Originally Posted by flashervn

Hi Simonk,
A friend of mine confirmed that the RCTimer 40A I am having are using internal crystal ( not sure its the same with the "internal oscillator" you are talking about). I tested again today and just like you ve described it, the jerky motor at 10% throttle only happened the first time. As the ESC heat up, its gone. So I think I am experiencing the "internal oscillator drift" you were talking about. Is there anyway to fix this and does this mean I should not use this ESC with your firmware as it wont work? Thank you.

Could you post a video of the problem? It probably has nothing to do with the oscillator unless it's just jerky because you're right on the edge of on/off at 10%. If 9% stops it and 11% is fine, then that's all it is; otherwise, it probably has to do with the FET switch timing, which newer code might improve, but that would be strange.

Guys,
I opened up another RCTimer 40A v2. In the back I found all FETs are with same marking(1718 BA1P4R)

The one I opened up before and flashed with tgy.hex had a mixture of (1718 BA1P4R) and (1718 BA1N35)

Originally Posted by simonk

Could you post a video of the problem? It probably has nothing to do with the oscillator unless it's just jerky because you're right on the edge of on/off at 10%. If 9% stops it and 11% is fine, then that's all it is; otherwise, it probably has to do with the FET switch timing, which newer code might improve, but that would be strange.

Hi Simonk,
Here is the video

20111215 171147 (1 min 7 sec)

Please ignore the noise. Its not from the motor but from the wooden base i attached it to

Originally Posted by simonk

It needs some pictures to really make sense (there are three basic layouts: P-FETs and N-FETs with NPNs on the P-FETs, all-N-FETs with NPNs on the high (originally P) side, and all-N-FETs with gate driver chips -- usually 8-pin things per phase). But for now, I pushed some updates I was working on to the README here: https://github.com/sim-/tgy (scroll down)

not so obvious. Sorry
Could one make annotations on these images to explain the layout?

What i found so far:
6679Z = MOSFET P-CH 30V 20A 8-SOIC
F7832 = MOSFET N-CH 30V 20A 8-SOIC
AMS1117 = 800mA LOW DROPOUT VOLTAGE REGULATOR

Thanks,
manu

Originally Posted by flashervn

Hi Simonk,
Here is the video

Hello! Ok, because the 8KHz PWM is still present, that means it is not exiting startup mode and "goodies" keeps getting reset. Which version is this? It is possible that it is a FET timing issue and maybe newer versions have improved it, but I haven't uploaded a build yet, just pushed the source code. I will get around to testing shortly and upload another version.

Originally Posted by simonk

Hello! Ok, because the 8KHz PWM is still present, that means it is not exiting startup mode and "goodies" keeps getting reset. Which version is this? It is possible that it is a FET timing issue and maybe newer versions have improved it, but I haven't uploaded a build yet, just pushed the source code. I will get around to testing shortly and upload another version.

Hi Simonk. This is the TGY.hex extracted from the tgy_2011-11-10_d3567e61.zip. I ll wait for your new release as I am not at all confident to mess with anything from the code. Cheers!

Originally Posted by alll

not so obvious. Sorry
Could one make annotations on these images to explain the layout?

What i found so far:
6679Z = MOSFET N-CH 30V 20A 8-SOIC
F7832 = MOSFET N-CH 30V 20A 8-SOIC
AMS1117 = 800mA LOW DROPOUT VOLTAGE REGULATOR

Thanks,
manu

Wow, that is some awesome soldering there.

FDS6679Z is actually a MOSFET P-CH 30V 13A 8SOIC, not N-CH. So, that board has one of the simple original designs, with three P-channel FETs and three N-channel FETs. In this arrangement, the P-channels always turn on/off the "high" side (eg: battery voltage) to each phase, and the N-channels always turn on/off the "low" side (ground) to each phase.

These N-channel FETs are "logic level gate", which means the gate mostly turns on at 5V (usually more at higher voltage, but 5V is enough). So, controlling the N-fets is easy from the MCU (AVR), because it already supports driving pins low (0V) or high (5V) with pretty good current. Most boards usually just have a low resistor (around 330R) on each N-fet gate pin, and then run that directly to the AVR. This slows the gate charge/drain slightly to avoid too much current at the drive pin, but the gate can usually still switch within a few microseconds. Also, the resistor can sometimes prevent blowing up the AVR if the gate shorts to one of the other FET pins, which almost always how they fail.

The P-channel FETs turn "on" by pulling down the gate from the source voltage (eg: battery voltage), but this is higher than TTL voltage, so they use an NPN transistor and resistors to accomplish the switching. These are those three parts on the left of your 5th picture which look like this: http://search.digikey.com/ca/en/prod...FSCT-ND/458971 The AVR pin is usually held low (0V) to the base of the transistor, not moving any current, causing the resistor from battery voltage to charge up the gate to match the source and stay there, making the P-ch FET turn OFF. When the AVR pin turns on, current flows from the AVR, through a limiting resistor, and out the transistor emitter to ground, and this current is amplified by the transistor, which drains the charge at the FET gate. The NPN and resistor values are chosen so that there is a pretty high discharge and charge current at the gate so that the the voltage drops to nearly ground relatively fast. If it's not fast enough, the FET acts as a resistor as it turns off/on and heats up, wasting power. The high side is usually not used for PWM and only for commutation, so it doesn't switch as often. It's not uncommon for the NPN transistors to heat up to 50-60°C whenever the motor is spinning, however.

So, that's how the FETs are driven on that board. You can see a similar schematic in the first post here: http://www.rcgroups.com/forums/showthread.php?t=766589

Medium-current boards with all-NFET designs do exist without dedicated gate driver chips. They're almost the same, but they have an extra diode butted to a capacitor on each high side N-channel FET to make the gate follow the source but filter the DC, allowing it to be pulled down by the NPN (when driven high by the AVR) or charged up to high (turning the N-ch FET ON) via the diode when the AVR pin is low. These boards start with the high-side FETs on until the AVR boots and sets the pins high to turn them off (the drive is inverted). If you watch the output voltage on any phase on a voltmeter, it will match the input voltage until the voltage rises enough for it to boot. The HK F-30A ESC has this design. Pictures: http://0x.ca/sim/esc/hk_30a_esc/. Similar "Mystery-nFET" schematic here: http://code.google.com/p/wii-esc/sou...c%2FSchematics That is part of ziss_dm's "wii-esc" tree, which you can browse at that URL.

High-current boards typically all have dedicated gate driver chips and no NPNs. These are usually three 8-pin chips dedicated to the task of turning on and off all N-channel FETs as nicely as possible, and sometimes there's an external voltage pump to get the gate voltage up to higher levels. The Mystery 40A, RCTimer 50A, HK F-60A, etc., have these, and look similar to this board: http://0x.ca/sim/esc/rctimer_50a_esc/. That same schematic directory has a couple like that as well. Some of the FET drivers like inverted inputs, others don't. It helps to check the AVR pins that run to the drivers on the original firmware.

So, that's it for FET/driving configuration. The rest is just the sense feedback, PPM input, etc. Those schematics all have a star-configuration voltage divider to establish a neutral and voltage dividers for each leg. All that matters after that is which pins they are connected to. You can usually recognize these sections on the board as three groups of similar resistors. Then you just have to figure out which ADC/AIN1 pins they go to.

The other chips on the board are for supplying 5V to the AVR and/or to the BEC output. Most usually use separate circuits for each, or don't have a BEC, etc. They're more commonly than not 7805-based parts (linear regulators) for the BEC, and then some lower current soic-8 5V regulator for the AVR (and thus FET driving). The chip in your second picture has "5.0" written on it (5V), which is a pretty good give-away.

Btw, the discussion thread from ziss_dm on the wii-esc tree is here, and also has some interesting info if you want to skim through: http://www.multiwii.com/forum/viewtopic.php?f=13&t=516

I scrawled over your board image with what I can pick out, but I can't tell where some of the vias go without testing. However, it looks to me like the "tp" (type 1) arrangement. The first clue is that the sense lines all run to ADC2,3,4 pins and not AIN1, which only tp (and now rct50a) have, so that excludes most of the other common arrangements. Then, the FET pins seem to be all on PORTD, also like tp.inc. Does this make sense?

Are you sure it's not just inaccuraccy in the RX/TX combo? It's not 100% precise - mine does the same (two rx/tx sets, turnigy plush, motors do not matter) - but it's just jumping from "throttle range" to "idle range" at one point, so it's constantly restarting.

Originally Posted by flashervn

Hi Simonk,
Here is the video
http://youtu.be/mh7Ri4AgKcc
Please ignore the noise. Its not from the motor but from the wooden base i attached it to

Originally Posted by zviratko

Are you sure it's not just inaccuraccy in the RX/TX combo? It's not 100% precise - mine does the same (two rx/tx sets, turnigy plush, motors do not matter) - but it's just jumping from "throttle range" to "idle range" at one point, so it's constantly restarting.

Hi, I ve ran it through both Rx Tx straight and through my FC which is a DJI WKM, same result. The one I posted in the video was through the FC

It is not accurate even directly from RX, so it will not be accurate from FC (and FC is doing it's own magic, so you can't know what it puts in here). It might be more accurate if you generate PWM directly on FC with a constant width, is that what you are doing?

Simonk i am embarrast you spent your time on my post already. I was prepared, based on your "readme", to further provide additional informotion on my post...
I still will, so others could also start doing it for other esc-layouts.

Again 1000x thanks
manu
PS: i really want to tame this esc!

I see a 8mhz oscillator there, make sure your fuses are set to intrc, so that it can actually be overclocked. I'd imagine running the code at 8mhz osc wouldn't have the proper effect.

Hey guys,

Just wanted to say that I flashed the original v1 firmware to my 30a RCTimer escs. It works great. The only thing I miss is that it doesnt have a 'soft start' or whatever. On my big 750kv motors, I have to slowly spool up the motors or else they cant seem to get turning. This is not bad as it kinda is a sort of saftey feature to prevent a sudden launch from the ground, but just thought I would mention it. Once in the air all is fine of course.

I know this sounds lazy, but I'm going to pull the new dad, new job excuse and ask: does anyone sell a 40amp esc already flashed? my turnigy plush 40's are great, but I'd love to try a flashed version.

Is anyone using a chip clip type hookup to the atmega? Got a nice little new esc with atmega and 16mhz resonator but no place to connect to other than soldering to chip pin. Can do it just trying not to