GGoodrum
1 MW
I figured it was time for a new thread, so I could stop cluttering up others with bits and pieces on this direct-drive stuff. 
Anyway, what I'm doing is using an AstroFlight 7-turn 3220 (kV: 97) to directly drive a 3-speed hub on a 20-inch West Marine Port Runner folding bike. Normally, I would use a Shimano Nexus-3 hub, but I found a Sturmey-Archer SRF3 hub already laced into a nice 20" rim at UtahTrikes.com for a total of $49, so I decided to try one. So far, so good. It seems to shift as smooth as the Nexus, and it is almost exactly the same size, so I'm hoping the internals hold up as well. We'll see.
First, here's some pics:
View attachment 6
View attachment 5
View attachment 4
View attachment 3
View attachment 2
View attachment 1
The drive part of this is really pretty simple. There's separate chains to the rear for the motor and for the pedals. Both rear sprockets are attached to the SRF-3 hub, via a standard Nexus 16T splined steel cog (also available at UtahTrikes.com...). I'm using a Nexus cog because they are dished slightly, which helps in the alignment of the chains. This steel cog is sandwiched between a 22T 4-hole 64mm BCD "granny gear" aluminum mountain bike sprocket, and a 94T aluminum #35 heavy duty go kart sprocket I got from gokartgalaxy.com. This sprocket already comes with a 2-1/8" hole and four holes in a compatible 64mm BCD pattern. The 16T cog lines up perfectly, so it literally just bolts together. I used some small 3/8" wide spacers I got from McMaster, to space the two sprockets apart a bit. This assembly then slides onto the splined hub and is held in place with the circlip. The big #35 sprocket is on the inside. The 22T sprocket ends up being perfectly aligned with the chainline coming from the crankset.
To keep the motor from driving the pedals, I replaced the front crank with an IPS crankset, which has an embedded freewheel. I'm having some adapters made, however, that will eliminate the need for having to add a FW to the crank. The adapter will replace the 22T sprocket with threads that will allow a standard BMX freewheel to be screwed onto the adapter. These come in sizes from 16T to 22T, so it will be easier to match a stock setup. Anyway, the adapters aren't quite ready yet, as my machinist guy needed to order a special bit to cut the fine-pitched 24 TPI threads, and I already had the IPS crankset and a 22T 64mm BCD sprocket, so I went with these to get this going.
The #35 sprockets and chain are much wider than even the 1/8" BMX chain/sprockets. I like this a lot.
Very robust, and will certainly handle the power/torque the 3220 puts out. The finer pitch (3/8" vs 1/2"...) allows for higher ratios, which is really what makes this direct-drive setup possible. I'm currently using an 11T motor sprocket (also from McMaster...), which gives a reduction to the hub of 8.55:1. These 3-speed hubs have 1:1 ratios in 2nd gear, and about 1:0.75 in 1st. That bumps up the reduction is 1st to about 11.4:1. Third gear is like "overdrive", and has a reduction of 1:1.33. This ends up giving unloaded top speeds of about 22 mph in 1st, 29 mph in 2nd and 39 mph in 3rd.
I've only done a couple quick test runs so far, but I already know there's way too much power/torque in 1st. It literally threw me of the back the first time I goosed the power in 1st. The acceleration is pretty amazing, but you quickly run it up to the unloaded speed, in like a couple seconds. Even starting in 2nd, I can get the front end to pop up, if I give it too much juice. I was only hitting 90A/4kW peaks, but this motor has twice the torque of the 3210 on my other bike. Just like on my original Cyclone 1000W setup, I can't see needing to use 1st except if I'm stopped at the bottom of a steep hill. Anyway, I need to take another look at the gearing for this beast. I think I'll try a 14 or 15T motor sprocket.
Initially, I'm using the same 15s4p a123-based pack I've been using on the other bike, but I'm just finishing up a new 12s3p 45V/15Ah LiPo-based pack, made from six 20C 6s-5000 Turnigy packs from HobbyCity.com. This pack has a 12-channel LVC board, which uses the 3.0V TC54 voltage detectors on each channel, and has an active cutoff section that uses four IRFB4110 FETs to cut power if a cell hits the LVC point. I will also use this active cutoff feature with a small kill switch I'll mount somewhere. Anyway, I'm still finishing up the matching 12-channel Charge Balancer that will be used to keep the cells balanced during the charge process. For charging, I'm using a small Vicor 48V/4.5A power supply, with the output tweaked to 49.7V. The balancer is set so that each channel charges to 4.13V. This will increase the longevity of the cells.
The controller is a standard Phoenix HV110 with the PWM rate set to 16kHz and the timing left in "Normal", which means 5-10 degrees of advance. I also disable the current limiting and brake functions, and leave the auto-throttle mode enabled. I'm using a semi-custom version of Fecther's throttle board, without the current limiting function. This unit also has the equivalent of a Dimension SHV BEC, and it has four big 330uF caps, to aid the controller's ability to absorb the big spikes we see with ebike-type loads. The throttle is a half-twist hall-type that was used on this bike previously, when it had a Clyte 5303. Finally, I wired in a WattsUp meter, and attached it to the top.
The motor mount is simply an angled piece of 3" x 3" x 3/16" aluminum that attaches to a U-shaped "foot" that is welded to the bottom bracket tube. This is normally used as a "resting" spot for the bike folded up. Anyway, this "foot" is sandwiched between the angle piece and a flat 3"-wide plate. The motor mounts to the angled piece, as does the controller. There's just enough vertical play to allow for proper tensioning of the chain. One of the reasons I picked this bike to do the first direct-drive setup is because it doesn't have a bottom chainstay that would get in the way of the chain. For a more "normal" installation, which I will be doing next, I have a different mount which attaches to the downtube and is up against the bottom bracket tube, to control the torque rotation. This puts the motor right in front of the crankset. More about that later.
Today I need to put some "dimples" into the motor shaft for the moor sprocket setscrews. I have a couple of "flats" that I hand-filed onto the shaft, but this isn't enough to handle the torque of this setup, as the setscrews just can't hold for very long. I'm going to drill some dimples, like Matt does, which should do a better job of holding. Once I get this done, I'll do some more test runs. Unfortunately, there aren't too many hills here on Mission Bay, where I'm at this week, but I can still do some more acceleration tests, check the top speeds, and check the temps. In the few "blasts" I did initially, the motor never got above ambient (about 78F...) and the controller was only slightly warm to the touch. I've got my little temp sensor widget, though, so I'll do some more thorough checks later today. I'll try and get some basic video, if I can figure out how to strap my iPhone to the bike.
All-in-all, I'm very happy with this setup. The installation is a ton simpler, and the big #35 chain and sprockets are perfect for this application. The hub works great, although I really don't need 3-speeds on a 20" bike, I think. This motor actually has all six wires brought out, so I'm thinking I may try a left-side direct-drive setup, with delta-wye switching, on a future version.
-- Gary
Anyway, what I'm doing is using an AstroFlight 7-turn 3220 (kV: 97) to directly drive a 3-speed hub on a 20-inch West Marine Port Runner folding bike. Normally, I would use a Shimano Nexus-3 hub, but I found a Sturmey-Archer SRF3 hub already laced into a nice 20" rim at UtahTrikes.com for a total of $49, so I decided to try one. So far, so good. It seems to shift as smooth as the Nexus, and it is almost exactly the same size, so I'm hoping the internals hold up as well. We'll see.First, here's some pics:
View attachment 6
View attachment 5
View attachment 4
View attachment 3
View attachment 2
View attachment 1
The drive part of this is really pretty simple. There's separate chains to the rear for the motor and for the pedals. Both rear sprockets are attached to the SRF-3 hub, via a standard Nexus 16T splined steel cog (also available at UtahTrikes.com...). I'm using a Nexus cog because they are dished slightly, which helps in the alignment of the chains. This steel cog is sandwiched between a 22T 4-hole 64mm BCD "granny gear" aluminum mountain bike sprocket, and a 94T aluminum #35 heavy duty go kart sprocket I got from gokartgalaxy.com. This sprocket already comes with a 2-1/8" hole and four holes in a compatible 64mm BCD pattern. The 16T cog lines up perfectly, so it literally just bolts together. I used some small 3/8" wide spacers I got from McMaster, to space the two sprockets apart a bit. This assembly then slides onto the splined hub and is held in place with the circlip. The big #35 sprocket is on the inside. The 22T sprocket ends up being perfectly aligned with the chainline coming from the crankset.
To keep the motor from driving the pedals, I replaced the front crank with an IPS crankset, which has an embedded freewheel. I'm having some adapters made, however, that will eliminate the need for having to add a FW to the crank. The adapter will replace the 22T sprocket with threads that will allow a standard BMX freewheel to be screwed onto the adapter. These come in sizes from 16T to 22T, so it will be easier to match a stock setup. Anyway, the adapters aren't quite ready yet, as my machinist guy needed to order a special bit to cut the fine-pitched 24 TPI threads, and I already had the IPS crankset and a 22T 64mm BCD sprocket, so I went with these to get this going.
The #35 sprockets and chain are much wider than even the 1/8" BMX chain/sprockets. I like this a lot.
Very robust, and will certainly handle the power/torque the 3220 puts out. The finer pitch (3/8" vs 1/2"...) allows for higher ratios, which is really what makes this direct-drive setup possible. I'm currently using an 11T motor sprocket (also from McMaster...), which gives a reduction to the hub of 8.55:1. These 3-speed hubs have 1:1 ratios in 2nd gear, and about 1:0.75 in 1st. That bumps up the reduction is 1st to about 11.4:1. Third gear is like "overdrive", and has a reduction of 1:1.33. This ends up giving unloaded top speeds of about 22 mph in 1st, 29 mph in 2nd and 39 mph in 3rd.I've only done a couple quick test runs so far, but I already know there's way too much power/torque in 1st. It literally threw me of the back the first time I goosed the power in 1st. The acceleration is pretty amazing, but you quickly run it up to the unloaded speed, in like a couple seconds. Even starting in 2nd, I can get the front end to pop up, if I give it too much juice. I was only hitting 90A/4kW peaks, but this motor has twice the torque of the 3210 on my other bike. Just like on my original Cyclone 1000W setup, I can't see needing to use 1st except if I'm stopped at the bottom of a steep hill. Anyway, I need to take another look at the gearing for this beast. I think I'll try a 14 or 15T motor sprocket.
Initially, I'm using the same 15s4p a123-based pack I've been using on the other bike, but I'm just finishing up a new 12s3p 45V/15Ah LiPo-based pack, made from six 20C 6s-5000 Turnigy packs from HobbyCity.com. This pack has a 12-channel LVC board, which uses the 3.0V TC54 voltage detectors on each channel, and has an active cutoff section that uses four IRFB4110 FETs to cut power if a cell hits the LVC point. I will also use this active cutoff feature with a small kill switch I'll mount somewhere. Anyway, I'm still finishing up the matching 12-channel Charge Balancer that will be used to keep the cells balanced during the charge process. For charging, I'm using a small Vicor 48V/4.5A power supply, with the output tweaked to 49.7V. The balancer is set so that each channel charges to 4.13V. This will increase the longevity of the cells.
The controller is a standard Phoenix HV110 with the PWM rate set to 16kHz and the timing left in "Normal", which means 5-10 degrees of advance. I also disable the current limiting and brake functions, and leave the auto-throttle mode enabled. I'm using a semi-custom version of Fecther's throttle board, without the current limiting function. This unit also has the equivalent of a Dimension SHV BEC, and it has four big 330uF caps, to aid the controller's ability to absorb the big spikes we see with ebike-type loads. The throttle is a half-twist hall-type that was used on this bike previously, when it had a Clyte 5303. Finally, I wired in a WattsUp meter, and attached it to the top.
The motor mount is simply an angled piece of 3" x 3" x 3/16" aluminum that attaches to a U-shaped "foot" that is welded to the bottom bracket tube. This is normally used as a "resting" spot for the bike folded up. Anyway, this "foot" is sandwiched between the angle piece and a flat 3"-wide plate. The motor mounts to the angled piece, as does the controller. There's just enough vertical play to allow for proper tensioning of the chain. One of the reasons I picked this bike to do the first direct-drive setup is because it doesn't have a bottom chainstay that would get in the way of the chain. For a more "normal" installation, which I will be doing next, I have a different mount which attaches to the downtube and is up against the bottom bracket tube, to control the torque rotation. This puts the motor right in front of the crankset. More about that later.
Today I need to put some "dimples" into the motor shaft for the moor sprocket setscrews. I have a couple of "flats" that I hand-filed onto the shaft, but this isn't enough to handle the torque of this setup, as the setscrews just can't hold for very long. I'm going to drill some dimples, like Matt does, which should do a better job of holding. Once I get this done, I'll do some more test runs. Unfortunately, there aren't too many hills here on Mission Bay, where I'm at this week, but I can still do some more acceleration tests, check the top speeds, and check the temps. In the few "blasts" I did initially, the motor never got above ambient (about 78F...) and the controller was only slightly warm to the touch. I've got my little temp sensor widget, though, so I'll do some more thorough checks later today. I'll try and get some basic video, if I can figure out how to strap my iPhone to the bike.
All-in-all, I'm very happy with this setup. The installation is a ton simpler, and the big #35 chain and sprockets are perfect for this application. The hub works great, although I really don't need 3-speeds on a 20" bike, I think. This motor actually has all six wires brought out, so I'm thinking I may try a left-side direct-drive setup, with delta-wye switching, on a future version.
-- Gary
That should be interesting, for sure.
