voltage regulators in a parallel?

[georgeecollins]

I have a 7.4v 4000mAh (30C) Li PO battery that I am connecting to twelve large servos.  The servos are designed to run at 6v, so introduced a voltage regulator with a fuse to be safe with the servos.  The problem is the voltage regulator has a 1 amp fuse and it does not take long to blow.  Also the circuit gets quite hot.  I'm pretty sure I am drawing more than an amp, maybe 3-5 amps?  They are 25Kg/cm servos. 

Voltage regulators get a lot more expensive as they get to 5 amps or more.  Would it be possible to take two voltage regulators and wire then in parallel to a single battery and a single servo controller board (putting the servos on different controller boards isn't very practical).  Is that crazy?  Could I do it with three voltage regulators?

I am not an electrical engineer and any advice is very welcome.  Thank you. 

[Soeren]

Hi,

Voltage regulators get a lot more expensive as they get to 5 amps or more.  Would it be possible to take two voltage regulators and wire then in parallel to a single battery and a single servo controller board (putting the servos on different controller boards isn't very practical).  Is that crazy?  Could I do it with three voltage regulators?

Paralleling voltage regulators is not a good idea, as they'll never be balanced (even if you trim them carefully, temperature changes will ruin that). First one pops since it saw the majority of the load, then the remaining two compete about who goes first.

Boosting a voltage regulators current handling, can be done with a PNP power transistor and a resistor - and with a few more resistors and a smaller transistor, you can make it short circuit proof.

Since you don't really need regulation (just a voltage decrease), you can get by with something like a power transistor a diode and a resistor, to "eat" the ~1.4V.

Or use something like the Micrel MIC2975x LDO regulator that handles 7.5A continuous with a guaranteed drop of 600mV max.

Or use 2..3 power diodes in series - to get the voltage drop needed.


No matter the method, the transistor (or diodes) will need to dissipate the power as heat, so a heat sink is mandatory... 1.4V@5A means you have to get rid of 7W of heat. You may be able to heatsink it by the chassis, if aluminum (Copper or silver would work as well )

[georgeecollins]

Thank you so much Soren-- that is excellent help.  I have downloaded the .pdf for the MIC2975x (although it seems hard to find a vendor for that part) and I will look up power diodes.  This will be something to learn about.  Those servos are pulling a really big hunk or aluminum, so I have a possible heat sync. 

I really appreciate your help! 

[Admin]

Each of your 'large servos' is likely to draw an amp, depending on what they are doing. Stall currents are probably 2 or 3 amps per servo. I'd bet they'd even drain a constant 0.5A (in total) when all are set to idle.

Although you can put regulators in parallel, as Soeren said if they are not well balanced there will be an electrical shorting. For example, a 5.03V and a 5.02V regulator will have a 0.01V electrical short. That may or may not, at 12+ amps, cause something to fry. Better to make multiple power buses, with one regulator on each.

More importantly, regulators burn away your battery power as waste heat to reduce the voltage - serious inefficiency!

Your *best* bet is to use a 6V battery.

Or if not, get a switching regulator that can handle high currents. They'll give you much better efficiencies:
http://www.societyofrobots.com/robotforum/index.php?topic=11559.0

[Soeren]

Hi,

Or if not, get a switching regulator that can handle high currents. They'll give you much better efficiencies:

Would you settle for "maybe better efficiencies"? 

Dropping 1.4V at 5A is 7W lost.
Loads power 6V * 5A = 30W
Total 37W
Efficiency is thus 30*100 /37 = 81.1%
(And it will obviously be the same whether using an LDO, diodes or whatever)
 
Don't assume that a linear regulator is always grossly inefficient

I have seen switchers of much lower efficiency, although they're not the typical case, but neither is the high nineties.
A buck switcher is not at its peak efficiency with such small voltage drops, so it will not be "much" better and may even be worse, very much depending on design.

The best efficiency will likely be with a chopper with a fast switch, but its a lot of work for an absolute (unattainable) max of 18.9% increase in efficiency, compared to a couple of diodes.


The 6V battery makes more sense IMO

[Admin]

Yeap, it really depends on his situation.

If he uses a regulator, it'll likely be a 5V regulator. So with a 7.4V battery fully charged you can expect less than 65% efficiency.

Looking at the datasheets of the switching regulators you can see efficiencies near 90% (depending on various factors).


Of course, the elephant in the room is servo efficiency, which is less than 30% for cheap servos . . . 

[Soeren]

Hi,

If he uses a regulator, it'll likely be a 5V regulator. So with a 7.4V battery fully charged you can expect less than 65% efficiency.

The Micrel LDO I recommended (and use a lot myself), is adjustable just like an LM317 (although not exactly like it, the f(R_adj-gnd) is opposite the LM317, so lower resistance means higher voltage), so the 81.1% is the realistic target (just like with a couple of diodes).

Looking at the datasheets of the switching regulators you can see efficiencies near 90% (depending on various factors).

Yes and one of those factors are input-output differential and I doubt you can find a 5A switcher above 85% efficiency at 1.4V differential (without resorting to expensive exotics)

Plain chopping at eg. 5..10kHz with a 81% duty cycle and a low ESR cap following, would be both cheap and efficient in getting rid of the excess voltage.

Of course, the elephant in the room is servo efficiency, which is less than 30% for cheap servos . . . 

You mean power out/power in(?) and yes, neither the motors nor the gearboxes are the Rosetta Stone(s?) of efficiency in cheap servos (not in expensive ones either), but that's hard to change under the current revision of the the Laws of Nature - perhaps some lobbying may help

[georgeecollins]

I spent some time reading about this and I came to the same conclusions.  It seems like what I would be doing would be converting a lot of the batteries energy into heat no matter what I did.   The advantage of LiPO batteries is that you can draw a lot of current quickly.  Any 6v battery (NiMH or LiFE) will not put out as much current. 

It seems like there would be a big advantage in breaking out the servos into two banks, which is possible but more difficult to control because I do not have a card that can keep servos on a separate power supply, although I might be able to make one.  It seems like a further hassle to try and coordinate two servo controller cards. 

I really appreciate the help. 

[Admin]

Of course, the elephant in the room is servo efficiency, which is less than 30% for cheap servos . . . 

You mean power out/power in(?) and yes, neither the motors nor the gearboxes are the Rosetta Stone(s?) of efficiency in cheap servos (not in expensive ones either), but that's hard to change under the current revision of the the Laws of Nature - perhaps some lobbying may help

The low efficiency results from the combination gearing, the gear material/design/lubrication, the onboard electronics components, the motor design, the control algorithm, etc.

For example, I've seen servos heat up simply by sending a neutral 1.5ms pulse to them. They aren't doing anything at all, yet drains tons of current to do it. Some servos will vibrate and overheat if you send a signal that has a slight bit of electrical noise in it.

Nature of course requires inefficiency that's avoidable, you're right, but most servos are designed to be cheap . . . not efficient or reliable.