ixxat USB-to-CAN compact & Sevcon DLD flashing

methods said:
Update:

* Confirmed -part number ending in 3 for SIN/COS.
EDIT
Doctorbass said:
Also Are you sure the Sin/Cos and A/B sin/cos are only decoded by the last digit of the serial number?.. i tought it was the last 3.. like xxxxxxxx201 (A/B UVW) and xxxxxxxx203 (for the A/B UVW Sin/cos) ?

-methods

Just to throw a spanner in the works on sin/cos compatibility I have two gen4's that came direct from test vehicles used at Ashwoods Electric Motors in the UK that were fitted with heinzmann pms120's and Ashwoods own ELMO-S112, both motors are fitted with sin/cos encoders. The part number of the gen4's I have is 634A8320.

Doc, Methods says part number and you say serial number, is that what you ment to say?

Given that it is a hardware difference I wonder if it is not the hardware version number available from DVT that would be the most logical place to identify sin/cos compatibility. Anyway I intend to trace the sin/cos input pins on a compatible model and compare the circuitry compared to non compatible board to see if it is just unpopulated parts or totally different circuitry.
 
kiwifiat said:
methods said:
Update:

* Confirmed -part number ending in 3 for SIN/COS.
EDIT
Doctorbass said:
Also Are you sure the Sin/Cos and A/B sin/cos are only decoded by the last digit of the serial number?.. i tought it was the last 3.. like xxxxxxxx201 (A/B UVW) and xxxxxxxx203 (for the A/B UVW Sin/cos) ?

-methods

Just to throw a spanner in the works on sin/cos compatibility I have two gen4's that came direct from test vehicles used at Ashwoods Electric Motors in the UK that were fitted with heinzmann pms120's and Ashwoods own ELMO-S112, both motors are fitted with sin/cos encoders. The part number of the gen4's I have is 634A8320.

Doc, Methods says part number and you say serial number, is that what you ment to say?



Given that it is a hardware difference I wonder if it is not the hardware version number available from DVT that would be the most logical place to identify sin/cos compatibility. Anyway I intend to trace the sin/cos input pins on a compatible model and compare the circuitry compared to non compatible board to see if it is just unpopulated parts or totally different circuitry.

Well you are right.. this is teh Serial number NOT the part number... any serial that match with SIN/COS motor have to end with 203 if i remember

Doc
 
Yes - we will make many mistakes as we work this out and show our work in real time.
Big difference between folks who figure it out then make a pretty post... vs working in real time :mrgreen:

I have in my possession both UVW and Sin/Cos controllers. Gen 4 size 4
Today we will find out once and for all if the hardware is identical, if they can be swapped, etc.

My plan is to first attempt to flash the UVW with DLD expecting Sin/Cos... if there are issues... I will open the UVW and compare it to the Sin/Cos at the PCB level.

Today I am also building Sevcon harnesses at the shop. We have one customer - low budget - and the rest goes to overhead expense.
I can make 10 harnesses at once that will take me an equal amount of time as building 2 at two different times...

Gah... how do I always get mixed up in this unprofitable work :mrgreen:
Only to payback the Master who spent so much time in training me.
I feel like I have peeled back most of the layers of the Sevcon... but we have a new riddle to solve.

EDS - and how that affects interpretation of the TDIO data coming out.
I have two controllers working on a Dyno. One reports good numbers and the other has significant offset.
Going to dig into that today as well.

I have many EDS going back years... sigh... complicated bunch of mess....

-methods
 
Doctorbass said:
kiwifiat said:
methods said:
Update:

* Confirmed -part number ending in 3 for SIN/COS.
EDIT
Doctorbass said:
Also Are you sure the Sin/Cos and A/B sin/cos are only decoded by the last digit of the serial number?.. i tought it was the last 3.. like xxxxxxxx201 (A/B UVW) and xxxxxxxx203 (for the A/B UVW Sin/cos) ?

-methods

Just to throw a spanner in the works on sin/cos compatibility I have two gen4's that came direct from test vehicles used at Ashwoods Electric Motors in the UK that were fitted with heinzmann pms120's and Ashwoods own ELMO-S112, both motors are fitted with sin/cos encoders. The part number of the gen4's I have is 634A8320.

Doc, Methods says part number and you say serial number, is that what you ment to say?



Given that it is a hardware difference I wonder if it is not the hardware version number available from DVT that would be the most logical place to identify sin/cos compatibility. Anyway I intend to trace the sin/cos input pins on a compatible model and compare the circuitry compared to non compatible board to see if it is just unpopulated parts or totally different circuitry.

Well you are right.. this is teh Serial number NOT the part number... any serial that match with SIN/COS motor have to end with 203 if i remember

Doc

That is what I'm trying to get to the bottom of, I just checked the two gen4's running sin/cos encoder motors and they have serial numbers 1206100319 and 1206100150 both part numbers are 634A83210. Another ES member has a gen4 running a PMS120 with sin/cos encoderthat he has running, I will ask him to check his serial/part numbers.
 
Alright guys... now we are getting down to it. :D

I have samples of both SIN/COS and UVW sevcons sitting on the floor behind me.
They look hard to re-assemble... so as soon as I can try to flash the UVW with a SIN/COS DLD/DCF we will know if it can do it. If it fails - then I will open the UVW and compare it side by side with the SIN/COS.

Does anyone have pictures of a jig they use for pushing the Sevcon mosfet board back into the heat sink? I suspect I could just "do it" but dont want to get into a repair situation if I dont have to.

All are welcome!
Please come and participate!

I am excited to interact with some folks who have more experience that we do.

-methods
 
Check it out... I fixed my first hardware problem:

https://endless-sphere.com/forums/viewtopic.php?f=38&t=88473#p1290426

Welded Contactor Error identified.
Could have been worked around by switching to contactor 2 or 3... (which I asked customer to test...)
That is the best field fix IMHO.

Anyway - shorted mosfet found, super easy fix.

-methods
 
subscribed!
 
When I got home tonight I saw my napkin sketch and thought... "Maybe the fet is actually ok and the suppression diode is shorted"
(There is a really big diode strapping the fet tasked with taking the brunt)

I will probably nip the source leg, re-measure, then either lift the diode or swap the fet.
Customer is keen to get his bike going...

The fet heat sinks are covered in a crusty hard green goo... I am really hoping it will reflow at 60C so I dont have to scrape it all off :?

-methods
 
Valuable Information:

Key Word: Encoder, encoder calibration, Sin/Cos, encoder tuning

https://endless-sphere.com/forums/viewtopic.php?f=10&t=88647

I have been PM'ing with this user regarding Sevcon business.
He took the time to outline the encoder calibration process.

We do a lot of this bringing the Zero motors (75-7) online with our calibration station.
For the Sin Cos a lot of time is spent:

* Shimming the sensor distance from magnet to ensure we dont rail SIN or COS (amplitude adjustment)
* Adjusting left and right for symmetry of the two wave forms (centering over magnet)

I have not seen evidence yet... but there is some slop in the 75-7 motors, taken up with a spring bearing, and if that slop "moves" the calibration (which is on the order of thousandths) would be thrown off. In a side load motorcycle application this is no issue. In a boat, or plane, or other application we may get different results.

Our mechanical team built up one motor for a flight application which had zero tolerances and a much larger thrust bearing. We have that motor on the Dyno today... got it up to 40KW then lost our software connection. We are getting up into the Power and Noise levels where things matter - so my first order of business today is cleaning up CAN wiring, power wire routing, etc... to break 50KW.

thanks,
-Patrick
 
Note:

* We came into possession of some plug and play UVW encoder replacements so we will do some side by side testing to see how those affect power and low RPM performance.

* We handed off the Dual-Inline 75-7 to Henry... who has/will be testing it in his Electric Insight. The timing of the mechanical motor assembly was perfect... so one encoder should suffice for driving both sets of windings... and Henry is using a Curtis.

* I got news back from testing Arlo and another did RE: isolating and duplicating one encoder for two Sevcons. Results were positive... so we may investigate that route.

* We integrated a Sevcon Clear View yesterday... and although people have spoken poorly of it, I read in the fine print that "features" are locked and can be unlocked with a simple process. I believe most settings of interest will be exposed to users. Now... to bring the cost down. FYI these units can be found on Electric Golf carts.

* We finished the BOM and cut lists for our "shorty harness" - this is basically the harness one needs to build a vehicle around the Sevcon with minimum complexity. The last step is to integrate CA compliance and a good quality shunt for "real" DC Current measurements (Sevcon guesses at battery current).

* We have 1pcs dead ixxat and 1pcs "good" ixxat (the expensive units). Today we will peek inside and determine which component fails in those programmers.

Off to work

-methods
 
sincos_support.PNGRe sin/cos support: within the help menu of DVT Helper is "Help on Sin/Cos Encoder", clicking that brings up the option to "check to see if this unit supports sin/cos" and that quotes does/does not and gives a hardware version. So I think that possibly the hardware version is the ultimate decider on sin/cos support. I can also confirm that the Gen4's on sale on EBay from Korea do not support sin/cos, not that they claim to in any case.
 
Well... it seems that is the case.
We can always hold out hope tho :mrgreen:

I will still open up one of the UVW's to see if its a totally different board or just a non-pop on that section. Someone in contact with Sevcon assured me that there was only one version of the hardware. Hmmm... maybe before or after a production year this changed... as it would make little sense to produce and track two versions for only a few dollars in parts. Crossing fingers.

As for testing...
We have a dyno script running that **very smoothly** ramps up RPM for a given load. It will be very interesting to see how the UVW stacks up to the Sin/Cos in this test.

Speaking of the Dyno -
Today I re-wired the hacked-ass CAN wiring (it was first built from scrap harnesses on a shoe-string... and we are slowly cleaning it up) and was able to break 45kW. With my PC 4' away I would lose communications above apx 35kw. Now I have a 10' tether and a proper CAN topology so no glitches or noise what so ever. Works great.

1:1 loading
two pcs 75-7
Regenerative setup
4500 RPM
45KW

I will admit that it gets a bit scary regenerating that much load at higher RPM. I bet we see 70hp by the end of the week.
(I need to sneak up on the RPM as we are still qualifying motors that were wrapped using different processes...)

As of now... these early production motors would be KILLER in an application that does not require >6K rpm
We have units running 6000 and above - but I saw some failures early on - so I am being extra cautious.
We are very sensitive about the reliability - especially at high RPM.

Craig is baking the units in now, and I have confidence that he will perfect the method (he says he has it dialed... but I am the master of wrecking isht). Its interesting watching him work... he is good at converging on optimal solutions instead of just throwing margin at things. I guess thats why he makes killer light weight carbon fiber bikes eh?

As for projects... we picked up two more. One is a very interesting Motorcycle application.
(I only share details like this at the bottom of dusty nerd-threads) :wink:

-methods
 
A note on Sevcon:

Our Dealer POC had a very positive interaction with them this week where they were "bend-over-backward" helpful. You have to be a F^C&ING genius to operate their stuff... but they definitely support it over email (for Dealers).

After approaching their gear somewhat blind... I can see why they dont really even try to help the regular DIY guy. Its... basically impossible.

So... yea... Sevcon stuff is working (thanks to the Master - it would have taken me 1,000 years to master it alone)

thanks,
-Patrick
 
Update:

We are now seeing failures with the Chinese IXXAT clones. If you can afford it... buy a genuine isolated IXXAT CAN box.
Our clone has become intermittent... tho to be fair... we use it a lot with the system running and not just for programming.

If you are just flashing DLD and DCF... a cheap one will probably last you.
If you are running a Dyno or live on the vehicle - spend the extra money. :idea:

-methods
 
Diagnosed a blown IXXAT Compact

It was not a Galvanic version... pay for the galvanic isolation.

Part 10.100 was the subject

1.01.0087.10100 USB­to­CAN compact (SUB­D9 plug)
1.01.0087.10200 USB­to­CAN compact (SUB­D9 plug); with galvanic isolation
1.01.0088.10200 USB­to­CAN compact (RJ45 plug); with galvanic
source: http://www.super-b.com/content/files/Files/Products/USB_CAN_Interface.pdf

Can transceiver in the corner had no 5V
VP 251 Can Transceiver
http://www.ti.com/lit/ds/slls545g/slls545g.pdf

Traced it back to a TO-23 on the other side of the board
No action, troubleshooting time up.

Even using the non-galvanic with a battery only laptop can still result in fubar.
Remember Sevcon runs 100V+..
Even on switch wires it 50V+
Big loops between wall, laptop charger, laptop, usb, dongle, DB9, CAN wires, battery negative, battery charger,


-methods
 
I have an Lipo4 24 kw go kart project underway and I am using the 80/350
Gen 4 with a Motenergy ME1012. I have the drive system 100% operational,
and after driving a Tesla model P100D I want to incorporate an automatic
regenerative system like the Tesla.
It has a two level system, that can be set on the control panel for high
And low level regenerative braking. When you let off the peddle there is
A point where you are coasting, then as you release the peddle completely
The regenerative kicks in at the level you have selected. This system makes
Most routine braking unnecessary!
Any ideas?
 
If the controller itself has an "ebrake" input, you could put a switch in the pedal that activates only when the pedal is fully lifted (not depressed at all), that then activates the regen via that input.

The easiest way to do that is a momentary switch that actually sits "above" the pedal, and is held closed (or open) by the pedal in rest position, but then is released as the pedal is pushed down.

(there may also be a way to do this via programming, but I'm not familiar enough with Sevcon to know :oops:)
 
Thanks for all the exciting work on Sevcons!

Do you have a link to the exact clone CAN adapter you recommend? I would love to just be able to handle my own motor encoder commissions and a few basic functions.
 
Hey nospin.... like your name.

There should be no problem recreating this feel in the Sevcon. I do not have a controller to connect to in front of me (so cant look at exact settings in DVT...) but it goes about like this from Memory:

* Settings are in H (Helper)

* An RPM can be set, which under that RPM, no engine braking occurs. I set this to 0 RPM for complete stop.

* A percentage of total thrust can be set, such that the amount desired is applied while off throttle.

* A percentage of total thrust can be set, such that regen is applied at a fixed percentage based on a regen switch, or, at a variable percentage based on a regen throttle.

* Priority of throttle, regen lever, and coast can be set... such that either Throttle or Regen is preferred... and such that the coast-brake (described above) or regen is preferred.

* I do not recall off hand if you can explicitly set a "coast" throttle percentage... but this should happen naturally. Off throttle will produce regen. If you apply only enough throttle to converge on the tipping point of thrust... it should equate to coasting. Easier with a speed control, need to see how it feels on thrust control.

* There are more settings, its not in front of me.

* I would start by setting Regen as high priority than throttle so that in can overcome throttle. If you dont do this then regen will be ignored any time throttle is applied and we cant use that variable input to achieve advanced functionality.

EDIT: You need to set regen input as priority over at least one of these two: On Throttle and coasting. IF regen does not have priority over one of those then it will be ignored at all times. There are only two logical states for throttle: Applied and not applied.

For drivability under high power you will want Throttle to have priority (long story... but picture jerky acceleration) but while testing... set it to regen priority for both. At an absolute minimum regen must have priority over coast braking. Think it through and it will be very clear.


* I would set Regen as higher priority that coast braking... as if you dont... regen will not be applied relative to user input when off throttle.

* I would set onset RPM to 0%

* I would try setting coasting regen to 10%, then 20%, then 30% and see how you like the system response.

* In order to have "two levels" as described, I would program Profile BASE to be one coast regen setting. I would program Profile 1 to be identical but with a higher or lower settings. Then a switch can take you between the two levels. There is a third profile that can be used for "safe mode" or whatever you choose.

* When I get back to the shop I will get a screen shot of all the settings under Throttle and we can look at other tuning options.

* You can tune the rate of change (or onset) of said regen braking in RPM/Second

* You can tune a few more behaviors that affect drive-ability - especially around higher levels of regen... if you have heavy regen set up you will want to ramp it up slower.

* My guess is that you will want to do more coasting than drag-braking when you integrate efficiency over time. You take a Velocity efficiency hit at all times ... but you also take an acceleration hit and deceleration hit... say 20%... so coasting is your most efficient. Dragging, then accelerating, then dragging, then accelerating will result in less range than coasting as much as possible. I know its obvious... but folks sometimes get overly excited by the through of regeneration and forget the efficiency of the transfer case, motor, wiring, controller, DC wiring, battery..... Stuff gets warm. That's lost power, proportional to applied level.

-methods

nospin said:
I have an Lipo4 24 kw go kart project underway and I am using the 80/350
Gen 4 with a Motenergy ME1012. I have the drive system 100% operational,
and after driving a Tesla model P100D I want to incorporate an automatic
regenerative system like the Tesla.
It has a two level system, that can be set on the control panel for high
And low level regenerative braking. When you let off the peddle there is
A point where you are coasting, then as you release the peddle completely
The regenerative kicks in at the level you have selected. This system makes
Most routine braking unnecessary!
Any ideas?
 
liveforphysics said:
Thanks for all the exciting work on Sevcons!

Do you have a link to the exact clone CAN adapter you recommend? I would love to just be able to handle my own motor encoder commissions and a few basic functions.

Sure:

At this moment in time, this search shows both clones and used genuine IXXAT.
https://www.ebay.com/sch/i.html?_od...XXAT+USB+CAN.TRS0&_nkw=IXXAT+USB+CAN&_sacat=0

Our testing results indicate the following:

* The clone boxes can do anything an IXXAT can do... using modern DVT software.
* The clone boxes may need to have the USB cycled from time to time
* The clone boxes: GALVANIC ISOLATION IS OPTIONAL... BUYER BEWARE
* You can get away without isolation if you are mindful of how you use the box. Dont create big ground loops. Dont leave the box connected under operation. Put it into Operational and disconnect.
* I would of course buy a genuine IXXAT used unit with Galvanic isolation any time possible.
* Later versions of DVT have no problem working with the original IXXAT, IXXAT Compact, IXXAT Compact V2 - they all work.... as do the clones.
* Python version of DVT is also compatible with all models including clones

You may consider grabbing a ClearView...
I used to hate them... but now that I can program them... I see the value for quick changes on the road.

For you... and racers... a laptop and dongle is the only way to go.
For tinker-types... folks just looking to feather out their regen settings... or try a different throttle... the CLearView will allow DCF changes. If you know the Index and Sub... you can directly modify the DCF in the controller. Some settings are even broken out and explicitly indexed... like throttle end points etc.

A cheat sheet is now needed for common indexes to allow for DCF changes on the road in the absence of DVT and a dongle.
It is still hard for non OEM to get a proper license.
Even for those with a proper license... its still a chore to get DVT up and running and working without a few moments of great frustration.

As far as repairing dongles... I have looked at only one, it was an IXXAT, it had shorting resistors across the isolation section, the CAN transceiver was not getting 5V, repair is possible but not worth the engineering hours.

-methods
 
To answer you specifically...

This is an IXXAT Clone similar to what we have tested:
http://www.ebay.com/itm/USBCAN-comp...71742940:g:sgAAAOSwNRdX5w6t&autorefreshk=true

The ad is a mess... but if you scroll down past the advertisements... it describes it as an IXXAT replacement.
Notice, like the true IXXAT, galvanic isolation is OPTIONAL... and since it is a Chinese seller... and the US beats Chinese sellers down to the lowest price always... the unit advertised most certainly does not have isolation. Send them and email and ask... bet it costs a bit more.

Further down the page is an IXXAT for $200 out of the UK.
I only do Ebay in the US... good luck. I know you are a seasoned ebay master.

-methods
 
Thanks for the links! Super grateful, and super great to hear that IXXAT doesn't have a monopoly anymore, your work here makes Sevcon's more accessible for the masses in a big way!

Thank you for this awesome contribution to the revolution!
 
After all those years of us slipping around on the icy parking lot known as Sevcon... it had to happen.

The controllers are good.
The controllers are designed around safety.
The controllers are affordable - especially used.
The controllers are perfect for a medium weight EV, light weight car, and all those other applications which are about to appear on our roads.

I really hope we can help to punch through all the fear and assumptions associated with the Sevcon.

The tools are not that expensive (crimpers and such)
The wiring is actually quite simple (Reference ThunderStruck-EV for the manual)

note: We adjusted our wiring to match that of the schematic listed in the above link so that customers who have things like a ClearView can go at it plug and play

The only real problem with the controller is how fussy it is with motor specifics.
The controller would be the epic win if it had a "dumb mode" where it would just "work" with a controller at say... 80% toward perfection. Tuners could then get it there.

I understand that companies like Curtis have auto-tune setups that extract the needed information and allow mating of controllers to motors.
One can only assume that Sevcon has something similar.

We will eventually converge on enough known good working applications that it will no longer be an issue... but at this exact point in time I can not say that "I can get any motor running on this" short of spending hours, days, or even weeks tweaking and tuning, borrowing tools, experimenting, etc.

One of the biggest issues is simply the STALE DATA issues with the user interface. My early days tuning Honda with Hondata on the road and on a dyno tells me that the biggest impediment to proper tuning is ACTUALLY having your changes stick so that you dont chase your ass.

My guess... many people fail to mate motors to this controller due to the changes they make and test not taking :idea:
The controller LITERALLY needs to be power cycled after every little change. It seems OCD... and it is deceiving because some changes "just take"... but others DONT.

Best practice (unless you read the DCF carefully)

* Set the controller in Pre-Op... as some changes will be ignored or throw an error in Op
* Make the change
* Confirm the change by reading
* Cycle power
* Take it out of PreOp
* Test
* Repeat

Doing this will greatly improve probabilities around successful mating of motor to controller.

I would gladly sit in a room all day getting motors running if there was a magic bag of money someplace to pay my bills :mrgreen:

Sadly there is not... so we do as much as we can for as little as possible and try to attract as many early adopters as we can and collaborate as much as we can.

Doc has been a great collaborator.
Biff has been instrumental.
Calfee has been very supportive.

Of course... some of our early customers have also been awesome in that they brought parts to us to explore... like the ClearView option.

FYI - We solved the ClearView. I posted up on how to use it to take a controller in and out of PreOp
IF you have a DCF to tell you what address a given setting is at... you can use the ClearView to change any of those settings on the fly without an IXXAT and Laptop

For that matter... if you are OG enough... you could sit down with minimon and set individual values.
Good luck! Lol... But it could be done.

ok - off to go sell a car to pay the bills.

-methods
 
Funny thing is, Borg Warner bought Sevcon recently, while at the same time SME has stepped up their controller game. SME has an amazing auto tune function and can be setup and learn a motor in a couple minutes.

Many good options in the stream for LEVs!
 
Yep - I will take any controller I can afford that will meet requirements.

There are enough Sevcon's out there to make it worth supporting them.

I just want to kick out roadblocks. No more enthusiasts with a pile of gear and at a dead end.
Its time for light weight enclosed vehicles.

2 wheels or 3 wheels fine by me.
Even 4 wheel NEV's ... they are speed limited to 35mph but only acceleration limited to the breaking point of the axles and safety features of the cab.

-methods
 
Back
Top