Searching \ for '[EE]: Olin's charger design' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=olins+charger+design
Search entire site for: 'Olin's charger design'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: Olin's charger design'
2002\04\18@131258 by Bob Ammerman

picon face
part 1 888 bytes content-type:text/plain; (decoded 7bit)

I've been fascinated by Olin's charger design.

I'm not really a hardware guy, but I wonder if somebody could comment on the
variation of Olin's design shown in the attached schematic (sorry for the
bandwidth!).

This version replaces 2 op amps (the voltage follower for the PWM signal and
the differential amp for the current sense) with a single op-amp. It does so
by summing the PWM output and the lower end of the sense resistor to the
positive input; and ground and the higher end of the sense resistor to the
positive input.

This should mean that the op-amp will try its hardest to ensure that:

Vpwm + Vsenselo = Vgnd + Vsensehi

or:

Vpwm = Vsensehi - Vsenselo

Thus the opamp should drive enough current thru the transistors to balance
the PWM voltage.

Did I analyze this right?

Does it have a chance of working?

Bob Ammerman
RAm Systems



part 2 8136 bytes content-type:image/gif; (decode)


part 3 144 bytes
--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@150201 by Tom Messenger

flavicon
face
At 11:58 AM 4/18/02 -0400, you wrote:
>I've been fascinated by Olin's charger design.
>
>I'm not really a hardware guy, but I wonder if somebody could comment on the
>variation of Olin's design shown in the attached schematic
>Did I analyze this right?
>Does it have a chance of working?
>
>Bob Ammerman

I quickly went over this as follows. Assume the top of D1 is at zero volts.
Assume the pwm voltage is one volt. Work it out, you get one volt across
the sense resistor.

Now assume the battery has charged up to where the top of D1 is at 3 volts.
Again, it works out.

I don't see yet where it will not work out. There is one small caveat
however. Since the pwm signal is around 10k impedance, there will be around
a 1% error due to the loading by the 1meg resistors.

If actually building this, I would check to make sure that it is stable,
frequency wise. Probably is, but worth a check.

Finally, this is extremely close to a Howland current pump configuration.
The exception is that your circuit appears (to me at least) to be easier to
analyze. I have been redrawing it to see if it looks more familiar; it
doesn't.  Very interesting.

Best regards
Tom M

ps If anyone sees an error here and can show that it doesn't really work,
by all means, please tell us.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@152523 by Olin Lathrop

face picon face
> I'm not really a hardware guy, but I wonder if somebody could comment on
the
> variation of Olin's design shown in the attached schematic (sorry for the
> bandwidth!).
>
> This version replaces 2 op amps (the voltage follower for the PWM signal
and
> the differential amp for the current sense) with a single op-amp. It does
so
{Quote hidden}

Yes, I believe it will work well enough for the purpose.  I evolved the
circuit the various sections of the circuit as I was designing them and
hadn't stepped back to realized that the diff amp and current driver could
be combined into one.

The only drawback is that the charging current now has a small dependency on
the battery voltage.  My original circuit buffered the PWM voltage thru an
opamp.  It is now connected to the battery voltage via a 2Mohm path.  This
is no big deal in case of a battery charger where 10% (or worse) current
accuracy is just fine.  The effect could even be eliminated in the firmware
since the battery voltage is known.

Nice work, Bob.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@160528 by Bob Ammerman

picon face
By the way, here is a link to the schematic of Olin's design  (which I
should have included in my original email -- sorry!)

http://www.embedinc.com/pic/bat/batt.pdf

The entire charger project is described at:

http://www.embedinc.com/pic/bat/index.htm

Bob Ammerman
RAm Systems

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@164714 by Tom Messenger

flavicon
face
Switchercad likes your circuit.

With a little modification, it can become a general purpose current source.
For my purposes, I removed the external transistor and added a negative
power supply. A DC sweep shows the output current follows the input voltage
in 2 quadrants. Thus, you now have a bipolar current source with seemingly
good compliance. Having needed these before, I have spent a fair amount of
time looking through apnotes and do not remember *ever* seeing this
configuration before. It'll go into my "cool circuits file" as the Ammerman
current source!

Best regards,
Tom M.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@170310 by Tom Messenger

flavicon
face
The Ammerman current source was short lived.

It is, indeed, the Howland current pump. I screwed that one up completely.
My memory had it different so I finally looked it up. For anyone
interested,  Horowitz & Hill, page 182 shows the traditional configuration,
which is what I remembered. Jung (IC op-amp cookbook), also page 182 shows
it modified as Bob implemented it.

None the less, it was a good observation on Bob A.'s part to see a way to
apply it here.

Best regards,
Tom M.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@171831 by Bob Ammerman

picon face
You better give Olin some of the credit. It was basically his circuit, just
slightly tweaked.

Bob Ammerman
RAm Systems


{Original Message removed}

2002\04\18@172631 by Dave Dilatush

picon face
Bob Ammerman wrote...

>Did I analyze this right?

Yup, pretty much.

>Does it have a chance of working?

Everything has a chance of working.  This seems to have a very
good chance of working: as others have pointed out, this circuit
is a variant of the Howland current source.

The only modification I'd make to it is to change R6 back to 20K
ohms as in Olin's design, to preserve his original PWM-to-current
scaling.

Getting REALLY nit-picky, I might recommend reducing the values
of R5, R8, R9 and R14 down to 100K ohms or so; the LM6132 is a
bipolar opamp (as opposed to a CMOS opamp) and has pretty
considerable input bias currents and offset current (50 nA worst
case).  Current accuracy will be a little better with the lower
resistance values (but, I admit, not spectacularly better).

If you do that, remember to make R14 lower than the other three
resistors by 15K ohms (the PWM filter's equivalent output
resistance) to keep everything balanced.

But like I said, that's a small nit.

One thing to watch out for- the LM6132 is a real screamer of an
opamp.  Its gain-bandwidth product is at least 10 MHz so careful
layout is a must: keep all connections to its inputs as short as
possible, or it can oscillate like a politician.

Dave

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@172839 by Bob Ammerman

picon face
Oh, and here is another possible optimization.

By changing the sense resistor from 10 to 1 ohm it no longer needs to be a
2W resistor.

You just have to use 100k resistors from the ends of Rense to the op-amp
inputs instead of 1M resistors (I think) to maintain the same scaling.

Of course, now the extra power is going to be dissipated in the TIP41 and
its heatsink instead of  Rsense.

Bob Ammerman
RAm Systsems

{Original Message removed}

2002\04\18@180143 by Bob Ammerman
picon face
Dave Dilatush gave great advice:

<snip>

> The only modification I'd make to it is to change R6 back to 20K
> ohms as in Olin's design, to preserve his original PWM-to-current
> scaling.

That was just a stupid mistake. I'll change it back.

> Getting REALLY nit-picky, I might recommend reducing the values
> of R5, R8, R9 and R14 down to 100K ohms or so; the LM6132 is a
> bipolar opamp (as opposed to a CMOS opamp) and has pretty
> considerable input bias currents and offset current (50 nA worst
> case).  Current accuracy will be a little better with the lower
> resistance values (but, I admit, not spectacularly better).

I really don't think I'd use the LM6132 for real. It is rather pricey and
overspecced for the requirements. As Olin mentioned, he used it only because
he had it on the shelf.

> If you do that, remember to make R14 lower than the other three
> resistors by 15K ohms (the PWM filter's equivalent output
> resistance) to keep everything balanced.

Makes sense.

> But like I said, that's a small nit.

> One thing to watch out for- the LM6132 is a real screamer of an
> opamp.  Its gain-bandwidth product is at least 10 MHz so careful
> layout is a must: keep all connections to its inputs as short as
> possible, or it can oscillate like a politician.

Again, I probably would use a slower amp.

Bob Ammerman
RAm Systems

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@180551 by Olin Lathrop

face picon face
> One thing to watch out for- the LM6132 is a real screamer of an
> opamp.  Its gain-bandwidth product is at least 10 MHz so careful
> layout is a must: keep all connections to its inputs as short as
> possible, or it can oscillate like a politician.

The 22pF cap from the output to the negative input pretty much takes care of
that.  As I mentioned in another post, I used the LM6132 because it was the
amp I had around that fit the other criteria.  It is overkill for a volume
production design.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\04\18@181435 by Olin Lathrop

face picon face
> Oh, and here is another possible optimization.
>
> By changing the sense resistor from 10 to 1 ohm it no longer needs to be a
> 2W resistor.
>
> You just have to use 100k resistors from the ends of Rense to the op-amp
> inputs instead of 1M resistors (I think) to maintain the same scaling.
>
> Of course, now the extra power is going to be dissipated in the TIP41 and
> its heatsink instead of  Rsense.

But that's exactly the problem.  This one I did consider carefully.  I chose
R3 to give the maximum voltage drop I could tollerate to minimize
dissipation in the pass transistor.

A better optimization along these lines might be to reduce R3 and then also
reduce the 10V regulated supply.  That would cut down total dissipation
because the 10V supply is being fed by a reasonably efficient switching
power supply.  I don't even have a heat sink on the FET and it only gets
mildly warm at full current.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


More... (looser matching)
- Last day of these posts
- In 2002 , 2003 only
- Today
- New search...