Cycle Analyst V3 preview and first beta release

Get real world experience and user feedback on the electric bicycle products.
Tench   100 kW

100 kW
Posts: 1033
Joined: Jul 13 2010 12:48pm
Location: Derby UK

Re: Cycle Analyst V3 preview and first beta release

Post by Tench » May 17 2012 1:20pm

I was thinking more of a guide to see where i am when i am in the menus as they are much larger now, and if i had a paper copy i could write down my settings at the side of each menu position, it would enable me to navigate straight to each individual setting easier when making adjustments.
Its come along way from the earlier version 2.2 and 2.25 that i also have.

Simon.
Project one http://endless-sphere.com/forums/viewto ... ilit=tench
Project Two http://endless-sphere.com/forums/viewto ... =6&t=37489
The uk's first Stealth Bomber

There is a box inside your head, inside amongst others are the words "Wont fit" stop looking for answers in the box, there are no answers only excuses.

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » May 17 2012 1:29pm

Tench wrote: ... if i had a paper copy i could write down my settings at the side of each menu position, ...
yep - I use the printable text file version to save settings before each reflash...
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
hjns   100 kW

100 kW
Posts: 1273
Joined: Aug 20 2011 11:05pm
Location: Basel, Switzerland

Re: Beta 14 firmware

Post by hjns » May 17 2012 2:30pm

OK, I flashed to B14 and it works on the bench. Have not yet tried it out on the bike. The B14 indeed solves the 10k thermistor issue. I have now an error of about 4oC with one thermistor. Good enough.

The 3rd digit for LVC allows me to set an LVC of 108.0V (3.6V cell level), which is great. The main screen shows my voltage correctly (1/3 used lipos at 117V), however the next screen shows only the 17V, and misses the 1st one hundred, as mentioned in a previous post by Justin. A bit weird, but not a real problem.
justin_le wrote: 1) The function of holding the left and the right button are swapped. So you hold the right to enter setup, hold the left for a reset. It took a bit for my old habits to get used to this but it does in the end make a lot more sense
Works like a charm - much more intuitive.
justin_le wrote: 2) Adds the ability to select which screens are shown and hidden, and you can choose one set of screens that appear when stationary and another set (typically a smaller set) to show up when the bike is moving. Right now there are 11 different display screens with various data and you change these by scrolling through with the left and right buttons, then holding the right button to toggle. Binary fashion, 1=shown, 0=hidden
Very nice. As I don't use the PAS thingies, I got rid of them.
justin_le wrote: 3) Attempts to display a bit more sensor data inside the setup menu so that it's easier to know what values to put in. So for instance, at the throttle input screen, you can see the actual realtime throttle voltage go up and down as you move the throttle, and then you'll know what to put for max and min throttle range values.
Very handy. I did not save my settings before flashing, so seeing the actual throttle voltages were great. I also took the chance to recalibrate the CA. Ended up with a rshunt of 2.385mOhm for my non modified Lyen 124115 controller.

justin_le wrote: I'm working on doing things so that the CA has a 12th screen that outputs a bit more diagnostic info on what limiting modes are coming into play which would hopefully shed light on these scenarios.
A screen which shows the temperature results continuously, instead of alternating with speed and Ah, would be great for my testing. I strongly believe that the temperature gauge is the one real hurdle for high power setups. For tweaking, it really is much nicer to have a continuous view of what is happening before any limiting is incurred by the CA. Now I have to rely on the CA limiting for 2/3rds of the time.....
Henk


All men dream: but not equally. Those who dream by night in the dusty recesses of their minds wake in the day to find that it was vanity: but the dreamers of the day are dangerous men, for they may act their dreams with open eyes, to make it possible. T.E. Lawrence

User avatar
Sacman   10 kW

10 kW
Posts: 750
Joined: May 27 2008 3:42am
Location: Corona & Irvine, California, USA

Re: Cycle Analyst V3 preview and first beta release

Post by Sacman » May 17 2012 5:13pm

justin_le wrote:
Yes, we decided to make the DC power port standard on all the CA's a few months ago but haven't yet got around to updating all of the documentation. The idea is to make it a convenient access point for those wanting to power small handlebar accessories (headlights, horns, handgrip warmers, stereos, GPS/ipod chargers etc.) without running an additional cable line back to the battery. Just a note that if you aren't using this, to leave the rubber cap on the DC plug so that it isn't at risk of shorting or corroding since the full pack voltage is present there.
Probably also a good idea to recommend a 3A fuse be placed at the CA's connection point to the battery to protect the wiring should an excessive load (or short) be placed at the end of that new DC plug.

User avatar
Diamondback   10 kW

10 kW
Posts: 525
Joined: Dec 15 2008 3:52am
Location: Sydney, Australia

Re: Cycle Analyst V3 preview and first beta release

Post by Diamondback » May 19 2012 10:25am

ColinB wrote:Hi,
This is a cool new development, and hopefully will draw in newbies like myself. I like how the kits offered at ebikes.ca are "open source" vs. Bionx. However, I like how the Bionx rides. Sounds like soon I will be able to have both. My brief tests weren't even though - comparing a heavy EZ bike with internal gears vs. a mid level hybrid with a 350w Bionx isn't exactly fair. But it was enough to tell which my preference was.

One question: I know the torque gauge adjusts the power based on your input, but is the amount variable while you are riding? For example, on a Bionx bike, when you are tired, you can press the "+" button to increase the assistance level. After you get to the flat you may turn it back down to conserve your battery. Will the CA be able to do this?
Colin

i too would be very interested to know the answer to this question.

Jason.
Ignorance can be solved, stupid is forever

User avatar
hjns   100 kW

100 kW
Posts: 1273
Joined: Aug 20 2011 11:05pm
Location: Basel, Switzerland

Re: Cycle Analyst V3 preview and first beta release

Post by hjns » May 19 2012 10:44am

ColinB wrote:Hi,
This is a cool new development, and hopefully will draw in newbies like myself. I like how the kits offered at ebikes.ca are "open source" vs. Bionx. However, I like how the Bionx rides. Sounds like soon I will be able to have both. My brief tests weren't even though - comparing a heavy EZ bike with internal gears vs. a mid level hybrid with a 350w Bionx isn't exactly fair. But it was enough to tell which my preference was.

One question: I know the torque gauge adjusts the power based on your input, but is the amount variable while you are riding? For example, on a Bionx bike, when you are tired, you can press the "+" button to increase the assistance level. After you get to the flat you may turn it back down to conserve your battery. Will the CA be able to do this?
Colin
Me thinks one could use the 3-speed switch for this, connected directly to the CA. The V3 also accept an AUX input, to select thingies on the fly.
Henk


All men dream: but not equally. Those who dream by night in the dusty recesses of their minds wake in the day to find that it was vanity: but the dreamers of the day are dangerous men, for they may act their dreams with open eyes, to make it possible. T.E. Lawrence

csm   100 W

100 W
Posts: 118
Joined: May 03 2012 12:42am

Re: Cycle Analyst V3 preview and first beta release

Post by csm » May 19 2012 10:59am

justin_le wrote:
hjns wrote:Hi all,
If you don't have access to a programmer and want to use the B12 firmware, you can either a) connector two of your thermistors in parallel which will effectively behave like a single 5K thermistor so that the ratio is right, or b) replace the 4.99kOhm pull-up resistor on the PCB (R1) with a 10K pull-up.
-Justin
Justin, have you made any video tutorial showing the installation of a Cycle Analyst (including shunt, thermisters, etc)? I am also interested in seeing a video tutorial on doing a firmware update on the Cycle Analyst.

If you haven't, please consider doing so.

User avatar
hjns   100 kW

100 kW
Posts: 1273
Joined: Aug 20 2011 11:05pm
Location: Basel, Switzerland

Re: Cycle Analyst V3 preview and first beta release

Post by hjns » May 19 2012 11:40am

csm wrote:Justin, have you made any video tutorial showing the installation of a Cycle Analyst (including shunt, thermisters, etc)? I am also interested in seeing a video tutorial on doing a firmware update on the Cycle Analyst.

If you haven't, please consider doing so.
Actually, I found the CA manual on this page very clear. There were some oddities in my own setup which I did not completely understand. When I asked the questions here on ES, I usually got an answer within 24h. This great forum covers so many details as could never be covered in a video tutorial. Similarly, all the different ways I see many ES users use their CA would be impossible to reproduce in a video.

Anyway, if you have a problem with installing the CA, I would suggest to open a thread and ask the question. My bet is you will get a useful answer within 24h. If only to refer to the correct page in the CA manual... :mrgreen:

Just my 0.02
Henk


All men dream: but not equally. Those who dream by night in the dusty recesses of their minds wake in the day to find that it was vanity: but the dreamers of the day are dangerous men, for they may act their dreams with open eyes, to make it possible. T.E. Lawrence

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » May 19 2012 2:18pm

csm wrote:I am also interested in seeing a video tutorial on doing a firmware update on the Cycle Analyst.
If you haven't, please consider doing so.
Please see this post for basic setup - see the links at the end for posts describing the reflash process, etc.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

lizardboy   10 mW

10 mW
Posts: 34
Joined: May 31 2011 6:19pm

Re: Cycle Analyst V3 preview and first beta release

Post by lizardboy » May 19 2012 5:34pm

Just tested the new code and the throttle cutting out issue remians exactly the same.

I turned off every limiting feature and its the same.

Changed to speed throttle and it works fine. Except for surging between full power and off to control speed but I'm sure I could tune parameters for this. Switching to pass thru and just using the speed control built in to the infineon is the best though.

If I Change back to the version you emailed me of b13 and it works great reflash to the version posted here for b13 and the problem returns, Update to b14 and the problem is the same.

The second current throttle ramps up to almost full motor power it cuts out. Its not a limiting issue. Unless the CA is throwing out 5 volt spikes that are tripping the infineons throttle protection circuit I have no answers of my own. I only know that you had it working till you changed whatever chaged between version b13 and b13b. Are you sure your test bench is presenting a full and proper load? Happens everytime for me within the first 3 or 4 seconds of operation. Should be pretty easy to duplicate.

I'm running a Mac 8T motor at 10S with a lyen (infineion) 9 fet controller. Should be a pretty standard setup. If I get a chance today I'll try disabling the controllers built in throttle protection. All the other limiting factors are set to the 99's and watt gain is at 50 as suggested.

Can someone else re-verify this still occurs with the new version?

Waldo   1 µW

1 µW
Posts: 3
Joined: Dec 07 2011 9:45am

Re: Cycle Analyst V3 preview and first beta release

Post by Waldo » May 20 2012 11:27am

Feature Request
~~~~~~~~~~~~~
It would be really great to have a way to send a complete edited setup parameter list to the CA, perhaps through the bootloader.
That would greatly speed up setting changes for testing and insure correct settings by eliminating human entry errors.
It would also be most helpful to have the parameter list included in the header of the log file to confirm that all parameters are currently set as expected.

Thanks for the most awesome development device ever,
Waldo

User avatar
Joe Perez   100 W

100 W
Posts: 187
Joined: Apr 06 2011 10:41am
Location: New York City

Re: Cycle Analyst V3 preview and first beta release

Post by Joe Perez » May 21 2012 12:02pm

So I've encountered kind of an interesting phenomenon with the temperature sensing- it seems to be influenced by the throttle input in a way I would not expect. I believe that this may be a hardware issue, rather than a software bug.

Background: I am using an LM35 temp sensor, which was provided by Cell_Man integrated into my new motor. The sensor draws +5 and GND from the hall board, and presents its output on a separate line. As such, I've removed R17 from the CA board, and connected the sensor output to the NTC pad. I've made no other modifications. The temp sensor is configured as Linear type, with 0° = 0v, and 100° / volt. I am running the Rev3_B14 firmware. I have my throttle connected to the CA's Thi pad group, and it is operating normally in "Pass-through" mode.

Now, the observations:

1: In normal operation, I observe the temperature indication to jump almost immediately to 90-100° when I apply full throttle, and then fall back to 20-25° immediately when I release the throttle. This movement is much faster (in both directions) than is believable given the thermal mass of the components involved.


2: In setup mode, the temp menu shows 0.25v at rest. If I gradually increase the throttle, it remains the same until I hit around 90% throttle, at which point it jumps to 0.35v in one swift motion. (Obviously the motor is not actually running when in setup mode.)


3: If I disconnect the temp sensor (such that the NTC pad is floating) and repeat this test in setup mode, the Temp indication is 0.74v at rest, and increases linearly to around 3.7v as I increase the throttle towards full.


4: With the sensor still disconnected and the CA still in setup mode looking at Temp, I attach a voltmeter (Fluke 77) between the CA's "NTC" pad and ground, such that I am looking for voltage coming OUT of the CA on the NTC pin. I observe a voltage of around 0.4v at rest, increasing linearly to around 1.8v at full throttle.


So, for some reason, the CA seems to be applying a large positive voltage to the NTC input pin which increases with throttle. I haven't determined definitively whether this corresponds to throttle in or throttle out, though in setup mode there should not be any actual throttle out (confirmed by the fact that the motor does not run), so I'm leaning towards some kind of cross-talk in or around whatever A/D converter is presumably being shared between all of the analog inputs.

The only thing I can think of is to try adding a pulldown resistor to the NTC pad. The fact that applying the voltmeter to the pad in test #4 above affected the readings makes me suspect that whatever voltage source is causing this phenomenon is quite weak, and can perhaps be dealt with just by sinking it through a resistor which will stay within the LM35's specified 10ma maximum current-source capability (400-500 ohms might be about right.)
The concept of "average" requires that 50% of the population be below it.

User avatar
Joe Perez   100 W

100 W
Posts: 187
Joined: Apr 06 2011 10:41am
Location: New York City

Re: Cycle Analyst V3 preview and first beta release

Post by Joe Perez » May 21 2012 11:05pm

Problem solved.

I grabbed a 390 ohm resistor out of the pile and stuck it between NTC and ground. With the sensor disconnected, this gave me an indication of approximately 0 degrees (with about +/- 0.5 degrees of noise), and with the sensor connected and the motor resting at room temperature, it indicated 22-23 degrees. Sidebar: a bit of averaging / low-pass filtering would be nice on these inputs if you can spare the processor cycles and memory to implement it. I'll see if I can find the space to put an RC circuit in there later.

I haven't been able to really load-test the motor to see how the sensor behaves as it heats up (I lack a decent battery at the moment, so I'm tethered to the workbench) however I did run the motor for a bit with the wheel up in the air, using the brake to put a bit of load onto it, and didn't see any abnormal behavior.

I've edited my posting a few pages back to reflect this, and will update the picture once I have a chance to tack a surface-mount resistor across the pads inside the unit; for the moment I'm testing with through-hole parts hacked into the harness externally.
The concept of "average" requires that 50% of the population be below it.

User avatar
justin_le   10 MW

10 MW
Posts: 2152
Joined: May 28 2007 3:27am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » May 22 2012 7:10pm

Joe Perez wrote:Problem solved.

I grabbed a 390 ohm resistor out of the pile and stuck it between NTC and ground.
Hey Joe, yes, if you removed the normal pull-up resistor and left the input floating, then the previous post results all mostly make sense. The CA samples the thermistor voltage input shortly after the throttle input, and so if there isn't another signal pulling it up or down then it will still be floating at whatever the last sampled voltage was at, which is why it appeared to change with the throttle.

One thing to note though is that you _can_ get issues with the temperature sensor output varying with load if there is a change in the ground reference of the CA and the temp probe ground (which is presumably the hall effect ground in the controller). So for instance, if the CA ground is located on a shunt upstream from the controller, then there will be a small voltage drop from this shunt to the actual controller ground plane when you have large amps flowing, and that could result in an increase in the apparent temperature sensor voltage by the amount that the CA's ground drops below the controller ground. You would also have this effect if you were to say run a set of lights from the CA's power port, which would cause the CA's ground reference to increase a bit.

That's why we'd generally recommend hooking up any temperature sensors to use the CA's ground pad if possible.

-Justin
Currently recovering from the Suntrip race on a back to back tandem solar powered row/cycle trike. 550 watt solar roof, dual Grin All Axle hub motors, dual Phaserunner controllers, 12 LiGo batteries, and a whole wack of gear.

Now back in Vancouver with my Big Dummy Frame (yes This One, thanks ES!) with Grin all-axle front hub, Phaserunner controller, and 52V 19Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

User avatar
justin_le   10 MW

10 MW
Posts: 2152
Joined: May 28 2007 3:27am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » May 22 2012 7:20pm

lizardboy wrote:Just tested the new code and the throttle cutting out issue remians exactly the same.

Are you sure your test bench is presenting a full and proper load? Happens everytime for me within the first 3 or 4 seconds of operation. Should be pretty easy to duplicate.
It's not just bench, we have 4 different staff ebike setups all running this firmware, two with direct drive hubs and two with geared eZee motors, and so far haven't been able to replicate anything like this even with a lot of trying. Can you confirm that when you are on the diagnostics screen that shows both the input and output throttle signals, you see the Vo signal drop to either 0V (or MinThrottle) whenever there is a cutout? This would confirm that it is definitely the CA and not the controller involved.

We did have a few cases where the MaxThrottle output voltage was set right on the edge of the controller's max throttle value, and that would occasionally cause an intermittent controller cutout, especially if there was a lumenator light on the DC-DC port that would cause the CA's ground reference to go up a bit. Dropping the default value to 3.55 or 3.60V for max throttle output totally resolved that.
Can someone else re-verify this still occurs with the new version?
Curious too if others have had throttle cutout issues similar to what lizardboy has mentioned with the B14 firmware? Any feedback could be really useful.

-Justin
Currently recovering from the Suntrip race on a back to back tandem solar powered row/cycle trike. 550 watt solar roof, dual Grin All Axle hub motors, dual Phaserunner controllers, 12 LiGo batteries, and a whole wack of gear.

Now back in Vancouver with my Big Dummy Frame (yes This One, thanks ES!) with Grin all-axle front hub, Phaserunner controller, and 52V 19Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

User avatar
justin_le   10 MW

10 MW
Posts: 2152
Joined: May 28 2007 3:27am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » May 22 2012 7:27pm

Waldo wrote:Feature Request
~~~~~~~~~~~~~
It would be really great to have a way to send a complete edited setup parameter list to the CA, perhaps through the bootloader.
Yes, that is possible through the current bootloader by deleting all lines in the .hex file associated with the firmware and only leavimg the eeprom ones. Similarly, you can update the firmware without changing any of the setup parameters by deleting the eeprom lines from the .hex and leaving the firmware in place. Rest assured there will be a nicer UI for doing this in the future! but the functionality is in place already.
Currently recovering from the Suntrip race on a back to back tandem solar powered row/cycle trike. 550 watt solar roof, dual Grin All Axle hub motors, dual Phaserunner controllers, 12 LiGo batteries, and a whole wack of gear.

Now back in Vancouver with my Big Dummy Frame (yes This One, thanks ES!) with Grin all-axle front hub, Phaserunner controller, and 52V 19Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » May 22 2012 7:42pm

justin_le wrote:...the serial output stream spec hasn't been finalized, but at the moment it sends the following:

Ah, Volts, Amps, Speed, Distance, Temperature, RPM, HumanWatts, Torque(Nm)

There should be a header on the start of the datastream that goes:

Code: Select all

Ah     V      A      S      D      T      R      W      N
Justin-
For purposes of performance comparisons it would be very helpful if there were a couple of additional columns that provided the Vi and Vaux values which would reveal the state of the operator controls. I suppose Vo would might better reveal the controller state, but Vi seems to reflect the operator input a bit better - no strong feeling between the two (having both seems like overkill considering the limited application).
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
Joe Perez   100 W

100 W
Posts: 187
Joined: Apr 06 2011 10:41am
Location: New York City

Re: Cycle Analyst V3 preview and first beta release

Post by Joe Perez » May 22 2012 8:38pm

justin_le wrote:The CA samples the thermistor voltage input shortly after the throttle input, and so if there isn't another signal pulling it up or down then it will still be floating at whatever the last sampled voltage was at, which is why it appeared to change with the throttle.
Ok. I had been assuming that each analog input had its own discrete A/D, but I guess you must be muxing them all into a single A/D with an analog switch. That explains things.

justin_le wrote:One thing to note though is that you _can_ get issues with the temperature sensor output varying with load if there is a change in the ground reference of the CA and the temp probe ground (which is presumably the hall effect ground in the controller).
Yeah, I agree completely, and this is actually how I drew the image several posts back when I first removed the pullup. My preference would have been to draw both power and ground for the temp sensor directly from the CA itself, but I'm limited by the fact that this is how the motor came wired from Cell_Man, and there aren't any spare conductors coming through the axle that I could use to fix it. (Besides which, I very seriously doubt that I could manage to get two extra wires through there if I tried- the phase wires are doubled-up, so it's an extremely tight fit as it is.)

justin_le wrote:So for instance, if the CA ground is located on a shunt upstream from the controller, then there will be a small voltage drop from this shunt to the actual controller ground plane when you have large amps flowing, and that could result in an increase in the apparent temperature sensor voltage by the amount that the CA's ground drops below the controller ground. You would also have this effect if you were to say run a set of lights from the CA's power port, which would cause the CA's ground reference to increase a bit.
Fortunately, I have the CA wired directly to the controller's internal shunt, and the CA's ground is equal to the controller's ground. Not as good of an arrangement as a completely discrete ground for the temp sensor, but since I'm not drawing any accessory loads across the CA's own ground wire (my headlights are powered independently) there shouldn't be much ground offset between the two. After inserting the resistor, I briefly load-tested it by using the rear brake to hold the system at about 10 amps while applying full throttle, and I only saw about 1-2 degrees temperature rise, most of which is probably attributable to actual heat within the motor.


justin_le wrote:Curious too if others have had throttle cutout issues similar to what lizardboy has mentioned with the B14 firmware? Any feedback could be really useful.
Before I figured out the temp sensor issue I thought I was having this problem (the temp sensor reading was floating to a level above max temp) but once I fixed that, I haven't experienced any such problems. I've only ridden the bike briefly thus far, as my sole power source consists of three old SLA batteries attached to the rear rack with tie-wraps until my new Cell_Man battery arrives, but so far it's been working perfectly.

lizardboy might check to ensure that he has temp sensing disabled in the setup. In looking through his last post where he listed all the setup parameters, I saw no mention of temp either way. (Assuming his pullup resistor is still in place, his temp would be pegged at maximum all the time.)
The concept of "average" requires that 50% of the population be below it.

User avatar
Joe Perez   100 W

100 W
Posts: 187
Joined: Apr 06 2011 10:41am
Location: New York City

Re: Cycle Analyst V3 preview and first beta release

Post by Joe Perez » May 24 2012 8:20pm

So I got the procedure worked out for how to reliably connect an LM35 temp sensor to the new CA.

As a bit of background, the reason I'm using this device is simple- it's what came with the MAC motor that I bought from Cell_Man. If you're doing this from scratch, then by all means use an LM335 or a 10k NTC thermistor. But for those not in control of what sensor is in your motor, a couple of internal modifications to the CA are required.

The first step is to remove R17, which is the 5v pullup resistor. Having done this, you'll find however that the Temp input on the CA still tends to float to a high voltage, as there is some leakage from the throttle input line. Thus, a pulldown resistor is needed on the Temp line to sink that voltage to ground. I experimented a bit and found that quite a low resistance is needed. With a 1k pulldown, I was still getting several degrees' worth of coupling. In the end, I settled on the same 390 ohms that I'd originally started with, which seems sufficient to keep the drift to within about 1 indicated degree at 20°.

I was still getting some jitter in the reading when the motor was actually running. My preference would have been to install an R/C filter, as is the common engineering practice when designing automotive ECUs and other such devices. The necessity of the pulldown resistor, however, prevents this. The combination of a pulldown and the series resistor of an R/C filter would create a voltage divider, and we can't have that. So I settled on a compromise, copying circuit from the LM35 datasheet, which uses a 75 ohm resistor and a 1uf capacitor in series, between the NTC line and ground.

Here's what the finished product looks like:

Image

The placement of the capacitor is not obvious in this view- it is standing up vertically on the circuit board, with one end on the R17 pad which connects to NTC, and the other end jumped over to the 75 ohm resistor resting on the ground pad. Here's a side view:

Image


With this design,I am seeing jitter of only about half a degree when running the system at 20 amps, which I find acceptable. The temperature display is quite jumpy (either a sample-and-hold or some actual averaging in the software would be nice) but it's entirely adequate for my needs. I haven't had a chance to really heat the motor up yet, however I expect that the magnitude of the jitter will diminish as temperature (and, thus, actual voltage on the NTC line) increases.
The concept of "average" requires that 50% of the population be below it.

lizardboy   10 mW

10 mW
Posts: 34
Joined: May 31 2011 6:19pm

Re: Cycle Analyst V3 preview and first beta release

Post by lizardboy » May 26 2012 2:55pm

edited never mind it didn't fix it
Last edited by lizardboy on Jun 26 2012 12:55am, edited 1 time in total.

User avatar
hillzofvalp   1 MW

1 MW
Posts: 1861
Joined: Dec 26 2010 2:09am
Location: Los Angeles, CA

Re: Cycle Analyst V3 preview and first beta release

Post by hillzofvalp » May 27 2012 11:01pm

I was just thinking today when I passed some cops: what if the CA had a hidden feature that just set all the trip statistics to say MaxS 25.1mph or MaxPower = 747W. That way when I get pulled over I can explain to them what I was recording and explain to them any discrepancy if necessary (they'd be baffled at how cool it was, more likely, and let me go).

I was reading the indiana laws today, learning that they go by the "Average power." In other words I can run 5HP for 1 second and .55HP for the next 9 seconds and still be at a max average power of 1HP. Maybe an Average power statistic is already included but I missed it.

I want to reiterate how useful it is to have a bar graph to display live readings on-the-go (mainly Power).

User avatar
hillzofvalp   1 MW

1 MW
Posts: 1861
Joined: Dec 26 2010 2:09am
Location: Los Angeles, CA

Re: Cycle Analyst V3 preview and first beta release

Post by hillzofvalp » May 27 2012 11:19pm

justin_le wrote:
That said, from a safety perspective it's a good idea to do as Teklektik suggests and use the fault protection and include a resistor inline with the 5V side of the throttle, so that if you do have a break in the throttle ground return it won't cause full power.

-Justin
In my experience this would be helpful. However, when mine went full throttle in testing with the V2.25 CA, it was because the throttle output to the CA broke (with current throttle enabled). I don't think the CA likes an open connection here. Maybe if the throttle output is not within itermmax/min it should be defaulted to whatever 0 throttle is (Aux threshold in my case). If it was covered already, I apologize for not having caught up yet.

lizardboy   10 mW

10 mW
Posts: 34
Joined: May 31 2011 6:19pm

Re: Cycle Analyst V3 preview and first beta release

Post by lizardboy » May 28 2012 9:36am

Just to be clear I'm not the only person reporting this issue; Teklektik has experienced as well

Okay so I spoke too soon on saying my problems were gone. I had the problem come back to me on a ride and while I was peddling my As- home I figured out how and what is causing the problem. When the throttle cuts out the CA has no output voltage. It snaps back to the minimum throttle settting and reseting etc. only makes it do the same things. Throttle input is fine. Also curiously it only did it after a charge cycle. This got me thinking about what happens when I charge the battery. I always disconnect the battery pack to charge. In my system I have one of those throttles with a button built in. I use this button to turn on/off the small red wire going into the controller. This alows me to easily disable the motor while I'm going up/down stairs or pushing/carrying the bike somewhere.

I'm able to simulate the problem almost every time by hooking up the power with the button off then activating the button. If I turn on the power with the button in the on position it does it as well if i just use the button and leave the battery power to the controller hooked up it works fine. However I have to add that when it quit on my ride last night nothing would reset it. I tried for 4kms before it finally left cutout mode and returned to normal. Must have reset 20 times to get 4 seconds of power each time. It does this in all modes pass-thru, speed, and current. Finally after many power cycles using the button it clears and works perfectly as long as battery power isn't interupted to the controller. If you cycle the power and don't touch the throttle at all for the 4 seconds then you'll find it cut out in your absence.



So more thinking ensued and I think I've got what the problem is. When I wired the throttle I didn't power it from the CA but instead left it powered by the controller. So my theory is that by turning the system on the way I am the throttle signal is live earlier in the boot sequence than expected throwing off the CA somehow. All the values display correctly when the cutout happens except that the output voltage returns to the minimum output.

So could this all be caused by my throttle wiring or did something change between the emailed version and now that makes this a sensitive issue? It's proably only a few msec. in timing but it explains why it always happenes on the way down the driveway. the only fix seems to be to cycle the power till it works. It does not do this with the version you emailed me not even once. So 100% its the software. HOw about some suggestions or more tests I can perform. The microcontrollers I work with have a real time debugger that allows me to look at all the variables while running. I could look furthur with more information or tools.... Maybe we need to record whats happening with an analogger?
Last edited by lizardboy on Jun 01 2012 1:27pm, edited 8 times in total.

User avatar
hillzofvalp   1 MW

1 MW
Posts: 1861
Joined: Dec 26 2010 2:09am
Location: Los Angeles, CA

Re: Cycle Analyst V3 preview and first beta release

Post by hillzofvalp » May 28 2012 11:25am

Just thought of something. Would it be effective to have the CA PWM the regen signal (mainly for higher power setups or with shunt modded regen power)?

User avatar
amberwolf   100 GW

100 GW
Posts: 25452
Joined: Aug 17 2009 6:43am
Location: Phoenix, AZ, USA, Earth, Sol, Local Bubble, Orion Arm, Milky Way, Local Group
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by amberwolf » May 28 2012 1:45pm

If you mean, PWM the braking signal? Probably not on most controllers--several of the ones I've tried havea notable delay both in engaging braking and in releasing it, so PWM would be on the order of half a second to a second per cycle, which unless you are braking on a very long downhill isn't going to have time to do much.

If the controllers have a physical RC circuit used to create that delay, then it could be modified or removed so that PWM would be possible at whatever rate the MCU accepts and responds to braking signals, but I don't know how fast that is, either. Probably not very.

I looked into this when I wanted to add a 555 PWM circuit to my ebrake levers a long while back, but never ended up trying it after I realized the delay problem.

Post Reply