Idea for Trickle Charger PCB

[_atari_]

After reading this and this thread I also built a trickle charger for my pimeta.
So far so good. Works like a charm. But building it was not so funny. I really like populating PCBs much more (I am lazy). So I wanted to build a very tiny PCB for the trickle charger.

All details are explained there. But the end is somthing like this:

TO220:

TO92:

Now I have the following questions:
1. What do you think about the PCBs? Any recommendations to make it better?
2. Which one do you prefer?
3. Is there any interest in PCBs? If it is enough I would perhp produce some boards.
4. Any clue for some kind of charge indicator (LED or something similar) ?
5. Do you think there should be a voltage regulator built in to adjust the maximum charge voltage?
6. Do you think there should be an filter circuit built in for unregulated or switching power supplies?

Best regards

Atari

 

[gates_2]

i'd be interested in obtaining several

 

[tangent]

Quote:

Originally Posted by _atari_ /img/forum/go_quote.gif

I don't like the resistor and diodes being underneath the regulator like that. Either change the regulator package to indicate that it's standing up (EAGLE library has both types) or move the parts out of the way if you really do want to lay the regulator down. Remember, the exposed tab on an LM317 is "live".

Personally, I'd probably lay the regulator down, and put a copper plane underneath it to act as a heat sink in light load situtations. See my PPA battery board for ideas on this.

If you didn't want to make the board bigger, then I'd move the parts to the back side and use surface-mount versions.

Quote:

All of those Molex connectors want to rotate 90 degress to the left.

Also, one of the resistor pads comes a little too close to the trace between the diodes for my taste. It's not like this is a particularly dense board, where you have no choice...

Quote:

Which one do you prefer?

If you could only offer one, it should be the TO-220 version. The TO-92 version is only strong enough to trickle charge AAA and smaller cells.

Quote:

Any clue for some kind of charge indicator

You could just put an LED and limiting resistor on the output of the regulator, in parallel with the battery pack.

Quote:

Do you think there should be a voltage regulator built in to adjust the maximum charge voltage?

No. That defeats the purpose of this kind of regulator, which is that its output voltage continually adjusts to the needs of the battery pack. This makes it unsuitable for charging lead and lithium battery chemistries, but that should not be viewed as a problem. Don't overdesign it. What you have here is a perfectly useful nickel chemistry trickle charger...it doesn't need to be anything more in order to be useful.

Quote:

Do you think there should be an filter circuit built in for unregulated or switching power supplies?

Batteries don't care.

If you're concerned about listening to the amp while it's charging, it seems to me that in the case of an unregulated supply, you don't have a problem. The ripple will still be taken care of by the regulator, because its output voltage is pinned at a relatively steady value. The long-term voltage does change, but the amp won't care about that.

For a switcher, yes, noise will still get through, but if you want to listen to it at the same time, you should just pick a switcher that has a noise signature you can't hear to begin with, not try to cure it in the charge circuit.

 

[_atari_]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif I don't like the resistor and diodes being underneath the regulator like that. Either change the regulator package to indicate that it's standing up (EAGLE library has both types) or move the parts out of the way if you really do want to lay the regulator down. Remember, the exposed tab on an LM317 is "live". Personally, I'd probably lay the regulator down, and put a copper plane underneath it to act as a heat sink in light load situtations. See my PPA battery board for ideas on this. If you didn't want to make the board bigger, then I'd move the parts to the back side and use surface-mount versions.

My original idea is very hard to model in eagle. It is meant that the bottom of the board, mounted with the "life" site to the case and fixed with a screw (isolated of course). The "original" mounting hole is then just a method to reach the screw - understandable?
I know it looks like a mess in the schematic - but it was intendend

Quote:

Originally Posted by tangent /img/forum/go_quote.gif All of those Molex connectors want to rotate 90 degress to the left. Also, one of the resistor pads comes a little too close to the trace between the diodes for my taste. It's not like this is a particularly dense board, where you have no choice... If you could only offer one, it should be the TO-220 version. The TO-92 version is only strong enough to trickle charge AAA and smaller cells.

I think I will throw away the TO-92 variant. It has too many disadvantages.

Quote:

Originally Posted by tangent /img/forum/go_quote.gif You could just put an LED and limiting resistor on the output of the regulator, in parallel with the battery pack.

Then you have to consider the current draw and maximum voltage in the calculation of the calculation of the resistor?

Quote:

Originally Posted by tangent /img/forum/go_quote.gif No. That defeats the purpose of this kind of regulator, which is that its output voltage continually adjusts to the needs of the battery pack. This makes it unsuitable for charging lead and lithium battery chemistries, but that should not be viewed as a problem. Don't overdesign it. What you have here is a perfectly useful nickel chemistry trickle charger...it doesn't need to be anything more in order to be useful. Batteries don't care. If you're concerned about listening to the amp while it's charging, it seems to me that in the case of an unregulated supply, you don't have a problem. The ripple will still be taken care of by the regulator, because its output voltage is pinned at a relatively steady value. The long-term voltage does change, but the amp won't care about that.

Shouldn't then the circuit look a little bit different? Something like this (schematics only):

instead of

But thousand thanks for your advices!

 

[tomb]

Just a couple of suggestions -

1. When Tangent says AAA cells and smaller, that still means 9V batteries, since they are made up of much smaller cells. So, there is some usefulness even in the TO-92.

2. If I were to do this again, it would include the 3rd diode. NeilR pointed this out in my Trickle Charger thread, and it was included in the original _Redrum thread. If you connect an LED-equipped power source, it will drain the batteries back through the LED unless the PS is unplugged. Many of us use power strips and simply switch on and off a number of different PS's connected into various amps at the same time. That won't work unless you include the third diode that prevents backfeed from the batteries.

 

[tangent]

Quote:

Originally Posted by _atari_ /img/forum/go_quote.gif My original idea is very hard to model in eagle. It is meant that the bottom of the board, mounted with the "life" site to the case and fixed with a screw (isolated of course). The "original" mounting hole is then just a method to reach the screw

I wouldn't do it that way, either. This energizes the case, which means you have to isolate all the other panel components. Or, you have to add an insulator, which removes some of the advantage of using the case as a heat sink in the first place. If you want to allow the board to be mounted to the enclosure, I'd just add a couple of mounting holes for the purpose.

Quote:

Then you have to consider the current draw and maximum voltage in the calculation of the calculation of the resistor?

Naturally.

If you don't want the LED to change brightness as the charging voltage changes, you can use some sort of constant current circuit instead of the current limiting resistor.

Quote:

Shouldn't then the circuit look a little bit different?

Yes, that's right. Obviously this alternative only works when the charging current is higher than the downstream circuit's operating current. But, since you can get from one version to the other just by removing a diode, it's worth considering it as a standard configuration option.

 

[_atari_]

It is much more complex than I thought (got no electrical eng. background - just a stupid computer science guy).

But to consider your suggestions:

Quote:

Originally Posted by tomb /img/forum/go_quote.gif So, there is some usefulness even in the TO-92.

TO-92 makes the most sense to me (size considerations)

Quote:

Originally Posted by tomb /img/forum/go_quote.gif If I were to do this again, it would include the 3rd diode.

Good Point

Quote:

Originally Posted by tangent /img/forum/go_quote.gif If you don't want the LED to change brightness as the charging voltage changes, you can use some sort of constant current circuit instead of the current limiting resistor.

Nope. I like the idea of "percieved charge level". It should be an indicator

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Yes, that's right. Obviously this alternative only works when the charging current is higher than the downstream circuit's operating current.

Can we assume that? I doubt it.

Lets do it in a structured approach:

Requirements:

It should fit between two 9V cells. So the space left is something like 20mm X 46 mm.

The biggest AAA I found has C=800mAh (there are 1000mAh - but we just assume they will just charge slowlier

) - so we need maximum 80mA charge current.

A TO-92 can give 100mA (pretty close to 80mAh).

A diode draws about 20-50mAh - death of the TO-92 variant?

The amp cannot be powered by the charge circuit. Typical 9V rechargable hav 240mAh to 300mAh - i doubt that an amp draws less than 24mA.

My 9V block has an charge end voltage of 9V - but it can get much higher.

Conclusions

Just using the TO-92 is possible but can be too close for some applications.

I do not know how to calculate the led resistor (because it is behind the LM317 to "display" the charge voltage. Any ideas how to solve this issue? Use a different approach?
I m more than happy with any help regarding this issue!

A second LM317 for the amp would be a good idea. Bu it uses space. Do we need some kind of capacitor to reduce ripples or should we just believe the amp can handle them?

I have designed some schematic sketch for the perfect solution.

Any suggestions?
My feeling sais we are quite close to the problem of overengineering - but not across this border yet

.

 

[tangent]

Quote:

Originally Posted by _atari_ /img/forum/go_quote.gif I like the idea of "percieved charge level".

I don't think the brightness of the LED is a good way to tell how full the battery is. What does 60% look like?

If you want a variable charge indicator, there are circuits to drive a bicolor or tricolor LED depending on an outside voltage.

Quote:

Can we assume that?

You don't have to assume it. You just offer it as an option: if you want to use an unregulated supply and your downstream circuit requires less operating current than the charge current, you can leave out D1 and use an unregulated supply. Otherwise, you should only use the wall supply for charging, not listening.

Quote:

It should fit between two 9V cells.

Terminology bug #1: there are no 9V cells, only batteries. A battery is multiple cells in series, giving a higher voltage. Cells are ~1.2V (nickel), 1.5V (zinc, alkaline, lithium alkaline replacements), 2V (lead), or 3V+ (other lithium chemistries, including all rechargeables). A 9V battery is anywhere between 6 and 8 cells, depending on chemistry and level of marketing BS.

Quote:

pretty close to 80mAh

Terminology bug #2: mAh is milliamp-hours, so it only makes sense to use it when time is a factor. Charge current is continuous as long as the supply is plugged in, so you just say mA. Rechargeables are spec'd in mAh because it's a unit of energy, and there's a quasi-linear relationship with time: Draw 1 mA from an 800 mAh cell and it should last for 800 hours. Draw 2mA, and it lasts for 400 hours. Etc.

The relationship is not truly linear, but we ignore this for practical reasons. You wouldn't really want to have to buy cells based on a third-order polynomial.

Quote:

A diode draws about 20-50mAh - death of the TO-92 variant?

No, because you really only need 1-5 mA to get acceptable brightness with any common LED. 20-50 mA is the absolute limit before destruction, not a goal.

Quote:

Typical 9V rechargable hav 240mAh to 300mAh - i doubt that an amp draws less than 24mA.

Some amps do draw less than 24 mA, and not everyone is going to use your circuit with 9V batteries only.

Quote:

I do not know how to calculate the led resistor

The output voltage of the regulator during charge will be approximately 0.9-1.35V per cell. It can be lower with a severely depleted battery, and it should never be higher unless you've miscalculated something.

Given that, the solution is just an application of Ohm's Law: minimum LED current = (cells * 0.9) / R. Or, R = (cells * 0.9) / desired minimum current.

Quote:

A second LM317 for the amp would be a good idea

Sounds like kitchen sink syndrome to me.

 

[_atari_]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif I don't think the brightness of the LED is a good way to tell how full the battery is. What does 60% look like? If you want a variable charge indicator, there are circuits to drive a bicolor or tricolor LED depending on an outside voltage.

All circuits I have found used voltage comparators - making everything bigger. If anybody knows a very simple solution - I am happy to learn about it.

In the meantime I just added two LEDs - one for power and one charge indicator - giving the abilit to use a bicolor led and get some "feeling" for charge. nothing very acurate but gives an idea if the LED is "quite low" or "quite bright" - better than nothing.

Quote:

Originally Posted by tangent /img/forum/go_quote.gif The output voltage of the regulator during charge will be approximately 0.9-1.35V per cell. It can be lower with a severely depleted battery, and it should never be higher unless you've miscalculated something. Given that, the solution is just an application of Ohm's Law: minimum LED current = (cells * 0.9) / R. Or, R = (cells * 0.9) / desired minimum current.

Thanks a lot for your help!

So I end up with this circuit:

With this tiny pcb (300dpi):

LED1 "shows" the charge voltage.
LED2 is for power indication.
R1 defines charge/LED current.


Did I get everything right? Any remarks?
I hope I got all the diodes right to protect/power the different parts.

Quote:

Originally Posted by tangent /img/forum/go_quote.gif Sounds like kitchen sink syndrome to me.

I am not really convinced. It would be a nice solution to used cheapo unregulated PSU as charger AND linear regulated power supply for your portable amp. I am basically thinking of throwing together the above circuit with half-a tread by tangent.
I twiddling around with it. It should be possible to make it real small for this application. I will post an update if I have worked out an circuit.

Do you think it is usefull or is it excessive?
Which one would you build?

 

[tangent]

Just a few more comments:

1. Your schematic should "flow" from left to right. Use supply symbols (V- or GND) to avoid having to wrap the LEDs back to the left side. The LED and associated resistor should just go straight down from the circuit point they monitor. Similarly, I'd rotate R1 90 degrees, and slide D2 and D3 over to the right to meet it.

2. Your web site address isn't going to print correctly, if at all. Silkscreen lines generally have to be at least 10 mils. Look through the EAGLE web site's script archive for scripts that will re-render all of the silkscreen lines onto a new layer. You'll probably end up with an unrecognizable blob in place of the web site name. You can either make it larger, or put it on one of the copper layers, which have higher resolution. Even so, it only looks like 1 or 2 mil lines, which is still below the minimum 7 or 8 mils resolution available for a reasonable price.

 

[_atari_]

Thanks to tangent for all the help.

I tried to consider all your remarks to design a preliminary final version:

This is the schmatics:

And this is the PCB:

But the whole idea of combining a TREAD with a trickle charger did work in my mind. I came back to it, since adding half a TREADS gives you the oportunity to use a cheap unregulated wallwart and still getting a decent power supply. A small one stop solution for my pimeta.
So I came back glueing a TREADS without the rectification circuit in parallel to the trickle charge.
The TREADS does not include an trimpot - so its voltage can not regulated that good as the real TREADS - but considering that the voltage should just be a bit higher than the final charge voltage of the batteries I consider it to be fine.

This leads to this schematic:

And this PCB:

I tried to add a GND plane and some usefull VCC plane. Note: There is a slight overlap between C2 and C3 - this will be fixed.

I still have some question:
Did I get al the diodes in the schematics right? I tried to check it - but it is quite complicated (at least for me).
Any remarks or hints what to do better?
Is the sizing and placement of the capacitors right? I tried to copy and adapt it from tangents TREAD. The sizes are estimates for AVX Tantalum, according to tangents recommendtions.
Is a trimpot for the TREAD neccessary? I have not yet calculated the final value of R2/R3 for my NiMH cells.
Which version do you prefer?

 

[tangent]

Quote:

Originally Posted by _atari_ /img/forum/go_quote.gif This leads to this schematic

A few more style points:

1. Never make connections at a crossroads. It's easy to visually confuse it with a cross-over. Obviously the big dot tells you whether it's a connection or a cross-over, but if you never make connections this way, you don't have to mentally remind yourself to look for the dot. If you see two wires crossing, you always know it must be a cross-over if you adopt this style. Just as an idea of how to fix one of these so you see the principle, I'd move the connection from D3's cathode to the first regulator's IN pin farther to the right, and leave the LED leg where it is.

2. I'd move R1 up to the same level as the regulator's OUT pin, and move everything connected to D1's anode to the right. This will fix a right-to-left flow problem, and a 4-way connection. You could move the LEDC subcircuit to the right then, too.

3. The voltage regulator circuit is messy. Add some space in there so labels don't overlap parts and nets, even up the GND symbols, etc.

4. C3: 11uF, really??

Quote:

And this PCB

I'd move the BAT pads to the left, by the IN pads. This will shorten the trace between the diode and BAT+. It will also let you move LEDC a bit farther from LEDP.

Also, I'd swap the positions of R1 and its neighboring diode.

Quote:

There is a slight overlap between C2 and C3 - this will be fixed.

It looks okay to me...

Quote:

Did I get al the diodes in the schematics right? I tried to check it - but it is quite complicated (at least for me).

It looks alright at first study, but I could have overlooked something.

Quote:

Is the sizing and placement of the capacitors right? I tried to copy and adapt it from tangents TREAD.

A better plan is to decide what parts you want to buy, get their datasheets, and look up their sizes. You'd be surprised how much size difference there can be among parts. Tantalums, for instance, often come in multiple lead formation styles, each with different spacings. Voltage tolerance differences can also change the size on you.

Quote:

Is a trimpot for the TREAD neccessary?

No, it's just a convenience. And, please, don't call it a TREAD. A TREAD is a circuit board, not a circuit design. What you have here is an LM317 voltage regulator.

 

[_atari_]

Thanks tagent and tomb for all you help.
I incorporated all your help into the final solution on my web site:
NiMH trickle charger

I hop at the end I got everything right. To be honest it was the first circuit and pcb I designed. Big thanks. I learned a lot by you!

 

[colonelkernel8]

Quote:

Originally Posted by _atari_ /img/forum/go_quote.gif Thanks tagent and tomb for all you help. I incorporated all your help into the final solution on my web site: NiMH trickle charger I hop at the end I got everything right. To be honest it was the first circuit and pcb I designed. Big thanks. I learned a lot by you!

Sorry for the thread crap, but "final solution"?

 

[_atari_]

Quote:

Originally Posted by colonelkernel8 /img/forum/go_quote.gif Sorry for the thread crap, but "final solution"?

Final for now

I just wanted to finish it to go to other tasks. It still has some improvement potential:

• It is my first PCB desing. There MUST be a lot of things to be done better.
• A constant current source for the power LCD could be used to make it independent from the input voltage.
• I could try to order some PCBs to see is it really works.
• There could be some tricks to enhance the ripple rejection in the power supply.
• Somebody a lot smarter than me could come up with solution to filter any artifacts of switching power supplies.
• There could be a simple way to set the output voltage to texactly the same level as the battery voltage.
• There could be a clever (=SMD
  
  ) way to incorporate a fast charging controller.
• There could be a way to make TO220 and TO92 LM317 usable on the same PCB foot print or even at a lower footprint.

As you see it is final.