Fantastic post mortem results Veggie ! Glad to see your 12V starter motor/generator vindicated, I have always thought it was a great setup you had. 

Next time you'll know to do an anti-corrosion plus wear and lubrication service on the motor/generator when you replace the battery.

Current limiting with a fixed voltage charge is well supported by most commercial PV charge regulators and is not usually not a problem even for older charge regulators.

LFP has to be regularly full charged if you have any hope of having an reliable state of charge indication and to do non-manual balancing.  The cell charge-voltage curve is flat with some variations until you reach around 90% of full charge.  Likewise, LFP cells should only be balanced by your BMS when approaching full voltage (voltage is rising appreciably), very unlike the lithium ion cells.  You can decide how full is full  if you like, but you must pick a value in the nearly full range where the voltage is rising substantially.  Most have so far found that even keeping the cells at full charge is not causing significant degradation over time.  See Will Proust's diysolarform.com to get some decent info. He does his technical homework well before presenting info.

I am not using LFP yet but just 5 months ago was working on an all linear, high voltage BMS design for 38 LFP cells in series with another EE friend.  In my research I found that the differences between LFP and Lithium Ion cells was causing trouble for many early LFP adopters with BMS hardware that was not well suited for them.  Balancing LFP by any means (active or passive, high or low current) at other than near full was always a mess.  The voltage-charge curve is just too flat, and instead of balancing, it was actually unbalancing.

Apparently large balancing current is not needed despite the more limited balancing time window, and 100ma to 1 amp seems to be working fine. The cells are just that good at least when young (under a decade old). 


Best Wishes,
BruceM

Hi Veggie,

That poor Listeroid in the video is so badly out of balance it's pounding itself to pieces.
A serious cruelty, it should be a crime.

An induction motor is a bad match for a starter, low starting torque and huge locked rotor current.
If it's free  and you really want to see it for yourself, just disconnect power manually when it begins to start to stop it from becoming a synchronous generator. You can do that by rpm sensor (I like the cherry hall effect gear tooth sensor for counting my spokes) and an Arduino Nano to do the disconnect automatically.  Or a thousand other ways...

Best Wishes,
Bruce

 Powder Monkey has an ideal, keep it simple application for thermosiphon tank cooling in that he is running 8 hrs a day in a no-freeze location.  No antifreeze needed and the long time for getting up to temperature and potential increased carbon build up are not a problem for his application.

My application was for short runs, typically 30 minutes. (Pump water while running washing machine, or pump air for the shop, or a 30 minute bulk charge of my battery bank.)   I also have the issue of sub zero temperatures in winter which also make the tank method a poor choice ($) as ajaffa1 wisely notes.  I opted for the thermostat and a (exhaust induced draft- no fan required) radiator with thermosiphon flow because of my  anticipated short run times and a design goal to not require generator operation when operating the air compressor compressor only.  (I could not find affordable cast iron radiators here in AZ or I would have used one with a thermostat.)  I have further reduced de-coking labor by adding a water/methanol feed system.  I keep waiting for the cold idle knocking sound which used to signal time (at 200-300 hrs) for de-coking but now it never happens. 

The perfect cooling system for the CS is application and location specific.  (And the desires and design goals of the builder/designer.)  XYZer created a lovely portable CS design for thermosiphon and fanless oversized radiator with convection only air flow. Standard automotive style cooling setups with pump and fan can be more compact and lighter weight while also being quite practical and economical.  Cujet's portable twin CS design is one outstanding example.

Best Wishes,
Bruce M

I agree, Cujet.  One safety weakness of my engine shed design is the shaft on the IP side which drives the ST-3 is unshielded.  Less of a concern since I mostly do remote starts, but still, the risk of clothes being grabbed by the shaft at a moment of thoughtlessness or instability is something I'd like to correct.  I had a PVC drain pipe with cap cover over it at one time but it was friction fit relying on some electrical tape and didn't work out long term.

I wonder if you or some others have already come up with a nice solution.  Your high quality workmanship is always inspirational!

Per Ade's request, here are some photos of my DIY single axis tracked arrays.

I have revised all my racks to (3) 2 inch, 0.10 thickness square tube rails for greater stiffness, having retired my original 800W rack and badly snail trailed 175W solar panels.  It was designed for seasonal tilt only with thinwall 1.5 inch rails and wasn't long enough for the 325 to 355 W surplus or used panels I can get locally for very cheap.

This is all part of my custom inverter upgrade, now with sufficient power to run my 1500W home cooler plus have enough power left over for all my other activities. The cooler works by chilling water for my in floor heat system, and that project was previously reported on here.

Under the solar panels you will see two aluminum boxes.  One of these is the Arduino Nano plus Accelerometer board for panel angle sensing, plus real time clock module and H-bridge board for driving the linear actuator.  The second box is unrelated to the tracker, and shorts out one of the 5 panels when there is excess voltage for the current load.  This pre-voltage regulator allows my simple custom linear PV current regulator to handle the new larger capacity input. The linear PV regulator uses 7 IGBT transistors, each with an op amp current sensing driver to evenly share the total current load.  My battery bank is nominal 125VDC, 10 100ah "marine" (lead calcium) batteries in series. 

The entire tracker system is powered by one very small 12V panel and a small AGM battery which is located 1.5 foot below grade in an 8 inch section of round galvanized duct with cap about 12 inches above grade.  The trackers are timed so that they all do their hourly movement staggered in time a couple minutes apart, keeping the max current demand to about 2 amps.  The real time clocks and tracker program of the Nano with specific data table for the individual array orientation are set via laptop and USB cable.  The manual motor drive cable can be used any time by just unplugging the 4 wire auto trailer plug at the Arduino tracker control box, and plugging in the manual cable.  Once the
Arduino is plugged in it checks current position versus table position for month and time of day, and motors the panel to the proper position.  The real time clock has a lithium cell to keep date and time when 12V power is removed.  As you can see in the photo of the tracker box internals, there is zero custom hardware, it is just wiring up of off the shelf components. There is no heat sink required for the H-bridge controller at the puny power levels (2A of 12V) required.  The Nano board provides the 12V to 5V linear regulation.  While the Nano is sleeping, power draw of the whole thing is < 5ma, if the Nano power indicator LEDs are clipped. 

The PV rack mechanicals are pretty simple and derived from older my seasonal tilt rack design.  I'm using 1/2 brass rods with steel split pins as the unlubed pivot points, with stainless steel washers to reduce pivot drag. Many here are very skilled in practical mechanical design  and will come up with there own solutions.  I am pleased with the results here.  It is very solid and does not bounce in the wind, and easily supports my weight and bouncing on the rack with very little movement.  I did all welding myself with my now 3 12V marine battery DC stick welder (with galvanized sheet strip resistor) and 3/32 6013 rods. It was a challenge but I did get the hang of it. MIG would be much better suited to the 0.10 inch wall square tube. 

At present to develop the data table needed for month, hour and position, I used a solar calculator service online to get the plots of sun azimuth and elevation for my location.  I built a scale model with digital inclinometer, and would set up the sun angle and azimuth pointing to the properly oriented scale solar rack, and then eyeball the best tilt angle of the array, reading the angle from the  inclinometer.  A very tedious process.  I then convert via table the angles to accelerometer units. 
A better solution would be to have a program which does this all for you given rack orientation and range of tilt, GPS coordinates and spits out the actual text data table to be inserted in the Arduino Nano program.  I have not done that as my programming skills are now quite diminished.

Having operated this for about a year, (the summer rack for a bit less) I am quite pleased with the cost, performance and reliability.  One design idea I have is to tilt up a version of the summer rack to say 45 degree inclination (I have a south facing hill handy) to do a "winter rack".  This would add to winter day performance but if the drive was changed to +- 90 degrees via gear motor and chain drive, it could also futher increase sunrise/sunset power all year.

I keep watching battery technology and may someday change to LFP or one of it's successors as prices continue to fall and performance continues to improve. 

Best Wishes,
Bruce M


 

Thanks. Cujet.  I just did an oil change and service and moved the compressor pulley inward 1/4 inch against of the off side gib key. With the new split ring collar now I'm feeling more secure. 

 The Picaxe monitor also shuts down if there is no air compressor or generator load for 10 minutes.  That has been very valuable for my memory problems.

There's now plenty of cheap, useful hardware not for the Arduino that lets you just hook up modules instead of doing your own PCBs.  I did a solar tracking project a year ago using them. I use the Arduino Nano with ADXL345 3 axis accelerometer and DS3231 real time clock module for a recent single axis PV solar "tracker".  Plus a BTS7960 H bridge for controlling a 12V linear actuator. The accelerometer lets you measure the angle of the solar panel, so I have the real time clock wake up the Nano (with LED's removed to save power) every hour, reposition the panel to the designated angle for month and hour, then go back into sleep to save power.  I have 3 single axis tracked PV arrays which have been doing the daily dance for a year now.  Nice to not need any active sun tracking, just a sealed box screwed underneath the rack.  If anyone here needs such a thing I'll make a post about the details. Very cheap and mechanically simple, unlike 2 axis...  I was not a fan of trackers before, but this about $130 per 18 foot rack using this actuator:
https://www.ebay.com/itm/275015055633?var=575299527943

Having much more PV power early and late in the day is a nice luxury. It was a fun low budget Arduino Nano project.

Best Wishes,
Bruce M

The internal wiring appears first rate.  Pity you have to mess with it.  It will be a chore to document well enough to remove and reinstall.  I hope the PVC wire insulation is still in good shape so you won't have to rebuild the wiring harness as well.