Honda Insight Forum banner
61 - 80 of 92 Posts

·
Registered
Joined
·
58 Posts
I'm in the PNW. Been patiently waiting for linsight to finalize from beta as while I'm quite handy with a wrench, am less so when it comes to the electronics side.
@Balto @Recovering_Gasaholic
I'm also in the PNW and will be installing two 47Ah conversions when they're available. Maybe we could make a party out of it in Washington state.

I still haven't received the last custom part required to ship these FoMoCo kits.
At this point even if I received them tomorrow, there isn't enough time to shoot and edit the installation videos prior to Thanksgiving... which I'm spending in Wisconsin with my wife's family.

So obviously these kits won't ship until early December. If you've been following along for a while now, this slipping schedule won't surprise you ;). It'll be ready when it's ready. I'm excited to share it.
Really appreciate all you're doing on this! While I'd love to have it done, life always comes first. My friend and I just submitted our purchase forms the other day and are sitting on a big pile of FoMoCo batteries to do the conversion.
Additionally, after watching your video on pack heating I was surprised you didn't consider an aluminum-clad PCB heater on the bottom of the FoMoCo packs. I have no idea what the G3 Insight modules look like, so those may be more challenging. From my experience, thermal management through the bottom of a lithium cell is very efficient as the can and jellyroll wicks the heat effectively. This would require a bespoke design for the FoMoCo packs, but wouldn't require the fans to be on and wasting BTU's recirculating air into the cabin. From what I've seen of your build, the vertical clearance of adding even a single PCB could be a factor though.
 
  • Like
Reactions: SkunkWorks

·
Linsight Designer
Joined
·
4,479 Posts
Discussion Starter · #63 ·
By negative logic, do you mean that power supply shutoff requires a line pulled high, so if what is pulling up the line loses power, the supply turns on?
Not that unsafe, but basically there's negative logic in how to control the output voltage. In both cases, if LiBCM turns off (or is unpowered), there's no risk that the power supply will turn on.

Note that I've worked around this issue entirely... it was just annoying.

I will be looking forward to the next commit to learn more about the issue at 120V vs 240V operation.
With Vin=120 volts, the output power is limited to either 900 or 1000 watts (datasheet claims one thing, boilerplate claims another). Regardless, in my testing at Vin=120, the supply outputs 1200 watts, which causes the input rectifier to overheat (since it's twice the current to yield the same output power). I'm working with MeanWell to better understand this limitation.

I can prevent this overheat condition from occurring by driving the current control pin, but that requires external circuitry (which I'm probably going to throw onto this heater PCB I'm designing at the moment). If I go that route, I might also move all the internal supply modifications (I mentioned previously) onto this PCB, too, so that we can just use the unmodified supply as-is.

For my LTO projects I am favoring redundancy on certain things. Any reason I should not employ both internal and external shut off circuits as independent shutdowns? (My place has taken a lightning strike before and I've seen what crazy things it can do to electronics, including leaving stuff half working.)
All LiBCM grid chargers have redundant control circuitry. That's why the "negative logic" issue I mentioned previously won't actually cause any issues... the redundant control mechanism might not actually work, but the primary one will. Again, though, note that I've worked around the "negative logic" issue... it no longer exists with the internal modifications I'm making to each supply.

Finding the insulator failure must have been fun....
It took a minute... I definitely wear ear plugs and glasses when testing these things... and in this case it was worth it.
 

·
Linsight Designer
Joined
·
4,479 Posts
Discussion Starter · #64 ·
After watching your video on pack heating I was surprised you didn't consider an aluminum-clad PCB heater on the bottom of the FoMoCo packs.
There is zero.000000 additional room above and/or below the FoMoCo modules.

I have no idea what the G3 Insight modules look like, so those may be more challenging.
For the 5AhG3 modules, I considered replacing the bottom metal insert with a PCB, but there are numerous safety issues in going this route. Creepage and clearance for the HVDC bus is pretty difficult. I'm still considering this as an option, but it would be difficult to do so safely.

From my experience, thermal management through the bottom of a lithium cell is very efficient as the can and jellyroll wicks the heat effectively.
I agree. Wish we had room.
 

·
Registered
Joined
·
2,166 Posts
With Vin=120 volts, the output power is limited to either 900 or 1000 watts (datasheet claims one thing, boilerplate claims another). Regardless, in my testing at Vin=120, the supply outputs 1200 watts, which causes the input rectifier to overheat (since it's twice the current to yield the same output power).
Really? So if you do something as simple as connect it to a few series LTO packs for charging (which will soak up gobs of amps), and nothing else, it will happily supply current until the input rectifier blows up?
I can prevent this overheat condition from occurring by driving the current control pin, but that requires external circuitry
In other words, you're setting the lower current limit. So, do they plan to upgrade the input rectifier/heat sink, or just drop the current?
In both cases, if LiBCM turns off (or is unpowered), there's no risk that the power supply will turn on.
Ah, so the supply enable input is pulled up internally, so that the supply turns on if no cable is connected. Did you change that input to a pull-down?

(actually I should hold these questions until mine arrives)
 

·
Linsight Designer
Joined
·
4,479 Posts
Discussion Starter · #66 ·
Really? So if you do something as simple as connect it to a few series LTO packs for charging (which will soak up gobs of amps), and nothing else, it will happily supply current until the input rectifier blows up?
Fortunately, there's a temperature sensor bolted to this input rectifier. In my testing (at room temperature), the input rectifier gets up to 110 degC before the supply shuts down. Still working with MeanWell to fully understand the behavior. Regardless, LiBCM will be able to control the output current so this won't be an issue.

In other words, you're setting the lower current limit. So, do they plan to upgrade the input rectifier/heat sink, or just drop the current?
The rectifier is bolted to the aluminum extruded chassis... it's on a short steel standoff, though, which isn't a great heat conductor. I'm thinking I might add a heat sink to the other side of the rectifier (so that I don't have to limit output power as much when input voltage is 120).

Ah, so the supply enable input is pulled up internally, so that the supply turns on if no cable is connected. Did you change that input to a pull-down?
The OEM configuration has the enable input hard tied (i.e. 0 Ohms) to +12V_AUX, which enables the supply. This is via a small cable jumper plugged into the isolated connector on the supply. My mod adds an isolated open collector to that configuration, so that LiBCM can control the supply_enable signal. LiBCM also controls the actual input power via a TRIAC.
 

·
Registered
Joined
·
2,166 Posts
@mudder thanks again for this tip.

Well, my order arrived.

I'm excited to finally be able to do pack-level load testing on my NiMH packs, lol

For the mass, the mounting slots are kinda small. Also nothing the four.screw posts on the bottom suggest 3 mm screws but they are all potted. So the top plate holding down the transformers seems important. It would be interesting to see how it fares on a vibration table in different orientations. I wonder if they have done that and have a report/data?
White Product Rectangle Textile Art


I am looking forward to seeing what can be done with the CAN interface. I have a Pi set up to do this; just need to find the protocol for this thing.
 

·
Linsight Designer
Joined
·
4,479 Posts
Discussion Starter · #68 ·
I'm excited to finally be able to do pack-level load testing on my NiMH packs, lol
A back-burner project I've got on the list is to create a "front panel" for this supply, so that the user can easily set the output voltage and current limits. Basically a super cheap Lambda Gen, except that its output voltage and current ranges are discontinuous down to 0.

For the mass, the mounting slots are kinda small.
The mounting slots fully encompass the bolt cross section (QTY4 4.5 mm diameter x 6 mm long). Also, the extruded aluminum enclosure is plenty thick to prevent deflection even under the most severe circumstances. I don't foresee any issues safely mounting it.

Also nothing the four.screw posts on the bottom suggest 3 mm screws but they are all potted.
The M4 mounting slots shouldn't be potted... all of mine have keepouts that prevent the potting from getting into the M4 mounting slot area. Maybe we're not talking about the same thing?

So the top plate holding down the transformers seems important.
MeanWell definitely has heat spreaders conducting up to this steel cover, but honestly not much heat comes up through them in my testing. Even with the steel cover removed entirely, the three cover-sinked components don't get very hot. Most of the heat gets sunk into the extruded aluminum enclosure, particularly the side with all the TO-220 junk bolted to it.

It would be interesting to see how it fares on a vibration table in different orientations. I wonder if they have done that and have a report/data?
Specified vibration handling is "10-500 Hz, 5G 10min/cycle, 60 minutes each cycle along X/Y/Z axes". That's pretty intense.

My biggest long term vibration concern is the one non-potted radial capacitor on one of the daughterboards. I've relayed this concern to MeanWell... in the meantime I'm just glueing this cap to the PCB since I already have the cover off to perform other electrical modifications.

I am looking forward to seeing what can be done with the CAN interface. I have a Pi set up to do this; just need to find the protocol for this thing.
FYI: CAN is only supported on the "UHP-1500-xxx-CAN" model... if you have the standard "UHP-1500-xxx" model, the CAN portion of the PCB isn't populated.
 

·
Registered
Joined
·
2,166 Posts
A back-burner project I've got on the list is to create a "front panel" for this supply,
I too thought about this, gutting some cheap supply for the panel meters. Thoughts shifted to a regular display and Arduino.
The M4 mounting slots shouldn't be potted... all of mine have keepouts that prevent the potting from getting into the M4 mounting slot area. Maybe we're not talking about the same thing?
I was talking about the four screws that hold the potted PCB to the plate. You can't see the screws because they are covered in putting material. You can see the inserts pressed into the plate, that they screw into, on the bottom of the supply.
MeanWell definitely has heat spreaders conducting up to this steel cover, but honestly not much heat comes up through them in my testing. Even with the steel cover removed entirely, the three cover-sinked components don't get very hot. Most of the heat gets sunk into the extruded aluminum enclosure, particularly the side with all the TO-220 junk bolted to it.
I was thinking how the top plate mechanically restrains the two transformers, not thermal restraint. But you found some vibration specs.
FYI: CAN is only supported on the "UHP-1500-xxx-CAN" model... if you have the standard "UHP-1500-xxx" model, the CAN portion of the PCB isn't populated.
I did not expect the CAN interface when I ordered it, and gleamed it from a skim of the datasheet this morning. Thanks for the correction.
 

·
Administrator
Joined
·
14,166 Posts
The CAN model is certainly interesting and moves us forward from the 1990's into current vehicle comms territory.

I might suggest adding CAN capability to REV (next) of the LiBCM PCB and Mario with Pegasus etc.
That must be pretty cheap to do assuming there are Arduino libraries etc

OBDIIC&C can speak CAN ;) so could control/program a charger maybe..
(Note I have not looked at the charger protocol etc as yet)

Lot's of things speak CAN now......
 

·
Registered
Joined
·
2,166 Posts
The curves for voltage and current control of the non-CAN/PD units start with full output at 0 V to about 0.4 V, then output is uncertain to 1V, then minimum output (not zero) ramping up to full output at around 5 V in. So that jump from full output to minimal ramping back up to full will be annoying.

For remote control I would need two DACs to control these lines and a digital output for on/off and another to cut off the mains and two analog inputs for voltage and current sensing and the need to employ a control loop to adjust the outputs, since the control voltage is only relative. And for my build I would definitely want a second set of eyes (another MCU) monitoring the voltage and current and able to pull the plug independently.
 

·
Linsight Designer
Joined
·
4,479 Posts
Discussion Starter · #72 ·
The curves for voltage and current control of the non-CAN/PD units start with full output at 0 V to about 0.4 V, then output is uncertain to 1V, then minimum output (not zero) ramping up to full output at around 5 V in. So that jump from full output to minimal ramping back up to full will be annoying.
The CC and/or CV inputs aren't pulled up to VCC, so when they're not connected they just sit at 0.0 volts. So the non-linear curve guarantees the supply will work if you don't connect the CC and/or CV control signals.

For me, the more annoying thing is that the CC and CV signals aren't isolated from the output rail. You actually have to generate the 5V VCC rail with an isolated switches, and then you also have to isolate the CC and/or CV signals (e.g. with a phototransistor). I would have liked to see MeanWell isolate these signals inside the supply, and also provide pullups to an isolated 5V supply. It makes controlling these supplies slightly harder, and for no reason.

Still a great supply.

For remote control I would need two DACs to control these lines and a digital output for on/off and another to cut off the mains and two analog inputs for voltage and current sensing and the need to employ a control loop to adjust the outputs, since the control voltage is only relative. And for my build I would definitely want a second set of eyes (another MCU) monitoring the voltage and current and able to pull the plug independently.
Given the above, you would need two isolated DACs ;). And yes, you'd need all those control signals to variably control the current and voltage.
 

·
Registered
Joined
·
1,480 Posts
Mudder, someone on the Facebook insight page mentioned that the 60s pack has issues with regenerative braking. How big of a problem is it for someone who doesn't plug in regularly and is it looking like it can be overcome or should I settle for 48s?

If I have to settle for 48s the 5ah would be the faster setup since it has similar horsepower and lower weight, correct?
 

·
Linsight Designer
Joined
·
4,479 Posts
Discussion Starter · #75 ·
Mudder, someone on the Facebook insight page mentioned that the 60s pack has issues with regenerative braking. How big of a problem is it for someone who doesn't plug in regularly and is it looking like it can be overcome or should I settle for 48s?
IMO, if you don't plug in regularly, then you should opt for the 5AhG3 pack instead. FoMoCo is primarily intended for consistent PHEV operation... you'll end up hauling around extra weight that isn't terrible useful unless you plug in.

...but to answer your question:
The IMA system is able to regen a 60S pack when you let off the gas and/or apply the brake. However, the typical background regen when the pack is low doesn't actually charge the pack. This causes the pack to tend towards discharged.

If you want the extra capacity - but don't expect to grid charge often - then I'd go with a 48S configuration so that background regen works properly.

If I have to settle for 48s the 5ah would be the faster setup since it has similar horsepower and lower weight, correct?
Right now the 5AhG3 configuration has identical horsepower to either 48S or 60S FoMoCo. Comparing apples to apples, a 48S FoMoCo setup weighs about 40 pounds more than a 48S 5AhG3 setup.

So yeah, if you don't plan to grid charge on the regular, go with the 5AhG3 setup.
 

·
Registered
00 Honda Insight 5MT, 01 Honda Civic 5MT, 85 Chevy K10 5MT DMAX SWAP, 97 F350 Powerstroke 5MT
Joined
·
46 Posts
61 - 80 of 92 Posts
Top