Battery Three way switch - any issues with this design?

pickworthi

100 W
Joined
Oct 1, 2020
Messages
129
Location
UK - Oxfordshire
I have three batteries on my e-bike, two main batteries in the front saddle bags, and a reserve in the back saddle bag. At the moment I plug and unplug as needed to change over. I want to see if I can make a three way switch, which will make the process during the ride easier. Switching will be at standstill, so not under load. Batteries are 10S 36V nominal (so 30V to 42V range). Maximum current is 15 amps.

I got a chunky (25A) single throw four pole switch for another project that got binned, so I would like to use that. This means I will be switching the positive battery inputs, with all three batteries connected to a common ground. I have attached a circuit diagram below - I can't find a 4 pole switch in Kicad - sorry. Just imagine an off position is present.

I can't see a problem with this arrangement, but since the batteries are connected to common ground, I figured it would be wise to ask people who have way more experience than me if there is any issue with this.

Thanks in advance for any views.

Screenshot_2021-04-09_10-24-46.png
 
You'll get longer range (higher actual Ah capacity) by paralleling for the whole ride.

Better current flow, less voltage sag and longer lifespan too.
 
Yes but.
I've measured parallel packs on my setup and they are measurably less efficient. In this case the reserve is 5AH, and the main packs are 21AH each - so plenty of riding before a stop for a change (and a snack!) and also those three in parallel will be way less efficient according to my measurements.

Anyway - since you didn't say that a common ground in my diagram should be avoided, I'm taking it that its OK.
 
yes all DC circuits should have a Common Reference, tied in with Negative Returns
 
pickworthi said:
I've measured parallel packs on my setup and they are measurably less efficient.
How are they less efficient?

I'm really curious to see what measurements and criteria you are using.


If you parallel them, then each one sees less load, so each one sags less in voltage, produces more watts (power), and creates less heat internally, and wears the cells out slower.


The diagram you show should work fine, as long as the switch itself (and the wiring and connections) can all handle the maximum currents that will ever flow, and that the switch is rated for the voltage placed across it.

If the switch isn't designed for at least as much voltage as the full charge voltage of the battery, then it can't guarantee disconnection (the physical gap between contacts may not be far enough apart). It is unlikely to ever be a problem...but that is one reason they have a voltage rating. :)

Since you are not switching under load, that's even less likely to be an issue...however, when you first switch on from the off position, if the caps in teh controller are not charged, there is a surge of current much higher than the usual for just a moment (you know that spark when you connect a battery to a controller?), and eventually that may wear the second contact (the first battery after "off"). If it's a physically largish rotary switch, contact welding isn't much of a worry, as you can apply significant force when turning it and would probably just push past any minor such problem if it ever does occur.

But...I have had a smallish toggle switch, and a few small slide switches, that were not rated for the use to which I put them, and eventually did weld the contacts together in such a situation, leaving the switch stuck "on", even though it was never switched under any load. With the slide switch and one of the toggles, it actually broke the metal contact slider off the switch mechanics when I tried to push it off, so it felt like it turned off but actually was welded on. :shock: I forget if this was on DayGlo Avenger or one of the early CrazyBike2 experiments (or maybe something else). :oops: Always fun to then start doing something with the system that you *think* is off...but isn't. :flame:

So now when I misuse a switch, I''m at least prepared for the eventual failure mode of it. :lol:
 
Thanks for your response.

amberwolf said:
pickworthi said:
I've measured parallel packs on my setup and they are measurably less efficient.
How are they less efficient?

I'm really curious to see what measurements and criteria you are using.

If you parallel them, then each one sees less load, so each one sags less in voltage, produces more watts (power), and creates less heat internally, and wears the cells out slower.

The method I use is to ride a set 50ish mile course around where I live. I plug a watt meter in line with the battery supply to the controller. All cabling is identical, except for a short parallel connector cable when using parallel packs.

I have two packs, each at 10Ah capacity (usable - not nominal).
Both packs balance charged (4010DUO) to 4.1V per cell at the start.
I ride the course with each battery connected on it's own in turn until they were both depleted (same low voltage cell alarms, the batteries have no BMS).
I then repeated with the packs in parallel. (On a different day, I'm not superman :) ).

Results:
Separately: The packs yielded 9.6ah and 9.2ah.
In Parallel: Packs yielded 17.2ah
If everything was perfect, I should have got at least 18.8ah. So difference (1.6ah) is an 8.5% efficiency loss roughly. Probably worse because of all the plus points you mention.

I believe that the loss is caused by the cabling - the packs are connected together over the two supply leads that go from each front pannier to the back pannier where the controller connection is. That is around 3 meters of cable battery to battery. Its very high rated (AWG 10 equivalent) but balancing flow over that length of cable must loose energy - at least that's my theory.

So, before you say "you can connect the batteries at the front" - yes I could, but it doesn't fit with the overall use I put the bike to. So I live with what I have :)

amberwolf said:
If the switch isn't designed for at least as much voltage as the full charge voltage of the battery, then it can't guarantee disconnection (the physical gap between contacts may not be far enough apart). It is unlikely to ever be a problem...but that is one reason they have a voltage rating.

It's a rotary switch from RS Components. Rated operational voltage: 400V. Rated thermal current: 25A, Rated short time withstand
current (3s): 100A
I said it was chunky :)

But thanks for the pointers, I'll keep an eye on the switch when using it.
 
pickworthi said:
Results:
Separately: The packs yielded 9.6ah and 9.2ah.
In Parallel: Packs yielded 17.2ah
If everything was perfect, I should have got at least 18.8ah. So difference (1.6ah) is an 8.5% efficiency loss roughly.
Yes, you should get the same or better. Noticeably better really only happens when there is enough load on a single battery to cause greater voltage sag and thus greater loss to internal heating of the battery, vs paralleled batteries. If there is not enough load on a single battery to create significant voltage sag then there is not significant internal loss of power to heating, etc., and paralleling them wouldn't change that much.

But at the least, you should get the *same* results.

The only things that I can think of to cause different results might be different total conditions causing greater loading, like higher headwinds or more starts from a stop, but those might have to be significant and very noticeable to you to make that kind of a total capacity difference, as they would have to cause significantly higher power usage causing significantly higher voltage sag over enough distance / time to cause that much loss. If both trips are at the same speeds and no pedalling (motor power only, since people don't pedal exactly the same over a trip and thus pedalling can greatly affect actual power usage in a test like this, depending on various conditions), and the same stops, starts, terrain, winds, and throttle usage, then the conditions should be close enough to get the same results from either test at any time the test is done. If conditions and usage aren't the same, then results might vary significantly even between separate trips of the same configuration.

What are the other stats from the wattmeter for each trip? (max and min of Watts, Volts, Amps, total Wh, etc--any info it provides).


I believe that the loss is caused by the cabling - the packs are connected together over the two supply leads that go from each front pannier to the back pannier where the controller connection is. That is around 3 meters of cable battery to battery. Its very high rated (AWG 10 equivalent) but balancing flow over that length of cable must loose energy - at least that's my theory.
Actually, it is *better* to run separate cables for each battery as far as possible, because then there is even less loss in parallel, as there is half the resistance in the wiring for the total.

If they were paralleled closer to the batteries, you would then have a smaller "pipe" for it to flow thru, by half, for more of the length of the path, and there would be higher losses, when used in parallel.

However, if it's 10g, then at the currents you are using them at, there is negligible loss. If you like, you can measure the voltage drop across the length of the cable at the max current you use them at, to see exactly how much it is.

But either way, since almost all the path is separate wiring, then that is identical losses to when using them separately, as well as parallel.

The only thing that is different between parallel and serial, as for wiring, is the parallel connector setup, correct? If that is a lower gauge, or has a connector-to-wire or connector-contact-to-contact issue causing higher resistance, then you would get voltage drop across that which you don't get otherwise. You could see that drop in the wattmeter display (if it is visible on the handlebars), during the highest current draws (hills or startups from a stop). If there's no significant drop across that cable (from the back of the battery-side connector to the back of the controller-side connector, to include contact-voltage-drops), then there's no significant losses in it either.

Another test that can be done is to leave the parallel cable in place for the non-parallel test, but connected to only one pack, and the controller (or rather, wattmeter, then controller). If this is done on each pack, then if it is a problem with the cable, you should see a difference with one or both packs this way, just like in parallel. If the problem is at either of the battery-end connectors, or the separate battery-end wires, the difference should be proportional to (the capacity difference between parallel and non) or the same as when in parallel. If the problem is in the common connection point or wiring, then the difference will be the same regardless of which non-paralleled pack is under test.


As far as "balancing flow" goes, I'm not sure what you mean. If both batteries are essentially equally capable, and the wiring is the same, then all the flow goes from both batteries to the controller. There should be no flow from one battery to the other, unless one is lower voltage (which won't happen if they are about equally capable, as both will drop in voltage by the same amount for the same load).


A possible issue: I don't see anywhere that you've said this is the case, but if the balancing leads are connected between the packs, then perhaps that is creating a problem, if some of the groups of each pack are imbalanced in capability (not static voltage) different from the same groups in the other pack. That could cause current flow between them under load, thru the balance wires, which are typically very small , and could waste some power as heat in those wires and the interconnects.


I'm too tired to think of anything else right now, but i"ll post back if I do. :lowbatt:
 
Thanks for the comprehensive reply.

The parallel connector is short - 2 x 12AWG to 1 x 10AWG.
Balance leads are not connected (they are on either sides of the front wheel with cell low voltage alarms attached).
Ride conditions were more or less identical.
The wattmeter is in the back saddle bag, so only end of ride numbers available. On the parallel ride a cell low voltage alarm went off at a higher pack minimum voltage than with the separate pack ride. Unfortunately I didn't note the value, I'll have to do it over again to get the values.
(Just for reference, I've set the low voltage alarms to 3.15V per cell. That gives me an overall pack end voltage at a bit above 32V. In another thread we discovered that these calls have good capacity down to 3.2V: https://endless-sphere.com/forums/viewtopic.php?f=14&t=110236 )

My reference to balance flow is my understanding that if one pack is stronger than the other (less internal resistance or better balanced internal resistance) the strong pack will transfer energy to the weak pack. I just can't see how packs wail auto-magically "know" how much each has to provide to stay in balance.

The end voltages may give a clue though. The packs are not identical. So, the weaker pack has one group of cells that gets to my low voltage alarm point faster than the other. This will take a lot more experimentation with different packs as well (the two 21AH packs I have are identical cells and were built by me one after the other, so a better match). Something to do during the summer!

Also, since this is all pedal assist, no throttle, with each separate pack I may be providing more power from my legs for longer (during each batteries lower voltage phase). My legs put in around 160-180 Watts under normal riding, more if pressing up a hill. Seems like that should help the parallel setup more though.

With all that said, when one is 50ish miles into a long ride, the psychological effect of switching in a fully charged pack really can't be overstated. So, I think I'll be keeping my rotary switch idea for those occasions. I can always parallel packs before connecting to the switch - so best of both worlds.
 
at a given voltage, drawing at a lower C-rate results in higher Ah capacity

regardless of resistance, heating losses etc

A 0.1C rate will show higher capacity than a 0.3C rate.

So, swapping out three 30Ah batteries sequentially

gives less range than one 90Ah battery

See Peukert's Law

Unless maybe if the weight load becomes a big factor and the "spares" are being carried by another support vehicle.

Of course there are a myriad of other factors involved, substandard wiring could be one if a given implementation does not give comparable results.
 
For me, Freud wins out over Peukert. Sorry. That mid ride hit from a fully charged battery is just too good. :D

I have now progressed from physically swapping batteries in and out of the battery holder, to swapping cables, to now turning a rotary switch. This final state is very nice. Beats fumbling with cable plugs next to the roadside.

I made a physical incarnation of my diagram by squeezing it into a junction box. I applied some surgical metalwork to my controller, and squeezed that in as well. Connectors are all removable, should I need to take it apart. I spliced a hall ACS712-20A sensor (for current measurement) and two voltage probe wires off the final connector, in readiness for the data logging circuit I'm currently working on.

Pictures (to save the 1000 words). Tested out on this ride (https://www.strava.com/activities/5143003981). Changed batteries at 29.1 and 53 miles. Works great. Switch has a nice positive and decisive throw. Should do, it wasn't cheap :D Controller does not appear to be over heating in the box - but my upcoming data logger will provide data on that.

PXL_20210416_132723019.jpg

PXL_20210416_132641393.jpg
 
Time passes, and my experience with my iCharger grows, so I have some data on single pack vs parallel pack discharge now. Since the discussion was in this thread, I thought it best to post the data here. Apologies for waking up a dormant thread if that breaks ES etiquette.

Since my last post I've moved my system on to a Cycle Analyst and Baserunner combination. Now I ride with a 500W power limit, which the CA adheres to across a whole battery discharge (thus increasing Amps as Voltage drops). I found a way to simulate this on my iCharger ( https://www.endless-sphere.com/forums/viewtopic.php?f=14&t=112842 ), so I did a 500W discharge on each pack separately, and then on the two packs connected in parallel. I've attached graphs at the bottom showing the data around the point that each pack reaches the 3.2v per cell low voltage cut-off.

From these graphs the results are:
- the PANDA pack discharges 8767 mAH to 32V
- the MH1 pack discharges 9987 mAH to 32V
- In parallel, they discharge 19678 mAH to 32V

That yields an additional 924 mAH in parallel over each pack singly, all at constant 500W discharge.

Thus, I was wrong in my assertion that the parallel configuration provided less capacity. Apologies.

If @john61ct is still listening, I have a question:
- Just about 1AH is not a very big difference - is this in line with what you would expect?

Thanks all for being patient with a newbie.

Screenshot_2021-08-30_12-53-46.png
Screenshot_2021-08-30_12-53-11.png
Screenshot_2021-08-30_13-09-34.png
 
I'm glad you tested and posted your results. It's common for people to make clams that go against common knowledge. If you read enough threads you will see.

Everyone here wants as many usable watts as possible. The biggest enemy is heat, turning your watts in to heat. If you discharge a battery to fast they get hot. Not only are you wasting watts it stresses the battery and shortens it life. Not to say discharging to lower levels try to get as many watts as possible. Paralleling is a great way to enhance happiness. Just make sure they are close in voltage or just bulk charge them to 80%.

This applies to controllers and motors as well. Get the best watts per mile when everything is below over heating.



by pickworthi » Aug 30 2021 7:38am

Time passes, and my experience with my iCharger grows, so I have some data on single pack vs parallel pack discharge now. Since the discussion was in this thread, I thought it best to post the data here. Apologies for waking up a dormant thread if that breaks ES etiquette.

Since my last post I've moved my system on to a Cycle Analyst and Baserunner combination. Now I ride with a 500W power limit, which the CA adheres to across a whole battery discharge (thus increasing Amps as Voltage drops). I found a way to simulate this on my iCharger ( https://www.endless-sphere.com/forums/v ... 4&t=112842 ), so I did a 500W discharge on each pack separately, and then on the two packs connected in parallel. I've attached graphs at the bottom showing the data around the point that each pack reaches the 3.2v per cell low voltage cut-off.

From these graphs the results are:
- the PANDA pack discharges 8767 mAH to 32V
- the MH1 pack discharges 9987 mAH to 32V
- In parallel, they discharge 19678 mAH to 32V

That yields an additional 924 mAH in parallel over each pack singly, all at constant 500W discharge.

Thus, I was wrong in my assertion that the parallel configuration provided less capacity. Apologies.

If @john61ct is still listening, I have a question:
- Just about 1AH is not a very big difference - is this in line with what you would expect?

Thanks all for being patient with a newbie.
 
The actual Peukert coefficient varies by specific chemistry used.

But all LI are much lower than lead for sure, usually under 1.2
 
john61ct said:
The actual Peukert coefficient varies by specific chemistry used.

But all LI are much lower than lead for sure, usually under 1.2

I thought it was time that I gave this chap Peukert a bit of serious study.

So I went here: https://en.wikipedia.org/wiki/Peukert%27s_law
...and my brain overheated.

Then I went here: http://www.smartgauge.co.uk/peukert2.html
...and my brain discharged below it's safe capacity.

I don't know if I can recover it :)
 
by pickworthi » Aug 31 2021 7:44am

john61ct wrote: ↑Aug 30 2021 10:11am
The actual Peukert coefficient varies by specific chemistry used.

But all LI are much lower than lead for sure, usually under 1.2
I thought it was time that I gave this chap Peukert a bit of serious study.

So I went here: https://en.wikipedia.org/wiki/Peukert%27s_law
...and my brain overheated.

Then I went here: http://www.smartgauge.co.uk/peukert2.html
...and my brain discharged below it's safe capacity.

I don't know if I can recover it :)

So does that answer your Three way switch question?
 
As far as I know, it is much wiser (as someone already mentioned here) to use batteries at the same time. In the event that they are builded from the same cells, they can even be charged at the same time from same charger. Even if they differ in capacity, but not in cells.
 
I have two batteries and swap them out, keep one in my room. Never have made the mounting to carry them both. but they are biggish. 20s 7p, 15lbs apiece. Bigger than others and not as big as some. So I just plug and unplug them. Working to start a another build so I can carry them both and will put them in parallel.
 
I'll just post my thinking here, since I do not believe that there is ever a single "correct" answer when dealing with engineering.

Yes, parallel batteries will provide greater capacity at a given discharge rate, and reduce current extracted from each battery.
But, in my case, with my batteries, and my power usage, the difference is not much. The approx 1AH measured in the graphs above amounts to between 2 and 4 miles on the road for me, depending on how much vertical distance is covered.

Set against this is the risk of failure within each battery. Failure can take many forms, both mechanical and electrical. I have experienced one such failure - a mechanical detachment of one lead.

Thus, being an engineer of the IT persuasion, I see this as a choice between redundancy and capacity. In disk terms - RAID 1 vs RAID 5. (I know, not an exact comparison, but illustrates the point.)

For my circumstances, and my batteries, I have chosen to balance the risk equation in favour of redundancy.

Others are of course free to asses their situations and make different choices.

As I see it, no one is "right" or "wrong" in the choice they make, as long as it is considered.
 
In the case of battery configuration in RAID 1, you, as well as in the case of HDD, will get a double gain (in the case of HDD, of course, it is less) in the discharge mode, however, in the charge mode, in terms of the amount of information / energy loaded, it will be more like RAID 0, and in terms of loading / charging speed it is RAID 1. Nick of what configuration in my opinion it is impossible to apply RAID 5 as an analogy. I would even say that using two or more batteries at the same time is more like JBOD. Naturally, provided that the batteries have an equal charge at the time of connection.
 
Silvaticus said:
In the case of battery configuration in RAID 1, you, as well as in the case of HDD, will get a double gain (in the case of HDD, of course, it is less) in the discharge mode, however, in the charge mode, in terms of the amount of information / energy loaded, it will be more like RAID 0, and in terms of loading / charging speed it is RAID 1. Nick of what configuration in my opinion it is impossible to apply RAID 5 as an analogy. I would even say that using two or more batteries at the same time is more like JBOD. Naturally, provided that the batteries have an equal charge at the time of connection.

It was just meant to be an illustrative analogy.
 
pickworthi said:
It was just meant to be an illustrative analogy.
This is just obvious. But the analogy itself is not so obvious. Each of the usage algorithms has its pros and cons. And only you can choose the right ones for you.
 
Back
Top