Honda Insight Forum banner

21 - 40 of 63 Posts

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #21 (Edited)
P1444 'High voltage short circuit' (This does not stop the IMA working)


I've probed around with this disconnecting stuff and seeing what triggers it.

It wasn't the BCM temp sensing stuff, NTC and PTC circuits. (A usual suspect with leaky nimh sticks)
It doesn't appear to be a leak from the battery terminals and cases/mounts to the chassis.

When measuring the leakage voltage it felt like a small capacitor..

I note when I disconnect the center tap earth for the switch board capacitor snubber/suppressor pack the code does not occur.

It's possible the higher (doubled) voltages have tipped a flaky snubber/rfi/suppressor pack over the edge, and it's leaking through one of the six capacitors from either the positive or negative side to the chassis.

It didn't happen with the 48 cell packs and this is one of the same two switchboards.

If I leave the centre earth tap disconnected what will happen?

The other two connections for the multi film capacitor suppressor block are across the main battery +/-.

I will probably change out the snubber/suppressor block later as I have some spares..

It's just a pain to get at.

The danger is if one has failed or is leaky and another fails on the other side we have a direct short across the pack and magic smoke.

It would be quite easy to make a new suppressor pack using brand new caps rather than another 25 year old set from another switchboard. I might do that.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #22
Fitting the Meanwell DC-DC PSU into the car.


Luckily I had an dead OEM DC-DC so was able to scavenge the heatsink base plate.

It fits onto the old plate and still leaves sufficient clearance for safety between psu and switchboard.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #23
More tinkering waiting for parts.


I added some acrylic insulating sheet to the Meanwell case.
I had some kicking around and it was just sensible to include it.

I also added an ignition controlled 12v 70A relay to isolate the Meanwell PSU from the 12v battery when off to prevent the small back flow parasitic drain flattening it.

Also added fake temperature resistor so the MCM thinks the stock dc-dc is in situ and nice and cool.
(Might swap this for an actual NTC glued to the Meanwell case so it can report the correct temp.)

Also adding the switchable 120V max power hack into the MCM. :eek:

This will only activate with WOT and when the rpm is above the IGBT drive changeover point, so about 2500 rpm upwards. It's much easier on the IGBT once it changes to the higher rpm drive waveform.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #24
Because access is a bit tricky once everything is bolted back in I may well add another precharge resistor or connection points in case the slower charging of the reservoir capacitors causes pre-charge contactor issues later.

At ignition on the MCM expects to receive confirmation the big capacitors have charged to a certain level within a certain time via the VPIN voltage signal.

As we have effectively moved that voltage window up by nearly 50%, it takes longer to charge them to that level, and the MCM might timeout with a pre-charge contactor fault if it doesn't fall within it's expected range.

I shall do some capacitor versus voltage versus resistance versus time calculations and see what we get.

Another potential issue is if the main contactor turns on too early when a large voltage gap still exists between the actual battery voltage and the capacitor voltage then it is also stressed by the inrush current and possible arcing.

I might have to add another resistor across the standard pre-charge resistor to increase the precharge current and therefore reduce the charging time.

Have to be careful not to kill the pre-charge relay by going too far.
But it's probably safe to assume it has a fairly big Honda safety factor and isn't already on the ragged edge.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #25
Snubber testing.

OK I have the parts now but want to test the old snubber I have removed from the switchboard.

I have tried capacitance which shows correct. 0.1uf on each side.

The caps don't appear to have any obvious leakage or physical problems.

The resistance measurement is infinity so no obvious electrical leakage.

I have tried applying 250V across the snubber but am not really sure what to expect at the mid (earth) point.

Any genius ideas on testing?

I wonder how exactly the car detects voltage leakage to the chassis?

86163
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #26
Snubber rebuild.


De-potting and rebuilding the switchboard capacitor snubber unit.
Note this snubber does not use series resistors, it's just capacitors.

Luckily the old OEM potting compound wasn't super tough,
so it was possible to carefully pick it out and then remove the pcb.

I used 6x brand new 1600v rated 0.033uf caps,
the originals were 1500v rated.

Re-potted using clear silicone sealant.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #27
I've ordered a dual gang 47k linear pot so I can adjust the voltage the car sees much more easily.
This pot has has two electrically isolated variable resistors mechanically linked so they stay together.

86177


One track will be the VPIN resistor, the other track will be the BCM Fooler pre-resistor.
So when I adjust it both change at the same time and the same rate and they keep in step.

I'm presently using a fixed 33k resistor for the VPIN and the pre-resistor, so this adjustable 47k part covers all the range I might need. It will be fitted inside the MCM with the knob external so it can be adjusted without opening the MCM.

Variable resistors aren't usually super accurate so I might have to calibrate it first.
For instance If one track is consistently 1k behind the other when moving up and down, I can just add a 1k resistor into the circuit of the one that is low/behind.

 

·
Super Moderator
Joined
·
8,223 Posts
I got one of those also, but I though better of using it since I only need one fixed value.

What do you think the tolerance for the two might be? I just read your comment that .1% is best for the fooler and Vpin lines.
 

·
Linsight Designer
Joined
·
1,899 Posts
This capacitor network is a high frequency snubber that references the IMA pack to ground. This prevents high frequency noise from the IGBTs from causing interference. You really should keep the center-tapped ground connected to the chassis. Also note that there are two of these capacitor packs... one on the junction board and the other inside the PDU.

The P1444 issue you're seeing could certainly be due to this capacitor network leaking... other possible causes:
-leakage inside the IGBTs
-leakage inside the larger capacitors (inside the PDU)
-insulation breakdown in the BLAC motor.
-fault inside the DCDC.
-leakage inside the MCM (you should measure more than 300 kOhm from both MCM connector E wires to chassis ground).

The proper method to test the capacitors for leakage is to test them while they have the full pack voltage across them. You'd place a small series resistor from each 'hot' leg, and then ground the capacitor bank. With the capacitors charged, you'd then place a DMM across both terminals on each resistor (one at a time). Using V=IR you can then calculate the leakage current. Ideally it should be pretty much 0 A.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #30 (Edited)
Thanks for the ideas John.


Just back from a fun 50 miles test run. No ima light or issues to report.
The rebuilt snubber might have cured that P1444.
But I will see how it goes over the next few days to confirm this.

Now I have the Meanwell working I was able to charge the LTO pack properly using IMAC&C P&P upto ~2.55V per cell or 245V with a long period of steady 20A regen.

20A might not sound much, but don't forget W = V * A, so that's about 5kw of regen for ten miles or so.

I added a grid charger 30A connection point for future use.
I also added a parallel pre charge resistor connection point in case an extra resistor is needed to speed up the filter capacitor charging.

You can see we still have good clearance between the Meanwell PSU and switchboard.
Everything is mounted safely and securely.

Just need the dual gang 47K pot to arrive now so I can build up the voltage adjustment and switched voltage hack modified MCM.

I was seeing around 95A of current at 230V under load so that's around 22kw of assist.
There will still be a bit more to come with the switched voltage hack and dual pot voltage adjuster.

Need to tweak a few OBDIIC&C CAN settings and keep and eye on a few different parameters on a longer test drive. MDM temperature, BVO, MDV etc etc

I also need a plug in a 12v display in the cigarette lighter so I can see the Meanwell is active, as the dash IGN warning light no longer operates of course without the stock DC-DC controlling it.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #31 (Edited)
Well the MDM is back out on the bench. A few things to sort.

I have added the temp 5k NTC sensor from the old OEM DC-DC to the Meanwell power transistors so I can see it's temp via the standard DC-DC temp parameter in the OBDIIC&C.

There are two temps sensors inside the OEM DC-DC one is a 20k NTC the other is a 5k NTC.
The 20k one is used internally by the DC-DC for its own control and sensing.
The 5k one feeds externally to the MCM for the DC-DC temp parameter. (That's the one I have used)

The noisy meanwell fan was annoying me as well.

It's loud and ramps up and down constantly depending on the psu load.
Even something as simple as foot on/off throttle changes 12V DC the load sufficiently to cause it to change speed.

So I have brought the 12v fan feed wire outside the Meanwell case and added some terminals so I can add a resistor inline with the feed to slow it down or switch it off entirely. It does not have speed sensing feedback so the PSU won't know what I am doing with it.

For now I have inserted a 15R resistor inline to see how that affects its speed.
I can monitor the psu temp now so I should not kill it.

I have also opened the MDM up and rebuilt the IGBT snubber with news caps.

Looking at the HV leakage I am seeing a resistance of around 200k as reported by the Honda Diagnostics. I'll check it again with the new snubber etc.

There are plenty of leakage opportunities as has been mentioned, this higher voltage pack seeks out any weaks or damp spots.

[EDIT]

All back in now ready for testing.
 

·
Registered
Joined
·
6,843 Posts
I was seeing around 95A of current at 230V under load so that's around 22kw of assist. There will still be a bit more to come with the switched voltage hack and dual pot voltage adjuster...
That's pretty crazy, like a fox. Hard for me to imagine what more than double stock max assist would be like... If this ends up working longer-term it makes a larger pack all the more enticing...

Very nice build, all packaged very well.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #33
The Meanwell fan is certainly much quieter, I'm going to up the fan resistor a little more and try 20R. :)

I'm implementing throttle and rpm based activation logic for the final power hacks so it doesn't engage until the motor is over 2000 rpm and into the sine wave drive zone. Easier on the IGBT.

I've been measuring battery current with a clamp on meter and checking it against the current hacked and compensated OBDIIC&C readings. The correct menu compensation value for the + 30/40 hack is 144
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #34 (Edited)
I have discovered a new standard engine Insight G1 phenomena! Torque Steer.. LOL

When the WOT/RPM max power is activated now at 2000rpm plus you get a massive assist boost and the steering wheel starts fighting you. Quite funny really, but not as insane as some of the cool K-Swap cars at Insight Fest

Also starting to notice the opposite to the 48 Cell low V conversion.

The 48 Cell car had mega powerful regen and more limited assist (But still around 15kw when tweaked to the max)

The 96 Cell car regens less but still manages ~8kw, but assist is much more powerful right up until the rev limiter. 20kw+ and still a few more tweaks to go. :cool:

The higher voltage just easily pushes those Watts and Horses through the motor even when rpm and back emf is high.

Anyway on more mundane matters, changed the Meanwell fan resistor to 22R and I can't hear it now above the road noise even with the IPU cover missing. So anything between 15-22R is probably fine with the IPU cover in place and the carpets back in.

Just done a 30 miles test and IGBT and Meanwell heatsink temps did not exceed 35C, batteries themselves < 20C. Ambient about 16C here today.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #35 (Edited)
Next time I have the pack out I will investigate reinstating the MDM fan outlet conduit and floor baffle etc. Like Natalya has done with hers.

I blocked mine up years ago, but if I can sneak some conduit under that pack in the OEM spot it would be better. These projects are never finished. :rolleyes:
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #36
In order to achieve/extract maximum power from the IMA when under load you have to convince it that the battery voltage is as near to 120V as you can get without going below that.

The BCM fooler and BCM don't really care too much about impedance as the inputs are very high resistance.
So fooling them with resistor dividers to bring the voltage down is very easy.

The VPIN ADC input is a bit different and as i'm experimenting with bringing down the voltage from stratospheric heights (240V) to really quite low (120V) I suspect impedance if I use a simple voltage divider will rear its head soon.

There is a 100K resistor to ground in the internal MCM VPIN circuit, so when we add a resistor into the wire we create a simple divider.

I'm currently using about 37k but if I push this much further I get a P1576 MDM voltage error.

Now the single error I have had so far may be just be a glitch, so I need to do some more testing.

But if the VPIN error is consistent with divider resistances over say 40k, then I will simply revert to using an op amp buffer after a voltage divider that then feeds the VPIN input from the relatively low impedance OPAMP source.

I have a few OPAMP buffers kicking about after the earlier 48V Cell experiments.

It's all never ending.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #38
How much power are you able to regenerate near top pack voltage (~240V?) and at what RPMs?
That will have to wait until I have completed all my mods and tweaks then I will do some figures.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #39 (Edited)
Is it really 4 days since the last update? Wow times does fly.

Anyway I have spent the last few days, checking, consolidating, testing, rechecking, testing, redoing, testing etc etc.
I'm after safety, consistency, reliability, performance and want to get everything sorted before the rolling road testing.

The packs have stood up well to the drives and they are still balanced to within about 25mv across all 96 cells.

What's new since last time.

1) Switched voltage hack. I have tried several different circuits for this, but now seem to be settling down towards a battery voltage (-60V) following normal mode, and a fixed 122V for WOT MAX power mode.

This new enhancement ensures the car sees just over 120V when maximum power is demanded via IMAC&C P&P.
This allows maximum current and today with the battery at 2.5V per cell at 30C I was seeing 117A at 225V under load.

That's just over 26kw! And it felt like I had a tiger by the tail.. :devilish:

Note with this high voltage setup the power is higher at higher rpm than ever before.


2) I redid the circuit boards for the various voltage hacks, concentrated a bit more on isolation etc, and used a conformal coating on the boards to remove damp effects. The short circuit code has not appeared, so that may be due to the rebuilt snubbers or the neater better laid out circuits.

3) I tweaked the OBDIIC&C code so I have an rpm activation limit in place. Under which you can't activate KERS mode. I currently have it set at 3000 rpm so we are well clear of the square wave drive zone before it kicks in.

4) I eliminated the BCM Fooler parasitic drain by adding an ign switch feed to the fooler +.
This is basically a 450V rated opto coupler which turns on when the ign does and powers up the fooler.
I use the same type of optocoupler to activate the HV 120V fixed voltage using a 120V zener.

5) I tested manually running the IPU fan at low speed when the ign was on, but even after a decent test run natural airflow made it really not necessary for now. Both the IGBT and meanwell heatsinks remained around ambient + 10C.

You can't really use warp 9 assist for very long round here unless you want to lose your licence, so the IGBT doesn't heat up much. Low level continuous assist will likely heat it more, but I will test that in due course on a long run.

Other observations and notes.

a) It's now possible to over rev the car especially in 1st gear.
Normally the ECM controls assist and ign/fuel etc, so when it reaches the rev limit it cuts both and everything is cool.

However now the ECM is only controlling the ign/fuel when you WOT.

IMAC&C P&P is controlling and potentially demanding 26kw of assist as the engine speed ramps up due to you booting it with WOT.

When it reaches the rev limit the ECM cuts what it can but assist is still pulling hard and it looks like it doesn't have an rpm limit built into the MCM firmware so it just goes on pulling. I over revved it today in 1st gear by about 500 rpm before easing off. Just have to be careful in 1st gear..

Also I don't want to break the driveshafts so I don't often give it WOT in 1st gear.

b) At very high rpm ~6000 the IMA will hardly regen at all. Even when full regen is commanded manually.

It comes back slowly as rpm falls, but then there is a dead zone for regen at about 5000 rpm.
Below that it comes back in getting stronger down to about 2000 rpm.

So contrary to normal behaviour, for the most powerful regen now you need about 2-3000 rpm.
 

·
Administrator
Joined
·
11,763 Posts
Discussion Starter #40 (Edited)
86261


This is the voltage fooling setup i'm testing at the moment.

I need to replace the 5V switching relay with a solid state changeover switch in the next version but I don't have anything suitable in stock at the minute. (Any recommendations?)

Remember my actual pack voltage range at 2-2.5V per cell is 192-240V.

This is a brief description of what is going on and assuming the HV battery is at 240V.

1) When the ignition is turned on 5V is applied to the circuit at the +5v points.

2) U1 (anti parasitic drain) turns on and allows the HV battery positive flow through to R3.
R3 (33k) forms a voltage divider with the 10x10k (100k) BCM Fooler (not shown) and this outputs approx 180V or 18V per BCM tap.

3) The 0-5V voltage from the MDM VPIN output flows through RV1 (adjusted to ~33k) and forms a voltage divider with R5 100k this gives the same % reduction as the HV side, and the output flows into the unity gain op amp buffer U3 and then into the MCM VPIN input.

4) Transistor switch Q1 is activated by the IMAC&C P&P IN going logic high when you want 120V for maximum power. U2 conducts and brings the 120V zener diode D1 into circuit which pulls down the HV output to ~120V.

5) The relay RL also activates and changes the op amp buffer input to a fixed resistor R4 which forms a divider with R5, and is calculated to output the equivalent of 120V for the MCM VPIN input.

6) When you disengage things flip back to stages 2 and 3.

You will see the HV side is now optocoupled and isolated by U1 and U2.

My workshop bench car with 0-300V psu has been invaluable for testing and developing a lot of this before it ever gets into the actual LTO vehicle. The workroom is also warm and dry with music and coffee etc, much more conducive for idle tinkering.
 
21 - 40 of 63 Posts
Top