Hyena
10 GW
Sweet, I should have known Justin would be all over such a thing
teklektik said:D'oh!Ypedal said:How cool would it be to wire up the e-brake cutoff to your trigger shifters,...
This is a much simpler solution than (1) or (2) that I mentioned in the post above about shifter/interrupters. I automatically shy away from involving the ebrake inputs because of DD regen considerations, but I didn't think this through - regen cannot be a consideration in the mid-drive case. The ebrake approach has the advantage of involving the CA in ramping/gain control.
Tidy and nifty!
teklektik said:Well - jerky starts aren't desirable but they can be 'normal' for a higher torque rig (eg MAC) that needs a little tuning.RC_guy said:I had to turn down the power gain to 30 to remove the jerkiness. Is this normal?
...
One thing I would like to tweek is that I find it difficult to give just a little bit of throttle. Even if I apply only a tiny bit, it gives about 15 amps. Once I get going then it will decrease and it is easier to modulate but, as soon as I get back on the throttle, it still wants to give more than I want.
Here's your earlier settings:
There are a few things in play here that can contribute to jerky starts:RC_guy said:Up Rate = 21 V/s
Down Rate = 50.1V/s
Fast Rate = 24V/s
Fast Thrsh = 3 Amps
Max Current = 45 amps
AGain = 80
- An aggressive FAST START setting for ramping.
Fast start can be tricky to set up properly so that it works well under all driving conditions - particularly if you are using PAS or looking for very low power getaways on bike paths or to walk the bike. Part of the problem is that the default setting of ThrO->FastThrsh = 3A is a whooping large power level to achieve before the ramping logic switches over to the 'slow' ramp defined by UpRate. I recommend that you set FastThrsh = 0 to start and address this fast start optimization after you are happy with the entire remaining setup and performance. This will disable the FAST START feature completely and the CA will rely entirely on UpRate.
Particularly for PAS, this can immediately remedy the startup surge. I used no FastRate for well over a year and only recently began using it with FastThrsh = 0.20A.
- High UpRate setting.
Let's look at what the various xxRate settings mean: The ramping is a clamp that prevents the throttle output from exceeding the configured rate. It is not in play for slow increases/decreases that do not exceed those rates. A typical throttle has a range of about 3v so a rate of 3v/sec means if you slam the throttle ZERO-to-WOT in PassThru, the ramping will hold back the throttle application so that it takes 1sec to traverse the 3v range. 6v/sec will take 1/2sec, and in your case, 21v/sec will take only (3v)/(21v/s) = .14sec (PDQ! - very little ramping at all).
Here we get into matters of riding style, vehicle weight, motor torque, and drivetrain type so there is no 'correct setting' and examples may be no more useful than the default setting. That said - for a MAC (torquey and with breakable clutch) I think you might turn your ramping down a lot - to something in the 1v/s - 5v/s region. That might be excessive, but you get the idea - you are presently way up there... (I actually run lower than that but my bike is heavy with two over-volted gear motors and I use the motors to jackrabbit from dead stops across intersections - no special driving practices necessary to protect the clutches - just slam WOT with no manual feathering. )
- High AGain setting.
You describe an overshoot siutation where the initial power seems high but then moderates when underway. That's a bit hard to interpret but in conjunction with your AGain setting of 80, it seems you are describing an overshoot situation. Here the CA looks at the 'error' between the current you requested by turning the throttle and the actual current and applies a correction to the throttle voltage to correct. The magnitude of that correction is controlled by the coefficient AGain. If this is too large, the CA applies a whooping big correction that causes an overshoot. If this is too small, the correction is very slow or possibly even insufficient to achieve the target. So the trick is to get a balance that seems to work 'best' generally. I would recommend cranking it down to 10 and increase it from there. At 10, the throttle will be pretty laggy, but the overshoot should be gone and you can quickly work back up to a more desirable setting.
- Non-linear throttle output.
In the end, the CA Current Throttle guarantees that the battery current will exactly track the throttle voltage - which for a perfect throttle is nicely linear with rotation (e.g. 25% rotation = 25% max throttle voltage). However, crappy hall throttles have non-linear rotation-to-voltage curves and so even though the CA performs as advertised, the throttle response has steps or deadish zones. Here you need to get a more expensive hall throttle like the Wuxing that has a single linear magnet (instead of two magnets and a non-linear magnetic field between them) or a resistive throttle like a Magura.
- Varied driving scenarios.
At the end of the day, even with a nice linear throttle response, you may find the throttle response insufficiently refined at the beginning for walking the bike or comfortable low speed operation. The problem is that for a multi-thousand Watt bike, even a linear throttle makes control of a few percent of the total power difficult or tiring.
Enter the CA 3-position switch. This allows you to cut the power for different types of driving situations. If your bike has a max current of 60A and your switch selects 25% current then the max current the CA will allow is 15A. With Current Throttle, a 33% throttle rotation will apply only 5A. This gives refined and easy to control operation. If you don't have one, this may help out low speed operation if that is presenting an unpleasantness. These switches can be flipped on the fly under power, so the inevitible 'Oh crap!' situation won't leave you strapped for power. I have a 3-position switch rigged for low power bike path use, full speed cruising (motor-won't overheat-at-WOT), and flank speed (aka boost! - motors will overheat if applied too long).
Although the analogy is severely flawed (current vs gearing), this is not unlike driving in 1st in the parking lot and shifting to 5th on the highway. The point being that there is some explicit operator selection of the drivable speed range...
Actually, yes. Changing the throttle curve is quite simple to do with a few resistors if you have a resistive throttle like a Magura. The procedure is described in this post. I never got any feedback on this and so didn't include it in the Guide as perhaps being a little too tweaky for the average user in spite of the simple construction.RC_guy said:Is there a way to change the throttle curve?
It seems you have lots of EM3EV parts so I'm guessing you got the V3 from there as well. Paul sets up a number of parameters in the V3 so it should arrive in a fairly workable state.efergy kb said:Issue:
- As per RC_guy post below; the initial acceleration is quite jerky. After it hits about 20km/hr - acceleration smooths out
- At 33% (Low speed switch); you can feel the sudden surges and loss in power in like 1 second waves until it hits 20km/hr speed..
- At 66% (Med speed switch); the acceleration seems to be better; there is a more gentlier and you can just notice the loss in power before the acceleration smooths..
My Setup:
MAC8T, 12fet 25A, 50V 12.3Ah Samsung Triangle Pack 14S6P)
- Have a CA v3, and 3 speed switch attached. Controller is limited to 33% Low, 66% Medium, 99%High speed. Also current is limited to 30A.
Cycle Analyst v3 settings: I have no idea so i thought i'd post as much as i ca
Setup ThrotIn 0.80V 13% Thru
Setup Aux Po 2.39V 74% Pwr
This matter of more powerful mid-drives came up in recent posts and although the V3 works pretty well with modest power mid-drives (eg EdgeRunner and StokeMonkey), really powerful units are an area where there is not a lot of experience.izeman said:i also play a lot with throttle ramping and different throttle mode (current, power, pass-thru) but am not happy.
the mid driver per se has a lot of lag. the motor has to turn, then there is the reduction gearing, and lastly the chain.
so if you are in a low gear and apply 2kW it's hard to apply that little throttle so the bike just won't flip over. if you add some throttle lag (slow up-ramp) it becomes better, but then it's hard to controll as soon as the bike is moving. twisting the throttle now is so slow that it takes too much time until power is applied. is there a way to tell the throttle act differently depending on speed? (and i don't think that speed-mode is the right throttle mode for that).
As mentioned in a recent post, it is fairly simple to get throttle mappings similar to what you desire.izeman said:so what is would like to see: some kind of exponential throttle signal. and some damping. if you turn it slowly and only a bit from the start there is very little power applied. if you turn it WOT you apply a lot of power immediately.
Sample curves that can be accomplished with a few resistors as described in that post:teklektik said:Actually, yes. Changing the throttle curve is quite simple to do with a few resistors if you have a resistive throttle like a Magura. The procedure is described in this post. I never got any feedback on this and so didn't include it in the Guide as perhaps being a little too tweaky for the average user in spite of the simple construction.RC_guy said:Is there a way to change the throttle curve?
The difficulty may be that although it is accurate, it may not be entirely revealing as to the effect.izeman said:i should have read that better. i don't know how often i flew over these lines and never may have understood it.
I see from your thread that you have an unusual Thun application in a hand cycle. I'm not sure if this introduces any special considerations on the Thun regarding operating torque range or hands vs feet - the hands construction does have longer effective levers to multiply the human torque and the minimum cadence assumption (55rpm) may not be spot on...Wheels_78 said:Can anyone tell me what the maximum torque assist level is on the CA3? Looks like it can be set to 99, but I'm assuming it's something less than that... or maybe not. Thanks.
The Thun can detect a torque of up to 200Nm which corresponds to a torque voltage of about 4.5v with a Trq->Offset of 2.5v yielding a 2v range for 0-200Nm. The V3 'Thun' setting uses a fixed Trq->TrqScale = -200Nm/V to yield the proper voltage-to-torque conversion (-100Nm for the base Thun conversion and a fudge factor of 2x to account for only measuring torque on the left pedal side).Unofficial Guide said:16.Torque-based PAS – Operation:
The CA takes the BB torque voltage, subtracts Trq->TrqOffst to normalize the voltage relative to the zero-torque signal level, and multiplies the result by Trq->TrqScale to convert to Nm of torque. Using this torque with the larger of 55rpm or rider cadence, it calculates the rider contribution in 'Human Watts'. It subtracts Trq->AsstStart from 'Human Watts' and multiplies the result by Trq->AsstFactr to determine the assist power to deliver. An external switch or pot hooked to the AuxPot input scales Trq->AsstFactr 0-100% to give adjustable assist based on the rider torque and cadence. The determined assist level is used to control the motor using the Power PID controller which is configured by PLim->MaxPower and PLim->WGain.
teklektik said:I see from your thread that you have an unusual Thun application in a hand cycle. I'm not sure if this introduces any special considerations on the Thun regarding operating torque range or hands vs feet - the hands construction does have longer effective levers to multiply the human torque and the minimum cadence assumption (55rpm) may not be spot on...Wheels_78 said:Can anyone tell me what the maximum torque assist level is on the CA3? Looks like it can be set to 99, but I'm assuming it's something less than that... or maybe not. Thanks.
Anyhow, to set the stage - the Guide describes the calculation:
The Thun can detect a torque of up to 200Nm which corresponds to a torque voltage of about 4.5v with a Trq->Offset of 2.5v yielding a 2v range for 0-200Nm. The V3 'Thun' setting uses a fixed Trq->TrqScale = 25Nm/V to yield the proper voltage-to-torque conversion; so for a voltage range of 2.5v to 4.5v the V3 calculates a maximum torque of 0 to (4.5v-2.5v)*(25Nm/V) = 200Nm -- exactly as specified.Unofficial Guide said:16.Torque-based PAS – Operation:
The CA takes the BB torque voltage, subtracts Trq->TrqOffst to normalize the voltage relative to the zero-torque signal level, and multiplies the result by Trq->TrqScale to convert to Nm of torque. Using this torque with the larger of 55rpm or rider cadence, it calculates the rider contribution in 'Human Watts'. It subtracts Trq->AsstStart from 'Human Watts' and multiplies the result by Trq->AsstFactr to determine the assist power to deliver. An external switch or pot hooked to the AuxPot input scales Trq->AsstFactr 0-100% to give adjustable assist based on the rider torque and cadence. The determined assist level is used to control the motor using the Power PID controller which is configured by PLim->MaxPower and PLim->WGain.
As described above, the CA works in the cadence and the Trq->AsstStart setting and arrives at a 'Human Watts' figure. This is multiplied by Trq->AsstFactr to derive the 'Motor Watts' of assist to apply (scaled by AuxPot Assist level if present).
At the end of the day, your final assist level is affected by two primary multipliers: Trq->TrqScale (which was chosen to scale the Thun output over the entire Thun torque range) and Trq->AsstFactr. AsstFactor has a max value of '99.99 times' (finally got to your question!). Assuming your rig can apply 2000W of battery power and using an Assist Factor of 100, you could achieve that maximum assist by generating 2000W/100 = 20W of human power. So - even though the range is limited to 99.99x, the achievable maximum assist is substantial - for feet.
Are you experiencing a shortfall in assist power because of hand propulsion (ie insufficient detected Thun torque or cadence)?
BTW - You should add a 'Hand Cycle' link to your signature - nice build with interesting considerations!
Nope - no limit - have at it!Wheels_78 said:2. I'm not seeing substantial assist until I'm in a higher gear than I'd like to be and really grunting on the cranks. This is hard on the joints etc. and as you may guess my arms and shoulders are pretty important to me. I didn't realize the trqscale multiplier could be mucked with but I presume that if I did that it will effect the accuracy of the calculated human watts etc. and I don't want to do that. Sounds like I just need to up the assist factor until it behaves as I'd like it to.
I was just wondering if there was a practical limit to the setting that was lower than digital limit would indicate. Sounds like there isn't.
You didn't explain the details, but you don't want to get into a situation where you are holding the assist level like a throttle - otherwise just use the throttle. But - you really do need to get the assist level control set up so you can diddle it on the fly.Wheels_78 said:...this is very close to what I want but I was still forced to rely on the throttle on hills if I didn't plan ahead and set the assist factor up.
...
Tonight I'm going to set the throttle up on the aux port to be an on the fly assist adjustment, ...
teklektik said:So - I'm thinking that since the 55rpm limit is not adjustable, you might trick the CA by reducing your pole count. Since the Thun pole setting is 8, you might (for the 30-70rpm example) reduce this to 4. Some adjustment of the start/stop thresholds would also be required. With this setting, when you row at 30rpm, the CA will think you are waling away at twice that or 60rpm. This would supersede the 55rpm limit and you would immediately be getting rpm-scaling as you rowed faster. This should make downshifting more attractive as the assist should increase. Since your 'human watts' figure is already bogus because of the 55rpm limit, this change will actually improve things since it will now be accurate but twice actual. Just a thought....
teklektik said:You didn't explain the details, but you don't want to get into a situation where you are holding the assist level like a throttle - otherwise just use the throttle. But - you really do need to get the assist level control set up so you can diddle it on the fly.
teklektik said:Looking at your build I'm thinking that adding a stub bar to the left rowing handle (similar to the one on the right that mounts brake, etc) might be a good plan and you could add the assist level control there. You might use a 1/2 twist or thumb throttle with the spring removed (see this post (use o-rings for drag)) but these have a short throw and the adjustment may not be as refined as desired. Another idea might be to use a friction shifter hooked to one of gwhy!'s hall throttle boxes.
This is wonderful news!Wheels_78 said:...the bike is a completely different animal...
mrbill said:Folks:
I just noticed an inconsistency in the human average power/energy calculation. The problem appears to have been present for a long time, but I didn't notice it because I never bothered to check the calculation myself until now.
The CAV3 (Prelim6) gives three data,
HWhrs
AvgHW
TripTime (that I assume is moving time and is the basis for the calculation) converted to hours
So, knowing any two of these one should be able to calculate the third.
HWhrs = AvgHW * TripTime
AvgHW = HWhrs/TripTime
TripTime = HWhrs/AvgHW
As I keep careful records of my rides, I have checked several of these, and for each of these rides the calculation is inconsistent.
Example from a recent ride:
The CAV3 (Prelim6) gives me the following stats at the end of a ride:
AvgHW = 164 w
HWhrs = 717.0 wh
TripTime = 6:13:47 = 6.22972 hours
Calculating AvgHW gives:
AvgHW = 717.0/6.229722 = 115 w
Calculating HWhrs gives:
HWhrs = 164 * 6.229722 = 1022 wh
Calculating TripTime gives:
TripTime = 717.0 / 164 = 4.3719512 hours = 4:22:19
Each of these calculations yields a figure that is significantly different from that displayed on the CA screen. I do not know which of these is in error, or for that matter which is correct.
Is this a known bug for which a fix is in the next firmware release? If not, could someone please independently verify this and report back to the forum?
Thanks.
delphinus said:I have the CAv3 working well, I got it from em3ev with the battery. Firmware is V3.0 Prelim6.
I'd like to connect to it from the computer to run the setup config utility. However I forgot to order the relevant USB programming cable. I already have a usb to serial adapter, so have wired up a 3pin earphone jack with GND, TXD and RXD to match the picture http://www.ebikes.ca/product-info/grin-products/cycle-analyst-3.html
When trying to connect with the utility the CA shows PC CONNECTION but it hangs on "Attempting to connect to the cycle analyst" and I get the error "Unable to read data from the cycle analyst". I have tried cycling the power and trying again to no avail.
Using a Win8 PC. Was able to use the same serial adapter to program my controller happily.
izeman said:no cal. that can't be done. shunt value is a global value. meaning it doesn't change based on profiles. i guess moving the ca from bike to bike or chaning the controller frequently just wasn't an expected usecase.
but you can program ALL values really fast with the ca programming tool. might come handy.