Golden Motor Regen BLDC 48V controller

[ZapPat]

Here's a graph of a test done by ES member silicium to see the GM regen controller's performance up and then back down a big hill.


PS: His nickname is silicium81 on the french forum - see his earlier post in this thread for the link to the french forum.

 

[silicium]

Yes, my french nickname is silicium81...

I wish I, too, succeed in changing the parameters of BLDC ...
There is an initial response here:
http://www.goldenmotor.com/e-Bike-DIY/BLDC%20Regen%20Controller%20Wiring%20Diagram%200807.pdf

At the bottom of the schematic, there is a dual display...

 

[ZapPat]

Silicium,

Have you found a way to get around the switching regen on/off problem? Or do you still have to completely stop before changing modes (between regen and freewheel)?

This seems to be the major problem with this controller

 

[silicium]

No, it's a problem

switching regen on/off is only for speed = 0

 

[Lanchon]

Are these 6 free-wheeling diodes!?

these seem to be TVS'es (transient voltage suppressors) in series with resistors, the generic 1.5KE62A type.

 

[Lanchon]

hi electrobent,

thanks for posting info on the golden motor 48V regen controller. I've been trying to find more details on the power stage of this controller but I couldn't find a thing. do you by any chance know the answer to any of these questions?:

1) what FETs are those?
2) do you know the value of the 6 resistors connected in series with the 1.5KE62A TVS diodes that were added to the circuit on the back?
3) also added are 3 small orange diodes that protect the high-side FET gates, can you make out the markings?

thank you very much!

 

[Lanchon]

and one more question:

4) could you tell me the capacitance and voltage rating of the capacitor marked in yellow? (far right)

thanks again!


PS. "1)" is the most important question by far; if anybody remembers it'd be great.

 

[electrobent]

I'll have to take it apart again to answer any of those questions--sorry

 

[silicium]

Edit: my regen controller is 36V model

Mosfet's: IRF2807
Value: 1uF/160V on my regen controller
R=0.1 Ohm
add diode: Yes

 

[Lanchon]

thanks for your reply electrobent, I wouldn't have asked that you open the controller.

and of course thank you very much silicium! if you opened your controller for me I'll be forever grateful. could you confirm this is the 48V controller and not the 36V?

and one more thing: the 3 small orange diodes I asked about in "3)", do you see any part numbers? (please only answer if you have your controller already open; this question isn't important enough to justify opening a controller.)

 

[silicium]

could you confirm this is the 48V controller and not the 36V?

my controller is 36V regen

diode part number not visible

 

[Lanchon]

ok, thanks. (darn it!) so I'm still needing to know what FETs are used on the 48V version, but the rest of the data is applicable to both version. thanks again!

 

[ZapPat]

Lanchon - I'm ready to bet that you will find the same 2807 FETs (75V) in their 48V version. There is nothing magic about the FETs in a regen controller anyways, it's mostly the power layout that is the problem.

 

[Lanchon]

I think that too, but I'm not willing to bet, I have to know. the problem is that the 75V limit is too close to 58V max voltage of 16 lifepo4 cells. resistance and inductance is various components could cause spikes above 75V during regen that could progressively degrade the FETs until failure. I can't test this, I have to relay on the experience of regen manufacturers.

please refer to this thread for more details:
http://endless-sphere.com/forums/viewtopic.php?f=2&t=6511

if anyone knows which FETs are used on the 48V version of this controller, please chip in.

 

[ZapPat]

I think that too, but I'm not willing to bet, I have to know. the problem is that the 75V limit is too close to 58V max voltage of 16 lifepo4 cells. resistance and inductance is various components could cause spikes above 75V during regen that could progressively degrade the FETs until failure. I can't test this, I have to relay on the experience of regen manufacturers.

please refer to this thread for more details:
http://endless-sphere.com/forums/viewtopic.php?f=2&t=6511

if anyone knows which FETs are used on the 48V version of this controller, please chip in.

I'm still ready to bet that your 48V version has 75V FETs also. The most clear indicator of this is that your TVS are 62V rated, and these are present only to protect your FETs. Another thing is that your main bus caps are 63V rated on both models. Spikes will be higher on your FETs than on your caps, but they shouldn't be too much higher, because that would be telling us that they are not really doing what they are suppost to. In my eyes that 0.1ohm resistor hack is a quick fix for a poor PCB layout. Make sure you add your high frequency film caps as close as possible across each FET pair's bus connections. I got some Epcos ECQ-B1104JF to use for this from digikey, but some people use AVX ffv3's for bypassing (very expensive though).

BTW, I have a regular 36V golden motor controller that I used on 48V, just had to change two power resistors, the rest was the same for both. They like to keep things simple by not having to many totally seperate production lines I would guess. It also has 75V FETs.

 

[Lanchon]

> I'm still ready to bet that your 48V version has 75V FETs also.

I'm as much of a scientist as I can be and avoid acting on gut feeling, I'm a sucker for evidence.

> The most clear indicator of this is that your TVS are 62V rated, and these are present only to protect your FETs.

the 62V nominal TVS can go up to 90V during a 23A spike like the one you'd get on max torque regen, so good design practice would dictate 100V FETs.

> Another thing is that your main bus caps are 63V rated on both models.

the big alu cans, yes, the smaller ones are 100V. my understanding is that alu cans are insensitive to spikes. at high frequencies impedance goes up (instead of down, as in an ideal cap) as they turn inductive, and so the spike is not really charging the capacitor, there's no energy transfer. this means two things: that the capacitor won't filter the spike obviously, and that the cap doesn't need to be specified for the peak voltage (AFAIK). so 63V caps unfortunately don't rule out needing 100V FETs.

> In my eyes that 0.1ohm resistor hack is a quick fix for a poor PCB layout.

IMHO they are there to prevent forward conduction on the unilateral TVS'es (FET body diodes will take almost all the current with the resistors, but not without them).

> film caps ... Epcos ECQ-B1104JF ... AVX ffv3

thanks, I'll have to research these capacitor types. though here in Argentina I doubt I'll find anything even medium-tech.

time has came to choose the power supply; I'll have to choose safety or performance, and without adequate data.

all in all I have to say that I'm 99% agreeing with you, I'm sure the 75V FETs would work. now if I could just be sure...

 

[electrobent]

OK the FETs say RF2807 across the top and then what looks like a diode symbol between an I and an R (International Rectifier?) with 751P on the same line. Finally it says "CO" and "UX" on the bottom.

The resistors are surrounded with heat shrink tubing and at this point I am reticent to apply an ohm meter to them in the circuit.

The capacitor says 160V but I could not make out its capacitance.

What does all this mean?

Sorry to take so long to reply.

 

[Lanchon]

thank you electrobent!

> What does all this mean?

that's apparently safe to regen on batts way up to 58V while running on 75V FETs, a thing that wasn't clear to me and for which a non-experimental answer was very difficult to find. means that I'll drive my power stage on 48V.

thanks!

 

[electrobent]

So you wanted my FETs to be 75 volt so that you could run a 36V controller with the same FETs?

I am still waiting on the dongle programmer thingy from Golden Motor but I guess their QC problems extend beyond just the products themselves.

If they'd post a 48V 20 inch front wheel I'd just buy that instead.

I have 66 nimh 10AH D cells in two parallel 39.6 volt packs and a 36V set up now. I guess I should have just bought the 36V regen controller but I figured I could add batteries to get to 48 and program it down so as not to fry the 36V motor and get better range.

So I guess I'd better not goose it past 48V nominal when I add extra cells to my strings?

 

[Lanchon]

if your FETs were 100V I'd be in trouble. but since there are no reports of busted 48V regen controllers I think I'm ok, specially taking into account that my power bus will be much better decoupled than yours.

your controller will fry if the bus hits a critical voltage somewhere between 58.9V and 65.1V. if you use it as non-regen, you'll be safe if you keep *max* voltage below 58V. however when you regen, the bus will be higher than the batt by an amount that's difficult to predict. 16 lifepo4 cells ("nominal" 48V) have a max voltage of around 57.6V, and that's dangerously close to your max. so I'd go very easy on the regen with a fully charged 16-cell pack. going beyond 16 cells would be dangerous, and I'd only risk 16 cells with cells of very good quality (low internal resistance) such as A123s. but maybe I'm being a bit paranoid. (again, without regen you'd have no problem with batt resistance.)

 

[electrobent]

So the danger is overcharging during regen? Couldn't you just check the battery voltage and then decide whether to send the regen current to the battery or a power resistor? Or have more than one set of batteries and run down on one and up on the other and then switch? I guess my first solution is not energy efficient and my second is not weight efficient. I should have left well enough alone--this thing is too fast already--but I can't resist the prospect of double the power and regen was gravy.

 

[Lanchon]

note that I edited my last post to read:

going beyond 16 cells would be dangerous, and I'd only risk 16 cells with cells of very good quality (low internal resistance) such as A123s.

which is what I originally meant to say but didn't.

> So the danger is overcharging during regen?

that's one danger; you need to go down a slope with fully charged batts for that. another is low charge acceptance, which results in higher voltage seen on the FETs. but what I was talking about is overvoltage of any kind, with causes that include inductance and resistance of various components.

> Couldn't you just check the battery voltage and then decide whether to send the regen current to the battery or a power resistor?

you actually check the bus voltage and the set up is called a break resistor. in my case I'd need a resistor with a power rating around a kilowatt which is something I don't want to pay for or carry around, or have it near me while it's being used for that matter. besides that you need additional power electronics and control logic. but this only protects you from a battery that can't accept charge, it does nothing to protect you from overvoltages secondary to parasitic inductances.

> Or have more than one set of batteries and run down on one and up on the other and then switch?

no, you just need to not fully charge the batt to begin with.

> I can't resist the prospect of double the power and regen was gravy.

if you want to go higher than 48V you just need a new controller. don't try it with yours, it'll fry. better keep it as a backup in case your new one should ever fail.

 

[fechter]

A brake resistor is one way to deal with excess regen. I'd make one by wrapping wire around the bike frame. The other approach is to simply limit the regen current based on bus voltage. Whenever the voltage gets up to the set maximum, the duty cycle of the regen drops to keep it below the set point. If your batteries got full on a long downhill, your braking would ease off and you'd need to use the friction brakes to slow down.

I think in normal riding, this would not happen very often, if ever.

 

[Knuckles]

A brake resistor is one way to deal with excess regen. I'd make one by wrapping wire around the bike frame. The other approach is to simply limit the regen current based on bus voltage. Whenever the voltage gets up to the set maximum, the duty cycle of the regen drops to keep it below the set point. If your batteries got full on a long downhill, your braking would ease off and you'd need to use the friction brakes to slow down.

I think in normal riding, this would not happen very often, if ever.

One of my favorite resistors is found right in the kitchen ... A heating element from the stove!

This element is a 40 ohm resistor that can handle plenty of amps.
Great for cooking your burgers too (while braking).

 

[The7]

One of my favorite resistors is found right in the kitchen ... A heating element from the stove!

This element is a 40 ohm resistor that can handle plenty of amps.
Great for cooking your burgers too (while braking).

Some maths here!

This 40 ohm stove element is for 240V.
And at 240V, the current is 6A and the power is 1440W.

For an ebike, if the regen voltage is 60V,
then the current is only 1.5A.