2QD Still In Progress, Now Using Curtis 1204-410 Controller

Since I have had so very little time to work on anything recently, I haven't finished the 2QD controller repairs or the physical repairs to the bike (wheel, chain).

The major problems were just the power output section (bottom two MOSFETs blown, gate resistors burned open as they were designed to). But one minor problem that kept anything from working right was a 9.1V zener diode that was only allowing 1.45V! So there was no internal power supply to run the entire comparator and feedback/control section, which is the majority of the controller.

I fixed that by replacing the zener with a 5.6 and a 4.1 in series, which while higher than the original still allows it to work. This got the main section working again, and now I just need to finish the power section as described a couple of posts back. I have wanted to use a different case for a while, so now is my chance.

Since it is taking a while to figure out and then make the holders for the FETs to line them up and keep them tightly against the case inside, and to polish the aluminum and the FETs for flatness and smoothness (for better heat transfer, since I have to use thermally conductive / electrically insulative pads, too), I decided to take a quick look again at the Curtis 1204-410 I have had for a few months with little time to troubleshoot.

I let it sit on the bench powered on (but not doing anything as it doesn't respond to input), for a while, and suddenly it started working, the motor attached to it began slowly spinning (the throtle was just a regular pot set to barely on). Apparently once it got warm enough (from me having the oven on to warm up the room), a connection was made well enough to start working.

There are six interconnect wires between the Curtis logic board and it's power board. Two carry B+ and two ground, and the other two carry signals. One is the PWM output to the FETs, and the other looks like a feedback from the FETs as it is time-shifted just a tiny bit late (which I can barely see at all with my old 531A), but identical to the PWM waveform.

The solder joints on them must've been flexed or vibrated enough to crack them, so they would only make connections good enough to work when warm or hot. Over 85F, anyway. Under that, they might work and might not. I reflowed the solder and now they always work.

Since it's a 36-48V 225A controller for brushed PMDC motors, it can be used in place of the 2QD, so for now I put it on CrazyBike2 and verified it works with the bike. I can use it until I finish the 2QD rehousing.

Now I need to build a new rear wheel, move the chain and some shifters and stuff from one of the spare bikes I got for parts from someone for Christmas, and finalize the new throttle control setup, and CB2 will be ready to ride more than just for tests.

I also cobbled together this temporary throttle lever and mounting, for using a powerchair's spring-return throttle on CB2.

The spring is so strong that without a pretty long lever, I cannot keep it pressed down for very long. Sorry the pic is so dark, but the flash keeps shining off the metal and the camera autodarkens the rest of it to compensate. :(

The aluminum bars are just end-pieces, uncut, off the rackmount fan enclosure from the same old studio-type Sony VTR the gray transistors from the last post are from. I'll probably wind up modifying a brake lever setup to use for this instead, as it will already be designed to clamp to the handlebar, and it has a long pivot arm.

Optionally, I considered using a lever-style shifter, but it is more complicated to set up for this, and it is also a much shorter lever.

UPS And PC Power Supply Parts Bonanza

A bit of salvaging of some old computer power supplies and UPSs (most of which were found in alleys or roadside on bulk trash days, over the last few months, but sat "intact" until now due to lack of time) netted me several pounds of parts (not counting the toroids or transformers!), including the stuff shown in the pics below.

Bunches of small value caps, diodes, zeners, transistors, mounting screws, heatsinks and clips (and thermal pads), and lots of resistors not yet taken off the boards (not in pic), several TL431A and some LM317


Some very old 100V+ Vce type transistors in PNP and NPN, off some boards from a Sony VTR (I never saw the VTR, just the boards)

They'll be used to replace some lower voltage 2222's on my 2QD so I can up it to 48V+ usability.

Bunches of small signal transistors, 50V Vce types, a bit better than the 2222 but not much.


A few beefy dual-diode (common cathode) units, one of them up to 40V at 50A!

Also about three times that many smaller TO220 versions of these, mostly in the 40V 3A range. All can be used with heatsinks to parallel unequally-charged battery packs in a pinch.

Of course, there are lots of large caps, too, in low and high voltages, and many form factors.

MOSFETs; lots of them, too, though mostly lower-voltage types, and generally fairly high RDSon.

Some IC's, too. Optoisolators, LM339 and LM324 chips, other single or double op-amp chips, some logic chips (74x series), some "house branded" chips in the UPSs which probably means they're ROMs or pre-programmed MCUs. Some standard PWM SMPS control chips, might be adaptable as controller chips.

Transformers, connectors, toroids, cables, wire, heatsinks, switches, LEDs, etc. Fans, too, but the fans are all defective. They might have usable hall sensors in them, though, as long as I don't need linear sensors (they're more likley to be switching types).

Most of the desoldering was done by carefully using a very small propane torch to heat the thru-hole lead areas, while gently tapping the PCB on the bench (outside, so any fumes would not suffocate me). Some things required a solder-sucker and regular soldering iron.

I should be able to build up my other 2QD controller out of what's there, plus have plenty left over for other projects for a while to come, such as a BLDC controller.

There are LOTS of electronics that have all this stuff salvageable if you have time to do it. I generally spend only a minute or two at a time, over days or weeks, and end up with what looks like "Christmas at the parts store". :) Even if only half the parts worked when done, it'd have saved me enough money it'd stil be worth it.

Blown MOSFETs Due To Chain Jam, Possible BMS Arrives for Li-Ion Pack

I have been having balanced karma of late--every time something really good happens, something equally bad compensates.

Yesterday on the way home from work the bike's rear chain (regular bike drivetrain part) somehow derailed and got tangled up, which actually pulled the entire right end of the rear axle and the derailer bolted to it off the dropouts and bent the rim up against the chainstay on the other side (the left end stayed bolted in), and then everything jammed and both bottom-leg MOSFETs blew.

The power meter registered over 3500 watts peak, 153 amps peak, which is a more than any of it is rated for--the MOSFETs I installed in my 2QD are only rated at 80A continuous, and can handle ~150A for *maybe* a second absolute max, with sufficient cooling (which they don't have in my setup, by any means). So it's no surprise they blew up.

All that only took about half a second as I was accelerating away from an intersection, crossing two lanes to get to a left turn I have to make to get home without going down a very busy stretch of road.

Then I had to drag the bike across the rest of the lane (while cars that weren't yet coming when this all started kept trying to go around me instead of letting me get out of their way!), up onto the concrete median, and spend an HOUR with traffic roaring by me, while untangling the chain enough to get the wheel off, unbend the rim and true it enough to get it able to roll. Then I had to take out around 6" of chain due to bent and broken links, and set it up to run only in lower gear in front since it was now too short for the big ring. Plus I had to take the motor chain off, since my pedals are linked to it and it is nigh impossible to pedal the bike without the motor running otherwise.

After that it was a heck of a ride home, at about 9MPH, which is just above the unstable speed of the bike, and I was totally worn out when I got home (and then had to go to a different job as handyman for someone else).

I didn't have the camera with me to take pics of the chain/etc, but wow, it was awful looking. It looked like someone had hit my rear wheel from behind and then stepped on it sideways, but it was just from being yanked so hard by the chain as it tangled.

The frame and dropouts all appear intact, but the wheel was seriously messed up. I will need to take it completely apart, restraighten the rim, and then re-lace it to get it back to anything close to really true.

The derailer cage was slightly bent but not badly, but the bolt that secures it to the little shaped nut that recesses into the dropout behind the axle had sheared thru as it was all pulled out. I am just lucky that I carry spare nuts/bolts in the toolkit, one of which happens to be the same thread pitch and diameter as that bolt--it is a bolt from a rackmount kit--it's so long that it prevents use of the highest (smallest) sprocket on the rear cassette, but as I had to pedal with no assist there was no way I was going to be shifting up that high anyway.

I guess the good news is that I *could* fix the problem on the road well enough to get home.


Here's some pics of the controller with toasty MOSFETs, before repair. I still haven't had time to go back and fix it yet; just replacing the MOSFETs and gate resistors (whcih also burned open as they are supposed to) didn't fix it. I'm too tired to try more now, after another day of work. Tomorrow and the next day I have to go help do a roof and some other things, so I won't be able to do it then, either. Fri and Sat I'm working again, so I don't know when I will find the time to finish troubleshooting it. Maybe Sunday. Should be easy, when I have time and am not too worn out to concentrate. (writing this post has taken several tries).

A close up of the damaged area. The discoloration around the screw on one MOSFET is actually the melted remains of the plastic insulator ring...it's still not shorted to the heatsink, but that insulator is totally destroyed from the heat.




This blown LiFePO4 BMS (sent by an Endless Sphere forum member) arrived today:

All that is known about it's problem is that a puff of magic smoke escaped from it somewhere, but not specifically where. With luck it's the MOSFETs, as those I have around here already.

Once I find and fix the blown part(s), then if I can figure out how to change it's HVC and LVC voltages to match those of my Li-Ion cells, I can use it as a BMS for my custom-built pack.

First, I just need to invent a time-machine that gives me extra time to do everything and still get enough rest....

Test Results of CrazyBike 3.1 Drivetrain

Remember I had been running on 24V, and I upped that to 36V?

So far the performance difference is pretty noticeable. Zippier starts with less current draw, both very important as the idea is to keep me from having to put any load on my poor knees (especially the kind of load needed to start from a stop, with a 120 pound bike plus me and cargo). Plus I need to keep current draw low so as to be able to draw more total power from the batteries over a longer period (Peukert's Law).

I didn't just change the voltage, I also changed the gearing for the motor, so that it still outputs about the same max speed to the rest of the drivetrain, but because of that it has better startup torque and gets to speed faster, and is in the high-current/low-speed region for a much shorter time.

So now, even with less total Ah on the bike (was 2x 12v 31Ah, is now 3x 12V 17Ah) it gets better results than it did before. I have not yet tested it's full range, but I think it will probably wind up slightly better range (was 15+ miles at around 15-16mph average cruising speed including sometimes generous pedalling, is now at least 10 miles at around 17-18mph average cruising speed with no pedalling, and only lost about 1-2mph over the last couple of miles).

I think that higher voltage, causing it to draw less amps for less time during acceleration, is combining with the Peukert effect to allow me to draw more from the lesser-capacity batteries.

At 24V the average cruising current was at a guess 15+ amps (didn't have a way to measure exactly, as I had no meter to do that). Startup currents were enough to pop the little 25 amp breaker I had installed, if I didn't help it start by pedalling pretty hard for at least a few strokes. At 36V the average cruising current is around 6-10 amps. Startup currents are over 20 amps (HF meter now used only goes that high), but it never pops the breaker.


The temperature of the heatsink never went more than about 11°F above ambient. When it was really hot on the way to my friend's place, around 1pm, it was about 112°F in the area, and that's the same thing the sensor read. On the way back home, about 2am-ish, it was around 80F when I started, and rose to 91°F for short periods during and just after high-current acceleration tests (starting up in the highest gear to put the most load on the motor). It cooled by several degrees within about 30 seconds or less, and stayed generally around 86°F.

So I guess I don't need a heatsink any bigger than that little bitty one for the small current draws this motor pulls with this gearing at these voltages. That's good, because bigger heatsinks generally weigh more and they take up more space, both of which are bad for this bike. ;)

Rigged for Testing

Well, the quick road test day before yesterday with the panel meter was half success, half failure, in that I found it only drew around 7 or 8 amps most of the time for cruising (at around 17-18MPH no pedalling), and pegged at 30+ at startup from complete stop for just about half a second before dropping quickly down to that steady rate.

However, after I parked the bike outside a store while I got some groceries, something must have happened to the meter, possibly because of direct sunlight heating it up (though I can't see why; there is no visible sign of a problem). When I came out it was stuck at around 6 amps, and I panicked at first as I thought something had shorted in the controller and was powering the motor even at a dead stop.

But after quickly yanking the main controller's Anderson connector (I haven't installed the battery cutoff yet), it stayed at 6 amps. I could tap the face and it wiggles, so it's not stuck, but it doesn't go to zero (it *was* at zero when I went inside, after turning it off and taking the key with me).

I hooked the power back up, turned it on and tested, and the system
itself still works fine, but the meter does not respond at all.

I unscrewed the casing and verified the movement coil is not burned open, but it doesn't respond to a controlled tiny voltage input either, not even from my multimeter on Diode Test. So something must be mechanically wrong; I suspect they must have used some adhesive to
secure the needle movement to the actual coil, which may have come loose from the heat of sunlight. I have to take the movement out of it to check that. There is nothing obvious wrong with it other than cheap physical construction.

Well, it looks like the movement coil itself has no continuity, so either one of the fine wires connecting it to the external red & black wires is broken, or it's burned open (which makes no sense since it still worked at shutoff, and had not yet had power thru it again).

Only a few hours and I already broke a new toy. :-(


In the meantime, I decided to ride the bike a longer distance (5 miles) on main streets on the way to a friend's to then carpool to a meeting all the way across the valley. I wanted to run it without pedalling at max speeds wherever possible, to test out the batteries and the controller heatsink, etc.

To that end, I wired up and zip-tied on a current meter (in the motor loop), a pack voltmeter, and a temperature-capable meter (with it's sensor down at the base of the heatsink). It looks awful, but I can glance down and read pack voltage and motor current draw at any moment, as well as controller heatsink temperature.

I thought about placing the thermal sensor on the MOSFETs themselves, inside the case, but decided not to, partly because I would have to dismount the controller, open the case, etc., and then do that again when I wanted to take it off after the test. Since I already verified that the MOSFETs don't get enough different temperature from the heatsink even at stall current for me to feel the difference with my fingers, then they should not be more than a few degrees different from it at most even though they're inside the case while the heatsink protrudes out of it.

If my theory is wrong, Murphy will notify me sooner rather than later, I'm sure. ;-)

I've also had a problem with the PDA and our intense midday sunlight, in that when it shines directly on the LCD, it darkens it and makes it difficult at best to read. Probably also damages it.

So I put this smoke-transparent plastic paper-guide-support from the back of some Epson inkjet in my junkpile onto the bars above it, in a way that shades it from most light above and at least a little to the sides, to see how well that might work but still allow some light thru for me to see it by, and not block my view of it.

I tried to get better pics of it, but it's just hard to see. Maybe it will show up better when I take the bike out on the driveway for some overall pics later.

It works pretty well, and unexpectedly it also helps keep certain road dust and whatnot out of my face, without blocking airflow to me. I'm not sure why that works, but it does.

2QD Rehoused, Brake Lever Motor Cutoff, Etc.

I didnt' get anywhere yet on the pedal freewheel, because something is sort of nagging at me about it but I can't figure out what yet. Safer not to start it till I do.

So today I just did a few cleanup things, such as putting the 2QD in a smaller case (the one I had was just so big that there's no good spot on the bike for it).

I also decided to test it out with 2 pairs of smaller TO220 Fairchiled MOSFETs as described in a previous post, instead of the NTY100N10 on each leg of the half bridge. They'll be much lower RDSon this way, so should stay cool even with a small heatsink, with the airflow I can generate near the rear wheel. I have a few of them from various things, so we'll see how they work out. If they blow up due to voltage issues, I'll put the others back on instead.

So there's what it looks like now, with the four smaller MOSFETs directly board-mounted, bolted to a heatsink pulled out of a dead PC power supply. It's small, but may be fully sufficient, since the other ones barely even got warm from operation on their heatsinks. Even if this one gets fairly hot, it should still be safe to run for testing for a while.

I wanted to use an old modem case that was made of aluminum, but I can't find the thing now.

But I remembered I had some similar-sized old Pactec enclosures I got from TriTek many years ago (they're not around anymore), when I was still a student at DeVry. All I can remember is that they were part of some super-deal of a grab-bag sale, and I thought I would use them all up quickly, but only ever used a couple.

I guess I've had these laying around for almost 20 years now, time to finally use one of them again. :) I drilled a hole *exactly* the same size as the OD of the main cap, since it's a lot taller than the case. I actually want that cap out in the air, because it gets more than warm doing it's work, and if it can be cooled a bit by airflow it will last longer.

Halfway installed into the case, you can see the directly-soldered on motor and battery wires, which lead to connectors outside the case. The throttle plug I didn't use an external connector for (not enough time today to dig around for one), so it's run right from the throttle/key up on the bars down into the case, and plugged into the 3-pin header on the PCB (uses the same connector as a PC case fan). When I have more time, I will cut the connector/wires off a dead case fan, desolder the connector for a fan from a dead motherboard, then make an extension cable that will stay as part of the controller with the external connector available for the throttle. Then I can take it off easily when I want to service it or whatever.

This is the fully enclosed 2QD hooked up to the bike, but not mounted yet.

The cap looks dumb but it should stay cooler, and wouldn't have fit in any other enclosure I had, anyway, except the one that's too large to fit on the bike properly.

It now fits neatly under the seat/cargopod crossbars. It's just ziptied for now, but I'll use a radiator hose clamp once I find one large enough in my junk.

It runs fine like this so far, though it does get warmer. I went around the block and it held up ok. We'll see about a longer trip tomorrow when I go to work. I'll just head out early enough that I have time to just pedal if it blows up or something. :-) I'm way too tired to go out and test it on the road right now--I'd probably crash.

One more thing you can see in this pic if you know what you're looking for is the lighting wiring.

The white wires to the black UPS battery in the tray go to the lighting system. I finally decided that the extra 7-14 pounds I was carrying in the one or two smaller SLAs is too much, and just hooked it up to this part of the traction pack for now. It makes a difference to the bike feel, not a lot but definitely noticeable not to have the other weight there.

While I was at it, I found one radiator hose clamp large enough to go around the battery that's on the cargo rack, but I forgot to take a pic of it. It's now secure enough I have no worries about it bouncing off during a ride anymore. Still gotta find some for the two on the white racks.

Also thinking about moving the cargo rack battery down to the midships area where the controller box used to be, where the two bike frames meet above the motor. Not sure it's quite big enough space, though. If it is, it'd improve the handling even more.


Another part of detail cleanup was to add the brake cutoff for the controller. All it is is a N.O. reed switch (under the foam tape sandwich protecting it, horizontally near it's bottom edge), wired in series with the throttle V+ wire (just like the keyswitch is).
Then a magnet off an old CompUSA name badge holder on the side of the head of a stripped-head bike-accessory screw.

I had to have some way to hold the magnet "upright" so it's field would activate the reed switch, and the smallest bit of steel I had just laying around that didn't have sharp edges I'd have to file was that screw (and others like it). I ziptied one end of the screw, played around with the positioning until squeezing the brake lever just enough to begin engaging the brakes also cut out the motor. Once it was adjusted I added Gorilla Glue with little moisture around it to secure it in place, then the zip tie can come off with no worries about losing it on the road.

That screw doesn't just give the magnet a handy place to stick upright, it also is doing something to shape the field, because without it, I can't get the reed switch to reliably operate for *on* unless the magnet is so close that pulling the brake lever doens't give enough distance to go beyond the hysteresis point of the reed switch and let it turn *off*. With the screw, it works at a nominal distance just fine.

A third thing I actually added a few days ago but forgot about was this little door-holder.

I dont' know what it was originally from, as I found it laying on the ground. It's got two little clips, one on each end, and a weakish long spring between them. I simply clipped one end to a hole already in the top cover, and the other to an existing slot in the side cover, and the tension of it holds the cover open no matter which way the bike leans when parked (it rests on the edge of a cargo pod, since I never put the kickstand back on), even when it's really windy.

It also helps hold it closed, although i don't particularly understand why.

Since it was there, I also took a pic of the repaired Sorenson power supply that I use to finalize the battery charge now that I have to swap the real charger around from pair to pair (it being for 24v and me now having a 36V pack).

It's current limited, so I can set it for 14.4-15.0V (per the battery's label for cycle use) and charge for a while first at the full 1.5A it can handle, then cut it down to half and then 1/4 of that current, as each battery gets closer to full charge (13.6V unloaded after sitting an hour or so off charger).

Kludgy and operator-dependent, but it works for now and only takes a moment to setup and change around, every so often.

I was trying to figure out how much power I am using in my 5-mile work trips, but I'm not exactly sure. The 24V charger (nominally a 3A charger) charges the first pair (A & B) for about an hour and a half at 2amps fading down to .25amps somewhere along the way (I tried to stay and watch it but got pulled away by the dogs every time so far). Then the uncharged one (C) plus one of the first pair (A) are hooked up and it charges for around 20-30 minutes at 2A, then quickly fades out to nothing, leaving C barely recharged and A closer to fully charged. Move it to C & B, same thing, maybe a few minutes less at 2A.

Moving around in pairs like that I let it top off as much as it can, but only takes a minute or less for the 2A charge to stop whenever A or B are in the loop.

So I use the Sorenson to finish charging C, which by itself sucks the 1.5A for maybe 30-40 minutes or more (again, got pulled away by the dogs), and then the battery voltage starts to catch up to the setting on the Sorenson and current drops way down to around 300mA or less, where it tends to stay until I disconnect it.

Unfortunately, with all that shuffling around, I'm not sure how to calculate what power I'm actually putting back into the pack!

Drivetrain v3.1 With "Bench" Test VIDEO!

Since now I am working with these new smaller and lighter 12V 17Ah batteries, I decided to change it to 36V from 24V. This will give me a bit more power than I had before, but requires I change the motor gearing, since it will be about 1/3 faster than it had been, at max.

I was already figuring out what chainring to add to the 2-ring set and how to do it, but this allows me to bypass doing that and leave it as-is.

At the moment I have decided to test it with one of the higher-RPM gearboxes, too. Originally 135RPM at 24V, I'll be using it at 36V and so getting a nominal max of 202RPM. Since that's about twice what I actually want, (90-100RPM), then I can use my newly-switched over chainrings more efficiently than I could the other way.

There's 52 teeth on the largest drivetrain input ring (the one with the cut-off crank arm), and 24 teeth on one of the motor rings I have setup, and 28 teeth on the other. Either one can be easily slipped on and bolted down, and the chain run over it. (Have to add or subtract links, though, because I don't have any other way to fix that right now; no tensioner/etc.). That's a 0.46:1 gear down in speed with 24:52 for 93RPM max at the drivetrain input, or 0.54:1 with 28:52, for 108RPM max.

The 28-tooth version would give me the best top speed, but at highest speed would also make me be windmilling my feet again, while the 24-tooth version would be just about perfect at max throttle as long as I didn't need any "overdrive" from the motor for an "emergency" speed boost, without changing gears. I'll test it with each and see how it feels.

This also moves my pedal chain to the smaller 40tooth ring (which required some indenting of the chainstay frame on the left side to prevent it rubbing on the chain), which combined with the front 41tooth ring gives me nearly 1:1 ratio again, same as I had before with the 3-ring set on this side, but now with the three ring set on the other side gives me my low gears back *and* still have high gears, as if the pedals were directly connected to the regular bike drivetrain, essentially. Pretty much what I was after.

For now, I'm going to go with the 24:52 93RPM max ratio, and see what happens.

Now I don't have to add a third small chainring to the outside of that 2-ring set, which was proving problematic due to location of inter-ring-connect bolts and whatnot blocking chain paths.

With chains on both chainrings right next to each other, the clearance is minimal, but there:

Above is a pic from directly over the top of the chainlines where they cross between the motor and the drivetrain input rings.

That one shows from in front, just under the pedal's chainring. The rustier chain is the one on the pedal's ring.

I took the above pic several times, but the only one you can clearly see the chain clearance in is this dark one, with the light-colored carpet showing between the chains. Smack dab in the middle of the pic is the motor's output hub with the 24 tooth chainring bolted to it (using the bolts and nuts from the 4-pole's wheel hub, since they fit better than the ones I'd scrounged up before).



Now, the moment you've all been waiting for: a new VIDEO! It's not exactly exciting, and it's sure noisier on the camera than it is in real life, but here you go.

EDIT: BLOGGER WON'T EVER FINISH UPLOADING, SO VIDEO WILL BE HERE WHENEVER IT DOES FINISH


There's a yellow DMM setup to measure DC motor voltage, and between the sampling rate of the meter and the PWM frequency, it causes it to be quite variable even though it is not, really. The red DMM is setup to measure DC motor current, which as a no-load current should be 3.6A at 24V according to it's label. It is actually much lower at 36V, as you can see in the video, even at full throttle and my highest gear (with the only load being the drivetrain and the wheel, with all their frictions and masses). The meter is only meant to take 10A, as it doesn't appear to be fused inside (just a shunt on the PCB to measure from). So I tried not to push it too long past that limit.

Even so, when I grab the rear wheel tread with my welding gloves to load the system (which sounds like a table saw because of the MTB tread!), it jumps to above 20A, out of range of the meter to even display. It doesn't slow the wheel down a lot, though it's noticeable, especially at the higher gears. In lower gears, naturally, it has less effect on the speed because there's more torque available to overcome my grip. No matter what gear it's in, at max throttle I still can't STOP the wheel even with both hands on the treads trying to grip as hard as I can, until the friction heat even thru the gloves makes me let go. About the same at the 2/3 throttle lowest setting I can go to right now (till I figure out a quirk described below), which is good.

I'm sure my grip is not the same as the load it will have when riding on the road, but it sure seems good enough at the moment. It *should* be better than it was before, because the weight will be less by around 10 pounds, with 3x 12V17Ah vs 2x 12V 31Ah, even though the max current draw from these is less.

Plus, I've removed the lighting battery, around 6-7 pounds or so, and will be running that off the traction pack. For now that will be just directly hooked to one of them. The automotive turn signals on one, the LED and CCFL lighting on another (so it doesn't flicker with the current drain from the turn signals). At some point I'll get a Roman Black SMPS in place for each one, and have the automotive incandescents replaced with LED grids, but I have had no time to do all the wiring yet.

Back on the new drivetrain topic, the noise is actually quite a racket compared to just a bike drivetrain, but this wheelchair motor is only meant to run at 24v and it's gearbox around 135RPM output, and I'm running it at 36V which ends up a lot faster than it was made for, and thus is even noisier than it was before. Unfortunately even without the gearbox engaged (if I flip the clutch lever off) the motor is still pretty noisy and whiny. Another reason I'd like to get the treadmill motor going with a first-stage grooved-belt drive, with chain as second stage. Should be less noisy than the gears in the gearbox are.

Even with the extra noise from the higher speed, it's still quieter than it was with my crazy derailer-hanger chain tensioner. I *will* need to change the little derailer wheel I have on the brake-lever-throttle-arm-to-be to a rubber wheel off a skateboard or rollerblade, though, because that is a big enough source of noise I actually had to tie it up out of the way to be heard clearly on the video! It's not that bad when riding, but in a closed room with a sensitive-mic'd camera, it's bad.

There is a quirk in the controller I have to figure out where if I run it at 36V, it starts out at around 24V motor output average, around 2/3 throttle, even if I have the throttle at zero, and even though I verified that at the throttle input pin to the comparators, it is at 0V, and should have no output at all. (0V throttle = zero pulse width; 4V throttle = 100% pulse width). I don't know what I did wrong yet, but I'll figure it out--it is probably a component I have in there that doesn't go as high as it is supposed to, and so is pulling down the rest of it. Probably one of my zener-kludges, since I didnt' have the exact values for them, and in at least one case had to series stack two to get the right voltage.

It also takes about three seconds to shut down after I switch off (cutting the throttle pot out of circuit). If it's at full throttle when I shut it off, it immediately reverts to that 2/3 startup throttle speed, then after a second of that it speeds way up to full again, then just stops.

It works perfectly at 24V. (and should also at 36V without any adjustments I can remember; then there are component changes for 48V operation since some transistors (as specified, anyway) might not take the fully-charged voltage of a 48V SLA pack (nearly 55V).

Battery Fairy Now Ships UPS

Of course, that's Uninterrupted Power Supply, not the destructive gorillas in the brown trucks. :-)

The battery fairy came by yesterday, with an old server-sized UPS with both the internal and external battery packs. Unable to sleep due to various things going on in my life right now, I spent the night until dawn or so disassembling the units to get to the batteries and check them out.

The external box just has 4 batteries in it, setup as pairs, two bolt-down wiring tie-in points, and a 100-amp fuse holder to fuse from the next box into this one. The Anderson connector shown on the back is to go to the UPS, then there is another internally-mounted Anderson connector to allow further daisy-chaining of these boxes to the UPS for lenghtened runtime.

It's a nice enough box that I may use it as-is on the trike, if I ever get that far with these batteries (they might not last long enough, as I don't know their age).

They slide out easily enough, with a string-pull on the Anderson connectors to make them easy to disconnect. I'll be leaving those string pulls on just in case. :)

Under a plastic taped-on-with-foam cover are the bolt-on terminals, recessed in each corner of the batteries. Much better than slide-on tab type battery terminals.

The batteries themselves came taped together by a layer of doublesided foam tape between them into 24V "RBC"s, with a 100A fuse bolted between them at one terminal pair, and a 600V/50A Anderson connector at the other.


Those connectors are as useful as the batteries themselves, because it means that now I will have enough of them to make easily disconnected "packs" for the bike. I did not have that option before, and had to bolt on the wires to them each time I mounted or dismounted the batteries from the bike, for fixing it or working on it, or even just to take them off and ride it without motor or batteries for any reason.

All but one pack read at least 23V unloaded; the last read 20V.

Even so they all ran my treadmill motor just fine, even when I put enough load on the shaft to draw about 8 amps (instead of the 1.3A or so it draws unloaded, at 24V).

One battery of the 20V pair is only at 9V, and may have a shorted cell or something. When I put the wheelchair battery charger on this pack, the charger light randomly pulsates at various brightnesses, instead of going off (dark) to show a charging cycle is in progress.

Monitoring the voltage across the low battery shows from 9V to 15V to 17V, while the other is steady at around 15V. So there is almost certainly something wrong with this battery, not likely to be cured by charge cycles.

During charging, maybe 5 minutes in, one of the other packs starts to make very faint bubbling sounds, but loud enough to just barely hear in the quiet room. I had to put my ear up to the side of each battery to figure out which one was doing it. Voltages seem ok on them both, so I'm not sure what's wrong. Current is only 3A max out of the charger, and 5A is the nominal max input current to the battery.

Another pack seemed to take charging normally, with no sounds/etc. After warming up a bit during charge, I unbolted the fuse and Anderson connector from the pack and pulled the two batteries in it apart, so I could read the label/specs of the individual cells. (The outer labels from APC only specify that they're a pack for use in APC stuff, etc, and to use only APC replacements yadda yadda. No numbers).

12V 17Ah by Sacred Sun company, (Shandong Sacredsun Power Sources Industry Co. LTD), max charge current 5.1A. I'm guessing around 13 pounds each, maybe 12.

I'm now letting that pack (as two cells) charge up overnight, then will do the same for the other three.

They're also easier to fit outside my cargo pods. I could even just strap them to the cargo pod-to-bike frame adapter, under the seat, without a battery box, at least for when I'm going to work, since the bike is inside instead of locked up outside where I can't keep an eye on it.

With 8 smaller batteries, of which possibly two might be bad (or will at least fail before the other 6), I have a lot of options.

They can be setup as a series pack for high voltage on that treadmill motor (rated at 120VDC/7000RPM/21A), as a 96V 17Ah pack, at 104 pounds, which would probably not work on *this* bike but should on the trike, and would make it a very powerful long-range vehicle.

They could be setup as a bunch of 24V packs, kept charged and ready to quick-swap out when I get home should I need to go right back out somewhere. If there are any places (like friends' houses) that I have to go where I might need to fully charge up in order to get home, I could leave a pack there on a trickle charger, and simply swap out for the one on the bike when I get there, instead of waiting for what might be several hours (it would also save me the several pounds of charger weight).

36V or 48V packs are also an option, with either one or two packs on the bike, for greater range. I don't know how well the wheelchair motors will hold up under 36V (nominally rated at 24V) but I have tested one of them at that voltage using the three larger 31Ah batteries in series, and it does run, although it is considerably noisier than at 24V.

The treadmill motor is by itself very quiet, but with the alternator fan on it's shaft it is like having a little bitty box fan on high. Moves a lot of air, though, and keeps the motor pretty cool at 24V and 36V. Probably at higher voltages, too, since it's original use at 120V running the treadmill didn't even have a fan on it, just some very minimal vanelike strips on the back of the cast iron pulley/flywheel that came on it's shaft. Those did not move much air at all, even at rated voltage (though they were much quieter). Still, it's a lot quieter than any of the wheelchair motors, and has more raw power potential than they do. Harnessing it's power is more of a challenge, since it has no easy-to-use gearbox built onto it like they do, forcing me to build my own systems for that (like the pulley-then-chain drive I tried it with before I got the wheelchair motors.


Besides the batteries and cabling, there's also a wealth of other EV parts inside the UPS:

More 100A fuses (which I think I'm going to use one of on each battery in case something stupid happens, since it *is* possible to connect batteries together directly in parallel with these Anderson connectors, and if one is fully charged and one is fully discharged, I can imagine some "interesting" things that could happen.

Interesting as in the Chinese curse and as in traditional Chinese fireworks. ;-) I prefer boring to that kind of interesting, and hopefully the un-hole-y kind of boring. (yeah, I can hear your groans from here, just be glad I don't do that a lot more).

Inside the battery box is the 100A fuse on top of this pic, with the internal Andersons to each battery, wired in series for 48V total in this box.

More of the UPS case itself, from the back, just prior to disassembly.

Off comes the outlet and breaker section, since it has these inviting arrows pointing at the screws on it.

Three panel-mount 30A thermal circuit breakers, push-to-reset. They're definitely handy. I think the one I have in the system right now is a 20A.

More Anderson connectors.

The front panel electronics. Not sure I have a use for them yet, but you never know.

The interior, main board, transformers, wiring, and a 120MM fan (24VDC 0.40A). Unknown CFM or sound level; standard 3-pin plug with power, ground, and tach.

Back of the board. It has spots for more MOSFETs and more relays, must be for the next model up or something.

24V 30A SPDT relays. Exactly like the ones used in my ScootNGo controller, except for having twice the SNG's current rating. I will probably use these as contactors for the system, since right now the ignition key only breaks the pot throttle circuit (as specified on the 2QD, which makes it turn off due to "pot fault" detection). That leaves me with no way to totally disconnect power except for disconnecting a battery manually at some point.

The full board. It's quite large, at least one and a half square feet, I'd say.
The four heatsinks each have 5 MOSFETs, IR630P on the front row (can't see the others to tell). Room for 3 more on each heatsink.

Room for more relays, too.

Those MOSFETs are a little hard to read, since they're so lightly printed. None physically looks blown, and at least one in each parallel set must be working, because the gate test works on all 20.

Some very tall caps, too. About 1.5" across, and about 5 inches tall, 75V @ 1500uF. 3 of them.

A whole board full of various kids of caps, chips, transistors, diodes, etc.


Many of the components to build a second full 2QD are on this board, perhaps everything but the LM339 (which I still have another one).