Five LM317's in parallel for battery charger?

[treez]

Hello,

Is the following 400mA NiCd battery charger, involving five LM217 current regulators in parallel OK?

Each LM217 is supposed to pass 80mA, but i wondered if parallel connection will make only one LM217 pass the 80mA, and the others will not be able to?

 

[betwixt]

There is a good chance the current will not be shared equally and even more chance the minimum load current specification will not be met. Even if they all carried the same current it would need the battery to charge at 25mA just to make the regulators happy.

Why use five at 80mA each when you can use one at up to 1.5A ?

Are they LM217s or LM317s? They are essentially the same device but you mention both.

Brian.

 

[treez]

Are they LM217s or LM317s? They are essentially the same device but you mention both.

LM217..LM317 was a typo...sorry.

Why use five at 80mA each when you can use one at up to 1.5A ?

...because this is just a charger for use in the lab, and we dont want to bother with a heatsinked 1.5A device and hassle with heatsink paste etc......its easier to just solder 5 TO220's in parallel.........the Vin is 15V, and the battery could be down as low as 0V, so five in parallel each carrying 80mA means theres definetely no heatsink needed for any of the five LM217's.

We need this charger for the lab because the sales and apps guys are constantly coming to us for fully charged batteries for the demo kits, and we cannot find any off the shelf NiCd chargers for five D cells in series.
We will get about 10 of these built , and we can leave the batterys on them overnight and they will be charged fully in the morning...(its a C/10 charger).

Min current is no problem as C/10 is 400mA for us.

Why won't they share the current?
I thought paralleling current sources was always fine?

 

[BradtheRad]

You can use one regulator IC, then add external pass transistors in parallel (as many as necessary).
All the transistors are controlled by the one regulator IC.

http://www.bowdenshobbycircuits.info/page12.htm

See especially:

LM317T Voltage Regulator with Pass Transistor
and
High Current Regulated Supply

 

[godfreyl]

Why won't they share the current?
I thought paralleling current sources was always fine?

Yes, they should share the current equally. I don't see a problem.

 

[betwixt]

In theory they would but I'm not sure the variance between them and resistor tolerance would guarantee that. What concerns me is that if one or more regulators drops below 5mA load it will fail to meet the worst case minimum for those devices. 15V in less 1.3V in the series resistor, less the overhead of the LM217 (~1.75V @ 80mA), less the charged cell voltage (5 x 'D' at full charge ~ 8V) less D3/FET/D5 drop (say 2.5V) onlly leaves about 1.4V 'spare' in the system.

Personally, I would use something like an MC33063 with the current limit resistor set to the charging rate. It's simpler, cheaper, much smaller and would work happily over a much wider input voltage range. No heat sink should be needed.

Brian.

 

[crutschow]

In theory they would but I'm not sure the variance between them and resistor tolerance would guarantee that. What concerns me is that if one or more regulators drops below 5mA load it will fail to meet the worst case minimum for those devices. 15V in less 1.3V in the series resistor, less the overhead of the LM217 (~1.75V @ 80mA), less the charged cell voltage (5 x 'D' at full charge ~ 8V) less D3/FET/D5 drop (say 2.5V) onlly leaves about 1.4V 'spare' in the system.

...........................

So the difference in current would be equal to the resistor tolerance, which is certainly no problem. :wink:

Why would they drop below their minimum load? They are current regulators and, with a normal battery load, they will always deliver their set current, provided the input and output voltage differential is sufficient.

 

[treez]

I was aso wondering if the control loops inside each LM217 would interact with each other and produce instability?

 

[crutschow]

I was aso wondering if the control loops inside each LM217 would interact with each other and produce instability?

The control loop has a very low output impedance and the load also has a low impedance so I don't seem much of a signal path between regulators to cause any problems.

 

[betwixt]

@treez - provided sufficient filtering capacitance is provided they shouldn't interact.

@crutschow - you are almost certianly correct, but personally I wouldn't adopt this method for the reasons I gave. The LM117/217/317 in constant current mode, work by delivering a constant voltage across a fixed resistance but they also have a 5mA (worst case) minimum current requirement. In order to deliver that current they need to sink it at the ADJ pin end of the load which is common to all the parallel circuits. It is unlikely but not impossible that four of the five regulators could leave the fifth with insufficient current. I accept that in real life the method of failure would be to increase that regulators output voltage which in turn would tend to bring it back in line with the others.

Brian.

 

[FvM]

In order to deliver that current they need to sink it at the ADJ pin end of the load which is common to all the parallel circuits. It is unlikely but not impossible that four of the five regulators could leave the fifth with insufficient current. I accept that in real life the method of failure would be to increase that regulators output voltage which in turn would tend to bring it back in line with the others.

I think you are misunderstanding LM317 operation. The minimum output current doesn't flow through the ADJ pin. ADJ pin current is 100 µA maximum.

Each LM317 will source the programmed current of 80 mA independent of the other's operation.

A point that hasn't been yet considered is the linear operation of the overvoltage shutdown circuit, possibly causing several W power dissipation of the MOSFET.

 

[betwixt]

No, I understand how they work. I wasn't referring to the current through the ADJ pin itself but at the "ADJ pin end of the load". The 80mA has to flow through the load resistor and then go somewhere, in this case to the battery being charged. Fully charged 'good' cells while still under charging conditions may not leave enough voltage overhead in the regulators for them to work properly. If the 400mA current cannot be maintained because there is insufficient overhead it raises the question of whether whatever the current actually is, is shared equally across the regulators.

I have no doubt the circuit will work but the design leaves perilously small safety margin and as you point out may still get hot for other reasons. I would still suggest a small SMPS with output voltage set just above full battery voltage and current limit set to the desired charging rate. There are of course components designed specifically for chargers which could be considered. The original design in my opinion is overkill in complexity and cost and does not satisfy all electrical requirements under worst case conditions.

Brian.

 

[FvM]

Beyond regulated output current, the minimum output current will be the sum of minimum output currents of paralled LM317, that's all. There's no uncertainty involved.
If the LM317 voltage falls short of specified droput value, the intended current possibly won't be achieved. Not risky anyhow.

The original design in my opinion is overkill in complexity and cost

I completely agree. The original poster seems to have a soft spot for overcomplex designs.

 

[crutschow]

So what if one or more of the LM317's goes below it's minimum current rating due to the load voltage getting too high. It will just cease to regulate and shut off, reducing the total output current, but that's not a problem for the battery.