Planning/build thread, DH frame, Mini-E, QS 205 w/ LG HG2

Hi all, this thread is for planning my second ebike build, my plans are still at an early stage. My first bike, which I am commuting on daily, is this:

https://endless-sphere.com/forums/viewtopic.php?f=3&t=66256

It is still performing very well, however I have caught the ebike bug and now must satisfy my desire for more speed, comfort, range, as well as some off-road capability.

The most critical factor is frame selection. I am looking at second hand XC and enduro mountain bikes, however the market here is both more expensive and illiquid than the second hand market in the US. I am particularly looking for feedback on what battery construction and mounting options are available these days, to try to find a winning combination of frame and battery.

I am reasonably set on using an Adappto Mini-E (will the new ones rated at 4 kW be out soon?) with the BMS, as it seems like the most well-integrated system available. The choice of motor is pretty open, but I may be constrained by dropout width, and I will prefer to run the DC bus voltage close to the limits of the controller (~84V pack).

As a starting point, between combinations of 4S LiPo and a custom LiFePO4 pack, what has the best option of fitting in a frame geometry such as this?

http://jedprentice.com/images/bike/epic-07-r.jpg

Specialised Epic Marathon, relatively narrow triangle but with no rear shock in the way.

At the moment I am leaning towards 24s Headway LifePO4, quite probably a custom pack.

This would deliver ~60 km/h on a 26" rim with a 6T, 6kV MXUS, if I can find one, and I would be happy to rely on field weakening to deliver more top speed if required (most likely not).

I could perhaps even go down to 20s Headway, that would make 2p 10Ah more viable than 1p 15Ah. Thoughts?

I have a donor!

https://imgur.com/a/UtMlL

It's a 2007 (?) Mongoose EC-D downhill bike. Some lovely componentry, Marzocchi 888, Fox DHX 5.0, and Hope Mono M4 brakes front and rear on 180mm rotors. Needs a bit of a clean up, and then to figure out if any of the components need servicing.

I have also discovered the wonders of QS Motor, and it looks like a V3 QS 205 50H would be a great match for the Mini-E. I may have to discuss my needs in the QS thread and see what they are offering for wheelbuilding:

http://qsmotor.en.alibaba.com/product/60244187355-801642985/QS_3000W_205_50H_E_bike_Spoke_Extra_V3_Type_Hub_Motor.html

I'm thinking that it could do well with an inverted-U type wraparound battery pack, the frame is aluminium so I'm not sure if anything can be directly welded to it, but I should be able to figure something out.

I am also interested in possibly machining some custom dropouts for better clamping, and maybe even extend the wheelbase. Is that a thing?

Cross-posted from my reply to the ever-helpful Ohbse:

---

When I was planning for LiPo, I had a plan for a top-mounted aluminium box, fitted to the top bar with either bolted angle brackets, or two form-fitting brackets that would be secured to the bottom bar with some kind of strap. I will have to work out some more mockups and see how a 20s10p pack could be built to fit around that configuration.

Did you need to order any extra cells for contingencies, or do much testing before you assembled the pack? It wouldn't be too much extra to order another 10. I will need to sort out a full BOM for pack assembly.

The LG HG2 looks good, I might look to order some soon. My goal is to be able to chuck maximum Mini-E amps at both the motor and battery uphill continuously without having to worry about overstressing either. The QS 205 50H motor looks potentially even stronger than the cromotor in this regard, with a really good copper fill and a huge stator.

If I end up being able to extend the swingarm with custom dropouts, and go to 19"x1.85 rims front and rear, the geometry will end up more closely resembling something like a supermotard, which would be desirable.

---

Waiting on a reply from Vito about ordering a QS motor on 19" rims. Thoughts guys?

Just ordered 205 LG HG2 from cl-rd.cn. $4 USD a piece. Some spot welding is in my future.

Nice 2.214 kWh thats heaps of battery capacity!

How was the shipping cost for all those batteries? I've been finding that shipping LiPo to NZ is getting more and more expensive.

Definitely going to keep an eye on this build

$138 USD for shipping via UPS, I bet you could do better than that with a slower shipping service though.

Yesterday I mocked up some possible battery pack configurations with cardboard.

https://imgur.com/a/DvWhc

For 200 cells, using hexagonal tight packing yields two square 200x160x70 box sections, or two 320x100x70 box sections. Using square 20mm packing with spacers yields 200x200x70 and 400x100x70 boxes respectively. Is heat dissipation likely to be a problem if using tight hexagonal packing at 2C discharge?

I mocked up the hexagonally packed arrangements with a few mm extra in cardboard, and taped them to the bike to see how they would fit while riding.

Depending on the eventual seat height, it seemed like the square sections would be liable to be hit with my knees, but the two long sections mounted on top of each other (measured as 150mm high) seemed like they would not present any problems.

I would like to use a motocross style seat if I can find one, but I can't locate anything being commercially produced to fit bicycle seat mounts. A previous owner had snapped the top of the seat post mount off, hence the seat post is currently riveted into place on the remaining part of the seat post mount. I might have to rivet a new adjustable post in its place, depending on what kind of seat I can find. I want the seat height to be maybe 100mm higher than where it is now, as well as allowing room for the battery packs to comfortably fit underneath, and perhaps even be integrated with the flat motorcross seat. Does anyone have any recommendations for these seats?

I'm going with a 5.5T 8.1 kV QS 205 50H V3 motor in a 19x1.6" rim. I think this will deliver the efficient mid-range torque production I would like, while field weakening can deliver more top speed if required.

5.5T:
8.3*70*pi*0.6*60/1000 = 65 km/h no-load speed in a 600mm wheel at 70V (3.5V max discharge), 75 km/h at 80V (HOC 4.0V).

5T:
9.0*70*pi*0.6*60/1000 = 71 km/h no-load speed in a 600mm wheel at 70V (3.5V max discharge), 81 km/h at 80V (HOC 4.0V).

Mini-E arrived today, 5.5T QS motor ordered.

In the past few days, I have also ordered halls/magnets for a controllable regen brake mod, a Domino 60 degree throttle, as well as 15m of 0.3mm nickel for the pack. Any comments on my earlier query about the viability of hexagonal vs. square/spaced packing for heat transfer?

Still to order:

19" tyre
Balance leads/connectors

Still to design and manufacture:

Battery box
Custom dropouts

I installed the unlimited free trial of Autodesk's Fusion 360, I will have to see how much my meager Solidworks experience translates into that. I will try to match the existing locations of the brake and shifter mounts. So far as I know, keeping a shifter is still a good idea even running single speed just for the chain tensioning, right?

Fusion 360 offers static stress analysis; I'd be interested in hearing from anyone who has experience translating dynamic ebike loads into static analysis, or any experience designing strong parts in general. I do have access to a manually controlled vertical milling machine, which should be up to the task. Also interested in recommendations for grades of steel to use. I will post some pics once I've laid out the required dimensions.

My LG cells arrived today!

I'm still designing the capacitive discharge spot welder I want to use to make the pack. It will have to be relatively hefty to handle 0.3mm nickel.

Luckily, I have a surplus of a few hundred 50V 3.3 mF electrolytic caps, which together should make a pretty saucy low-ESR pulse capacitor bank, ~660 mF at 20-40V ought to do it, roughly 500J available.

I also have a bunch of 1.9 mOhm 60V Infineon MOSFETs. In my current design, I will parallel about 15 of them. With a proper snubber design, there is no reason to ever blow MOSFETs in pulsed discharge applications!

https://imgur.com/a/KbhEA

Simulations with relatively large total series inductances of over 1 uH show that current is nicely shared between the paralleled FETs; the welding probes are connected at either end of each bus. The FETs will bridge two copper busbars directly (hence the 20 microohm simulated resistance between them). At a 10 mOhm snubber resistance, each 100A translates into an extra peak blocking volt. Peak FET currents are 62A for a 800A pulse, giving peak power dissipation of only 12W or so in each FET during the pulse.

Hopefully this should be able to sustain practically continuous duty when connected to a DC supply so long as the probes and connecting leads are bulky enough.

Sounds bloody awesome mate you should be able to do some damage with that setup!

Regarding heat dissipation in the batteries, im sure others on ES have packed them hexagonally without issues and on higher discharge rates. If you know the internal resistance of the battery you can calculate the power loss P=I^2 R

Some progress to report. Most of the major bits have arrived, including the awesome Domino throttle, and a cheap domestic 63A circuit breaker. The motor will be a while longer, and the nickel strips from Aliexpress have shipped but could take an indeterminate amount of time with China Post.

Other miscellaneous items coming ex Aliexpress:

Balancing leads
Halls, magnets for regen brake conversion
Keyswitch (nice find, can remove in either off or on position: http://www.aliexpress.com/item/Key-Operated-Security-Barrel-Switch-Lock-Pole-Single-SPST-ON-OFF-2-Position-Key/32268499109.html )

I've got the free version of Fusion 360, and after a bit of a learning curve and time spent tracking down where they had hid all the required but seemingly basic functionality such as aligned dimensioning, I've modeled the existing removable dropouts on the bike, which should allow me to make meaningful comparisons about size and strength with the integrated FEM static analyser.

I've also made up at least some of the capacitor bank I will use for a spot welder. Sitting at around 350 mF (100x mixed 3.3 and 3.9 mF), or about 260J at 40V. I will get started with the rest of the circuitry so I can evaluate how much stored energy I will really need at various voltage levels and load types. I'll chuck in a few reverse protection diodes and bleed resistors to the bank before I get too carried away.

https://imgur.com/a/qHKjq

(hey admins, any chance of adding resize html to your img tags?)

Welder beginning to take shape, 420 mF capacitor bank sorta wired up with reverse protection diodes and (very slow) bleed resistors.

15x BSC028N06NS 2.8 mOhm 60V FETs soldered to my protoboard, with holes under the gate pads that I will attach to the common gate drive plane on the underside with 10 ohm gate drive resistors. Then it's just wiring up the snubber and gate drive, and sorting out the pulse controller. Might do some initial testing with a digital function generator, then put an Arduino or something on it. I'd love to get all fancy with an LCD display, we'll have to see if I can be arsed.

https://imgur.com/a/KCloi

Went into the lab on a fine Sunday to work on the welder, made some good progress.

I tested the gate drive circuit, hooked up to the 12 BSC016N06NS FETs (3 empty slots on the board due to removing some while debugging, thought 12 will still be plenty for now) in parallel, it made light work of them, driving them at 7 V for a 250 ns rise time and 100 ns fall time.

I made up some welding leads from triple braided speaker cable, hooked up to 8 mm terminals.

Using a 10 amp current-limited source, I made some kelvin measurements of the voltage drop across the whole circuit including the leads. 27 mV total, yielding 2.7 milliohms (!!!) total series resistance. This neglects the ESR of the capacitor bank itself, and the extra lead that will be required from the negative terminal of the ban to the common-source busbar.

Measured busbar-to-busbar, the voltage was 3.8 mV at the 7 V gate drive voltage, or 3.4 mV at a 12 V gate drive voltage, giving a total switch+busbar resistance of 0.34 milliohms. Based on the nominal Rdson of 1.6 milliohms, 0.13 milliohms of that will be the FETs, and the rest will be in the busbars and solder joints.

Pulse testing to begin next week, I'm going to need to get myself an arduino or something. Also find some welding probes that will hopefully bolt straight into those 8mm terminals.

https://imgur.com/a/0zHgh

subscribe

Awesome work sir!

Regarding probes, I am using a couple short lengths of of 5mm copper rod clamped to 0 gauge leads. These work very well, they require the tips to be touched up once every 60 welds or so in my experience, few strokes of a file and good as new.

if you want to try using 0.1mm copper tabs, i have had great success with 8mm copper probes with a short piece of tunsten rod, 1.6mm dia, set into the ends.

the power required to get a good weld must be quite substantial, and needs to be in a very short pulse

My word, that pack looks beautiful. 12p? I think 22s12p has to be the sweet spot for for a Mini-E build designed for robustness and longevity. At 22s, I would go for an even slower wind (6T QS 205?), 90V HOC at 4.1, 88V at 4.0, and a bit less under load to max out the efficiency of the controller without hitting the voltage limits. Sweet sweet efficient midrange torque.

I have still not been able to source a 19x2.75 Shinko 241 or equivalent in New Zealand.

http://www.torpedo7.co.nz/products/MITYXNNM5/title/michelin-starcross-ms3-front-70-100-19

These would fit a 19x1.6 rim right? Bit knobbly for street riding, but I want to know if I'm on the right track.

Any tips or dealers would be appreciated!

Welder progress hit a bit of a snag, I am going to have to beef up and reorganise my gate driver system to be able to hold the parallel gates down more firmly in the presence of extremely large di/dt commutations. On the bright side, this has not so far resulted in any hardware failures. All the testing so far was with dumping 1-2V into a hard short, producing expected peak currents of 300-400A. I will need to sort out a CT or something to take the current measurements, the 50A passive current probes will not cut it.

edit: got an SR244 in 2.75x19 from Ebay with mildly exorbitant shipping. Looks like a better tread than the 241 for road use.

Trade me had some cheap tyres back when I was looking. There were a bunch new CST 90/90R19 for $10 each. maybe not what you're looking for but they make a decent road tyre

Those look pretty nice, I will have to see how the 244 looks once it's on and see what kind of clearance I will have remaining, it might get a bit tight.

Tyre has arrived. It looks like it will be a fine fit if I go with the option to extend the wheelbase with the custom dropouts. Waiting on the motor.

You can see the rough geometry I will be going for here, as well as a mockup of the proportions of the battery box. I should be able to get it fabricated from aluminium quite cheaply. I haven't quite figured out what I want to do for securing the box to the frame, or how to secure the lid that will be form-fitting over the top and sides (I think it could be nice to salvage a locking clasp of some kind on the front face, and hinge the rear?)

https://imgur.com/a/ynRRW

Need to finish the welder, got bogged down with the gate drive. Commutating thousands of amps is never a walk in the park. I'm looking forward to proving some ideas about weight distribution and handling!

Looks great! Battery location is ideal for handling, additional wheelbase should let you use the full capability of the mini e. thoughts on controller location? Phase leads should be as short as possible was always my motivation.

Don't get to caught up in overbuilding the welder - if my cobbled together proto board mess works for thousands of welds then I'm sure your first pass will be more than enough!

Unlike mine yours could end up being very capable off-road, I do get asked a lot how well mine goes in the dirt weirdly

Not sure about controller placement. Possible options:

1) Mounted on top of the battery box, or possibly inside an extended height battery box, thermally bonded to the enclosure for heatsinking. This would mean a battery box that is more like 200 mm high instead of 160 mm, which either means I definitely have to go for a new, taller seatpost (is currently riveted in place), or an exposed controller if mounted on top. Inside would be cleaner, and I would expect that if well bonded to the lid, the thermals would still be ok. Controller stays high and dry, which is good, but long-ish phase wires are required.

2) Under the front 'wheel arch'. Probably tricky to mount, good airflow, but also long phase wires.

3) Behind the seatpost. Could be quite clean if I figured out a good mounting option, maybe integrated with the mounting of the battery box itself. Short phase wires, but long runs of pretty much everything else.

Phase currents will not be that high, as the motor is 8.1 kV, and it will only be 5-6 kW max. I can also run arbitrarily thick wires if I need to.

Subd