Honda Insight Forum banner

201 - 220 of 220 Posts

·
Linsight Designer
Joined
·
2,371 Posts
OK, based on the feedback thus far, I've decided to ship all LiBCM modules with full 48S/54S/60S hardware support. This will make LiBCM ~$100 more. The initial LiBCM units will only support 48S, with a future firmware update unlocking full 60S support.
 

·
Registered
Joined
·
1,174 Posts
I think the big key is going to be getting the IGBTs upgraded to handle higher voltages and getting the system to play nice with those voltages. Peter has pushed to 175 amps which from my few years in electronics should generate similar heat at all voltages the difference being the horsepower since heat is current squared times resistance neither of those change when going up in voltage if current is kept constant. Getting up to 400v (probably the highest the motor will take) would be 94hp at those same amps albeit briefly. If we're looking at 84s being the maximum then at 344v for 4.1 per cell we could push 80hp. As long as we're not nudging too close to the breakdown voltage and keeping the runs short enough that it isn't developing a lot of heat I can't see it greatly lowering the service life of the IMA motor.

I think a k sight would still be beating us in the quarter unless we add a turbo but I think if at least 80hp can be pushed from the motor and it's all bottom end torque that with my mutant trans we'll be pretty close to a k series to 60mph. I think with 5psi we beat them to 60 and if we go whole hog Julian style at 10 to 15psi then we can probably match them in the quarter.

I'm definitely pretty excited about the 60s on board support. It will still allow expansion boards to like 84s eventually?
 

·
Registered
Joined
·
1,036 Posts
I am pretty far out of my swim lane, but I believe that saturation (10 pounds of Guess in a 5 pound (iron) bucket) could be a limiting factor, and a sufficiently large magnetic field may also lead to demagnetization. I'm sorry that I can't point to an article but I'm walking my dog in the rain. In other words, it would be worthwhile to build a test bench to learn these things. It would be wonderful to have hard data captured by going beyond the limits in a controlled, well instrumented way. It would require fairly expensive test equipment, but that can be sold on eBay after the research is complete as easily as it can be aquired from eBay.
 

·
Registered
Joined
·
1,174 Posts
When they have the setups that can really start pushing the motor available to install I'll probably have a spare insight and also a 3rd ima motor. The 3rd IMA motor will be going off to a motor rebuild shop for some extensive testing. Breakdown voltage, etc to see what the hard limits are. I may even have it rewound to take more after they inevitably fry it. Unless it can just straight up take like 5 or 600volts it's going to be fried when they test it so I see if I have a safety factor at 344 or more so likely I'll have a rebuilt beefed up version if that happens. I know there's a lot of talk about using a different IMA motor but I think that would be far more work than rewinding one. Might cost a lot more but it would be far less work to have it rewound and I have money, don't have a huge amount of time.
 

·
Linsight Designer
Joined
·
2,371 Posts
I think the big key is going to be getting the IGBTs upgraded to handle higher voltages and getting the system to play nice with those voltages.
That will certainly become a limiting factor as we push past Peter's present experimental results. However, there are many additional problems to consider, too.

from my few years in electronics (175A) should generate similar heat at all voltages the difference being the horsepower since heat is current squared times resistance neither of those change when going up in voltage if current is kept constant.
It's certainly true that circuits with no reactance (i.e. entirely resistive elements) don't consume more power as you increase voltage (but keep current constant). Unfortunately, this theory doesn't hold true for non-linear circuits, which includes both downstream IMA subsystem. The IGBT drivers are firing just above the audible spectrum (maybe 25 kHz). Each time a switch fires, there's a period where the IGBT is neither off nor 'on' (i.e. in saturation).

As the IGBT PNPN junctions transits across this 'active' region, power through an IGBT scales considerably more than linearly (due to Miller Capacitance, tail currents, etc), but not quite O^2. In fact, IGBTs are notoriously bad at generating wideband RF emissions (aka 'nasty interference'). Therefore, IGBT drivers intentionally slow down the gate drive charge, which makes IGBTs stay in this active region longer.

In the IMA motor itself, as soon as the field coils have saturated, any additional power you sink into it is converted entirely to heat (in saturation, the IMA motor is essentially a dead short). The higher you go in voltage, the less time it will take for the inductive coils to saturate (Ldi/dt). Therefore, you end up needed to fire the IGBTs even more frequently, which causes them to sink even more power. And of course all that effort is wasted, since the additional energy is getting turned directly into heat. The only way to prevent this would be to rewind the IMA motor to make it more inductive (so that it doesn't saturate as quickly), thus helping to give it a higher flyback voltage. A better method would be to redesign the rotor to account for the increased magnetic fields at said higher voltage. This would include both a thermal and a field-focusing redesign.

Getting up to 400v (probably the highest the motor will take) would be 94hp at those same amps albeit briefly.
The stock IMA motor absolutely will not tolerate 400 volts, or if it will, it won't be a motor for long. To prevent saturation, that would require increasing the switching frequency roughly 10x, which would increase the IGBT switching losses roughly 10x, or else if you didn't increase the power there, then you'd need to increase the power sunk as heat in the motor itself. It's unlikely the motor could safely tolerate this additional heat - without permanent damage - for even a few milliseconds (if I had to guess)... certainly the IGBT driver would not (but of course you could move up to a larger IGBT driver in the same Mitsubishi series; that's only a mechanical kludge to get working).


If we're looking at 84s being the maximum then at 344v for 4.1 per cell we could push 80hp.
LiBCM's 84S support (i.e. with QTY2 external daughterboards) isn't designed for standard lithium cells, but rather for LTO (and maybe LiFEPO4), both of which have considerably lower voltage ranges per cell.

LiBCM's safety creepage & clearance were designed for a 220 volt operating range, which might be safely pushed to 250 volts, but above that you'd need to space the various components further apart to guarantee safety isolation persists. Certainly LiBCM will not safely work at 300 VDC, for MANY MANY reasons, even if you swap out for higher voltage rated components.

IMO you need to limit your LiBCM plans to not more than 250 VDC, or else you need to accept the risks that you could breakdown the galvanic isolation elements and thus electrocute yourself. I absolutely will not endorse anything beyond 250 VDC, and do not recommend using LiBCM above 220 VDC (because that's what my design calculations were).

Will LiBCM fail at 300 VDC? Probably not, and honestly the IMA motor and IGBTs are much more likely to fail first. At 300 VDC, you'd need to entirely rethink the entire insulation environment... for example, you'd need to change all the HVDC cables to a higher rated insulation factor.


As long as we're not nudging too close to the breakdown voltage and keeping the runs short enough that it isn't developing a lot of heat I can't see it greatly lowering the service life of the IMA motor.
I strongly suspect Peter's existing experimental data is already grossly pushing the IMA system's limits. Even if you replace everything in the IMA bay, my guess is you'd still need to rewind the motor itself to prevent destroying it in a matter of seconds.

I think a k sight would still be beating us in the quarter unless we add a turbo but I think if at least 80hp can be pushed from the motor and it's all bottom end torque that with my mutant trans we'll be pretty close to a k series to 60mph. I think with 5psi we beat them to 60 and if we go whole hog Julian style at 10 to 15psi then we can probably match them in the quarter.
I don't have much experience between the two, but will offer that rather than turbocharging, you'd be much better off supercharging the engine with a duplicate IMA system driving the spindle. You'd basically tie the relevant control signals in parallel with the existing IMA system (driving the IMA motor), and then spool the supercharger with a BLAC-driven motor.

I'm definitely pretty excited about the 60s on board support. It will still allow expansion boards to like 84s eventually?
Right now 54S is already beyond the original scope of the project. If 54S ends up working (almost certainly the case), that doesn't guarantee that 60S will work (or work well). Let's first get the 48S design out and field-proven, then we can cross that bridge later.

I'm super glad you're excited to push the boundaries. I'm just in a different headspace than you are right now trying to finish the 48S design.

Keep in mind LiBCM's primary design goal is to provide a simple, safe, drop-in lithium solution for insight owners. I suspect 80%+ of LiBCM purchasers will keep it bone stock. LiBCM is certainly a good test bed to build on, but my goal right now is to ship the "base" version. I'm excited that you're excited to push the IMA to the limits, but most people aren't going to want to risk destroying their car. You and a handful of us will eventually explore that path, but that is a LONG way away in regards to the engineering effort I'm willing to expend for these niche cases. I'm certainly down to devote more effort to pushing the limits in the distant future.
 

·
Registered
Joined
·
1,174 Posts
Ok, well don't get me wrong 54 or even 48s is still pretty exciting, the power increase is quite drastic already. Even if we end up limited to going up to 220 or 250v there's a bit left on the table still. I'm hesitant to run higher amperage but from what you're saying about issues with running higher voltages then it looks like amperage is the way to go but for very very brief times. I'll have to look into some way to monitor temperature if I'm going to be trying to push up amperage beyond 200.

It looks like from your test drive there's not too much voltage droop at 70a, it was around 20, right? It will be interesting to see what it droops once you get all of Peter's stuff integrated to do 140a. It looks like you should be pretty close to the 30hp mark on the IMA which is very exciting since I think they run closer to 13hp factory.
 

·
Linsight Designer
Joined
·
2,371 Posts
I am pretty far out of my swim lane, but I believe that saturation (10 pounds of Guess in a 5 pound (iron) bucket) could be a limiting factor, and a sufficiently large magnetic field may also lead to demagnetization. I'm sorry that I can't point to an article but I'm walking my dog in the rain. In other words, it would be worthwhile to build a test bench to learn these things. It would be wonderful to have hard data captured by going beyond the limits in a controlled, well instrumented way. It would require fairly expensive test equipment, but that can be sold on eBay after the research is complete as easily as it can be aquired from eBay.
I actually have a homemade desktop dynamometer, including all the test equipment. Specifically, I have a Prony Brake Tester, which I primary use for sub-kW spindle motor testing (to analyze the market's offerings). Certainly the mechanicals would need to get larger to handle the additional power, but all of the existing instrumentation will work as-is. You're absolutely correct that it's not cheap ;)... fortunately, the test & measurement company I used to work for sold their hardware at cost... if anyone wants to use National Instruments hardware for this project, I probably have it (or something equivalent).

I am not going to take on this project in the next six months. If you're interested, you should look on youtube at some of the existing DIY Prony Brake testers. Many people use a caliper with brake pads to dump the heat... if someone with a welder and some technical know-how wants to copy an existing designs, I'll provide the instrumentation and software.
 

·
Linsight Designer
Joined
·
2,371 Posts
Ok, well don't get me wrong 54 or even 48s is still pretty exciting, the power increase is quite drastic already. Even if we end up limited to going up to 220 or 250v there's a bit left on the table still. I'm hesitant to run higher amperage but from what you're saying about issues with running higher voltages then it looks like amperage is the way to go but for very very brief times. I'll have to look into some way to monitor temperature if I'm going to be trying to push up amperage beyond 200.
IMO, you'd need to replace the entire junction board, the lithium battery modules (which have fusible links), the HVDC cables, and the IGBT driver in order to push beyond 200 A. I really think 40 horsepower is already pushing the limits, and I believe Peter has nearly gotten there with his various hacks over the years. As Sean mentioned, the correct way to test this will be on a test stand with instrumentation.
 

·
Registered
Joined
·
1,174 Posts
Well, even 30hp plus my transmission upgrade already drastically changes the car for the better, I think I'll be pretty happy when I get my hands on it, I've just always liked seeing where the limits are.
 

·
Linsight Designer
Joined
·
2,371 Posts
As the IGBT PNPN junctions transits across this 'active' region, power through an IGBT scales considerably more than linearly (due to Miller Capacitance, tail currents, etc), but not quite O^2. In fact, IGBTs are notoriously bad at generating wideband RF emissions (aka 'nasty interference'). Therefore, IGBT drivers intentionally slow down the gate drive charge, which makes IGBTs stay in this active region longer.
Followup: One thought to reduce the IGBT switching loss would be to dig inside the PDU's IGBT-pre-driver module. We could replace the series gate-charge resistors with lower-valued components, which would decrease the switching time, and thus the heat generated inside the IGBT driver. For example, if the OEM R_gate is 40 Ohms, we could replace those with 20 Ohms, which would approximately halve the IGBT's switching losses. If you were sinking 1300 watts before in the IGBT driver, that would drop it down to something like 850 watts. Those mJ really add up when you're switching tens of thousands of times per second! At these frequencies, switching losses likely account for 80% or more of the heat generated in the IGBTs.

Note that this will almost certainly increase RF emissions... you could attempt to reduce emissions slightly by placing a ferrite bead around each IGBT's emitter lead... this inductance would slow down the turn-on/turn-on time slightly, but would eat most of the noise. The closer you put it to the IGBTs, the more effective it would be. It is very likely possible that by modifying the IGBT-pre-driver, you could use the OEM IGBT driver much closer to its limits without causing it to overheat.

Technically this doesn't require FCC approval, since components in a sealed metal enclosure inside a vehicle are exempt from mandatory testing. This is a carve-out for the automotive industry, which for many decades used mechanical distributors which were absolutely bad neighbors in RF-land. And yet the carveout is so loosely written that it still applies to pretty much all electronics permanently installed in a motor vehicle... but auto manufacturers would get raked over the coals if they didn't seek RF approval... we the casual tinkerer hobbyists retain the legal protection, though.

The second highest IGBT heat generating component is the tail current, but there's nothing we can do about that.

A note on IGBT cooling: The MCM lets the IGBTs get really hot. For example, during auto-stop, the PDU fan is always disabled, regardless of PDU temperature. Odd that Honda would sacrifice temperature for cabin noise, but ok. To solve this problem, I recommend wiring the PDU fan to always remain at the low speed setting (e.g. by bypassing the fan relay directly to the ignition). With the PDU fan now always on at low speed, the next recommendation is to rewire the MCM's PDU_FAN_LOW lead to the fan's PDU_FAN_HIGH relay. This will cause the PDU fan to turn to high speed whenever the MCM commands it to low speed. I like this configuration.
 

·
Registered
Joined
·
1,174 Posts
That's interesting. I don't want to take your time away from the main project explaining things that are honestly over my head. I'm kinda like a kid who understands a few words here and there when it comes to electronics, I know basic circuits but igbts start going outside of the scope of my knowledge pretty quickly.
90899
 

·
Registered
Joined
·
1,174 Posts
Is the PDU fan the main battery fan? If so I have mine upgraded and set to always blow high. I might have a switch added for winter though as I don't think that's ideal when it's like -20 out thought I guess it takes in cabin air so it might actually heat it?
 

·
Linsight Designer
Joined
·
2,371 Posts
PDU fan is the squirrel cage fan bolted to the PDU, on the left side of the IMA bay. Remember that it's technically an 8 volt fan, so make sure to power it through the OEM series resistors.
 

·
Premium Member
Joined
·
281 Posts
I decided to buy another set of four modules last Monday 6/7. They just showed up here Friday 6/11.
I guess you could say I'm all in LIBCM!:)
IMG_20210611_112330148.jpg
 

·
Registered
Joined
·
15 Posts
Can some one please measure a 12S module and post it. If I can't find a 12S module I will hack a 18S that I have. Thank you.
Can someone who has the 12S battery measure the internal length of the side plate? I believe this is the only dimension we need, when cutting and welding the 18S side plate to 12S size.

And also inspect the notches (pointed to by the vertical arrows below) to see what they are for. Thanks.

View attachment 90858

Looks like QTY2 more 18S modules are in my future.

George
 

·
Registered
Joined
·
200 Posts
90925


As I still don't have the 12S measurements myself, I thought of another way to get it.

Mudder had provided individual cell measurements (taken from post 48 link)

90924


The width of 1 cell (plus plastic spacer) is 15.6mm. So the width of 6 cells (18 minus 12) and spacers is 93.6mm.

So I think what might work is, a 93.6mm strip could be cut from the middle of the 18S sideplate. Then the sideplate pieces welded back together.

Caution: Proceed at your own risk as I haven't tried the above procedure yet.

And Mudder did say about the measurements that "I make no guarantees that this is accurate".

So when you take your 18S module apart, do your own measurements. Would suggest making a cardboard replica of the 18S sideplate, cut a 93.6mm strip from it and check for fit before cutting the real item.

George
(edited for typo)
 

·
Registered
Joined
·
123 Posts
Discussion Starter · #218 ·
Since a 54s configuration leaves only 6 cells on the last LTC, I'd imagine there are some interesting considerations in trying to keep the design open for future use of 60s. (Assuming these 6 cells were equally divided amongst the two internal LTC MUXs.)

I have been curious how BMS manufacturers like Orion handle unpopulated cell connections in their systems which accept any numer of cells, outside of multiples 12.
 

·
Linsight Designer
Joined
·
2,371 Posts
LTC6804 doesn't care how many cells are connected, as long as the most positive and most negative cells are powered. If you look at the LibCM schematic, page 2, you can see how I took advantage of this fact to automatically configure the LTC IC to power itself from either 42S or 48S... no jumpers required. It's a pretty clever circuit.

Note that LTC recommends tying all unused cells together, but this isn't actually required because all the cells are already tied together with a ladder-resistor network.

The same configuration will be used for 54S and 60S... for the 54S option, the specific connector that goes from the ribbon cables to the OEM lithium connector will tie cell 54+ to C60+ on the LiBCM PCB... and everything else will just work.
 

·
Registered
Joined
·
123 Posts
Discussion Starter · #220 ·
Thank you for the Honda Insight, LiBCM, & other electronics lessons, both on the forum and in video. Always looking forward to them.

Viewing the schematic, I think I understand. 6 cell stack powers LTC without going though extra series resistance due to C54+ physically tied to C60+. No voltage potential exists when a measurement is made on any of the unused cell channels, as all are connected to C54+ due to their PMOS have 0V on the gate.
 
201 - 220 of 220 Posts
Top