held off buying the MMM for my new 1/8th buggy, sounds like it would work great in there.

[glassdoctor]

I ran mine in my Fusion 1/8 buggy at a RC Pro Series race at Real RC in KC. I ran it with a Neu 1509 2y on 4s and it ran really nice and smooth, and temps (was about only 70* that weekend) were very good... similar to what I got with a MMM and what my buddy was getting on his RX8 system. Previously I ran a MM with Koolflight Ubec... tweaked to install very cleanly in the buggy without a mess of wiring. The MMPro was a dream to use.... the same clean install without any work.

I was planning to run it with a sensored motor in my 1/8 and also race it with sensored motors in 1/10 buggy, but mine didn't survive the weekend at the big race. I believe it blew a brake fet... died coming into a corner hard on the brakes. So my MMPro went back to the bat cave with Castle Mike. I finished the weekend with a MMM, and it's been in my car all summer since then. I didn't get a second gen beta unit, so that's all the experience I got with it.

It's an awesome little speedo and I can't wait to get another one. I think it will be good for a light 1/8 buggy that's set up for racing, which means not overpowered 50mph type of setup. In that case, the MMM would be a better choice. The MMPro size fits AWESOME in my Fusion, and I don't run a crazy fast car so it's perfect for me.

Although.... after running a MMM in my car all summer, I have grown to like the install and look of it in the car. But when the MMPro arrives, the MMM won't see it coming