FLIPSKY new 20s 100A tiny controller (vesc based)

[atkforever]

Guys I'm planning modding this controller in order to increase phase current.

IMO 120A is weak for a 100A DC rated controller, I mean it's less than its little brother flipsky 7550 that is capable of 160A burst.

So I've identified a mosfets that could suit that application :

IPP026N10NF2S : https://www.infineon.com/cms/en/product/power/mosfet/n-channel/ipp026n10nf2s/

100V vds and 180A id, with equivalent rdson.

Is it a good choice and does it need other mods (resistors?).

Software needs to be edited ?

regards

 

[Vbruun]

I think youre down the wrong track here. The controller only uses 6 mosfets and the current is almost at the package limit. I think you are Better off designing another controller or buying another controller.

 

[badgineer]

So I've identified a mosfets that could suit that application :

IPP026N10NF2S : https://www.infineon.com/cms/en/product ... 26n10nf2s/

Still a TO220. weirdly I could not find the package limited drain current in the datasheet at first glance, but I still think it's there since it's a TO220 limitation, afaik. Anybody? )

BUT, probably a lot more modding is necessary.

VESCs measure the phase current, so they can limit it. They can't limit it higher than what they can measure. Probably the implementation will not be able to measure much more than 120 by design, so you would need to mod / change for example the shunts. (current is measured by measuring the voltage drop over the shunt of known R, afaik)
This is just an example... probably there's more to do, but I have zero hands on experience with vescs, just looked at the schematics for 10 minutes

So my 2 inexperience cents: Don't expect an easy job with modding the thing for more phase amps.

Br,

IMO 120A is weak for a 100A DC rated controller,

Yeah, but it's good for a FOC capable 100$ (+tax & shipping) controller. At least Imho.

Just the "no firmware upload" is a serious bummer.

 

[atkforever]

I think youre down the wrong track here. The controller only uses 6 mosfets and the current is almost at the package limit. I think you are Better off designing another controller or buying another controller.

for the price modding this controller could be a great option.

"the current is almost at the package limit" ; we identified the first bottleneck for the 120A which are the fets, we still don't know what's inside

Just the "no firmware upload" is a serious bummer.

yeah I love the way they lock an open-source project :lol: chinese probably not communists ?

 

[badgineer]

but I still think it's there since it's a TO220 limitation, afaik. Anybody? )

As my physics teacher in high school used to say, "fist think, then ... shut up!"

CSD19536KCS - TO220 100V ESC, good RDSon and total gate charge, and explicitly stated 150A package limited drain current.
So no, not all T0220 are the same.

 

[badgineer]

yeah I love the way they lock an open-source project chinese probably not communists ?

Actually this is more capitalism in its most extreme form (f*** rules! let the market decide!) than communism. This is coming from somebody who lives in an ex-communist country (Romania).

Br,

 

[mxlemming]

but I still think it's there since it's a TO220 limitation, afaik. Anybody? )

As my physics teacher in high school used to say, "fist think, then ... shut up!"

CSD19536KCS - TO220 100V ESC, good RDSon and total gate charge, and explicitly stated 150A package limited drain current.
So no, not all T0220 are the same.

Don't believe those ti specs. Fiction. Zombies did some serious tests on to220... Iirc 75A resulted in high temperature rises. I blew the legs off the ti csd19536 TO263 version.

The firmware will be updateable. Even if they don't release the source, it would take barely an afternoon for me to reverse engineer a VESC header file for it. They've released everything else for the other ESCs so I can't imagine they won't for this.

 

[badgineer]

Don't believe those ti specs. Fiction.

Hi. Thanks for the warning. I've read that in a few other threads on this forum (mostly something like "datasheets are made to look good. Take them with a grain of salt"). And, even if the limit is actually *really* the one stated, it's probably still not a good idea to run things at the limit, at least not while also expecting reliability.

However, I was just retracting my statement (that all TO220 have the same Package Limitation). Which, in hindsight was a stupid thing to say.

3 low side shunts is a good solution. Easiest way of getting 20+s voltage capability.

About this: I was looking at the schematic the other day, and... would this not limit the information the VESC gets about 1 phase to just 50% of the time (when the low side FET is open), as opposed to the Shunts located on the actual Phase wire, where you get 100% of the time information about what the current is up to? Would this not decrease the potential of the VESC for achieving good motor control by starving it of 50% of the phase current information?

And, a bit more off topic, is this the max opamp common mode voltage the only limitation in the 75V VESC? If yes, assuming we design and build an ESC around the 75 300 schematic, replacing the 75V FETS with say 100V FETs get us a max 90V ESC( as limited by the max common mode Voltage of the opamp?)

 

[mxlemming]

Don't believe those ti specs. Fiction.

Hi. Thanks for the warning. I've read that in a few other threads on this forum (mostly something like "datasheets are made to look good. Take them with a grain of salt"). And, even if the limit is actually *really* the one stated, it's probably still not a good idea to run things at the limit, at least not while also expecting reliability.

However, I was just retracting my statement (that all TO220 have the same Package Limitation). Which, in hindsight was a stupid thing to say.

3 low side shunts is a good solution. Easiest way of getting 20+s voltage capability.

About this: I was looking at the schematic the other day, and... would this not limit the information the VESC gets about 1 phase to just 50% of the time (when the low side FET is open), as opposed to the Shunts located on the actual Phase wire, where you get 100% of the time information about what the current is up to? Would this not decrease the potential of the VESC for achieving good motor control by starving it of 50% of the phase current information?

And, a bit more off topic, is this the max opamp common mode voltage the only limitation in the 75V VESC? If yes, assuming we design and build an ESC around the 75 300 schematic, replacing the 75V FETS with say 100V FETs get us a max 90V ESC( as limited by the max common mode Voltage of the opamp?)

90V is the absolute max at which it'll fail. Ringing and switching noise are substantial with through holes at 120A and will push an extra load of voltage above your bus voltage and regen can add well.

Re. The FOC, you don't need that many samples. The current measurements are just for feedback, regulation... The algorithm will run many many pwm cycles without a current measurement before error occurs as long as your motor is sufficiently sinusoidal and has a non tiny inductance. Normally FOC controllers will sample at dead center of the low side FETs freewheeling period. The current is then the average current in the middle of the ramp down period and is the same as during the high side FET freewheeling period.

You lose some degree of protection against shorts and reduce some ability to do advanced things like sensorless by inductance tracking, but really there's no major disadvantage.

 

[badgineer]

The FOC, you don't need that many samples. The current measurements are just for feedback, regulation... The algorithm will run many many pwm cycles without a current measurement before error occurs as long as your motor is sufficiently sinusoidal and has a non tiny inductance. Normally FOC controllers will sample at dead center of the low side FETs freewheeling period. The current is then the average current in the middle of the ramp down period and is the same as during the high side FET feeling period.

Interesting, thanks for the information. Low side shunt, then, seems a quite nice solution.
I still wonder why all the VESC6 have the shunt on the phase wire (except the 100v one?). But that's quite off-topic .

Now I'm just curious to see if it's going to be programmable (as in firmware update) or not.

Br,

 

[mxlemming]

The FOC, you don't need that many samples. The current measurements are just for feedback, regulation... The algorithm will run many many pwm cycles without a current measurement before error occurs as long as your motor is sufficiently sinusoidal and has a non tiny inductance. Normally FOC controllers will sample at dead center of the low side FETs freewheeling period. The current is then the average current in the middle of the ramp down period and is the same as during the high side FET feeling period.

Interesting, thanks for the information. Low side shunt, then, seems a quite nice solution.
I still wonder why all the VESC6 have the shunt on the phase wire (except the 100v one?). But that's quite off-topic .

Now I'm just curious to see if it's going to be programmable (as in firmware update) or not.

Br,

When you've got a 10$ mcu and 20$ of gate drivers and FETs, what's another 3$ of INA240?

They are better. Work better at the extreme of modulation and can allow full Clarke transform rather than the simplified 2 phases input version. This enables cancellation of common modes and lower noise.

TI make an AMC 1300 but that's like gold dust at the moment and requires a floating power supply... Technically can use the bootstrap...

 

[Woly]

Now I'm just curious to see if it's going to be programmable (as in firmware update) or not.

Got mine yesterday. I didn't find the ST interface exposed so if the existing firmware doesn't support downloading then some mods will be required.

 

[Woly]

It is using MDP10N027. 100V 12A 2.8mOhm.

 

[Woly]

That should be 120A for the mosfet.

 

[mxlemming]

Now I'm just curious to see if it's going to be programmable (as in firmware update) or not.

Got mine yesterday. I didn't find the ST interface exposed so if the existing firmware doesn't support downloading then some mods will be required.

Isn't the empty jst connector an SWD port ? I see gnd something clk 3v3

So that would be programmable like usual.

These pics confirm low side shunts.

Could you read off the gate driver and opamp chip marking?

Looks pretty sensible and standard from what we can see!

The shunt resistors aren't going to like being boosted to much more current though. I'd say 120A on single shunts is pushing it already. They look like 2512 0.5milliohm resistors and 120A will dissipate 7.2W in them. There's a big busbar next to them to sink that heat but they're normally only rated up to 3W.

 

[Woly]

Oh, yeah I think you are correct. That pin is labeled DIO so they must be SWDIO and SWCLK.

 

[badgineer]

Got mine yesterday.

Thank you Woly for the teardown pics and info!

Isn't the empty jst connector an SWD port ? I see gnd something clk 3v3
So that would be programmable like usual.

Really hope you're right on this. This would make an excellent little controller for the money, but non-firmware upgradeable kinda defeats the purpose for what I personally expect form a VESC-project based ESC...

The shunt resistors aren't going to like being boosted to much more current though. I'd say 120A on single shunts is pushing it already. They look like 2512 0.5milliohm resistors and 120A will dissipate 7.2W in them. There's a big busbar next to them to sink that heat but they're normally only rated up to 3W.

Going out on some speculation here. I always *assumed* the 120A Phase current limit is a limit on RMS current through that phase wire. In case of low side shuts, the current passes through them only 50% of the time, when the low side mosfet is on. Meaning that would put the RMS current through the shunt at 50% of the "real" phase current (60A for the max of 120A). Or is my assumption wrong again?
That would still put the current through the shunt at the rated limit, but not 2x over

Br,

 

[mxlemming]

Going out on some speculation here. I always *assumed* the 120A Phase current limit is a limit on RMS current through that phase wire. In case of low side shuts, the current passes through them only 50% of the time, when the low side mosfet is on. Meaning that would put the RMS current through the shunt at 50% of the "real" phase current (60A for the max of 120A). Or is my assumption wrong again?
That would still put the current through the shunt at the rated limit, but not 2x over

Br,

No, you are absolutely right here. The average power dissipated will be half my estimate. Good spot.

 

[Ianhill]

The anti sparks ive had so far have failed well below rated, i got the new 300amp version and id say ot will do 60amp its got 10awg cables on it.

Water cooled version rated 200amp 26s with 8awg cable i run it 16s 60amp got myself 3 miles and noticed i couldn't power off fets had blown.

Im dubious of this controller why is it much cheaper than the 7550 but more contionus rated, in my view the new 300amp antispark is suited to the 7550 they both use 10awg no chance of feeding that 300amp with out emulating chernobyl.

To be fair to flipsky they helped with spare fets and reimbursed so they not a joker company but use your own wisdom around max ratings.

Im interested in the new led controller seems a decent way of lighting a small ride and looking beast while your at it.

 

[john61ct]

Could an over temp de-rating or cutoff control be added to help test for real-world ampacities?

 

[Woly]

The drivers are EG3112. The marking on what I think is the op amp is |18jd.

What's left appears to be for the power supply section.

There is a temperature sensor epoxied to the heatsink and labeled on the board as NTC.

 

[mxlemming]

The drivers are EG3112. The marking on what I think is the op amp is |18jd.

What's left appears to be for the power supply section.

There is a temperature sensor epoxied to the heatsink and labeled on the board as NTC.

18jd is an INA181A1opamp from Texas instruments. Gain 20. A good cost effective choice. It'll read up to about 160A before saturating then.

The EG3112 who knows. Probably a replica of an Infineon part. Datasheet in Chinese. Probably fine.

 

[mxlemming]

The drivers are EG3112. The marking on what I think is the op amp is |18jd.

What's left appears to be for the power supply section.

There is a temperature sensor epoxied to the heatsink and labeled on the board as NTC.

Sorry to hassle you again, what's the DCDC converter chip? Probably the big one next to the inductor.

 

[badgineer]

The EG3112 who knows. Probably a replica of an Infineon part. Datasheet in Chinese. Probably fine.

Seems google made less daring assumptions about *my* language skills and mercifully handed me an English datasheet . Can't read much into it though. Ebike controllers ans BLDC Drivers are among the explicitly stated applications for the driver, which is good news.
https://datasheetspdf.com/pdf-file/1417749/EGmicro/EG3112/1

Br,

 

[mxlemming]

The EG3112 who knows. Probably a replica of an Infineon part. Datasheet in Chinese. Probably fine.

Seems google made less daring assumptions about *my* language skills and mercifully handed me an English datasheet . Can't read much into it though. Ebike controllers ans BLDC Drivers are among the explicitly stated applications for the driver, which is good news.
https://datasheetspdf.com/pdf-file/1417749/EGmicro/EG3112/1

Br,

Nice find. I didn't look very hard, the 8 pin gate drivers tend to all behave pretty similarly. And the data sheets rarely comment on quality.

Slightly worrying is they have no under voltage lockout.

I'm more interested in the low side topology, I built a VESC based controller with low side shunts and it really hates low inductance motors. Everything checks out perfectly but... Still just behaves a somewhat badly.