Honda Insight Forum banner
101 - 120 of 202 Posts

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #101 ·
I've been testing a new idea today and had to go shopping for essential supplies.
So I took advantage for a brief road test to the shop and back.

Basically the idea is implement a completely fixed voltage system.
I did this when I was using supercaps and it worked very well.

There are a few caveats.

The fixed fake voltage has to be >150V or the car won't IMA start.
Probably something around 155V will give good reliability and performance.
It bypasses start up and shut down checks the system does regarding stuck or faulty contactors.

It's easy to implement and does away with the HV connection from the actual battery to the BCM Fooler.
The BCM fooler gets it's HV from being reverse fed from the MCM HV connection.

We implement a 150V voltage source using 3 x 48 dc-dc converters inside the MCM and attached to the MCM HV connector from the inside.

We implement a simple relay controlled RC charge/discharge circuit for the VPIN detection with the switch on rise of voltage and hold it at 150V or whatever we set.

I'll post a video later of it on the bench.
Ran out of dc-dc converters as blew a couple yesterday due to spikes and back emf etc.
Solved with a flyback diode across them.
 

·
Super Moderator
Joined
·
9,229 Posts
Sounds like a viable option, but is it better in some way than the previous approach using the voltage doubler??
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #103 ·
I'm still exploring all the various voltage fooling options during the lockdown.
But the inability to do decent/any test drives is really hampering progress on that.

Today's video is on the 2nd LTO pack tail/connections. A few little hints and changes from MK1.
I'm shortly going to install the temp sensors I removed from the old NIMH IMA pack.


Still lots of bench and software work to do.
 

·
Super Moderator
Joined
·
9,229 Posts
I'm still exploring all the various voltage fooling options during the lockdown.
But the inability to do decent/any test drives is really hampering progress on that.

Today's video is on the 2nd LTO pack tail/connections. A few little hints and changes from MK1.
I'm shortly going to install the temp sensors I removed from the old NIMH IMA pack.


Still lots of bench and software work to do.
I think driving is safe to you personally. All the U.S. lockdowns allow driving for food. Just "shop" at a more distant store. Ops, I forget - you are a retired policeman;)

On that tall space you show, I just cut them with a hacksaw and dress them with a file to be flat and smooth. You can check you work with a quick caliper look. I noticed that they are some sort of special compound, not like the run of the mill AL spacer, but don't know if it matters.
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #105 ·
Yes I could have cut them down, but the Insight spacers were redundant and a perfect fit.
Seemed easier than trying to get a really flat surface after a manual cut with hand hacksaw.
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #106 ·
Been busy recently but a bit of tinkering today on my bench test setup.

Watching the ALF & RLF OBDIIC&C flags as i turned the HV up and down.
Assist and regen limit flags. 1= Disabled. 0 = Enabled.

Basically.

No load/assist/regen battery voltage higher than 175V = RLF 1 (Easily repeatable)

No load battery voltage lower than 120-140V = ALF 1 (Varies)
This varies quite a bit and might be more sensitive to dc-dc converter load etc.
 

·
Registered
Joined
·
7,843 Posts
^ I've been looking at those lately as well, in the stock setup, currently looking at 'Alf'. Does your bench setup include tap data? For instance, are the flags responding to these total voltages or to the underlying tap voltages that correspond to the totals?

On Rlf, did you try to vary the nominal state of charge at the same time? I think that can make a difference. Might be the same with Alf...

On Alf, I've noticed a few things the past couple days:

-Alf can trigger quite some time earlier than a neg recal (or bona fide empty pack happens).
-It seems to trigger repeatedly and possibly at a set interval if assist continues after the first trigger;
-assist seems to be throttled commensurate with repeated triggers. For example, the first trigger results in throttled current, the second trigger more throttle, the third even more, and so on.
-None of the triggers I've seen have been due to low total pack voltage...

I've been thinking that being able to intercept the Alf and Rlf commands, and issue our own, would be useful management tools.
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #108 ·
I've been thinking that being able to intercept the Alf and Rlf commands, and issue our own, would be useful management tools.
I think they originate in the BCM so the 'BCM Interceptor' is probably what you need ;)
I set those flags to disable assist/regen.
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #110 ·
^ I've been looking at those lately as well, in the stock setup, currently looking at 'Alf'. Does your bench setup include tap data? For instance, are the flags responding to these total voltages or to the underlying tap voltages that correspond to the totals?

On Rlf, did you try to vary the nominal state of charge at the same time? I think that can make a difference. Might be the same with Alf...

On Alf, I've noticed a few things the past couple days:

-Alf can trigger quite some time earlier than a neg recal (or bona fide empty pack happens).
-It seems to trigger repeatedly and possibly at a set interval if assist continues after the first trigger;
-assist seems to be throttled commensurate with repeated triggers. For example, the first trigger results in throttled current, the second trigger more throttle, the third even more, and so on.
-None of the triggers I've seen have been due to low total pack voltage...
I don't use or log/view tap data for my bench setup as I'm using a BCM fooler with 0.1% tolerance resistors.
They are balanced very closely, certainly well under the BCM tap variation detection sensitivity limits.

No I didn't fiddle with SOC at the same time. I'll try that in due course.

With my bench system I can vary the voltage from 0-300V, so I may see some strange behaviour that a working car with good IMA won't.

Pack voltages consistently below 120V or above 175V etc etc
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #112 ·
Useable LTO voltage range.

Been looking at SCIB LTO specs online and I see Toshiba says they can go from 1.5 - 2.7V (Nominal 2.3V)

So although there isn't perhaps a lot of capacity left below 2V it seems a bit premature to cut the pack until a cell gets down to at least 1.5V under load...

The under load bit gives a nice IR related safety factor, as they will always rebound to higher voltage than the minimum if you stop discharge once a cell gets down to 1.5V

That gives a useable voltage range for 48 cell LTO from 78 - 130V

So my plan for now is to set my CAN OBDIIC&C to give an alarm when a cell reaches <1.55V or >2.65V
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #113 ·
Grid charging the pack. Onboard or offboard? The LTO packs can take serious charge power.

As we have 240V 13amp ~3kw domestic power sockets in the UK it seems sensible to max that out when plugged in.

I have several ZIVAN NG3 chargers from previous Lithium projects, so I need to see if I can repurpose one of those.

Or alternatively make my own using ex server or meanwell PSU's like before.

Onboard Charger. Convenient when travelling as can be plugged in anywhere with normal or EV sockets.
Weight maybe 5/10kg or so. No external high voltage/current DC connections needed.

Offboard Charger. Less weight. Not so convenient can only charge where charger is located.
HV DC current/voltage connectors required.

Best of both worlds? Build the charger into a small portable box. Doesn't have to be carried if not going long distances, HV DC connections can be inside the car. Like normal grid chargers but meatier.
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #114 ·
Now making some serious progress with the 48 Cell LTO setup.

All a lot more stable and no errors on another essential shopping test drive.
I also now have the OBDIIC&C CAN combined device setup working properly.

The OBDIIC&C LTO CAN monitoring firmware and the standard OBDIIC&C IMAC&C P&P firmware etc have all been merged into one!

You only need a single OBDIIC&C device with PIC18F2680 & CAN mod to do all the normal OBDIIC&C stuff, IMAC&C stuff, LTO stuff and now theEV assisted Cruise mode. It does have a few tweaks to help with stability but i'll do a video on them in a few days. Extra resistor and caps etc. Nothing heavy or expensive.

It's working really well and was a pleasure to drive today.
I've posted a video on the other thread showing briefly the cruise mode in operation.

Basically you accelerate to your chosen cruising speed then press the button to engage C (Cruise mode)
The system will then use assist and regen to regulate your speed trying to keep within 1 mph of your target.
You can feather your foot on the throttle to keep the car in lean burn and the IMA will take up the slack.
You can adjust the maximum permitted levels of assist and regen Cruise is allowed to use in the Menu.

If you touch the brakes/clutch etc Cruise disengages.

Obviously if you feather the throttle too much, even max assist won't maintain speed and you will drain the LTO pack very quickly as it ramps up to the max trying to help out.

But if you juggle the throttle and assist in a delicate balancing act trying to maintain an average 15-20A assist then your pack will last about an hour and you will be doing stellar mpg. Grid charge at work and your laughing.

Other points to note on these recent tests.

1)
I've seen assist currents upto 120A at 110V (13kw) at various RPMs so the system seems quite capable of delivering good assist power even with the reduced voltage.

2) More interestingly I have also seen regen currents at almost 100A at the same sort of voltage. (11kw)
So the reduced pack voltage clearly gives extra regen potential. Good for those in hilly areas to recapture energy.

3) Cell balance seems very stable although my drives are very limited at present.
You have seen my ideas on how to address that issue anyway in previous posts but i'm still waiting for some parts to arrive for that experiment.

4) Some lesser used functions have been trimmed from the OBDIIC&C firmware to allow the LTO CAN stuff to fit.
I had to compromise on the display layouts so we lose decimal points, but you can use your mind's eye to put them back in again anyway. It also stops screen corruption and makes it easier to handle processes in the background.

5) I do not have the FSC (IGBT short circuit flag) disabled in this setup but satisfyingly no IMA errors were triggered even with the +40% current hack and 100A regen. So hopefully that's not needed.

As I make progress I will do further videos detailing the new OBDIIC&C setup and operation.

Here are today's videos from the other thread.



In the last vertical video the amps glitch was fixed and the screen layout rationalised.
 

·
Registered
Joined
·
7,843 Posts
^ Nice progress...

You mentioned that you're thinking about setting the low voltage cutoff at around 1.5V cell level. Food for thought:

-given that you saw 120 amps at 110V (2.29V cell level), that the usable voltage range is pretty flat and narrow, and that voltage will tank fast approaching empty, 1.5V seems low. Even heavily loaded and approaching empty, I doubt there will be much energy left around 2V, so it seems like going lower might invite more risk/stress than necessary...

-Earlier you mentioned using Rlf and Alf with BCM Interceptor to enforce high and low cutoffs. Do you use those with the LTO setup?

-Based on what I've been seeing looking at those parameters, they seem tied to assist and regen throttling, not only or necessarily outright disable, i.e. it seems like they could be used to throttle assist and regen to enforce high and low voltage cutoffs, such as when a cell approaches a set voltage or approaches empty and the discharge slope becomes steeper, rather than disabling outright.

If that's true you wouldn't need to have a somewhat riskier lower voltage cutoff in the interest of maintaining 'wiggle room' for higher power demands; rather, you could have the cutoff higher, power demand would taper as it approached/reached the threshold; cell voltage would stay above ~2V...

Testing my small Toshiba SCiB cells, I see that they tend to heat up most toward end of discharge. My cells are a bit different than the larger, 'high energy' ones, but I imagine a characteristic like that might be similar for both. These cells have a nominal voltage at 2.4V. Temp doesn't change much if at all, even at high discharge currents, until voltage drops below nominal, and below about 2.3V temp increases about twice as fast -- and they're empty very shortly after.
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #116 ·
🚃
All sensible stuff. EQ1.. It's all early days. :unsure:

Yes I will probably use a 2V cell limit for starters, as I gain experience I can nudge that down 100mv at a time if i think there is enough headroom and time to kill assist with a flag/imac&c etc before cells plunge off the cliff and below 1.5V

I have a BCM Interceptor installed for testing but presently it does not see live voltages so can't help autonomously with those flags.
It's just used to fake SOC at 75% and temperature at 25C for maximum performance.

However it can/could respond to external logic inputs and set the flags, but I'm not using that yet.

The updated CAN OBDIIC&C can of course do what's needed. :)

It knows the live cell and pack voltages and can use IMAC&C P&P (if fitted) to command regen/assist autonomously overriding what the cars wants if a cell has a problem etc.

That's all to be added into the software over the coming days/weeks.

At the moment is just beeps and flashes when a cell/pack approaches or goes out of range etc..
I then back off or manually correct the issue with the IMAC&C P&P functions.

Lots of possibilities spring to mind for later.

IIRC As the LTO discharge graph just shows a slow drop in voltage associated with soc reducing..

It might be possible to program a BCM Interceptor to drive the OEM SOC Gauge in a way similar to the NIMH BCM based on resting pack voltage. We know the MCM automatically starts charging below 60/65% or whatever, we could drive the SOC in such a way as too take advantage of that for our own purposes.. Scaling our LTO voltage range inside the BCM Interceptor to mimic classic OEM behaviour.

I did do some of this before years ago in previous conversions, I'll just have to dust off the code etc.

Time for you to stop dithering and jump aboard the LTO express. :)
Then you can help out with your analytical skills and testing...
I can't do everything or explore every avenue or rabbit hole.

LTO is $300 a pack, and two needed for 48 cell setup. ;)
Never going to get much cheaper IMHO.
 

·
Super Moderator
Joined
·
9,229 Posts
All very interesting and constructive conversation from both folk.

From my testing results on the PL8, doing 8 cell segments at a time, there is very little capacity below 2V. In addition, and maybe more important, the cells of these USED LTO packs in many cases don't stay balanced very well below 2V.
Post #139 in this thread talks about some of this PL8 testing:


As you mentioned above, it is probably wise to keep to 2V minimum on discharge in non BMS mode. At such time as you can trigger assist stop on individual cell voltages, it might be appropriate to extend the lower limit a little at a time as you are doing. Even then, I don't think you will gain much capacity, because the lowest capacity cell will frequently trigger a stop by around 1.8V. The PL8 does a nice job watching individual cell voltages and it frequently triggered a discharge stop around 1.8V on the USED cells.

I may have saved some data from the PL8 showing the <2V tails of the 8 cell segment testing, but I am an "old line" guy and I have a tendency to just keep pencil notes ;)
 

·
Super Moderator
Joined
·
9,229 Posts
Time for you to stop dithering and jump aboard the LTO express. :)
Then you can help out with your analytical skills and testing...
I can't do everything or explore every avenue or rabbit hole.

LTO is $300 a pack, and two needed for 48 cell setup. ;)
Never going to get much cheaper IMHO.
Truer words were never spoken! If a a Gen 1 aficionado really loves their car, has a good serviceable example, and cares about having IMA functionality, there has never been a better time to kick the NiMH nuisance.

I'm even considering converting Dogbite. The recent BAT auction shows that prices for even tip top low mileage examples has collapsed and there is going to be scarce funds among many buyers going forward. Why shouldn't I just do version 3.0 in Dogbite and enjoy it in my very old age????
 

·
Registered
Joined
·
7,843 Posts
Time for you to stop dithering and jump aboard the LTO express. :) Then you can help out with your analytical skills and testing... I can't do everything or explore every avenue or rabbit hole.
It has crossed my mind, to jump aboard, but I bought small, new LTO cells ages ago to build a pack, before the FIT option became an option, and I have a hard time letting go of that 'dream', or simply seeing those cells go to waste. Back then I thought being able to buy new, small, inexpensive cells that could be a direct replacement for the NiMH, without much fuss, would be the best option for Insight owners. That's probably the main reason I took that dive... But then, I started seeing new promise in the stock cells and got re-sucked into understanding all that. And then the FIT option came online...

I think underneath it all my strongest compulsion/propensity is to save, conserve, to make the most out of the least, so bringing anything into my life that isn't absolutely essential is a major hurdle for me... More batteries might fall into that category...

The extent to which a replacement pack, whatever it may be, could bring new life, with years and years of future use, to an Insight that would otherwise likely end up as scrap very soon, is compelling. The extent to which reconditioning stock packs and maybe some management tweaks can do the same is probably a hair more compelling to me... I do see, however, that a more thorough/detailed and successful reconditioning regime is not something that's going to catch on. So yeah, that's a dead end... I've just got so much inertia on that front...

In general, it all seems pretty pointless to me, now, whether I use stock NiMH, FIT LTO, my small SCiBs, etc. I don't really care how my car performs, per se, as long as it works, keeps working, and doesn't become a time-suck that doesn't give anything back (such as knowledge, 'recreation', etc.). Originally when I bought the Insight I was thinking I could use it as a learning bed to transition to a full EV, and that it would be a good candidate to convert. Back then there weren't any EVs around and I thought making one would set a good example. But, it turned out to be more complicated than I imagined, and then EVs started to come to market, and I also realized that 'setting a good example' was pretty dumb in the first place. Et al... So, my larger Insight Dream fizzled...

So yeah. I guess you're right, a lot of 'dithering'...
 

·
Administrator
Joined
·
13,083 Posts
Discussion Starter · #120 ·
No issues to report.

Converted a LB CVT MCM today and added all the bells and whistles.

1) Fixed 150V hack.
2) IMAC&C P&P.
3) +40% amps.
4) BCM Interceptor

Testing it now in my manual car.. Can't tell any differences at the moment.

Added battery temperature alarm detection to the code.
This cuts both assist and regen if triggered and drops bars on the SOC gauge to 4.
 
101 - 120 of 202 Posts
Top