Re: (ET) DIY Smart Battery Charger (David Roden)

["David Roden" <etpost drmm net>]

On 21 Jun 2015 at 14:16, David Erskine wrote:

> The Picaxe especially is inexpensive [the 08M2 costs some $4] and has
> many features [ie. built-in analog-to-digital conversion]; as well as
> an extensive forum. If interested smoke out their site. 

For quite a while, I've been intrigued by the Raspberry Pi.  Well, your 
Picaxe is about 1/10 the cost!  But maybe the two aren't comparable; I 
really don't know enough about it.  Yet.  ;-)

I know about as much Basic as I do Ruby or Perl -- enough to make minor 
modifications to existing programs, but not enough to really write 
something 
from scratch.  Mostly I speak C (not C++) and Bourne shell.  So that might 
nudge me toward the Pi, regardless of cost.

> My ET uses a 20-inch length of #6 gauge copper wire to sense operating
> current. But "Sparkfun" offers a 100+ [have to look it up] amp current
> sensor for some $20. Is that worth it? 

For an IU or IUI charging algorithm, I think you need more precision than 
the ET "shunt." But there's another alternative that might not -- see 
below.

I think the big challenge in a project like this is *outside* the micro. 
It's designing the electronics that actually handle the higher power 
charging current.  The GE charger is good for, what, maybe 20-25 amps?  
(The 
rule of thumb says that initial charging current for a lead battery should 
be at least C20/10.)  So at 45v we're working with something over a 
kilowatt.

The whole package also needs to be able to withstand mains surges of 
perhaps 
10kv (more?  guessing here), and mustn't radiate too much noise back into 
the power line.  Commercial charger makers deal with these, or they don't 
get UL and/or FCC certification.  Homebrewers need to look after them for 
safety and good relations with the neighbors. ;-)

As for charging algorithm, there are lots to choose from.  This is partly 
because there are different routes to the destination (a full charge) and 
partly because there's a bit of art to the science of battery charging.

I searched for a long time, trying to find an online source that described 
and compared the major charging algorithms.  I never found one, so finally 
I 
decided to write my own.  (It took a while.  Sorry.)

http://www.evdl.org/pages/chargeprofiles.html

I'm kind of partial to DV/DT charging, but since the ET charger lives on 
board the tractor, it might be a good candidate for amp-hour tracking.  

Amp-hour tracking counts the amp-hours you use while discharging.  When 
you 
charge, the counter counts backwards until it reaches zero.  Then usually 
you have it add 5% to 10% more amp-hours to account for inefficiency in 
charging.  

It's simple in principle, as long as you have the smarts on board the EV, 
but there are a couple of complications.  

One is that you might have to figure in battery self-discharge.  However, 
with flooded golf car batteries you can probably guesstimate this and just 
chuck in a little more overcharge.

Another is that your amp-hour monitor has to be consistent.  You don't 
have 
to count the exact number of amp hours, but counting up and counting down 
have to be equally (in)accurate.  This sounds easy, but remember that 
discharging amp-hours usually zip by faster than charging amp-hours.

Back in the late 1990s, this tripped up Pivco (later Think Mobility) in 
their prototype Citibees.  Their batteries tended to get undercharged over 
time, with declining range.  I think this happened because their amp-hour 
counters weren't fast enough, or didn't sample frequently enough, to catch 
peak driving currents.  For sure, they undercounted amp-hours used, so 
they 
failed to replace all of the amp-hours actually used when they charged.  

Your Picaxe processor would need to be fast enough to properly catch high-
current amp-hours as they flew by.


David Roden - Akron, Ohio, USA

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = 
Note: mail sent to the "etpost" address will not reach me.  To send 
me a private message, please use the address shown at the bottom
of this page : http://www.evdl.org/help/
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =