VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Report your hardware problems here. Include as many details as possible with pictures to get the best possible help and feedback.
Brushless Devil
Posts: 3
Joined: 31 Mar 2017, 12:27

VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby Brushless Devil » 31 Mar 2017, 14:35

Hello everyone,

First of all, to congratulate Benjamin Vedder for this outstanding project.
Very,very nice opensource work :) !

I'm a mechanical engineer, and do electronics for passion.
I'm in the hobby scene, since mine 10 years old, and i'm 35.
I've designed and built RC boats since i remember,and RC helicopters.

When i saw this outstanding project, i thought to mine self...I have to do it...
Ordered 2X 4.12 hardware boards and started soldering.

Mine boards are not complete, and for inimaginable for what might seem , i can´t get the "reversed" USB connector..
I live in Portugal, Europe, and the only house that sells them is Farnell, and charges 20€ shipping for 1€ connector.
I refuse that..Prefer to use "standard" mini usb connector and make a "mirror" pin from a standard USB cable...


To everyone, and to share what i know (i have built mine first Brushless speed controller) 16 years ago..
First..mine programming know how is very limited..I know how to edit and compile files for Atmel and Pic micro controllers, but stays for there...It´s out of mine league..I know how things works..I understand algorithms, but codding that to micro controller language is something i don´t know how to do.

But let explain everyone something about mine initial experience in assembling and understanding VESC:


DRV 8302 - In mine humble opinion is not operating in the "optimum range zone", and that might lead to some of the problems all off you might have.
This chip has different and Independent sections inside the same chip.
The buck converter is independent of the gate drivers and current amplification.
The buck converter "inside" the DRV8302 is a TPS 54160.
It provides a Max output of 1.5A, with an input voltage from 3.5V to 60V and an output voltage from 0.8V to 58V.
In vedder esc is delivering 5V for providing power to the 3.3V voltage regulator (TC2117).

According to TPS54160 Webench TI Simulation http://www.ti.com/sitesearch/docs/unive ... 0#linkId=1 Assuming :

Vin Min (4S Battery) Almost discharged 4x3.3 volt= 13.2V
V Max (60V) the limit of DRV8302 = 60V
Vout - 5V the same in VESC
I Max OUT (DRV8302 Maximum 1.5A) = 1.5A

The lower bound result for the L1 inductor is 47uH ...That is assuming a compromise between "footprint" and efficiency , that is 84%.

If we demand maximum efficiency, wich is the possible for TPS54160 we get 88% and a L1 inductor of 120uH!!!

If i try to select 22uH for L1, the design doesn't even let me, because it's "out of range" considering the voltages input!
Some people say with having problems of the DRV8302 smoking at 12S, this could be a reason..Don't know!

Of course that changing the L1 inductor values automatically, according to the VESC schematic, R19,R21,C19 and C21 have to be changed, since these components control the "clock" of the buck converter.


Mosfet Selection.

In Mosfet Selection we have 2 primary losses:
- Conduction
- Switching

In condution mode the most important value is Rds on, wich is the resistance that the MOSFET has fully conducting

In switching mode, the most important factor is gate charge, wich is directly connected to PWM and switching frequency..The MOSFET Builds UP heat when changin from ON to OFF and from OFF to ON state.

Almost 95% the lower the Rds on of the MOSFET, the higher the gate charge of the MOSFET.
Mosfet are controlled, not by "voltage" like most people thinks of, but by "capacitance".
To Turn ON the Mosfet the gate has to be charged and this parameter is seen on the MOSFET data Sheet has "Total Gate Charge" wich is expressed in nC (nano Coloumbs).

The higher the switching Frequency, and depending of the PWM, the higer charge/second has to be accumulated, and delivered to the MOSFET gate for it to Turn on, and to Turn it off.
..Yes..because charge has to be delivered to the gate to turn the Mosfet On, but it has also to be discharged in order to turn it off...
If you never tried, if you touch with your finger in the gate of the Mosfet, it will start conducting,even if you take your finger from the MOSFET gate....And only after connecting the gate to GND, for discharging it, the MOSFET will Turn OFF.

Why we are limited,concerning the max number of mosfets we can use,paralleling mosfets, and increasing switching frequencies and FOC mode!??...Because the gate drivers don´t have enough juice for adding more mosfets, or limit our MOSFET selection because of the low mA it can sink for the gate of the MOSFETS!

Why some of us have seen in other forums, someone "asking" and paying for someone to change that part of the driving stage, for eliminating the DRV8302 and getting discret components that to the same, but have more "power capabilities"?..That's the reason why.


Assuming an switching frequency of 20Khz, wich i think from what i read is the switching frequency of the VESC,and that rise time (Time it takes from the MOSFET full OFF to FULL Conducting) is 0.4% of the PWM period we get:

dT = 1/ 20 KHz * 0.4% = 200 nS (200 nano seconds rise time)

Let's assume the Total Gate Charge of the VESC BOM MOSFET IRFS7530 wich is:

Typical Total Gate Charge = 236 nC
Max Total Gate Charge = 354 nC

I'm not going to the worst case scenario, even though , "Dimensioning calculus in Engineering" is always worst case scenario plus security margin :)..

354nC + 236nC / 2 = 295 nC

I (Amp) = 295 nC / 200nS = 1,475 Amp...yes you have read correctly. :shock: :shock: :shock: :shock: .That´s the Peak amp draw that has delivered to the gate of the IRFS7530 MOSFET, for turning it from full OFF to Full ON in 200nS..

Of course that reducing the switching Frequency, will decrease the peak amp draw, and the same applies to the MOSFET gate charge.
If the gate charge is lower it needs less amps, for turning it ON in the same space of time.


For everyone what wants to try an alternative MOSFET for the VESC that has the same voltage limits (60V) and have a lower Rds ON and lower gate charge, i suggest the FDBL0110N60 wich you can get at MOUSER and other companies, and it's made by Fairchild Semiconductors.
It´s the MOUSER product page http://pt.mouser.com/ProductDetail/Fair ... GFGg%3D%3D

It have a Typical Total gate charge of 170nC versus the 236 nC of the IRFS7530 and a maximum of 220nC Vs 354 nC.

170nC / 236nC = 0.72 = 1 - 0.72 = 0.28 * 100 = 28% Lower (Typical)
220nC / 354nC = 0.621 = 1 - 0.621 = 0.379 * 100 = 38% Lower (Maximum)

The package is different from the D2PACK 7 Legs of the IRFS7530, but can be easily replaced, because the foot print of the legs is 95% the same dimensions.
In the VESC is perfectly viable.

Other thing concerns cooling them.

The FDBL110N60 has a slightly lower Junction to Case Thermal Resistance, then the one of the IRFS7530, altough the IRFS7530 has better junction to ambient thermal resistance then the FDBL110N60, because it has "more exposed area for dissipation".



The VESC speed controller with a 200AMP "continuous" amp draw, and i mean "continuous" for a RC boat, a RC car, or RC Heli ..4 to 5 minutes, would be the dream of any RC hobby passionate.

This is, in anyway to criticize what has been made, because is outstanding, but only to show some trades that could be improved, and some alternatives and options!

If i could ever develop a project like Benjamin Vedder did....Never in mine life ;) !

But if i can contribute...Yes i can!

Hope all of you enjoy, and share your opinions.

Best Regards ;)

rew
Posts: 937
Joined: 25 Mar 2016, 12:29
Location: Delft, Netherlands.

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby rew » 01 Apr 2017, 12:35

Thanks for your contribution. I think you have a good grasp of what's going on. I knew that there was a separate die for the step-down part of the DRV8302, but I didn't know its part number. Thanks for finding that.

You write:
Of course that reducing the switching Frequency, will decrease the peak amp draw, and the same applies to the MOSFET gate charge.
where you say "switching frequency" you mean increase switching speed. (the peak amp draw does not depend on how many times per second you do that).

When you reduce the switching time, it becomes possible to reduce the deadtime. And reducing the deadtime will also improve the switching losses.

When the top-side mosfet turns off, the CPU delays turning on the bottom side FET for a small time, to make sure that the top side fet is turned fully off before the bottom one starts conducting. If there is still current flowing through the motor, once the topside mosfet turns off, will cause the bottom mosfet diode to start conducting. So you'll see a 0.6V negative voltage on the drain of the bottom mosfet during this time. Tuning this parameter will improve switching losses as well.

Oh, one more thing. When you divide the charge by the time to move that charge, you do not get the peak current. If your current waveform is precisely a block, then your peak will equal your answer. The peak is AT LEAST what you calculated, and usually more because the waveform cannot be precisely a block.

i82
Posts: 2
Joined: 06 Apr 2016, 08:05
Location: Madrid

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby i82 » 07 Sep 2017, 10:57

Sorry to be so newbie, I have 2 VESC 4.12 pcb to be assembled and have yet to buy all the components.
I already have an Enertion vesc for my single motor setup longboard, but I had overheating issues, could not climb any hill and replaced it for a 150A ESC (that makes me climb any hill, but general response has nothing to do against VESC).

So, I still have these 2 pcb to build. I would love to use directfets and big dissipators, but since my PCB's do not allow for that, I'm looking for the best FET I can use. Would the proposed FDBL0110N60 fets improve over the "standard" IRFS7530 ? Would there be a better option?

I ordered some copper dissipators for my current VESC's FETS so I will try to alleviate the heating issues and use it again, but I have limited hope. I am thinking even on changing the FETS on this VESC if there's a clear benefit to be gained.

Thank you for your answers.

rew
Posts: 937
Joined: 25 Mar 2016, 12:29
Location: Delft, Netherlands.

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby rew » 08 Sep 2017, 19:36

Once you go above say 50A, there is no way around dissipating significant amounts of power in the mosfets. At an ideal 1mOhm FET resistance, you'd have 50mV, or 2.5W in the mosfet. With the configuration of the VESC, that's about the (long term) limit. The reason is that the FET will dissipate to the PCB and there is another mosfet on the other side instead of a heatsink....

ThierryGTLTS
Posts: 85
Joined: 09 Aug 2017, 11:10

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby ThierryGTLTS » 09 Sep 2017, 09:44

I've just discovered that the DTC is fixed by a resistor between Pin 7 and GND.

According to datasheet of DRV8302 (and DRV8301 too), you have 50nS with 0 ohm, and 500nS with 150K, and a linear derating between those values.

So with a 10K, you set it to 80nS :!: :!:

When you take the IRF7530 datasheet, and you have a look at the turn On/Off Dela,y and rise/fall time (page 3) you have larger figures.

Brushless Devil has made calculation with 200nS, and we need about 1.5A of drive current.

Perhaps we have to increase the DTC to be more safe.

Have a Nice W-E.

Thierry

pf26
Posts: 293
Joined: 28 Mar 2016, 14:37
Location: FR Valence

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby pf26 » 10 Sep 2017, 09:45

@i82: some people did manage to cool the VESC4 Fets at way higher amps. http://vedder.se/forums/viewtopic.php?t=622#p3773 !
I am quite confident the VESC4 could feed the motor with 80-100Amps continuously, simply using heatsinks directly pasted to the current FETs and very good ventilation. Still not a perfect solution because of the high thermal resistance of the Fets, but there is not so much heat to be dissipated either.

@Thierry: there is an additional dead time configured in the STM32.
I think the rise time of the FET gate voltage is a consequence of gate resistor and possibly also of the driver pin and power supply internal resistances / decoupling. It would be interesting to measure the voltage at gate resistors to see the actual max current deliverred to the gates. With 4R7 gate resitors, it might exceed 2A at the very beginning of gate charge.

ThierryGTLTS
Posts: 85
Joined: 09 Aug 2017, 11:10

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby ThierryGTLTS » 10 Sep 2017, 14:00

@pf26, thank you for your quick answer.

Do you know what's the additional deadtime programmed in the STM32 :?:

I don't need 150A pk, so I wanna change the MOSFET by TI's one (csd19536) with much lower Qt.

That's why it could be interesting to change the deadtime to fit those FETs.

Have a Nice W-E.

Thierry

User avatar
Ecyclist
Posts: 30
Joined: 24 Jun 2017, 01:26
Location: San Diego, CA

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby Ecyclist » 28 Dec 2017, 21:48

Thank you all for sharing your knowledge I read this thread a couple of times and learned a thing or two, but I'm still going to ask you a simple question.
Can I use heatsink with proposed FDBL0110N60 MOSFET or I will have the same problem with cooling as with your typical VESC MOSFET?
Are there other options if comes to that?
Is it difficult to desolder FETs from VESC?
I would like to be able to pull about 70A continuously out of VESC and have it inside of the case protected from the elements and no forced air cooling.
I'm a mechanical engineer and just peeking into electric or electronic worlds. At this point, I made some progress and built pretty decent battery pack.
Battery case 1.JPG
Battery case 1.JPG (82.18 KiB) Viewed 1206 times

Battery case 2.JPG
Battery case 2.JPG (90.65 KiB) Viewed 1206 times

Battery case 3.JPG
Battery case 3.JPG (101.81 KiB) Viewed 1206 times

User avatar
TheFallen
Posts: 79
Joined: 26 Oct 2016, 10:49
Location: UK
Contact:

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby TheFallen » 29 Dec 2017, 00:39

Pretty much the same problems. That MOSFET isn't that much better, it and the original have roughly the same on resistance, gate charge and thermal properties.

if you're only after 70A have you looked into this style of VESC: https://www.ebay.co.uk/itm/VESC-New-4-1 ... 2975246657

User avatar
Ecyclist
Posts: 30
Joined: 24 Jun 2017, 01:26
Location: San Diego, CA

Re: VESC DRV8302 Problems MOSFET Swapping Considerations to all!

Postby Ecyclist » 29 Dec 2017, 02:55

TheFallen wrote:Pretty much the same problems. That MOSFET isn't that much better, it and the original have roughly the same on resistance, gate charge and thermal properties.

if you're only after 70A have you looked into this style of VESC: https://www.ebay.co.uk/itm/VESC-New-4-1 ... 2975246657

Thank you for the link. I was not aware of that model of VESC.
I would be happy with 60A continuous, but I would like to have a little headroom. It looks like the VESC on eBay is a little better, but...
Is it difficult to desolder FETs from VESC? Sticking a copper plate under three top FETs and heatsinking them might do the trick. Is this a crazy idea?


Return to “Hardware Problems”

Who is online

Users browsing this forum: No registered users and 0 guests