Honda Insight Forum banner

1 - 13 of 13 Posts

·
Registered
Joined
·
612 Posts
Discussion Starter #1 (Edited)
I am building a grid charger with logging capability based on the three Meanwell 48V supplies plus one 350 mA constant current 48V LED driver design.

I hate to add another post to the several thousand? posts already, but I've decided to, as I need to get this built and my car on the road.

Questions:
  • Is measurement in volts sufficient or should I try to measure tenths of volts?
  • Is there anything to gain by making the output current adjustable?
  • What's the preferred connector for the harness that can keep fingers away from the 170V on either end?
  • What precautions are people taking to make sure the vehicle is not accidentally sitting at 120AC or 170VDC potential? (Optoisolation? Trusting the airgap in the power supplies? Plugging the device into a GFCI wall socket?)
 

·
Registered
Joined
·
612 Posts
Discussion Starter #2 (Edited)
I'm going to try to answer my own questions, then ask more questions.
Questions:
  1. Is measurement in volts sufficient or should I try to measure tenths of volts?
  2. Is there anything to gain by making the output current adjustable?
  3. What's the preferred connector for the harness that can keep fingers away from the 170V on either end?
  4. What precautions are people taking to make sure the vehicle is not accidentally sitting at 120AC or 170VDC potential? (Optoisolation? Trusting the airgap in the power supplies? Plugging the device into a GFCI wall socket?)
  1. I may be able to measure at 1/10th volt if I put the constant current supply at the bottom of the stack and measure only its voltage.
  2. Making the output current adjustable is one step beyond making it switchable, which is necessary for switching out the charger and switching in a discharge load.
  3. Not Anderson Power Poles.
  4. Still need to think a lot about safety. I switched the charger from a metal to plastic chassis. Still concerned about the possibility of a fault putting the car at 120 VAC or 170 VDC relative to the garage floor. GFCI is not sufficient; they can be damaged by lightning or the device plugged into a non-GFCI circuit
So what happens if the charger, which is putting out at least 144 volts (3x 48V + whatever at 350 mA), is connected to a battery that has been discharged to 100V? It's very late, and I can't seem to noodle through this simple question.
 

·
Registered
Joined
·
612 Posts
Discussion Starter #3
I want to elaborate on my last question, because I have not found an answer in the forum posts yet, and I think it is the most important question: how does this grid charger design deal with deeply discharged banks, whose voltage at a 350 mA charge current would be well below that contributed by the constant voltage supplies.

The grid charger design I'm using is the one that employs three 48V constant voltage supplies and one 350 mA constant current supply which can develop up to 48V, all in series.

I'm going to round up 48V to 50V to make the math simpler. I am also assuming that the constant current supply will adjust its output voltage to whatever is required to the highest that will not exceed 350 mA, which could be anywhere from just above 0V, up to its max output voltage of 50V. (Which is the definition of a constant current supply, but I'm assuming, to simplify this discussion, that it does not drop out at some low positive voltage until it is just above zero.) (Note: after working through this, I wonder if this is, in fact, it reduces current as it approaches zero volts and if this is what makes this solution work even for deeply discharged banks.)

What I understand:

When the steady state (that is, has had a moment to stabilize after everything is turned on) charging voltage is above 150V:

Each constant voltage supply will supply 50V at 350 mA. This current is set/limited by the constant current supply.

The constant current supply's output voltage will adjust to equal the steady state charge voltage @ 350 mA minus 150V. In other words, if the steady state charge voltage at a particular moment is 158.4 volts, the constant voltage supplies are delivering 50V each for a total of 150V and the constant current supply is providing 8.4 volts.
As the batteries fill, the charge current will remain maxed out at 350 mA but the constant current voltage output will increase, potentially (no engineering pun intended) to a maximum of 50 V (200V total).

Two things I don't understand:

1) The voltage across and behavior of individual cells as they fill, and how the charger's ability to deliver 350 mA to a combined maximum of 200V might impact individual cells. It would be helpful to know what is going on at the micro level within aging cells to answer this question. I am starting to see how "balancing" might work: with each cell getting the same amount of current, a cell with higher internal resistance than its neighbors will have to dissipate more power (v=ir and p=iv so p=iir); "weak" cells get a larger share of the charging energy? (Can our 350 mA charger damage a cell that might otherwise be saved or would the Insight have already thrashed it? If a car has sat for a long time, can its first run damage a weak cell that might otherwise be brought back with conditioning?)

2) What happens if the bank demands more than 350 mA at a voltage below 150V?

It is this last question that has me most puzzled, because if I deep-discharge the bank to 100V, certainly if the charge current were limited to 350 mA, the voltage would not rise much past 100V for a while. But the constant voltage supplies want to put out 150V. The constant current supply can't go negative??????

On this last question, what do you see in practice? How does this design deal with deeply discharged batteries?

Does the constant current supply actually start reducing output current as its output voltage approaches zero, such that the total battery voltage rises to 150V but its current draw is some very small safe level?

Or?
 

·
Registered
Joined
·
612 Posts
Discussion Starter #4
Another question. Is there enough room between sticks of cells to thread strings of tiny diodes to sense abnormal temperature rise of a single cell during charging (diodes can be used as temp sensors, their forward voltage drop varies with temperature) or would that block fan airflow? Could a pack be disassembled and said diodes expoxied to cells (no electrical connections) without having to break rivets or welds?
 

·
Premium Member
Joined
·
3,421 Posts
1) battery voltage will be limited based on current input, i.e., voltage rises in response to input current at a voltage higher than pack, and 350mA will only drive it so high. Rare that you'll see a reasonably healthy pack go much higher than 178V. Doesn't matter if the supplies CAN deliver 200V or 2000V. The battery can't voltage can't increase that much with 350mA input

2) The bank demands nothing. The "bank" is a battery. Batteries are a source. If you apply a source of higher voltage to the pack, current will flow into the pack, and its voltage will rise in accordance with the input current and its internal resistance. If the battery voltage rises to meet the source voltage, current flow will reduce.

What happens with this configuration is that the constant voltage supplies discover they can't operate at their set voltage, but they don't give up. You'll get the various power supplies pulsing at random. Sometimes, you'll see slightly higher current values as the 48V PSU fire off. This will help raise the pack voltage.

Eventually, when pack voltage has risen from this pulsing as well as NiMH's tendency to increase in voltage towards nominal, all supplies sync up and can provide their voltage at the 350mA limit of the LED PSU.

I avoid the issue by using only 2X APC-35-350 with their output in series. Provides roughly 56-200V of constant 350mA current. Note that the lower limit is fuzzy as they won't sync up at the minimum voltage, but I've seen full current applied at under 100V without issue.

The design in question works fine. Hybrid Automotive has sold thousands of units based on a similar design.

So... you want to attach 120 diodes to cells and add wiring to monitor it? Have you actually looked at the variation in forward voltage drop with temperature?

Do you have the PTC strips on your sticks? If so, you can simply monitor the resistance of that circuit as it was designed to alert the BCM if a single cell became hot.

Note that Honda removed the PTC strips from the 04 HCH1 and on. I speculate that the benefit wasn't worth the cost.

Mike's Genesis One charger had PTC circuit monitoring.
 

·
Registered
Joined
·
612 Posts
Discussion Starter #6 (Edited)
Thank you for taking the time to read my lengthy posts!
1) battery voltage will be limited based on current input, i.e., voltage rises in response to input current at a voltage higher than pack, and 350mA will only drive it so high. Rare that you'll see a reasonably healthy pack go much higher than 178V. Doesn't matter if the supplies CAN deliver 200V or 2000V. The battery can't voltage can't increase that much with 350mA input
OK, so if the bank is healthy, the voltage increase will stop before the constant current supply reaches its maximum voltage, and it will hum along burning off the 350 mA as heat.

Did I read somewhere that as individual cells become conditioned, the final bank voltage (with charger still active) may be lower compared to earlier charges where conditioning is incomplete?

2) The bank demands nothing. The "bank" is a battery. Batteries are a source. If you apply a source of higher voltage to the pack, current will flow into the pack, and its voltage will rise in accordance with the input current and its internal resistance. If the battery voltage rises to meet the source voltage, current flow will reduce.
This helped me realize that one should create different circuit models, one for when the battery is a source (discharging) and one when it is a sink (charging).

What happens with this configuration is that the constant voltage supplies discover they can't operate at their set voltage, but they don't give up. You'll get the various power supplies pulsing at random. Sometimes, you'll see slightly higher current values as the 48V PSU fire off. This will help raise the pack voltage.

Eventually, when pack voltage has risen from this pulsing as well as NiMH's tendency to increase in voltage towards nominal, all supplies sync up and can provide their voltage at the 350mA limit of the LED PSU.
Hmm. "Pulsing at random" and "don't give up" tells me that when the battery voltage is sufficiently low, one or more of the supplies sees its output reverse and then all bets are off: we are operating in uncharted territory. The supply behaves according to how its protection circuitry has been designed. Which is going to differ between manufacturers, and models within manufacturers, and possibly even different versions of the same model, with some of them blowing up, their designers having not considered that a power supply output would be attached to a reversed battery.

Do I have this right? If so, I would want to test this before hooking it up to the Insight and the only way I can think of doing so is unprintable lest someone read it, attempt it, and end up in the hospital if the supply's protective circuits aren't designed to defend against this. (which, it seems, they probably are)

I avoid the issue by using only 2X APC-35-350 with their output in series. Provides roughly 56-200V of constant 350mA current. Note that the lower limit is fuzzy as they won't sync up at the minimum voltage, but I've seen full current applied at under 100V without issue.
Wish I'd found this design earlier; I would have saved a few bucks.

The design in question works fine. Hybrid Automotive has sold thousands of units based on a similar design.
Fortunately I bought the Meanwell's specified in Mike's block diagram PDF so I will probably be OK.

So... you want to attach 120 diodes to cells and add wiring to monitor it? Have you actually looked at the variation in forward voltage drop with temperature?
I instrumented an alternator with a few diodes to see where it generates heat. Using an Arduino I was seeing the MSB increment one bit for each degree increase in temperature, roughly. Recently I built a similar circuit to measure temperature of an LED lamp which will become a dome light (I'll post on that later; it has a super-bright mode but normally adjusts brightness based on how dark it is outside so as not to blind the driver). On this one it took several degrees of change before the lowest bit flipped, but that's enough to tell if something is OK or too hot.

It probably is not practical. It might be better to just measure voltage of the string at the bottom and top and three points in between, equal number of cells measured. Any difference will indicate at least one cell being off, and then current control can be used if necessary to keep the errant cell's voltage from becoming too great and forcing it to dissipate too much power. This might be also be useful during discharge to ensure that no one cell is reaching a point where it might reverse. It would work for a bank that only had a few questionable cells.

I have a feeling that this has been rehashed in the forum dozens of times already and that I have it wrong.
Do you have the PTC strips on your sticks? If so, you can simply monitor the resistance of that circuit as it was designed to alert the BCM if a single cell became hot.

Note that Honda removed the PTC strips from the 04 HCH1 and on. I speculate that the benefit wasn't worth the cost.

Mike's Genesis One charger had PTC circuit monitoring.
I don't know. I haven't pulled the battery out. I have the car in pieces already so I didn't want to add to the mess yet.

I think that what might be best for now is to start off simple, wire up the grid charger, put together a simple logger that reads the output voltage, and get some experience under my belt. But part of me wants to know the state of the battery before I begin. Maybe I should yank open the battery compartment and see what I am dealing with, space and connection-wise.
 

·
Premium Member
Joined
·
3,421 Posts
OK, so if the bank is healthy, the voltage increase will stop before the constant current supply reaches its maximum voltage, and it will hum along burning off the 350 mA as heat.
The bolded portion is irrelevant. Even a horribly damaged and disfunctional battery will behave the same way.

Did I read somewhere that as individual cells become conditioned, the final bank voltage (with charger still active) may be lower compared to earlier charges where conditioning is incomplete?
You have it backwards. As voltage depression is removed from the cells, the peak voltage increases.

Hmm. "Pulsing at random" and "don't give up" tells me that when the battery voltage is sufficiently low, one or more of the supplies sees its output reverse and then all bets are off: we are operating in uncharted territory. The supply behaves according to how its protection circuitry has been designed. Which is going to differ between manufacturers, and models within manufacturers, and possibly even different versions of the same model, with some of them blowing up, their designers having not considered that a power supply output would be attached to a reversed battery.
No we aren't. You're repeating something that's what... 6-8 years old and has been done literally tens of thousands of times on thousands of cars

Provided you have a diode in series with the battery to prevent backflow from the battery, you'll be fine.

Your concerns may be valid, but there are thousands of examples where they have been demonstrated to be unfounded. You are overthinking it.

I think that what might be best for now is to start off simple, wire up the grid charger, put together a simple logger that reads the output voltage, and get some experience under my belt. But part of me wants to know the state of the battery before I begin. Maybe I should yank open the battery compartment and see what I am dealing with, space and connection-wise.
Yes. This. Just monkey-see/monkey-do, and you'll likely be fine.

Concerning the bolded portion, you don't even know how to determine this, nor is there a way to do it without establishing boundaries and defining what the "state" is and then repeating it AFTER the process to see how things have changed. Do you want to measure all 120 cells? Do you want to measure the capacity of all 120 cells, 20 sticks, or what? Do you want this to take months or days?

If you're looking for a recommendation, using the link in my sig, measure the 10 tap voltages 24 hours after the car has been off (no power to or from the battery). That will give you a 30,000 foot view of the battery's "state." The larger the deviation in the 10 voltages, the worse the "state" of the battery. Can I quantify it? Nope.

For additional diagnostic/data gathering juice, taking those 10 measurements at 144, 132, 120V and 96V can give you an idea of out the reconditioning process affects them. Also, measure discharge time to those same target voltages and note how they change with each cycle.

If you know your discharge current, you can compute pack capacity based on discharge time.

EDIT: Read the "Everything I have to say about Grid Charging" link in my sig
 

·
Registered
Joined
·
612 Posts
Discussion Starter #8
Your concerns may be valid, but there are thousands of examples where they have been demonstrated to be unfounded. You are overthinking it.
The advice to block current from battery to charger with a diode - thank you for that tip. I will definitely do that.
Concerning the bolded portion, <about wanting to know state of of the battery before charging it the first time> you don't even know how to determine this, nor is there a way to do it without establishing boundaries and defining what the "state" is and then repeating it AFTER the process to see how things have changed. Do you want to measure all 120 cells? Do you want to measure the capacity of all 120 cells, 20 sticks, or what? Do you want this to take months or days?
I should have said that I want "baseline measurements" and your tip to measure the 10 tap voltages should be good enough for now. I will have to do that sooner than later.

If you're looking for a recommendation, using the link in my sig, measure the 10 tap voltages 24 hours after the car has been off (no power to or from the battery). That will give you a 30,000 foot view of the battery's "state." The larger the deviation in the 10 voltages, the worse the "state" of the battery. Can I quantify it? Nope.
Will do.
For additional diagnostic/data gathering juice, taking those 10 measurements at 144, 132, 120V and 96V can give you an idea of out the reconditioning process affects them. Also, measure discharge time to those same target voltages and note how they change with each cycle.
I would like to look at plots of the voltage of those taps over time, but that may have to wait.
If you know your discharge current, you can compute pack capacity based on discharge time.


EDIT: Read the "Everything I have to say about Grid Charging" link in my sig
Done. And I do have some hall-effect current measuring devices in the junque box and a voltmeter with a DC current probe should I decide to just get it done.

Thank you for spending the time on bringing me up to speed (at least a little).
 

·
Registered
Joined
·
612 Posts
Discussion Starter #9 (Edited)
I want to know more about how the classic grid charger design might work when recharging a bank that has been discharged to a voltage well below that of the sum of the supplies in the charger.

Original design thread, first post having a link to the pdf of the block diagram: http://www.insightcentral.net/forums/modifications-technical-issues/14496-grid-charger-balancer.html.

I ran a test using a 12V gel cell in great need of a charge or with a shorted cell (resting at ~7V when I pulled it out of the junque pile, and charging at about 10.7V at 350 mA after a few minutes.

I wired an RS-25-12 (12V version of the 48V constant voltage supply) and an LPC-20-350 (350 mA constant current up to 48V supply) in series.

I then connected the battery up to the supplies individually, and then both together.

Voltages, battery not connected:

RS-25-12: 12V
LPC-20-350: 48V
In series: 60V

As expected.

With the battery connected to:

RS-25-12 alone: 12V 1.2A (within reason; this battery is shot)
LPC-20-350 alone: 10.5V 0.35A (within reason; supply is limiting current)
Supplies in series: 11.3V 1.0A (interesting!)

So with the supplies in series, I'm not seeing current regulation at 350mA! In the last configuration, if I measure the voltage across the supplies individually:

RS-25-12: 12V
LPC-20-350: -0.7V

So the constant current supply is seeing a negative voltage across its terminals. Going into this experiment, I expected one of these supplies to "go negative". I was unsure which would, and by how much, because I knew one or more would be operating out of specs where any protective circuitry would be kicking in.

For the final experiment, I left everything connected in series, but removed mains power from the RS-25-12, leaving only the constant current supply powering the battery. Here again I did not know what to expect. Here's what I saw:

RS-25-12: -0.7V
LPC-20-350: 11.5V
Battery: 10.8V
Current: 350 mA

This tells me several things:
- Supply output voltage may "go negative" if the battery voltage is sufficiently low - it's simple math, actually
- The constant current supply may go offline leaving only constant voltage supplies operating, and the current to the battery may rise to the limit of those supplies
- We are in uncharted territory, namely, the "out-of-spec" behavior of the supplies when reverse biased, which is handled by protective circuitry whose design and operation is opaque to us and not listed on the spec sheet
- The charging voltage and current of the battery used for this test is within the capacity of the constant voltage supply, so we can't draw conclusions about what would happen if the load is greater than what the constant voltage supply can provide. Therefore we cannot draw any conclusions about potential for damage to supplies and resulting hazards from this test.

I had seen in other threads that people are complaining of issues when using brands other than that specified in the original design that uses Meanwell supplies. Perhaps this is due to other supplies having less robust protective circuitry that doesn't work in a way that supports the behavior of the original design without breaking.

Has this - the behavior of the supplies charging a bank at a voltage less than the sum of the constant voltage supplies - been discussed in past threads? I searched for a while and didn't find anything, except that in the early threads, Mike said he discussed the design with TRC Electronics. Perhaps I need to kick this over to that thread and ask there.
 

·
Registered
Joined
·
612 Posts
Discussion Starter #10
What happens with this configuration is that the constant voltage supplies discover they can't operate at their set voltage, but they don't give up. You'll get the various power supplies pulsing at random. Sometimes, you'll see slightly higher current values as the 48V PSU fire off. This will help raise the pack voltage.

Eventually, when pack voltage has risen from this pulsing as well as NiMH's tendency to increase in voltage towards nominal, all supplies sync up and can provide their voltage at the 350mA limit of the LED PSU.
Thank you! I finally searched the original thread http://www.insightcentral.net/forums/modifications-technical-issues/14496-grid-charger-balancer.html for the word "pulsing" and found the relevant posts:

http://www.insightcentral.net/forums/modifications-technical-issues/14496-grid-charger-balancer-17.html#post146169

Mike and Peter describe the problem with pulsing supplies.

So this design is *definitely* relying on the behavior of the overcurrent protection mechanism. My test suggests that the constant current supply goes offline first (not conclusive). It also suggests, and measurements across a powered off supply bolster, the idea that they have diodes across the output which conduct when a reverse current is applied.

So when a supply goes offline, it appears that the voltage across the terminals reverses to the diode drop voltage, and if it's the constant current supply which goes offline first, the supply falls out of "current regulation" and the others' output will rise to meet the load. And if the load is too great, they'll enter a pulsing mode where they shut down and then try to restart.

I have not experienced this personally so I am guessing that this is what is happening.
 

·
Registered
Joined
·
612 Posts
Discussion Starter #11
I have been working on a design that keeps the constant current supply operating with a positive voltage across it, and powers up constant voltage supplies as needed, ensuring that the sum of constant voltage supplies is 8 volts less than the battery voltage (8 volts is a safe amount above the constant current supply's cutout voltage.)

The end of this post supports this:

http://www.insightcentral.net/forums/modifications-technical-issues/22125-grid-charger-parts-list-3.html#post232159

I believe the constant voltage supplies already have diodes across their outputs internally but I am not taking chances and am adding my own so that when they turn off they only go at most 0.6-0.7V negative and are not harmed. The diodes should be able to handle quite a bit more than 350 mA.

It should be possible to turn the constant voltage supplies on with an Arduino-controlled relay on the mains. This all needs to be built with isolation between the high voltage section and any microprocessors that control it, and the rest of the system (12V supply, mains voltage, etc.)

Looking for safe connectors I learned about J1939 connectors used for OBD on heavy vehicles. These are nine pin connectors that can be locked together if you don't get the version designed for diagnostic tools which has no locking mechanism. The pins appear to be recessed on plug and socket, but not enough to make me completely comfortable. However it appears that as you push the connectors together, a central pin connects first (that will be the safety ground), then maybe 6 more connect, then two appear to be deepest and connect last.

There is a need to ensure that there is no high voltage on those pins until the connectors are together, and likewise, cut power before the connectors come apart. Those two deepest pins might be good for that; they could be sending data between controllers on the in-vehicle harness and in the supply that cause them to provide power only when the data connection is established and cut it immediately when it fails.

Now J1939 is not found in my garage but may interest someone who gets their hands on the car should I be unable to control that outcome. So breaking one of the pins in the plug and filling the corresponding hole in the receptacle should help ensure that some curious soul isn't able to plug into it with their truck OBD reader.
 

·
Registered
Joined
·
30 Posts
I don't think this is explored much because batteries are expected to be in a normal voltage range, and most power supplies don't misbehave terribly outside of normal ranges. Don't worry about the charger: worry about the pack!

You're overthinking the connector a bit, I'd say. The J1939 circular connectors look stout, but why not an old-school 97-series? You can't get much more make-first/break-last than a metal threaded shell, and it's rated for this sort of thing. It's also cheaper and very common on the used/surplus market. A third connect/disconnect level is for signaling a harmless connect/disconnect, not for safety.

Or put the grid charger in the car and hard-wire it like everyone else. Do that first just to test it.
 

·
Registered
Joined
·
612 Posts
Discussion Starter #13
The J1939 circular connectors look stout, but why not an old-school 97-series?
Thank you for the tips. I've used the 97-series in the past, but had forgotten its name. It is a great connector. The J1939 appears to be designed so that fingers don't accidentally touch pins on either end. Is there a 97-series like this or do they all have exposed pins on one end? I'd prefer to use a unique 97-series over a connector used for OBD on large trucks, but prefer more to have no exposed pins.
 
1 - 13 of 13 Posts
Top