Blown Controller

[heathyoung]

So I've fixed my "LYEN edition" controller. What I did is below.

Nope, there are not three, but four additional chips. Original mosfet drivers were driven via 120Ω resistor, which is madness. So I upgraded it a bit to 3A sink/source drivers. Also, original discrete low side drivers had an inverted signal, so I needed another hex inverter chip.

Yeah, they are a craptacular gate drive design. Absolute amateaur hour design, it would have cost bugger-all to implement proper gate drive but when you sell them (12 FET) on flebay for $33 bucks each, you gets what you pays for. Sigh.

Those 18/36 FET controllers are just mindless control-c, control-v based design (aka cut-n-paste). The Lyen edition stuff - well, its just rebadged Keywin/Ecrazyman/Shenzin Sucteam Technology (teehee sucteam) - no real improvements apart from some QC cleanup...

 

[dnmun]

can someone explain to me what i am looking at in circuit's controller.

the ceramic caps in place of electrolytics, for the S/D? i just looked and the charge pump caps are electrolytic.

and each little wire goes from a comparator circuit inside that huge DIP?

and also the gate drain circuit has a wire from the pnp transistor to the phase wire? very delicate work.

but why ceramic caps there too?

 

[heathyoung]

Well, he had taken the gate drive signals off the microcontroller, inverted them, and run them to proper gate drive IC's rather than the descrete original gate drive.

I also noticed that the electros are missing. Maybe not installed yet?

 

[circuit]

I also noticed that the electros are missing. Maybe not installed yet?

Yes, I removed electrolytics to get access to other components. It would be extremely hard to work with those lying around.

By the way, another thing to notice - these discrete drivers take their ground from anywhere which is closest. Makes me wonder if the person, who designed this, has any knowledge of what he is doing. Looks like its really just a cut'n'paste without giving a crap.

One more thing. I have spun the motor and deliberately wiggled my hall sensor connector to loose a contact a bit. And yea, the motor started to jump around, supply wrong phase time to time and do some serious regen. I was doing this on benchtop with lab psu, but imagine this on quality battery pack.
That's a serious software bug!


When you see how and what these are made from, a natural question rises by how much overpriced they are. I bought this from LYEN for like $200!

 

[nebriancent]

one thing i love to use when taking off a solderd through hole component is solder wick

all it is is a flat braided copper wire with an arsload of flux on it

to use simply get your iron hot ... place said solder wick over connection point ... push wick into connection point with soldering iron tip and like magic all the solder disapers from the hole and you get another hard inch of solder wick to cut off and dispose of

 

[dnmun]

the idea behind using the gate driver was the delay in the hall signal on the scope?

so the transistors that drive the gate are slow to turn on?

and so the driver IC helps by turning on the mosfet faster by pushing more current to the gate sooner?

and does it drain the charge off the gate too or does the pnp transistor do that still?

 

[circuit]

the idea behind using the gate driver was the delay in the hall signal on the scope?

so the transistors that drive the gate are slow to turn on?

and so the driver IC helps by turning on the mosfet faster by pushing more current to the gate sooner?

and does it drain the charge off the gate too or does the pnp transistor do that still?

No, there are no transistors left, only the power mosfets. These are fully driven by chips, up and down. I have not compared it under load, but previous discrete drivers must be like 30 times slower. This quicker switching lowers the losses.
However this is not related to phase delay in my situation. The delay is due to misaligned/misplaced hall sensors in the motor. In other words, I hane not only a crappy lyen controller, but the motor as well. Damn, I'm tired of this chinese $#¡7.

 

[amberwolf]

another hard inch of solder wick to cut off and dispose of

Or save for reinforcing/beefing up high-current traces.

 

[nebriancent]

we have a wiki???
when did this happen
and btw i love the idea of using old solder wick to beep up traces .. sounds about as good as me going down to the hardwear store to buy solid strand house wiring to strip and solder in for the same thing (10 gauge or better for me usualy 2-3 strands wide if i can get it in)

 

[amberwolf]

we have a wiki???
when did this happen

Mrvass's first date of title page creation is signed as "01:26, 26 August 2011", so less than a month after you got here.

 

[h0tr0d]

So I've fixed my "LYEN edition" controller. What I did is below.

Nope, there are not three, but four additional chips. Original mosfet drivers were driven via 120Ω resistor, which is madness. So I upgraded it a bit to 3A sink/source drivers. Also, original discrete low side drivers had an inverted signal, so I needed another hex inverter chip.

It spins well now (crystalyte x5). However here is what I've noticed:

This is unrelated to previous problems.
Channel 1 shows phase voltage at 100% throttle, channel 2 shows signal from one hall sensor.
What we see here is a 16 degree delay in hall sensor signal. Whoa! I'm not going to use THAT!
And the darn thing is unadjustable.

EDIT: when spinning in reverse, there is a 16 degree advance. Which is better (than delay), because it compensates motor's inductance at high RPM. But I can't mount my rear motor in reverse?!

Could you reverse the stator, making the wires exist through the sprocket side?

 

[circuit]

Could you reverse the stator, making the wires exist through the sprocket side?

Thought of that, but it's a rear wheel hub, so reversing is not as straight forward as I would like. I'd need a press to reverse the axle... Maybe some day...

 

[nebriancent]

swap out 2 of the phase wires and revurse becomes forward....

 

[nebriancent]

^had a problem with one of my controllers going into revurse when it would rain untill i potted the whole frocking thing in conformal spray and desoldered those nasty revurse leads off at the board