Fechter's Mini Meanwell Limiter Board

[cor]

I plan to have the monitors switch off with the key switch (I made a little circuit with an opto), the only thing I see is that the cheaper copy of the celllog called BVM-8s does not seem to have an upper voltage limit, only a low voltage alarm and a cell-to-cell maximum delta alarm.

[Degull]

I'm reviewing the design of the mini limiter board now.

It looks like by adding 2 resistors and a transistor, I can make an interface for CellLogs or HVC boards that will drop the charge current to zero when active. These parts can be left off the board for those who don't need this function.

Does this sound like a useful feature?

Another thing I'd like to test is to replace the output diode with a red LED and see if it is bright enough to see. This would provide the diode function needed in addition to indicating when the limiter was active. My guess is the LED would light up during bulk charge getting gradually dimmer as the pack approaches the CV point, then go out. This would give a visual indication of when the pack was done.

You are taking an already useful board and making it better! Both those features would a great addition and well worth any wait!

[cor]

Hi Richard,
Lets do a quick calculation of the current through the LED-as-diode:
Suppose this is a 48V supply charging a 12s LiPo pack (so the supply is set to 51 for 4.25V per cell) and the pack is charging away at 4V per cell (48V).
The voltage divider for the OVP in the NES-350-48 is 44k Ohm fixed and a 2k + 1k pot. That means that the current through the LED must make up the difference between the 44k resistor at 48V output and at 51V output - this is the current at that point of charging.
The voltage across the resistor is 2.5V less than the output, so the current is (48-2.5)/44k respectively (51-2.5)/44k or 1034 and 1102uA, the difference being 68uA. I am not sure how bright a red LED is at this current level, but I doubt that it is more than a faint glow, hard to detect in daylight. It is two orders of magnitude smaller than typical LED current.

Also note that the LED will be OFF (zero current) while the charger is still blasting away at max current, only has reached its max voltage set point.
It may be easier and better to indicate end-of-charge when the output from the opamp falls below a specific voltage.

[fechter]

Thanks for running the math on that. I can see that approach won't work.

What would work is to just drive the LED off the output of the op amp through a resistor. When limiting starts, the output of the op amp will be 2.5v plus about .6v for the diode, plenty to drive a LED. This will only take one extra resistor and the LED, both of which could be optional.

[cor]

Correct. And instead of fading off while the charging is catching up to the max voltage set point, the output of the opamp stays at 2.5 + 0.6V as long as it is limiting the charge current, no matter the output voltage (if you have no or only little resistance between opamp and OVP point). Once the current limit is lifted because the pack has reached the max charger voltage and the charger starts tapering off, also the LED starts fading because the opamp output is representative of the current (amp'ed shunt voltage) so if you give the opamp a moderate gain as for example suggested by S.B.D. here:
http://endless-sphere.com/forums/viewtopic.php?f=14&t=36114&start=45#p528393
then you will see the LED not completely go off as soon as the current limit ends, but instead it will slowly fade with the current falling, for example it is at max brightness when the opamp output is 3.1V (for example representing 7A as with a NES-350-48) and extinguish when the charge current falls to about 3A (1.3V on opamp output)
Of course you can add one more resistor from LED to +Vcc and have the LED stay on a little longer, for example extinguish completely by the time the output current goes to zero. This pull-up also helps with the max current sourcing capability of the opamp. Luckily super-bright LEDs only need a few mA.
To get this effect, at the moment that the opamp output hits 0V you want the resisor divider from opamp via LED to +Vcc to present about 1.3V on the LED, so if the Vcc is 12V and you use a 1k resistor to drive the LED then you want about 8k2 to +Vcc. This last resistor is essentially a current source that continuously injects 10.7V/8k2 = 1.3mA into the LED. The opamp can add to it (when limiting it will be about (3.1-1.3)V/1k = 1.8mA sourced) or subtract from it, at 0V output the opamp is sinking all of the 1.3mA so there is nothing left for the LED.

[fechter]

Here's what the schematic looks like so far. I still need to check some of the resistor values.

One of my design objectives is to use the fewest parts possible. I looked at using the other half of the op amp instead of the PNP transistor but it ends up needing more resistors/diodes and would probably take as much or more board space. I'm still considering that route.

To avoid overheating the regulator, the total circuit current needs to stay below about 10mA.
High efficiency red LEDs are easily visible at 1mA.

In this configuration, the LED will indicate CC mode and go out at end of charge. If HVC activates, the LED will light up extra bright and the output voltage of the supply will drop to minimum.

Mini Limiter Board schematic V3.jpg (37.48 KiB) Viewed 381 times

[cor]

Hi Richard,
Two remarks:
The LED will never extinguish, because it is back-fed from the 2.5V reference and in fact, the presence of the LED will make the output voltage of the MeanWell shoot up due to the loading of the 2.5V reference (OVP). So, it is required to have the LED at the other side of the diode to isolate the supply from this effect.
The other remark is that the opamp will cease to drive as soon as the current is no longer limited. That is what I liked about S.B.D.'s use of a fixed gain opamp that has the trimpot in the feedback (gain setting) so you can tune it to start pulling the OVP up when the current through the shunt gets amplified enough to go 1 diode drop above the 2.5V reference. The side effect is that when the current tapers, also the voltage slowly tapers and thus the LED slowly dims with the current, instead of hard switching off.
I did not incorporate the last suggestion, but I changed your drawing to remove the first effect. I was also able to replace the transistor with a resistor, the only possible side effect might be some noise on the current limit if the HVC line is long and picking up noise:

[fechter]

Ah, yes, thanks for the double check. I see the LED configuration will be a problem like I showed.

The HVC positive line may be pulled to ground by a CellLog, so I want to accomodate that if possible. I could use the transistor to pull up the positive input of the op amp.

Due to my 'minimum parts' constraint, I was trying to avoid using a feedback resistor on the amp. Since the inputs are in the millivolt range, it seems to work fine without one.

I think I could just throw another diode in series with R1 and solve the problem.
I also increased R5 to limit the current when HVC is active. 1M drives the LED at about 3mA using a 2N3906.

Mini Limiter Board schematic 3a.jpg (39.54 KiB) Viewed 350 times

[fechter]

In testing, I can see that my HVC line will be excessively sensitive to noise pickup, so not a good approach.

Using the HVC to pull up the positive amp input will work fine other than we can't use a direct connection to a CellLog. Maybe this is not such a big deal, as you need to use isolators for more than 8 cells anyway. It certainly keeps the parts count down and I don't think noise pickup will be a big issue with a 100 ohm resistor to ground. This should work fine for any optocoupled HVC board like Methods' or Gary's.

[teklektik]

Using the HVC to pull up the positive amp input will work fine other than we can't use a direct connection to a CellLog. ... This should work fine for any optocoupled HVC board like Methods' or Gary's.

I think this might work to preserve CellLog compatibility while providing a separate indicator (red) if HVC is asserted. The extra indication wasn't really important, but the bi-color LED offered a means to OR together the op amp outputs without adding yet another component - as it is it adds 3 resistors, the LED, and uses the other half of the op amp chip...

mini limiter board schematic 3a-4.jpg (32.27 KiB) Viewed 320 times

EDIT: revise R3, R6 (whew - finally cleaned up all the sloppy edit artifacts )

[cor]

If you connect pin 5 (+ input of 2nd Opamp) to the OVP point (which is always 2.5V) then you do not need to split the 160k resistor in two, saving one more resistor.

[teklektik]

!

[fechter]

If you connect pin 5 (+ input of 2nd Opamp) to the OVP point (which is always 2.5V) then you do not need to split the 160k resistor in two, saving one more resistor.

We think alike! I think the OVP point will work to feed the op amp, thus saving a resistor.

It's possible that could induce some kind of oscillation when HVC activates but even if it did, I don't think it would cause a problem.

Going back a few schematics to cor's, it would be possible to use a single un-isolated CellLog by placing a diode (LED?) across the HVC pins and programming the CellLog alarm to be normally closed instead of normall open. The alarm line would then pull the line to ground during normal operation and the resistor would pull it up upon alarming. This would have the added benefit of being fail-safe on the alarm line wires. If a connection goes open, it would terminate charge.

I do like having a separate LED for indicating HVC versus limiting. It's bad practice to reverse bias the LEDs though, I'll have to check the reverse voltage rating. I've damaged them with reverse voltage before.

[teklektik]

... It's bad practice to reverse bias the LEDs though, I'll have to check the reverse voltage rating.

hmmm - I've cooked a few doing that as well - I think LEDs are commonly rated at Vr = 5v so this doesn't look good with reverse 12v... too bad - I kind of liked the HVC indicator.

[teklektik]

... and programming the CellLog alarm to be normally closed instead of normally open.

Maybe I missed it, but I don't see a way to directly configure the CellLog alarm for NC operation. Are you thinking of pressing the CellLog LVC setting into service as a NC HVC?

[fechter]

I'll have to double check the instructions but I'm pretty sure there is a setting for NC alarm output.

[teklektik]

fechter - One last try to get CellLog direct connect compatibility and an HVC indicator - this time with a 2-lead bicolor LED to circumvent the high Vr issue. The red HVC indication relies on U2/7 going high to suppress the charge current causing U2/1 to fall to Gnd. U2/1 and U2/7 have the reverse relationship during normal limiting operation giving a green indication.

mini limiter board schematic 3a-5.jpg (38.7 KiB) Viewed 339 times

[fechter]

Excellent, yes that should work. The 3 legged bicolor LED might be OK at 12v too, I just need to check the datasheet. Now to work on revising the layout to see if it all fits in the same size board. I can always make the board a little bigger if needed.

[teklektik]

I did a few spot checks on the 3-lead part to no avail - maybe you will have better luck. The 12v Vr oversight is looking like a fatal design flaw...

Also - I think the LED colors should be reversed to R=limit, G=HVC similar to the Voltphreaks charger; red means 'wait' - bulk charge is underway, green means bulk charging has terminated.

[fechter]

I agree on the colors.

Maximum reverse voltage on the LEDs I used in the past is 5v.

There may be another issue with the two wire LED; the output of the 358 can only pull down to about 1v at 2-3 mA. If one amp is hanging around 3V and the other one is 1v, that doesn't leave enough to light up the LED. There are some other amplifiers with rail-to-rail output that could possibly overcome this but there would still be a big difference in the current between the red and green sections.

More head scratching....

[fechter]

OK, I think this one addresses all the problems but takes quite a few parts. I'll have to see if I can squish it into the layout.

By using separate resistors for the LEDs we can also get the current matched between them a bit closer.

Mini Limiter Board schematic 3e.jpg (38.64 KiB) Viewed 298 times

[teklektik]

Cool - too bad about the parts count, but the extra features are well worth it.

[fechter]

I pretty much had to start from scratch on the layout, but I think it turned out pretty good.
I made the board just a few millimeters taller than the ver.2 boards to fit everything.

I was able to place the adjustment pot near the top so it is easier to reach.

The LED is also along the top, and the leads are oriented to make it easy to bend at 90 deg. if desired. This might be good for the clones with horizontal terminals. It would also be possible to mount the LED on the 'bottom' of the board since the board sits upside down on these supplies.

All the resistors and diodes (except the zener) are in a nice row, which makes construction faster and easier. You can stuff all the resistors and solder them all in one pass.

If you don't want HVC interface, you can leave out a diode and two resistors.

Mini Limiter V3 layout top.jpg (65.91 KiB) Viewed 289 times

Here's what the top side layout looks like (sort of):

[teklektik]

Slick!
Sign me up for 2 when you do a board run...
( Very cool that you undertook these enhancements. )

[NeilP]

Not trying to knock any of the work at all, but what does this board do different to the original single pot current limiter? I see the HVC function, but what about charge monitoring for end of charge shut down? like the three pot?

Maybe this is just a personal thing, but that function is more useful than the current limiting. Current limiting can be done on the board, much simpler with the pot/resistor mod. To my mind, if someone is going to the trouble,effort and expense of building the board it has to offer more than you can get with 20 minutes work and less than a couple of dollars worth of components ..I a pot and some wire to do resistor mod (S or NES series) ...

I see the HVC can do that , but it requires an external input..cant that be incorporated in the board??