HV Battery Display

[IamIan]

This idea isn't really new... but this is my take on it, and I wanted other feedback and thoughts.

We know the on dash instrumentation panel lies to us about the current going into and out of the HV battery...

Some have added LED indicators to show real time the current coming in during regen and out during assist from the HV battery.( included with MIMA discussed elsewhere viewtopic.php?f=4&t=7188 )

And modifications to the Regen Assist indication level has also been discussed under:
Building a more accurate regen/assist guage
viewtopic.php?f=4&t=7076

And sometimes we know the SoC display is not correct... especially if we are doing the grid charing:
viewtopic.php?f=4&t=3996

So it would be useful if we had a display to tell us a more accurate idea of the electrical power during assist or regen... like the LED displays... I would prefer a digital display myself but the LED display with MIMA has worked well .... and to have a better idea of the current amount of Ah in the battery.

The Cycle Analyst described here:
http://www.ebikes.ca/drainbrain.shtml

Seems to fit the bill for most of this.

It does up to 200V... which we should not see above that from the insight even during max Regen.
It can be set for over 400 Amps... the Insight under full Assist doesn't do over 100 Amps.
It will track up to 199Ah ... more than enough.
Resolution is under 1Amp which is plenty for during driving.

The only modification needed that I see is that it is designed for use with a Shunt for the Current Sensing...
But the hall effect sensor on the HV Battery pack gives us a similar signal... we would just have to modify it slightly to fit the range the Cycle Analyst is expecting from the Shunt we will not be using... or make it think the Hall Effect signal is a Shunt resistor signal...

Thanks to the work of others we know the hall effect current sensor on the insight is powered by the +/- 12V feed and has an analogue DC voltage output proportional to the current flow, ranging from +2VDC at full 100 amps of assist and -1VDC under full 50 amps of regen.

It would seem to me that in order to modify the Hall Effect Current signal to match the Cycle Analyst all we need to do is rescale it... we want it to swing positive and negative so the Cycle Analyst can properly track the assist and regen... but it would only be expecting a swing from +200mV to -200mV from the 100 Amps of Assist to 100 Amps of Regen... so we will need to scale down the +2V and -1V down to +200mV and -100mV...

I was thinking after I get this setup for the Insight's HV stock battery... I will setup another one to go with my booster battery to track the two separately from each other.

Of course this is tracking the electrical flow back and forth and is not accounting for the less than 100% efficient charge and discharge efficiency of the batteries.... unless we also include that efficiency into our scaling device... but I think that adjusting for the cycle efficiency of the batteries would be an upgrade to work on for latter and the basic unit would still be better than what we have now... especially as a first step.

What do people think?
Any better ideas?
Any important detail you think I am missing?

[uhtrinity]

Something like this could be helpful to track exactly how much charge is left. Even though the car is detecting additional Ah's in my setup it still isn't perfect, especially if the system doesn't recal or I have more charge than the system expects. I still have the 3 bar soc bug when I get closer to empty (140 - 145 volts). Sometimes I can go a long time before the system cuts juice, other times it cuts sooner. Only when I force assist for a long time will it go to zero bars and attempt a recal. I see this becoming more of an issue if I can charge my packs 10Ah, when the system is expecting < 8Ah.

Rescaling the shunt voltage values doesn't seem like it would be to hard. Couldn't a simple voltage divider be used on the voltages from the hall sensor? I have the extra hall sensor and it would work great right on the main in/out lead going to the motor controller. The charger feed could also be threaded through the sensor to keep track of soc when the other systems are shut down. However, would the hall sensor be sensitive enough to pick up charge at 500mA or lower?

[IamIan]

Originally Posted by uhtrinity

However, would the hall sensor be sensitive enough to pick up charge at 500mA or lower?

In the High Voltage and Current unit we would be using it is listing a current sense resolution of 0.1Amps or 100 mA.
I has a calibration setup procedure for fine tunning.

The unit only uses 7mA when the back light is off.

It is listing that it stays within 3% over the operating temperature range of -10C and 50C.

so it seems that it will measure at least to the nearest 100mA... but the ability to do a calibration for the current should allow to correct for most issues I would think.

Just keep in mind that the batteries are not 100% efficient and 10Ah in does not mean 10Ah out... some of the energy will be lost to heat and such... unless this charge / discharge cycle efficiency is taken into account during the rescaling of the hall effect sensor's signal.

[retepsnikrep]

Good idea, I'm looking at something like this home made using a PICAXE pic with my PHEV project.

For those with MIMA fitted it would be reasonably straight forward to tap off the already inverted current sensor output. Mike's MIMA pic uses a 0-4v input, 2v being the 0 current point. This is what I intend to do when building my own AH counting system for my PHEV system. Others could copy the basics of that circuit and modify it to provide the outputs they need.

I agree a simple voltage divider should take the standard Insight current sensor output down to the level reqd for the bike brain.

Why not just add a shunt into the main + line at the top heavy duty connectors. Might not like the idea of the main pack V+ going inside the car but it's the easiest option. Double insulate the wires. I assume the device is powered from the HV battery anyway is that correct?

I'm happy to work on a Insight specific PICAXE project battery fuel guage/monitor system if you like, as I will incorporate the code into my phev system later anyway. Thinking caps on.

What do we want to monitor display?

1) Battery current in and out. (Use Mikes MIMA current sensor output conditioning circuit to get this into pic from Insight sensor)
Or use our own hall effect sensor so BMS can be left operating 24/7 monitoring grid charging etc

2) Battery voltage (Need to opto isolate this, ideas?)

3) Battery temp? (One probe just to give us and idea)

4) SOC (Calculated from voltage/current/time etc)

PICAXE/LCD Display circuit to be powered by normal 12V car system. If not backlight then drain is pretty minimal.

[IamIan]

Originally Posted by retepsnikrep

Good idea, I'm looking at something like this home made using a PICAXE pic with my PHEV project.

glad to hear it... more good news

Originally Posted by retepsnikrep

Why not just add a shunt into the main + line at the top heavy duty connectors. Might not like the idea of the main pack V+ going inside the car but it's the easiest option. Double insulate the wires. I assume the device is powered from the HV battery anyway is that correct?

A shunt could be used... it would be the easiest method .... they recommend a 2mOhm shunt resistor for 100Amp detection... I did a short / quick search and did not find a Shunt at 2mOhm that would handle 100 Amps... anybody already know of someone who sells them to save me the searching??

The down side for me with adding a shunt is that we already have a hall effect sensor in place giving us a signal of what we want... so no need to add resistance... even 2mOhm will generate up to 20Watts of wasted energy as heat during use in the Insight.

You are correct that it takes its ~7mA of power from the HV supply... and that would end up getting routed to the display... even if it were just a sense wire and would be 24 gauge or thinner ... if it shorted to anything you can used the thinness of the wire as a built in fuse.

That is one benefit of us building our own is that we can opto isolate the HV instead of running it up front in thin little wires.

Originally Posted by retepsnikrep

I'm happy to work on a Insight specific PICAXE project battery fuel guage/monitor system if you like, as I will incorporate the code into my phev system later anyway. Thinking caps on.

more good news
Glad to hear it

Originally Posted by retepsnikrep

What do we want to monitor display?

1) Battery current in and out. (Use Mikes MIMA current sensor output conditioning circuit to get this into pic from Insight sensor)
Or use our own hall effect sensor so BMS can be left operating 24/7 monitoring grid charging etc

2) Battery voltage (Need to opto isolate this, ideas?)

3) Battery temp? (One probe just to give us and idea)

4) SOC (Calculated from voltage/current/time etc)

PICAXE/LCD Display circuit to be powered by normal 12V car system. If not backlight then drain is pretty minimal.

My wish list of things to monitor / display would be very similar to your initial list:

1> Current In ( Regen ) and Out ( Assist) , Ideally not only up to 100Amps but also down low enough to properly register the smaller currents people might grid charge with at least accurately down to 100mA.... that by itself is no small task in my limited experience... I like the Hall Effect sensor ... because it doesn't add any extra waste resistance and the Insight already comes with one in place for the stock pack.

2> Battery Voltage also useful... at least the total HV ... and if someone is feeling exceptionally ambitious I would say the voltages of each subpack.

3> Battery Temperature... at least toward the center of the pack where the thermal images show it gets hottest... If we can do it without messing up the BCM it might be nice to tap off the temperature sensors & PTC strip already in a pack... the PTC strip allows us to know if any one cell goes into thermal failure and gives us a nice average pack temperature reading.

4> SoC ... this gets tricky very quickly to do accurately... so many different variables go into it
Voltage is nearly useless for NiMH except at the top 90%+ and bottom 10%- SoC...
Current flow in and out doesn't take into account the cycle efficiency...
Efficiency unfortunately changes at different current rates... it is not the same from charge to discharge and at different temperatures, or even different SoC, or pack age, SoC is not the same as Ah capacity, etc ,etc ....

To avoid making it overly complex we could try to get a reasonable approximation that we then recalibrate from time to time like at oil changes every 3 or 4 months or something.... We can use Voltage and temperature as failsafe upper and lower end markers... then use Ah counting in the middle adjusted for a general cycle efficiency... I would think that should be close enough to give us a fair idea of the SoC ... and shouldn't get too far out of whack between recalibration at oil changes.

All together it still looks like a very big project to me...

that was why I was tempted by the high voltage and current model of the Cycle Analyst as it does most of this already... even if not exactly the way I would want it... such as opto isolated and such.

[IamIan]

I was just reading through the Cycle Analyst manual...

It can be programed with the resistance of the type of shunt you are using... and supports shunts as low as 0.0728 mOhm in the high current mode...

If I find a 0.08mOhm 100+Amp Shunt then that is only a maximum of 0.8 Watts of wasted resistance as heat even under full 100Amp assist.

So that makes the shunt option a bit more appealing... even if the hall effect sensor is also still appealing... and the specific made unit from retepsnikrep is the ideal.

[uhtrinity]

and if someone is feeling exceptionally ambitious I would say the voltages of each subpack.


Well, using the existing wiring you could monitor 10 subpack pairs, or in my case 20.

[retepsnikrep]

Originally Posted by uhtrinity

and if someone is feeling exceptionally ambitious I would say the voltages of each subpack.


Well, using the existing wiring you could monitor 10 subpack pairs, or in my case 20.

I'm probably going to go with 40-45 TS 50ah LCP cells if I can afford them, I may not monitor each individual cell V with my pic master BMS but each cell will have a minimum analogue over/under V detection and balancing with an opto isolated master bus to tell the monitoring PICAXE if there is a problem. If people are going to keep/add to the nimh pack system with additional buddy packs then I can see it would be nice to monitor them (subpacks) if possible.

I have pondered an individual picaxe powered monitor for each TS cell or nimh subpack pair with a balancing capability. They could communicate over an opto isolated serial bus with the master pic as in my case. That is pretty straight forward IMO.

To implement on a standard IMA pack, each subpack pair powers via a low dropout 5v reg and some filtering /smoothing an 08M Picaxe. It measures the voltage of each subpack using a Voltage divider to bring voltage down to 0-5v input range 10bit resolution. it could also have an output controlling a mosfet 1 watt balancing resistor across the same pair of subpacks. The picaxe could communicate with the master PIC using an interupt driven cascade output onto the opto isolated serial bus.

The master pic sends an interput to Pic 08M (1) it drops it's data onto the serial bus and sends an interupt to the next 08M pic in the chain which then drops it's data etc. The master receives this lot in sequence and any failure to recieve indicates a problem etc. System could be expanded adinfinitum almost Master pic analyses data and displays/acts on it as necessary.

So a bms for the Insight pack on this basis would require 10x 08M picaxe and 1x18X master pic or similar. Master Pic could also monitor the PTC pack and thermo sensors, current in/out and pack V if reqd. Just brainstorming.

[IamIan]

I got a response back from the Cycle Analyst people I thought others might find useful :

Blue = them
Green = Me
------------------

From: "Ebikes Information" <info@ebikes.ca>
To: "Ian G" <ian_p_george@yahoo.com>

Hi Ian and thanks for the inquiry.

#1> My understanding is that it will operate fine bellow 200VDC.
The battery pack I am using has a nominal voltage of 144V and the highest applied voltage it sees during charging is 180V. So the Cycle Analyst should be fine with it.


Yes, the high voltage model would be OK with this.

#2> It is my understanding that the shunt it is attached to determines how it counts the Amps. The highest discharge rate for my application is 100Amps. The manual I read suggests a 2mOhm Shunt for reading 100Amp loads. The manual also says you can calibrate / program the Cycle Analyst for the specific shunt you are using.


That is right, you can program in any resistance between 0.078 to 9.99 mOhm. For 100 amps, I'd recommend a 0.5mOhm shunt. 0.08mOhm is quite low and would result in higher offset and drift errors.


#3> I already have a Hall Effect sensor on my battery project that swings from +2V at 100 Amps of discharge to -1V at 50 Amps of charging. Can the Cycle Analyst use this signal instead of a shunt if I step it down into a range of +200mV to -100mV?


Not easily. It needs to be a ground referenced signal with very low imedance, so you'd need to step it down to +200mV with a resistive divider and then follow that with an op-amp buffer. Hall Effect sensors have pretty low dynamic range and large offset errors, so resistive current sensing is generally preferred when doing amp-hour integration.


#4> It is my understanding that the Cycle Analyst counts the current flow in Ah in both directions Charging the battery and Discharging the battery. Does the Cycle Analyst display a total for each direction or a net total of the two combined?


The main screen shows them combined. The regen screen toggles to alternately display the forwards and negative amp-hours.

#5> It is my understanding that there are different versions of the Cycle Analyst. Will any of the high voltage and high current models do what I am looking for or will I need to look for a specific model of Cycle Analyst to meet my needs?


Yes, but unfortunately there aren't any high voltage models on hand, so it would have to wait until I prepare another batch of these. The stock model is good only to about 110V, it wouldn't handle your case.


----------------------------------------------

So a 0.5mOhm 100+Amp Shunt will do the trick the easiest ... only wastes ~5 Watts of energy even during 100 Amps of Assist.... But they are out of stock... so I guess we will see how far we get with the retepsnikrep suggested more specific built device for now... the ideal retepsnikrep monitor might end up being available before the High Voltage High Current Cycle Analyst is.

[IamIan]

In a recent minor curiosity search I was doing about ways of decoding the CAN communication in the Insight ,
( low on my list stuff but entertaining to read about from time to time. ) , I found something a little interesting:

$250 for USB to PC connection Hardware and software of basic unit Does OBD-II

$230 for Enhanced Honda CAN expansion,
http://www.autoenginuity.com/Honda-A...tems-List.html

Together it would cost $480 + Shipping... a bit high for my budget right now... but nice to know that the work has been done and that there is a product out there ready to buy... and plug into a laptop or car PC....

Kind of gives me visions of a Knight Rider like dash with a car PC and small LCD display interfacing with and displaying things like... the CAN information ... the Car PC could have a GPS hook up... movie , music , etc...

If I do this much I would have put some videos of KITT interactions in the car PC.

------------------

While It may not crack the Honda CAN...
The BMS work fellow Insighter , "retepsnikrep" / Peter , has done looks really nice too:
viewtopic.php?f=4&t=7665

And, Peter's work has the additional nice feature of balancing functions.