Crystalyte Controllers - Repair and Modification information

[Drunkskunk]

Thanks! good to know. I haven't found a switching DC-DC that would do up to 85 volts,function as low as 29 volts, and have an output around an amp. I figure I would have the controller feed an accesory battery, and run everything off that. Full current will only be needed a few seconds at a time.

[solarbbq2003]

anyone thought of using transient voltage suppressors to protect fets from voltage spikes, was some discussion on it for dual speed motors, but might make controllers less prone to fet damage

[fechter]

I'm not sure if they make them in the right voltage?

Anyway, it's better if your main capacitors are good enough that you don't need them. The energy can be recovered instead of dissipated.

You could get some benefit by adding super low ESR capacitors across the existing ones. Something like multilayer ceramic or polyester ones, less than 1uf. These can take up the very fast spikes better than the aluminum cans.

[solarbbq2003]

thats very interesting, so without removing the old caps you could add the 1uf caps?
so the function of the three caps on the fets is to dissipate the voltage spikes?? I never realised that was wondering what they were for!!!!

[fechter]

The main capacitors have to shuttle huge amounts of current back and forth. If they are undersized, they will heat up and sometimes explode. The aluminum electrolytics are pretty good at carrying the bulk of the charge, but they are not really good at very high frequencies.

Without the capacitors, the voltage spikes from the PWM switching would most likely destroy the FETs immediately. The batteries behave sort of like capacitors also, but their effective series resistance would be very high at the switching frequency, partly due to the inductance and resistance of the battery wiring.

I've had a couple of controllers where the main cap blew first, immediately followed by a FET failure.

Multiple smaller caps work better than one big one. More lead wires make for less resistance.

[nemo]

I alway like to read this part of the forum, thanks to all of you!
Last time you helped me with recognizing zenner diode 13V. Does any of you have a layout of other diodes on the board? I can provide the quality photos. I don't have the original controller any more so I'm unable to confirm what is zenner and what is only a regular diode. Most of them are in smd. I think those close to mosfets are also zenners. My cloned controller worked on lower voltage just fine but then one day blew and I need to get to the bottom of the problem.(building another 3!) That 13V zenner will definitely help. I didn't even use zenners close to mosfets and later on read somewhere in datasheet that zenners on the gate should protect the gate and are readily used there.As well as small 5-10 ohm resistors.(they protect the driver chip in case of the failure of the fets)
So if you have an idea what diodes are where I'll provide the photos. I'm still planning on drawing the schematics for all but recently I'm too busy with other projects. Also I'm not really sure that there really is interest for somebody to clone the controllers as I did. If I'm wrong let me know. (this may speed the schematics drawing)

[fechter]

Sure, post a pic. Most of the other diodes are just regular diodes. The larger ones are used in the driver bootstrap circuit. As I recall, the smaller ones are in parallel with a resistor to make the FETs turn off faster than they turn on.

Ah, here it is:

Brushless Controller Output stage.jpg (27.99 KiB) Viewed 2006 times

[nemo]

I've located some pictures from my pc of the original board.
Apart from bigger diodes that I can read and 1 smaller next to a big resistor that is 13V 1.3W zenner. There are 6 smd diodes (smd 4148?)paralel to 100ohm resistors going to 2n5551 as on your drawing. Then 1smd on the picture(unknown) and from the other side one smd called D4(unknown). All other parts I know. Thanks you!

[fechter]

I can't get to the one on the bottom easily since the heat sink is back on.
The one on top measures like a regular diode. I'd give you 90% odds the one on the bottom is the same. Something with a 50v rating should be fine.

[nemo]

Thank You, I'll try to confirm this from the schematic when I finish it. Nemo

[momo]

I'm wondering if anyone got their hands on the complete schematic for the controller?

[rkosiorek]

here are the circuits for the 15V regulators used in the Crystalyte controllers. first the one used in the original analog controller.

Crystalyte Analog Controller 15V regulator

15V REGULATOR.jpg (12.06 KiB) Viewed 1817 times

and now the switching regulator used in the newer V2 controller

Crystalyte V2 Digital Controller 15V Switching regulator

Low Voltage Switching Supply.jpg (25.61 KiB) Viewed 1820 times

hope this helps.

rick

[giveahoot]

Thanks to fechter for this great thread, and for so generously sharing information. I've learned so much from all your knowledge this past year. It makes one appreciate the limitless possibilities of this hobby, with a better realized ability to tackle them.

[bobmcree]

D4 is a regular diode like a 1N914 or 1N4148.

[ngocthach1130]

Hmm today i decided to do some wire maintenance on my bike when i decided to mod the controller while i have it disassemble anyway. I notice the wires on the new crystalyte 72v 35A controller is puny. They're like 14guage wires. I think i'll up them to 12 or 10. As people have stated these are populated with IRF4310 so i'm going to replace them with IRF4110. Someone asked this before about this controller having 6 Mosfet even though it have 12 slots. I did a quick look and i THINK there is gate resistor connection already provided for the unpopulated spot. Anyone have different opinion? Do you think i'll gain any any benefit to adding 6 more? I need to know since i have exactly 6 IRF4110 left from other mod and i'm thinking of ordering more. Don't want to waste time and money is it won't make a diff. Thanks ahead for any reply. BTW it's the white PCB version.

[bobmcree]

Hmm today i decided to do some wire maintenance on my bike when i decided to mod the controller while i have it disassemble anyway. I notice the wires on the new crystalyte 72v 35A controller is puny. They're like 14guage wires. I think i'll up them to 12 or 10. As people have stated these are populated with IRF4310 so i'm going to replace them with IRF4110. Someone asked this before about this controller having 6 Mosfet even though it have 12 slots. I did a quick look and i THINK there is gate resistor connection already provided for the unpopulated spot. Anyone have different opinion? Do you think i'll gain any any benefit to adding 6 more? I need to know since i have exactly 6 IRF4110 left from other mod and i'm thinking of ordering more. Don't want to waste time and money is it won't make a diff. Thanks ahead for any reply. BTW it's the white PCB version.

you are certainly better off with twice as many fets, whether they are 4310 or 4110. I have not seen the new controller yet, but i have a couple coming next week.

if you measure with your meter from the empty location's gate pin to the gate of the populated location you should see 20 ohms if there are 2 10 ohm gate resistors installed. i would not recommend mixing the 4310 and 4110, and the 4310 has a higher Miller charge than the 4110 so 2 4310's would need more drive current than 2 4110's, but if they are using the same boost circuit for the high side fets as the older controllers it should be fine with either part. If you are not running at over 50A you might keep the 4310s and just add 6 more of them. With 12 4110's i have quite a few controllers running around now set at up to 85A.

[ngocthach1130]

It's too late anyway. I already clipped off and desoldered the 4310. I was going to add the 4110 anyway. Just a decision of adding 6 or 12. I think the lower resistance of the 4110 is better since I'll be changing the battery wire and the motor wire to larger 10guage wires. This might result in higher potential current traveling through. At the same time I would probably use the same controller for 99v once in a while. So the robustness of the 4110 will help. This will be for the occasional speed fix.

[Jozzer]

Hi all, i'm just repairing a dual motor controller, that appears to use the 20A miniboard (6 fets). Although the controller appears to work OK, the motor wont spin up from all positions. Sometimes it spins, other times you have to roll it back to make it fire up. Any idea's?
Thanks:)
Steve

[fechter]

Hi all, i'm just repairing a dual motor controller, that appears to use the 20A miniboard (6 fets). Although the controller appears to work OK, the motor wont spin up from all positions. Sometimes it spins, other times you have to roll it back to make it fire up. Any idea's?
Thanks:)
Steve

That usually indicates one hall sensor is not working. Could be just a bad connection.

[Jozzer]

Hmm, wasn't halls sensors at all, and by the time we got through a batch of 10 controller repairs, we found this fault on 2 controllers. One thing that will cause it, is the gate leg of the mosfet not soldered in properly, another was a driver chip. Strange to see the same fault with 2 different causes...
Another symptom is slightly noisier running. (rumbles and such)

[Deepkimchi]

Was there any problem with using these instead of 4110 Fets?

Risson presented these on page 15 but nobody made a comment..

http://www.st.com/stonline/products/lit ... 80ne10.pdf

Have great screw connectors instead of the spindly legs.

DK

[fechter]

Was there any problem with using these instead of 4110 Fets?

Risson presented these on page 15 but nobody made a comment..

http://www.st.com/stonline/products/lit ... 80ne10.pdf

Have great screw connectors instead of the spindly legs.

DK

Looks like it would work if you can figure out where to stuff them inside the box.
As I recall, those are very expensive too.
I'm also not sure about the gate drive requirements. It says 21nF input capacitance, which would be much more than a 4110, but it might be OK.

[disadvantage]

Risson presented these on page 15 but nobody made a comment..

http://www.st.com/stonline/products/lit ... 80ne10.pdf

Have great screw connectors instead of the spindly legs.

Ooh, I missed that post. Those are a good find.

Thank you for remembering that post!

Edit: $40 bucks each @ mouser.com -- ouch!
http://www.mouser.com/Search/ProductDet ... 0kWg%3d%3d

[rkosiorek]

better have deep pockets too. minimum order is 34pcs @ $40 ea = $1360

[rf]

My acceleration often fades momentarily while powering up a hill. Power fades several times going through about 15 or 20mph. The fade lasts for less than a second. I'm pushing my controller a bit and I'm wondering if I've damaged it. It's still quite reliable. And never gets more than warm (it's well ventilated.) But I live up a steep 2-mile hill that takes about 6 minutes to climb.

I have a Crystalyte 48-volt, 20amp controller that I use with an A123, 20s4p pack. I've overridden the 20amp limit on the controller and set my CycleAnalyst's current limiter to 28amps. The mosfets in the controller are IRF2807s, rated
at 75-volts which seems sufficient for the roughly 66-volts the pack produces. I just switched from a Crystalyte 408 motor to a 5304. I'm pretty sure the fading was happening with the smaller motor too, just less pronounced.

Do I need a heftier controller or mosfet replacements or something?

Richard