Dshot question

[Hozza]

So how can you tell if your esc';s are capable of running dshot600?

[Cheredanine]

Well depends on what escs mate:
Blheli32 - all are
Kiss - depends on firmware version
Gemfan - ditto
Blheli_s - depends in chip, bb1 can not, bb2 can (very few bb1 were made) and also age, early designs that predate dshot have a cap on the signal line that can smooth it to unreadable, you can just remove he cap and you are fine
Blheli and older - nope

[Hozza]

I know they are BB2 and when I put fc to dshot600 the motors spin up in betaflight, just not sure if they have the signal cap.

[Cheredanine]

If they spin up then you are fine mate. It has been a long time since anyone put the cap on the signal line

[Hozza]

Ok cool, just wanted to make sure that nothing would burn out,it';s not my quad just replaced a fc for a friend.

[Cheredanine]

No mate, not gonna burn anything either way

[mo_miah]

mines running oneshot, i kinda understand the theory behind dshot, but what is the difference in how the quad flys between oneshot and dshot?

[atomiclama]

Speaking from my poor flying ability  ::) The only difference I noticed was due to the uniformity of the control. Not having to calibrate the ESC with dshot gave a more leveled response on punch outs and drops. Even with calibrated ESCs my hex has a slight change in attitude when I punch or drop.  This wasn't there in dshot. But my ESCs can't handle dshot, I get glitches/ticks in the motor.

[Cheredanine]

Fundamentally, PWM (pulse width modulation) signals including one shot and multishot are an analog signal.
They are transmitting a value from 1000 to 2000, they do this through timing, the signal wire goes high (5v) for and ammount of time and this defines the value
For PWM 1ms = 1000, 2ms = 2000
Oneshot125 125us = 1000, 250us = 2000
Oneshot42 42us = 1000, 84us = 2000
Multishot 7us = 1000, 25us = 2000

Note you need quiescence between each signal to make it distinct (I.e your signal wire needs to go high to start and low to stop the signal, defining its length, you need time it remains low between each signal, last time I measured on OS42 this was 11us)

There are a number of issues with this approach, these are as follows;
1. Particularly the slow protocols limits the speed with which you can update the motors, take Oneshot (oneshot125) sending a value takes 250us + 11us quiescence, so 261us (millions of a second)
This means the fastest you can run pid loops, or more accurately update the motors with the output of a pid loop) is about 3.8khz. I.e. you can’t run 4khz pid loops without causing massive problems at the top of throttle.  We want fast pid loops because that is how you get smooth flight
2. Timing, look at multishot, we have 18 millionths of a second to define 1000values, the difference between each value in theory is 1/18000000000 of a second, slight differences in timing between the flight controller and ESCs means you get inaccuracies h(this why you calibrate - to get rid of instead timing differences) even when you have calibrated, when you switch the signal wire voltage (up to 5v and down) this takes a finite time, if other things are pulling voltage, the time it takes is variable, again introducing inaccuracies.
3. Granularity - in order to run the faster pwm protocols and not get completely nerfed by point 2, the steps are made bigger than one value, I.e. if your throttle changes from 1000 to 1001, the motor speed won’t changed, it has been a while but I think os42 uses 4 points as one step from memory (it is easy enough to find out, use the motor tab to increase the value sent to a motor one point at a time and listen to change of pitch) that means instead of 1000points of throttle, you have 250 and the difference between each of them is four times what it should be) multishot will be worse
4. Validation - there is no validation on pwm. The esc sets the motor to what it thinks the F.C. says without any check

Dshot - dshot is a digital protocol. The number following it is the frame rate (number of Kbits per second).
Dshot is a 16 bit signal, so 16bits are used to transmit one signal, this includes not only a numerical value between 0 and 2000 for the throttle making the granularity not 1000 (which as above the faster PWM do not actually manage) but 2000, I.e. the throttle value is 8times more accurate than oneshot42 (assuming my recollection of the stepping value is correct)
Because a 16 bit signal can carry far more than 2000 values it can also carry pre defined commands, and the dshot protocol include 2 parity bits, which confirm if the signal is correct, allowing the ESCs to identify and disregard erroneous or damaged signals.

Dshot therefore provides a more accurate, more reliable signal, which does not further compromise granularity or accuracy further still for faster speed esc updates

That is not to say you will notice a difference, it is perfectly possible to get a perfectly smooth flying quad on oneshot, but in most cases dshot will be smoother, the question is how noticable is the difference.

I know pilots who have switched and didn’t notice a difference, but I know many who switched and did. It depends on how good your quad, the build, it’s components and your soldering is - essentially all things that effect pwm, and how good a pilot you are (do you feel the difference in throttle granularity)

For my money the biggest difference in feel recently is blheli32 allowing you to up the motor pwm to 48khz

But in general, use dshot, it is much more throttle granularity (which the F.C. will notice even if you don’t) and it makes it far more likely you will have a smooth running quad, there is no down side, only up

[hoverfly]

Thank you , that has cleared up a couple of things i was pondering over. 

[mo_miah]

so a nice new shiny helio spring with some blheli32 esc running dshot 1200 and 32/32 will make my quad smooth as butter compared to my current SPF3 and blheli oneshot125 combo :)

did want a complete new chameleon build but think i need to save money, this rc hobby isnt cheap