 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Solar : cheap adjustable Voltage Sense relay / co
  Page 2 of 2 |
|||||
| Author | Message | ||||
| john61ct Newbie  Joined: 07/02/2019 Location: United StatesPosts: 27 |
Thanks so much for joining the discussion Mike. And wow, I'd really appreciate some 101-level resources so I can understand more, googling MOSFET for example it's all Greek to me. But if you guys keep talking, and I keep googling at some point it will make more sense I'm sure. I've come across systems where smaller than usual wire gauge is used to reduce / limit current, always wary about this idea from a safety POV especially in a mobile living or marine context, but lots of clever people do it, and I guess fuses can act as a failsafe? Ideal would be the maximimum current allowed through the regulator (term here for "per cell black box") to also be a variable setpoint - is that at all practical? Ideally this device - or at least its design - would be adaptable to service larger capacity cells, say 180 or even 300+ Ah. With LFP the SoC affect on amps accepted is not that marked as it is with lead, and the cells in a given pack will usually be at a similar SoC anyway. .5C accepted across the regulator really is just fine for me, even up to 1C. Upstream charging energy will often come from ICE sources. If running solely for charging purposes, limiting runtime via "faster charging" is critical. Another option, if per-cell current regulation isn't practical, would be the supply buss gets current-limit throttled upstream at the pack level by the central CPU, based on the number of currently closed/connected relays. So for the x16 25Ah example, a 200A-available starting point would get scaled back by ~12A for each relay that gets opened/isolated. |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
In this case - per cell LFP charging - voltage will be 3.7V as a "very theoretical" maximum. I practice 3.5 will usually be a bit higher than the relay-open setpoint. |
||||
LadyN Guru  Joined: 26/01/2019 Location: United StatesPosts: 408 |
John, if you could, can you update the v0.1 of the diagram with you posted with a newer version that captures the blackboxes in light of the new information you have? That is a cheap way of doing it. That method is actually using the wire gauge as a resistor, so the smaller wire gauge drops power (but heats up). I actually like this idea as it keeps things simple, but let's make it even simpler: we will use the rated wire gauge but use power resistors to dissipate the power. So, Let us use power resistors for now to manage/reduce the current. This is the V0 prototype that will be very inefficient but will allow us to start putting the system together. We will iteratively improve the system and make it more efficient. Like replace the power resistors with MOSFETs as you and I learn more about MOSFETs - OK? practical from an engineering perspective. actually necessary from an engineering/proper design perspective. practical from a monetary perspective? that's your call. Let's table that current for now. For now, what's the minimum you will be happy with? yes, so to ensure faster charging we have to ensure the power from the main bus gets routed properly and the system losses are minimal. To achieve that, we will have to limit the current per cell (even if the cells themselves will not be hurt by high current), otherwise the most power hungry cells will drain all the power while the other cells are starved Again, this is a design consideration. If you don't care about this scenario nor care that cells might get charged at different rates, we will take that into consideration and design accordingly. In general, we want to minimize the current flowing through the system as that directly affects the conductance losses and hence the efficiency of the system. Mike (and others), please weigh in with your considerable experience on our thoughts. Help is much appreciated for us newbies. |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
I don't know of any changes that are visual, could you clarify? the wiring paths are the same right? I could write up a text re-statement as concisely as I can, if you think that's needed? > use the rated wire gauge but use power resistors to dissipate the power Could these be swapped out easily? Maybe mounted on a bit of inline wire with Anderson connectors at each end? > We will iteratively improve the system and make it more efficient. OK > practical from a monetary perspective? Well just like energy efficiency, let's be iterative, what is doable / ideal while prototyping, may prove too expensive to deploy in production; so then, we kludge workarounds. > Let's table that current for now. Sure, long as the need to scale is kept in mind. One workaround for large Ah cells, is to bulk charge "rough" still wired as a xPyS pack, up to say 95% SoC, depending how well they stay balanced that way, and then do the per-cell "fine finish" per-cell charging for just the last 10-20Ah, to restore close as possible to a well-balanced state. > For now, what's the minimum you will be happy with? 8-12A? At least that, with 20-30Ah cells, is in the ballpark as an in-production minimum. I mean, we **could** start with baby cylindricals like A123 cells, what 2-3 Ah? Which would put rates at say 800-1500A per cell. I'm just afraid then components will be very different, and maybe test results won't extrapolate well when moving up to the bigger sizes, maybe cylindrical LFPs have different behaviour responses than pouch/prismatics. > we will have to limit the current per cell (even if the cells themselves will not be hurt by high current), otherwise the most power hungry cells will drain all the power while the other cells are starved My comment: >> With LFP the SoC affect on amps accepted is not that marked as it is with lead, and the cells in a given pack will usually be at a similar SoC anyway. was to explain why I do not believe your & Mike's scenario will be an issue. Remember, each cell has an independent wiring pair to the two busses +/-, no current is passing "through" one batt to get to the next. If we find such imbalanced flows do become a problem, let's deal with it then. So IOW, limiting current is IMO only needed to the extent we see per-cell rates going much over 1C - to protect cell longevity - while the average per-pack target is .3 -.6C The infrastructure - wire, terminators, CP etc - should be sized to handle say 2C, so cell longevity is the only concern wrt high amp rates, not over-temperature / fire safety. |
||||
yahoo2 Guru Joined: 05/04/2011 Location: AustraliaPosts: 1166 |
Hi john, if you just want cell balancing for a small 12 volt system (with no individual cell monitors that will trigger a shutdown) I would use four of these. https://evparts.com.au/ev-power-bms/lfp-voltage-balance-board.html#  they are about $6 AU each and they start bypassing at 3.5 volts. they are only 67 mm long, if they dont fit solder a couple of short wires with ring terminal connectors on the ends and tidy it up with some heatshrink and cable ties. I like the wire extension because I can easily check that they are functioning with a DC clamp meter. You can use your esp32 or whatever to open a relay/contactor/latching relay on high or low pack voltage. I must be getting old and grumpy, I just think if it is your first LiFePO4 battery bank and it is only 4 or 8 in series, go with something that works and start fiddling with version 2 once you have your head around everything. the only problem I have seen with these units is blowing the fuse on the first charge because the cells are out of whack, you have got to sneak up on full charge the first time. I'm confused, no wait... maybe I'm not... |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
Well thanks for that yahoo2, I had seen those before, but will refresh myself on their details perhaps for other future projects. But no, I am more interested in exploring and testing this balance-charging design concept, rather than out of any short-term pragmatic need for another working bank. I am only buying the cells in order to pursue that idea. Again, the focus is on the charging, yes achieving as precise a top balanced result as possible, but as a side effect of the normal charge cycle, that balancing is not the main goal, to be achieved using other more conventional methods. Plus, at lower currents (as with solar), I do not intend to get over 3.5Vpc in normal operations anyway. If/when the buss is allowed to go that high, I certainly do not want it to hold it there for the length of time required to redistribute charge between high and low cells. Remember, there is no Absorb / CV stage to this charging profile, only CV, charge to a setpoint and disconnect the cell, and Resting will always be well below 3.5Vpc. I also do not want to leave any per-cell gadgets in place during normal usage discharge cycling. Again, these cells are not staying assembled in any givem pack / bank xPyS arrangement, "just a bunch of cells" each being independently charged, through the single pair of wires and the relay in between (dis)connecting each cell separately to/from the buss, no other gadgets or connections while charging, otherwise each cell isolated, standalone. And no, the eventual production systems will not be 12V, nor small neither in cell count nor Ah capacity - but all that is not germane to the topic at hand. Finally, I want this scheme to accommodate cells at wildly divergent SoC at the start of charging, so that current limitation won't work, even if that balancer were in alignment with the other criteria. I do like the pricing though! I've seen balancers go for USD $25+ each, without any trigger voltage required, but limited ampacity as well, to only 6A. Edit - these units only allow at most 400mA balancing current (???!!!), RC hobby chargers for tiny cell packs routinely exceed that, and even then can take forever! The goal here is to be able to charge at .5C or even higher, and not add **any** extra waiting time for balancing, just be able to charge and go as fast as possible. Yes, that is really all this thread is about, how to accomplish just that, but at a per-cell level with whatever N count cell collection. Maybe starting with just 16 cells, but as needed perhaps working up to hundreds charging concurrently. Apparently besides a mechanical relay, those MOSFETs (transistors?) are available in the 10~20A range, with a "gate bleed resistor" (???) would also do the job? and perhaps an opto-coupler (??) if good isolation turns out to be required (maybe if the cells were to be left wired in series? or multiple busses fed from separate PSUs? dunno). |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
Ok, now that you have had a look at some offshelf solutions and found them to be lacking, it's becoming more apparent that we need to build something ourselves. Now is a good moment to look at DC rated relays. Please, dedicate some time to look around for them and if you could, please, write about your findings and their price on a post here. I have some homework to turn in next week so I have had no time to work on drawing an updated schematics but for now my thoughts are this: 1. You need a smart switch that can monitor/limit its own current. This very well may be a DC rated relay + uC or something else entirely. 2. This smart switch needs to be a scalable design. The V0 version of our switch should be able to handle say upto 15A but it should use a design that would allow it to scale up to 200A and we will keep this in mind throughout the design process John, what are your thoughts on this and what parts and tools do you currently have available to start constructing this switch? I am thinking of either a simple N-Channel based board or a DC rated relay to begin with after you have done some research on DC rated relays and we have some more information to work off of. Is your bus voltage going to be 3V? You will have to start learning about heatsink design, have good tools to measure temperature from 20 degree C upto 100 degree C quickly and log them to a computer for analysis. Yes, MOSFETs are a kind of transistor. Yes, a means to bleed the gate would be necessary. I don't understand yet whether a resistor would work for our application but I don't see why not. We would possibly end up using a transformer and/or opto-coupler for the isolated psu to drive our MOSFETs |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
Will do, but apparently the cheap-Chinese stuff is not properly rated by the makers. So a contact ampacity of 20A may only support 10A continuous or even less for reliability? Do y'all think that is being too conservative? > You need a smart switch that can monitor/limit its own current. This very well may be a DC rated relay + uC or something else entirely. (TIL "μC" = microcontrollers like Arduino SoC similar? rPi more PC-like but not real-time) By "smart switch" does that describe the whole "per cell regulator" we are designing, or just the relay component functionality? > This smart switch needs to be a scalable design. The V0 version of our switch should be able to handle say upto 15A but it should use a design that would allow it to scale up to 200A and we will keep this in mind throughout the design process It seems maybe only the voltage sensor (and if there is a per-cell microcontroller) will **not** be sensitive to the current rate. > what parts and tools do you currently have available to start constructing this switch? Assume nothing. I think I ordered a clamp ammeter I'd need to dig up, and I have various power supplies but at higher voltages. What is an "N-Channel based board"? as in mounted with multiple relays, or a microcontroller? I have bought some intro-level electronic texts from O'Reilly and the "Make" series, so hopefully will start to understand some of these terms. Recommendations for other resources and basic tools would be appreciated. > Is your bus voltage going to be 3V? As stated, the charge buss will be ~3.5V If you mean for relay signaling, and powering the control devices, I'm open. My perception is that 5V is a good default, stepping down to 3.3V as needed should not be hard? I've seen LiPo chargers that can be embedded to make our devices portable, that will be important since grid power will rarely be available in production use cases. > You will have to start learning about heatsink design, have good tools to measure temperature from 20 degree C up to 100 degree C quickly and log them to a computer for analysis. Noted, I'm sure will be lots of such sub-topic learning projects. Thanks |
||||
| LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408 |
I am really unsure that we will end up purchasing a DC relay and am biased that we will end up making a N-Channel MOSFET design BUT we MUST first give DC relays a chance. So assume the cheap-Chinese stuff do what they advertise to do. We note down the specs and the price. That's all we do for now. To help understand what our options are. We are beginning with the relay component. So the "smart switch" I spoke of above is really that - a relay but it can also control current through it. Correct. We may end up with a per-cell microcontroller but I am hoping to multiplex multiple cells per microcontroller. IF that ends up being too complicated, we will throw money at the problem and simplify with per-cell microcontroller. Hmm, in that case a $100 used analog oscilloscope will be the best bet. You can use it as an ammeter, voltmeter, signal tracer, the works. Will be very helpful. Infact I am getting one for our project. My brother is going to purchase a Tek 2115A analog oscilloscope for me for $60 tomorrow and the seller has 2 to sell. He had 5 earlier this week but they go fast. If you want one please let me know and I will get another for you. Let me know quickly though as he sells fast and he only has one on hold for me until tomorrow. These won't work 100% and will need something to be fixed or another. I think guaranteed working ones can be got for $300 or so? You can have a look and let me know if you find anything. yes please. I will depend heavily on you to design the heatsink properly and would like to offload all that learning to you. I would like to contribute with the N-Channel MOSFET design and share bits of the uC dev with you. If you are not familiar with Arduino, you can get started with that too using the ESP32 board. I have a very good feeling about this project! I am excited! |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
Yes please, trying to figure out how to get you a PM so you can text me for any other time critical stuff. |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
Got PM thanks. request thread for relay recommendations Will use that to post possibilities as I gather them. Please also use that thread to clarify / correct any misunderstandings on my part, or feedback on other relay-specific issues. Other sub-topics and "big picture" discussions here. Thanks |
||||
| john61ct Newbie Joined: 07/02/2019 Location: United StatesPosts: 27 |
Well "MUST" may be overstating it, I just know even less about MOSFETs than relays. In my googling people talk about them being cheaper than mechanical relays (are SSRs within that category?) Older tech does not mean simpler or more reliable I guess, those are the other key priorities to me. In any case it seems we should design the physical side, so that the sensor(s) and (maybe per-cell) μC can remain as is, but the current-specific relay and (whatever limits current) are in a swappable module, maybe the two modules snap together to form our per-cell controller. Note I assume the current controlling component does not need (for now) to actually measure (sense, report) the current level? But a "future dev" thought is that other sensors, not just amps but e.g. temperature and internal cell resistance, would be worth exploring. Presumably they won't need to be swapped out with different cell sizes. > throw money at the problem and simplify with per-cell microcontroller. I've come across smaller/older Arduino boards under $5 each. I am not at all familiar with them, nor am I a coder, but pretty OK with ICT stuff in general from a sysadmin / networking POV. I'd like a design goal for N cell count to scale pretty high, ideally 100+. So, even with a multiplexing microcontroller, it will have some limit of N cells (12? 16?), right? So I think a whole-system controller will likely be needed to coordinate changing the setpoints from one session to the next, if more than a few μC are involved. > you can get started with that too using the ESP32 board. Is it the "analog outputs" we need for controlling the relays? Is that same as "GPIO"? A board with lots of those better than fewer? > I have a very good feeling about this project! I am excited! Great, me too! |
||||
| Page 2 of 2 |
|||||
 Print this page |
