My new Custom GMAC build

AHicks

Well-Known Member
Region
USA
City
Snow Bird - Summer S.E. Michigan, Winter Gulf Coast North Central Fl.
Different strokes, got it. It would be a giant priority/reason to buy here, but that's me. Sorry for underestimating your understanding and research.

As a MAC 12t owner for the last couple of years, I do agree the changes made in the design of the GMAC (torque arm, axle, side cover) are great reasons to go GMAC as well. I would just question the value of the locked up clutch if you aren't going to use regen.
 

CityExplorer

Well-Known Member
Very exciting stuff here. I'l like to see a more simple optical based based brake-lever for regen, it would not be as dynamic/sensitive as a load cell, but more reliable, and maybe easier. I would think one could get 5-10 levels with individual optical sensors. A fully variable one based on intensity probably could be done, but that again I think would be less reliable. A system that takes into account the current motor RPM would be the best I think and prevent full lockup. Maybe CA already does this with their Throttle control. I'm not sure what tech is in the Thumb throttles, is it just some resistance control?
 

linklemming

Well-Known Member
As a MAC 12t owner for the last couple of years, I do agree the changes made in the design of the GMAC (torque arm, axle, side cover) are great reasons to go GMAC as well. I would just question the value of the locked up clutch if you aren't going to use regen.
Agreed, if the virtual freewheeling feature wasnt available, I would have never considered the GMAC, Hopefully Justin will offer a clutch version in the future. Justin seems really focused on regen which is the last thing I care about. Others have commented that they have contacted justin directly and he had a few prototype clutch unit assemblies available.
 

AHicks

Well-Known Member
Region
USA
City
Snow Bird - Summer S.E. Michigan, Winter Gulf Coast North Central Fl.
Agreed, if the virtual freewheeling feature wasnt available, I would have never considered the GMAC, Hopefully Justin will offer a clutch version in the future. Justin seems really focused on regen which is the last thing I care about. Others have commented that they have contacted justin directly and he had a few prototype clutch unit assemblies available.
It was my understanding the GMAC would use the same clutch the MAC motors used. You would certainly want to confirm that prior to buying one though....

From the pictures I've seen, they do appear to be very similar! If that did work out, would be a great mod for your application.
 

AHicks

Well-Known Member
Region
USA
City
Snow Bird - Summer S.E. Michigan, Winter Gulf Coast North Central Fl.
AHicks,

Curious if you have tried the KT field weakening opensource stuff available on endless sphere.
Honestly, even though I have a pretty good grip on most things mechanical and electrical, the open source conversations on ES are above my pay grade. I could likely sit down and read and reread it enough to comprehend what they are doing, but I just haven't had the need. I AM a big fan of the KT controllers and displays though.
 

linklemming

Well-Known Member
Finally got my rear brakes(shimano deore 6100). My original order is still stuck in IL, this is my second order. Brakes installed without a hitch. I can post pics tomorrow.

As with any new bike, there is always chasing down all the squeaks.

Only seemed to get squeaks when I was in the saddle so greased everything on the seatpost/seat mount. Also tightened the crank arm to the spider just to be sure.

About 10 miles into a ride tonight, the squeaking was so bad, I decided to turn around.

Tried taking the crankarm off on the drive side and spent about 1/2 hour trying to get a crank puller on. It appeared the threads on the crankarm were messed up, in the end I just think it was paint overspray on the threads. I finally just got the best feeling for being in the threads although the puller wasnt 100% straight and just committed to forcing/screwing the puller in(I dont recommend this but I was out of options). I was lucky and got past the paint overspray. I think the squeaking was coming from paint overspray on the inside of the crankarm square taper interface causing a less than perfect fit(despite being on tight), it just didnt look like the arm was on the square taper the same as on the non-drive side.. I could see where due to this less than perfect fit the aluminum on the inside of the crankarm was marred up a bit. Spend a few minutes with acetone, xacto knife and files to cleanup the taper interface on the inside of the crankarm. Im pretty sure I caught it in time (the square sections were still square). The crankarm finally looked like it bolted on to the correct depth on the BB.

Overall, I am more than impressed with this bike/build. Not much else to do other than enjoy the ride and play with CA settings.

This build was meant to be a more offroad capable version of my Juiced CCX(with marathon Plus MTB tires). It has exceeded my expectations and is a blast to ride. I only really liked riding the CCX flat out and this bike is fun flat out or just cruising, has better offroad grip and can get to 32mph+ on the throttle if I want. Its also 50# as opposed to the CCX at 65# Also nice and quiet, on par with my Brose bikes untill you get over about 750W. Most of my riding is with 1x or 2x assist topping out about 500watts...so its basically silent most of the time.

I will/do miss the 52V 19.2AH battery on the CCX(this build has a 52V 13.5AH). I will likely be getting a second battery for this build.
 
Last edited:

linklemming

Well-Known Member
Another ride and the drive side crankarm is still giving me issues. Brought a long Hex wrench on the ride and stopped several times and the crankarm bolt was loose several times. The third time I tightened it seemed to have worked as it didnt loosen anymore from that point on. Came home and put a torque wrench to it and tightened the crankarm bolt to about 380inlbs (spec is 305-391inlbs according to park tools). Fwiw, this is pretty friggin tight, quite a bit tighter than I was using before. The crankarm, never acted like it was rounding off on the square taper bb axle, but was just made noises and the mounting bolt wasnt fully tight. . Alot of controversy on what to grease and what needs to be dry. I put grease on the square taper, on the crankarm bolt threads but not on the crankarm bolt washer. This was from sheldon browns website with inputs from jobst brandt.

Hopefully that will fix the crankarm loosening, if not I will be getting some new crankarms.

Changed the CA to have assist levels at 0.5,1.0,1.5,2.0,2.5,3.0,3.5,4.0,4.5,5.0. I really like riding in 1.0,1.5 and 2.0 unless I want to just go fast and then it doesnt really matter what setting it is.

Im using a 52V 13.5ah pack with about 700wh, I seem to be averaging 15wh/mile for a range of about 33 miles. Since my longest rides are typically 35-40miles, this seems perfect. Im usually ready to get off the bike after 30 miles.
 
Last edited:

AHicks

Well-Known Member
Region
USA
City
Snow Bird - Summer S.E. Michigan, Winter Gulf Coast North Central Fl.
Re: power levels, just an idea if you don't mind? Electric motors make more torque at lower rpms (from lessons learned in radio control personally, and born out, easily visible using Grins spreadsheet), so I beleive when it comes to assigning power levels, rather than use linear values (5,10,15,20, 25, etc.) exponential values will give more linear, more progressive motor response. Find something that works good at speeds where you struggle to keep your balance as a starter, then make smaller changes as you progress through the first half of the values, making up for the smaller changes made in the second half of the values - if that makes any sense. To illustrate, KT controls uses uses 5 steps (one for each PAS level), starting at 13%, then 20, 33, 50, and finally 100%. These can be split easily of course if you need 9 or 10 values. I tried this idea when programming my Ultra, and couldn't be any happier. There are no big holes or jumps in available power as you range up and down through the PAS levels, even while monitoring the watt meter in the display. Noteworthy, with the Ultra programmed with this idea anyway, is the great progression in available power through the first few power levels. Kinda like having a close ratio manual transmission in your car. Seems like there's always a value that's "just right" for anything you want to do. Only takes a minute to set up/change, so it might be worth a try with your rig. If you decide it might be worth your trouble, holler with your riding impression. -Al
 

linklemming

Well-Known Member
Re: power levels, just an idea if you don't mind? Electric motors make more torque at lower rpms (from lessons learned in radio control personally, and born out, easily visible using Grins spreadsheet), so I beleive when it comes to assigning power levels, rather than use linear values (5,10,15,20, 25, etc.) exponential values will give more linear, more progressive motor response. Find something that works good at speeds where you struggle to keep your balance as a starter, then make smaller changes as you progress through the first half of the values, making up for the smaller changes made in the second half of the values - if that makes any sense. To illustrate, KT controls uses uses 5 steps (one for each PAS level), starting at 13%, then 20, 33, 50, and finally 100%. These can be split easily of course if you need 9 or 10 values. I tried this idea when programming my Ultra, and couldn't be any happier. There are no big holes or jumps in available power as you range up and down through the PAS levels, even while monitoring the watt meter in the display. Noteworthy, with the Ultra programmed with this idea anyway, is the great progression in available power through the first few power levels. Kinda like having a close ratio manual transmission in your car. Seems like there's always a value that's "just right" for anything you want to do. Only takes a minute to set up/change, so it might be worth a try with your rig. If you decide it might be worth your trouble, holler with your riding impression. -Al
Totally agree!!

When I played with the opensource TSDZ2, you could assign the actual assist# for each level. I ended up doing like you describe

Doesnt appear to be the case for the CA, it evenly distributes the assist so to get the resolution I want down low requires alot of unnecessary steps up high.

If 3 assist levels was enough I could just assign each preset (there are 3) to a set assist level and program the CA to switch presets instead of assist levels with button presses. I tried this and didnt like it
 

AHicks

Well-Known Member
Region
USA
City
Snow Bird - Summer S.E. Michigan, Winter Gulf Coast North Central Fl.

linklemming

Well-Known Member
I played around with the CA some more (added 2 more presets).

One of the nice things about presets is you can have it default to any one you choose on powerup.

Here is the setup:
Power up to preset 1
Preset 1(legal) - 20mph, 10 levels of assist, up to 3x assist, up to 750 watts - basically a class 1 ebike with a little more max power
Preset 2(speed) - 28mph, 10 levels of assist, up to 4x assist, up to 1000watts - basically a class 3 like my Juiced CCX(The CCX is probably more like a 5x with 1000watt power limit)
Preset 3(unlimited) - unlimited mph, 10 levels of assist, up to 5x assist, up to 1500ish watts - stupid fast above 4x assist

Note that the names in parenthesis are predefined and cant be changed, there are other names as well like offroad.

The drive side crankarm seems to be holding now that I torqued it down to proper specs. I will likely order another torque sensor/bb from grin (with crankarms) just to have a spare

Here is a pic from todays ride after being snowed in for two days.

Things to note:
1.) The neoprene/velcro sleeve to hide all the wires up front. Initially I laughed when I saw this but it works very well. I had planned on making a custom wiring harness but not sure I will anymore.
2.) I used 6mm spiral wire wrap for wire from the CA to the phaserunner under the top tube, encapsulating the rear shifter cable as well except at the frame attachment points.
3.) I used 6mm spiral wire wrap for wire from the Torque sensor to the CA under the downtube tube, encapsulating the rear brake housing as well except at the frame attachment points.

I am also a big fan of all the CA displays which includes wh used, wh/mile and the battery gauge which isnt just based on voltage but wh consumed (you input your battery specs like voltage and ah). FWIW, My juiced CCX seemed to be the same way. With this 700wh battery, 80% discharge is 560wh. I typically ride up to 500wh, its nice to see the wh so you can judge the real battery capacity left.

As I have mentioned before, I usually ride at 1x,1.5x or 2x assist and at these power levels, the GMAC is quieter than the smart sam tires. I REALLY LIKE THIS EBIKE😁😁😁😁😁

final-troll.jpg
 
Last edited:

AHicks

Well-Known Member
Region
USA
City
Snow Bird - Summer S.E. Michigan, Winter Gulf Coast North Central Fl.
Nice clean install. Good job!
 

linklemming

Well-Known Member
Looks really good. Are you planning on mounting a water bottle cage somewhere?
Funny you mention that, when I was looking at the pic it dawned on me that there was no room for a water bottle in the typical spots. I could always mount the phaserunner (the silver box on the seat tube) on the top tube and move the battery forward. Or even mount the phaserunner on the bottom of the downtube.

Also, since this is a bikepacking bike, there are TONS of potential water bottle mount points (24 bolt locations) although many would not be convenient.

That being said, I have been riding with a camelback backback since they first came out in the 90s.
 
Last edited:

Ravi Kempaiah

Well-Known Member
So here is the ride report summary: WOW

Seriously, I couldnt be happier with this setup.

Frame: More than stiff enough for the power Im putting thru it. This frame is rated for a 300# rider with 55#gear. I have owned/ridden flexy touring frames and this frame is solid. Maybe too solid.

Tires: Smart Sam 26x2.35. I really like these tires. I dont do alot of pavement riding (just ride pavement to get to more dirt). Better offroad traction than my marathon plus MTB.

GMAC+CA+torquesensor: I am TOTALLY HAPPY with this setup. Not as responsive as my Brose bikes....but really close. While I love the Brose bikes when going really slow and uphill and in technical terrain (eMTB), all of that is less important as you get to higher speeds and gravel rides which I built this bike for. Definately better than my juiced CCX as far as responsiveness. Everything is adjustable thru the CA but I am pretty happy with most of the stock settings. I initially had it setup with up to 4x assist of human power with 4 assist levels. With Just pedaling input, it would top out about 26mph on level ground, With throttle on level ground I got it up to 32mph (1550watts), up to 40mph on a 3% downhill. I think I will go up to 5x human power and 5 levels of assist. Might take off the throttle and leave in my camelback as it just seems like the most boring way to ride an ebike...IMHO. An interesting observation compared to my Juiced CCX

Great build!
Was curious to see how responsive this setup was. I am glad you are thoroughly enjoying the setup.
On one of the Zen models (shakti that will go on sale next year), we are using a similar torque sensor and controller and we have found it offers very decent performance.
 

linklemming

Well-Known Member
Great build!
Was curious to see how responsive this setup was. I am glad you are thoroughly enjoying the setup.
On one of the Zen models (shakti that will go on sale next year), we are using a similar torque sensor and controller and we have found it offers very decent performance.
Its not as responsive as my two brose bikes (class 1 and class 3) but more responsive than my Juiced CCX.

I am still using mostly stock CA settings but to be honest, it seems good enough.

It feels alot like a TSDZ2 conversion I did with opensource software. I did look at that code and the assist was based soley on user horsepower (force x cadence). I think systems like brose have an additional boost feature to make it more responsive. The opensource software also had that but I never played with changing it as the TSDZ2 became useless after about 1k miles due to bearing slop.

I have found that the more responsive a torque system is, the more I enjoy running it at lower assist levels and getting more of a workout. While my CCX was a great bike, I only ever enjoyed riding it fast. I can see from my garmin logs that I am burning more calories on the same rides with this build. If I really need to go fast, its only a few button pushes away.
 
Last edited:

Johnny

Well-Known Member
Its not as responsive as my two brose bikes (class 1 and class 3) but more responsive than my Juiced CCX.
It may be programming, Asi controller seem to have 1khz sampling rate for the pedal sensors which is on par with the mainstream mid drives like bosch.

Can you monitor your cadence/torque / motor output from the display? It may be nice to see if there is any delay in those outputs.

Your build looks very nice and clean btw.
 

linklemming

Well-Known Member
It may be programming, Asi controller seem to have 1khz sampling rate for the pedal sensors which is on par with the mainstream mid drives like bosch.

Can you monitor your cadence/torque / motor output from the display? It may be nice to see if there is any delay in those outputs.

Your build looks very nice and clean btw.
It might very well be the programming but there are still a few things I can play with although I am happy with the way it is

That being said, its very good and the differences could be considered subtle. Its definately responsive just not the connected to my brain feeling of the brose. It might just be the difference between adding power thru the drivetrain vs thru the wheel hub. At higher speeds, this isnt as important although it still feels nicer on the brose. If I dont think about it and just enjoy the ride, its the last thing on my mind.

Now that I think about it, thats how the last gen bosch systems felt to me...just lacked that connected to my brain feeling.

The ERIDER torque sensor does appear to be a quality unit as it measures torque from both pedals with 18 pulses per rev. Since it measures from both sides, the CA manual recommends averaging for 1/2 of the pulses or 9 pulses. If it only measured on 1 side then CA manual recommends doubling the pulses so 4 times the averaging and more latency. I could try reducing this(torque assist averaging) although it might introduce pulsing of the motor if it becomes too responsive due to less averaging. There is also the start threshold setting.
passettings.PNG


Note that the CA only communicates throttle outputs to the phaserunner so the throttle output settings might be affecting this, notably the PAS ramp up rate
thsettings.PNG


In the end, Im totally happy with with the settings as is but will likely play with these in the future (because I can :cool: )
 

Johnny

Well-Known Member
18 pulses per rev. Since it measures from both sides, the CA manual recommends averaging for 1/2 of the pulses or 9 pulses. If it only measured on 1 side then CA manual recommends doubling the pulses so 4 times the averaging and more latency. I could try reducing this(torque assist averaging) although it might introduce pulsing of the motor if it becomes too responsive due to less averaging.
Interesting, as you said the more pulses you average the more lag that is introduced while the output being smoother. I would imagine it pulsate if the sensor signal is too noisy. Also 18 sounds a bit low.

I hope you share your experience with smaller averaging numbers (1, 2 )

Do you know how many times Brose samples the torque sensor btw?