[saraengelstad]

I may have found a way to answer my own question. V- always goes to the ground plane, yes?
 
If so, than I am guessing that this is the ground plane:

 
I am sorely wishing for a test point, so I came up with the idea, why not add one to the ground plane? That should be my V- test point. So, since I'm not populating R19 or R21, and R19 seems to connect so nicely with the ground plane, why not put a test point on the point where R19 would go to V-... right there on the yellow circle? (Line indicates the other hole R19 would go through, if it were present.
 
Here's my work:


And here's my test point...
 

 
 
What's wrong with my logic?
While I wait for advice, I'm going to retest using my new toy.
Cheers everyone!
-Sara

 

[saraengelstad]

No, I have problems. I put in the 9K1 in R13 and used my new test point to test. I got unexpected values.
 
I have no idea if it is because the battery pack and wall charger are different,or what, but here are my test results:
 
S1 is unjumpered. No ICs on the board.
S3B to V- with PLED ON: 10.5 V
S3B to V- with PLED OFF (it takes a really long time to dim all the way down): 1.4 V
 
With IC6 on the board; Pins 1-3: 5.2V     Pins 1-2" -5.2 V
 
Step 8:
PLED ON, wall supply connected; S2B to V-:  .9V
PLED OFF, wall supply connected; S2B to V-: ~ 15.3 - 15.6 V.  CLED Flashing.
 
R14: ~.65  dropping to ~.5 when LED flashes
 
R14 and the regulator are very slightly warm.
 
Are the values off because of my
850 mA batteries and my 18V supply and a 9k1 resistor in R13?
 
Are the results then OK, or did I make a critical error?
 
I'm going to leave the thing run off the batteries and drain them while I'm away and try again later, hoping to hear back on my crazy findings.
 
Could it be that I wired the toggle incorrectly?
 
 
Thanks guys.

 

[saraengelstad]

Just in case, this is how I wired the toggle. The blue shrink leads connect to S2A/S2B and the red shrink leads connect to S3A/S3B.
 
It took some trial and error to get it to work, and I went with the first thing that allowed me to choose between blinking CLED and solid PLED.
 

 
Off to work, now. I am so grateful for the help.
 
-sara

 

[tangent]

I don't know where to place my probe to find "V-".

 
It's the big, thick trace on the bottom side that connects to W- and B-, then wraps around the front side of the board. You can also get to it via the in-board pin of RLED.
 

Different places that look likely according the the schematic and board layouts give me different readings,

 
The schematic is unambiguous. In fact, my observations about W-, B- and RLED are nicely concentrated on page 2, in the "Virtual Ground Power Supply" schematic fragment.
 
You can trace this out on the board layout image. The only difference is that the "V-" logical signal label never appears.
 
I've updated the board layout image to the current version. Previously I had a v0.5 image up there. There are only small differences between the two. I decided to update it more to remove a source of uncertainty than to fix a problem.
 

I take this to indicate that the cap is discharging,

 
Bingo.
 
 
Quote:

No, I have problems. I put in the 9K1 in R13 and used my new test point to test. I got unexpected values.

 
I'm going to ignore those measurements until you confirm you're measuring against the real V-, or re-measure. You don't need to post them if they turn out to be spot-on after all.
 
 

My calculations with these values gives me the result 9.6

 
Please show your work, Miss Engelstad.
 
I get something quite different:
 
R13 = (18 - 1.4) / 0.085
R13 = 16.6 / 0.085
R13 = 195
 
So, a 200 or 220 ohm resistor would be the smallest I'd use. The stock 1K resistor you replaced would still have been okay, it would have just taken ~5x as long to fully trickle charge the battery that way. In practice, that's not going to be a problem in most cases. The trickle charger exists more to keep the battery topped up if you leave it connected to the charger overnight, so it's 100% ready to go when you unplug it to use it on battery the next day.
 
Your 9.1K isn't going to be a serious problem. It just slows charging down even more, making it more of a "charge maintenance" resistor than a "trickle charge" resistor.
 
I estimate that you can go up to at least 16k and still have a useful trickle charge path, since that will still be greater than the pack's self-discharge rate.
 
If you don't want to desolder this resistor again or don't have anything in the low hundreds on hand, a quick and easy option is to solder tack the 1K you removed on top of it, giving ~900 ohms.
 
 

V- always goes to the ground plane, yes?

 
In an analog circuit, V- and ground are almost always different things. LNMP is no exception.
 
Your biggest clue was the IG pad hanging off that plane, which you can see from the schematic doesn't connect to V-.

 

[saraengelstad]

Quote:

"It's the big, thick trace on the bottom side that connects to W- and B-, then wraps around the front side of the board. You can also get to it via the in-board pin of RLED."

 
I have some experience doing this, so I feel okay with it but I would definitely want you to weigh in on it first. How about if drill a small hole through that trace, near RLED, and placed a permanent test point there? Would that disrupt anything?

 

[saraengelstad]

"The schematic is unambiguous. In fact, the my observations about W-, B- and RLED are nicely concentrated on page 2, in the "Virtual Ground Power Supply" schematic fragment."

I apologize, the schematic is indeed unambiguous, the problem occurs somewhere in the chain after the schematic, after my eyeballs, likely in the eyeball to back of head transfer process. But hey! I'm learning, right?
 

"You can trace this out on the board layout image. The only difference is that the "V-" logical signal label never appears." 

That was the picture, sign, AND guy in a lab coat pointing that I needed since this where my blockhead phenomenon was kicking in.
 

"I've updated the board layout image to the current version. Previously I had a v0.5 image up there. There are only small differences between the two. I decided to update it more to remove a source of uncertainty than to fix a problem."

Great, I already printed it, I'll watch for an updated version. Thank you for the updates! 
 

"I'm going to ignore those measurements until you confirm you're measuring against the real V-, or re-measure. You don't need to post them if they turn out to be spot-on after all." 

Me too.
 

"Please show your work, Miss Engelstad. "

Ha! I did on paper, I just didn't post it. I should have because I knew I was off by some factor of ten.  Here is where my error was:
 
I used 85 instead of 0.085 because the part selection guide reads:
"The trickle charge current shouldn’t be more than 0.1C for NiMH cells. (E.g. 70 mA for 700 mAh cells.)" If I'd worked the formula there, I would have seen how to properly work the problem with different values. I blew past it the first time because didn't think I'd ever need to because I have very little experience with rechargeables and didn't even know they come in different mAh values. I got lazy. Here was my calculation:
 
R13 = (18 - 1.4) / 85 = .19
That didn't make any sense, so I tried recreating what might have been used with other battery.charger combos and got:
 
R13 = (16 - 1.4) / 70 = .2085
 
R13 = (12 - 1.4) / 70 = .1514
 
So, from that... I don't know what the hell I was thinking. 
 

"So, a 200 or 220 ohm resistor would be the smallest I'd use. The stock 1K resistor you replaced would still have been okay, it would have just taken ~5x as long to fully trickle charge the battery that way. In practice, that's not going to be a problem in most cases. The trickle charger exists more to keep the battery topped up if you leave it connected to the charger overnight, so it's 100% ready to go when you unplug it to use it on battery the next day.
...
If you don't want to desolder this resistor again or don't have anything in the low hundreds on hand, a quick and easy option is to solder tack the 1K you removed on top of it, giving ~900 ohms."

I'm okay desoldering.... now. After many tears and frustrations and monies spent on desoldering guns and many destroyed boards, I finally learned to use braid. They key was your liquid flux. That's how I'm able to rework so much of my stuff. I build boards, look at them, get all obsessive, then decide, "CHANGES!" and redo them... I really like your boards, they are actually reworkable, the traces don't just lift off the boards like tinfoil. Man, I hate that. End digression.
 
 

"In an analog circuit, V- and ground are almost always different things. LNMP is no exception."

Well that nicely illustrates how much I know. I had that concept Perfectly Backwards.
 

"Your biggest clue was the IG pad hanging off that plane, which you can see from the schematic doesn't connect to V-."

I'll keep that in mind in the future, thanks.

 
 
Thank you Tangent. You get the prize today.
 
 

 
 
 

 

 

 

[saraengelstad]

If I were to drill that hole to place a test point, I was thinking the place to do it would be ~2cm  to the left of C1, making sure I only hit the bottom trace, and not the top one. 
Still a bad idea?

 

[saraengelstad]

Quote:

So, a 200 or 220 ohm resistor would be the smallest I'd use. The stock 1K resistor you replaced would still have been okay, it would have just taken ~5x as long to fully trickle charge the battery that way. In practice, that's not going to be a problem in most cases. The trickle charger exists more to keep the battery topped up if you leave it connected to the charger overnight, so it's 100% ready to go when you unplug it to use it on battery the next day.

 
I put in a 330R for R14 and these are my results, testing against V-:
 
Battery Voltage with no ICs installed from V- to S3B: 10.1V - PLED LIT
Toggle PLED OFF: 2V
 
IC6 Installed, PLED on: 5V and +5V between pins 1/3 and 1/2.
PLED OFF: .5 V between pins 1./3
 
Plugged in wall supply, S2B to V- .6V
 
CLED flashing.... R14 measures between .6V and .45 volts as PLED flashes.
 
In the time it took to write this, PLED is on steady.
R14 = .46V
 
S2B - v- = 17.2V
 
I think that the switch may be set up incorrectly, and / or perhaps the values in R16 and R17 are wrong for the 18V supply:?
R16 = 56K
R17 = 10K
 
I checked the rest of the resistor values, and they are all perfect except R10. It is a 47R because I didn't have a 47R5 on hand.
 
Ideas?
 
I went through my collection of wall-warts, but all the ones with the right connector are all 24V.
I have a (very) poor man's bench power supply that will give me 12V, but I would have to rig a connector, which I can do. Worth it?
 
Also, I have the power supplies I built that this thing is supposed to test but they are all intended to run at higher voltages themselves.
 
Suggestions?

 

[tangent]

Quote:

I don't know what the hell I was thinking.

 
You're clear on where 0.085 comes from now, right?
 
The 'C' notation is very common with rechargeables. It's so well entrenched that it refuses to go away even though it is inherently confusing, despite attempts to replace the notation.
 
(1C is the [dis]charge rate that gives 1 hour, meaning it combines amps and hours, but that is an inherently bogus thing to do where "C" is used, because we're not expressing the multiplication of two measured quantities. Instead, what we're doing with this "C" business is pretending that amp-hours are linearly convertible, which they aren't. The amp-hour curve changes shape as a function of [dis]charge rate, and it's different for charging and discharging. But it's useful to pretend we can treat rechargeable battery amp-hours as linear, so everyone just copes with it as long as we all understand that it's just a form of engineering handwavery.)
 
Quote:

If I were to drill that hole to place a test point, I was thinking the place to do it would be ~2cm  to the left of C1, making sure I only hit the bottom trace, and not the top one. 
Still a bad idea?

 
Yes, I think so. After drilling the hole, you have to work out how to scrape away enough solder mask to get a solid joint. You can't just bond to the hair-thin rim of copper you expose with the drill. That's not enough for a strong bond; certainly not strong enough to be hanging test probes off of. Even if you did get a good electrical connection, you've increased the resistance of the V- trace, since solder isn't as good a conductor as copper.
 
Me, I'd just jam a probe into some likely place, like the elbow of the RLED lead.
 
If I absolutely had to create a hard test probe point on the LNMP v1.1, I'd probably surface-mount a wire loop test point to W- or B-.
 
You can also alligator-clip to the solder lugs on the DC input jack. One of those is V-.
 

I think that the switch may be set up incorrectly

 
 
Looks fine to me.
 
 

perhaps the values in R16 and R17 are wrong for the 18V supply:?

 
Supply voltage doesn't have anything to do with it. It's purely a question of battery voltage, or as the docs put it, cell count. The excess voltage gets burned off in the LM317.
 
 

I went through my collection of wall-warts, but all the ones with the right connector are all 24V.

 
Go ahead and try one. You need something that will put out 0.3 A or more, with the 4.7 ohm R14.
 
That could explain your current problem. If your 18V supply isn't studly enough, it will give these sorts of weird voltage numbers as it staggers under the load.
 
 

I have a (very) poor man's bench power supply that will give me 12V, but I would have to rig a connector, which I can do. Worth it?

 
For an 8-cell pack, 12V is probably just a bit on the low side. If it were, say, 12V unregulated and rated for 1A or so, it'd probably be fine, since it would develop 13V+ when "lightly" loaded by the LNMP.

 

[saraengelstad]

Quote:

You're clear on where 0.085 comes from now, right?

Yes, thank you, I was not converting the units correctly.
 
Quote:

The 'C' notation is very common with rechargeables. It's so well entrenched that it refuses to go away even though it is inherently confusing, despite attempts to replace the notation.

(1C is the [dis]charge rate that gives 1 hour, meaning it combines amps and hours, but that is an inherently bogus thing to do where "C" is used, because we're not expressing the multiplication of two measured quantities. Instead, what we're doing with this "C" business is pretending that amp-hours are linearly convertible, which they aren't. The amp-hour curve changes shape as a function of [dis]charge rate, and it's different for charging and discharging. But it's useful to pretend we can treat rechargeable battery amp-hours as linear, so everyone just copes with it as long as we all understand that it's just a form of engineering handwavery.)
 

I'll have to look at this more closely later, because I feel like I'm not quite catching it
 
Quote:

That could explain your current problem. If your 18V supply isn't studly enough, it will give these sorts of weird voltage numbers as it staggers under the load.

OK, I switched power supplies to a ... uh oh, maybe this is my problem.
I tried a 24V AC adaptor that puts out 24V AC and got some oddball results. I'll dig up a DC. Oops. Dang, now I have to go and remeasure everything...
 
I have a suggestion for the test procedure. In step 7, I would have the user measure DC volts from V- to S3B with PLED OFF and then again with it ON, instead of the other way around (or at least, I read it that way). When I tested it with it ON first, I had the capacitor discharge problem throwing me off, but if I did it the other way, I didn't have that problem. Then I retested and saw the cap discharge phenomenon so I knew I was high speed, at least on that test.
 
With the AC power supply, I passed tests 1 (PLED OFF and ON and OFF again, yielding 0V, 10.5V, and discharge voltage)
Passed test 2 (IC6 pins 1-3 = 5.2 V, pins 1-2 = -5.2V)
 
Test 3 :
Passed measured DC volts from S2B to V- = 0V
-------
 
This is where things started to go sideways. I am assuming it's the AC converter. I'm going to try a DC and repeat.
 
At least this is more fun than unbricking a router that had a brownout during a firmware updated. That WAS NOT fun. I'm going to feel some real accomplishment after this.
 
Sigh. If I make it through. This is difficult stuff. Thank you for your help, Tangent. You are helping me learn things that will benefit me for the rest of my life.
 
Go MacGyver!
-s

 

[saraengelstad]

Should I just ditch the 850 mAh batteries for 700s?
Is that the problem?
 
The power supply I'm using is 18V DC 1.4 A. Maybe I made a mistake somewhere, I'll retest with this one, just in case I got them mixed up.
 
Or, should I get a different supply and different batteries, or ....?

 

[tangent]

Should I just ditch the 850 mAh batteries for 700s?

 
No, they're fine.
 

The power supply I'm using is 18V DC 1.4 A.

 
More than adequate.
 
It's clear from your +/-5V results across the amplifier section power rails that that is probably all working. All we're concerned with here is the operation of the charger, which is a fairly simple circuit.
 
Here's a thought: plug in the 18V supply, ensure that 18V is getting past D1, then continue to measure DC volts between V- and all the nodes in the circuit:
 

1. both sides of R13, R14
2. B+
3. D2 / IC4 / R15
4. R16 / R17
5. R18 / CLED 

 
It might help to print the schematic and write the measurements in colored ink. With a bit of thinking and Ohm's Law (V=IR) you may see something obvious.

 

[saraengelstad]

Should I put in all the ICs first? Because the testing sort of says what order to install and test them.
You're making me work!

 
0
 
It'll be an early morning for me.

 

[tangent]

We're only testing the power supply, and all that should be soldered down. If you're using IC sockets, they should only be in the amp section.

 

[saraengelstad]

I always socket everything. So desolder the sockets and solder the PS ICs in directly? Could that be my problem???