linear voltage regulators in parallel

[hul]

under what conditions can linear voltage regalators be put in parallel? Small resistor on each output per the attached diagram? How small?

Thanks, Hans

[Chippie]

The small current share resistors need to be around 0.1 to 0.5 ohm in value, 0.2 ohm is shown in your picture...The wattage is dependant on load current, 1 watt rated should suffice.

[lazy-b]

a New Linear Voltage Regulator with Low-Voltage Drop with same Pin Configuration and same Package as LM 317 is now available its 5 Amps with 7.5 Amps Peak the part no. is LX8383.

Been using this with my SJM Helicopter, cause the Built-in Voltage Regulator of Brushless ESC alway over-heated.......it works flawlessly.

just check the google for its data sheets.

[hul]

what happens if the load sharing resistors are too small? One reg runs into the current or power dissipation limit and the next one controls voltage?

Thanks, Hans

[Hell-e-Guy]

what happens if the load sharing resistors are too small? One reg runs into the current or power dissipation limit and the next one controls voltage?

Thanks, Hans

This Application note will best explain the concept of load sharing resistors:

http://focus.ti.com.cn/cn/lit/an/slva250/slva250.pdf

Hell-e-Guy

[hul]

I understand the concept.
Most linear regulators are pretty much fool proof, current limited, overtemperature protected etc. What happens if the resistors are too small? Any disadvantage other than the load not shared? Instability?

Thanks, Hans

[rich smith]

under what conditions can linear voltage regalators be put in parallel? Small resistor on each output per the attached diagram? How small?
Thanks, Hans

No resistors required at all. Aluminum runs, bonding wires. and junction resistance will provide all the "equalization" required.

PS Look at the back of virtually every low cost ESC on the market.

[Dan Baldwin]

I understand the concept.
Most linear regulators are pretty much fool proof, current limited, overtemperature protected etc. What happens if the resistors are too small? Any disadvantage other than the load not shared? Instability?

Thanks, Hans
The reason for the load sharing resistors is that there is some variation in voltage from regulator to regulator. The higher voltage regulator will will end up taking most or all of the load. If the regulators are tight tolerance, or if they are hand matched, very little resistance is needed for load matching.

Dan

[hul]

they won't start hunting if their output voltage is close?

I don't really mind if one regulator does all the work and the others are just hanging around waiting for the occasional load spike. Can I leave the resistors off with no ill effects in this case?

Thanks, Hans

[rich smith]

they won't start hunting if their output voltage is close?
I don't really mind if one regulator does all the work and the others are just hanging around waiting for the occasional load spike. Can I leave the resistors off with no ill effects in this case?
Thanks, Hans

Theory is fun but in the real world parallel works fine. Initially one or two may get warmer than others but none get really hot as current increases. Again I suggest you look at the back of a regular ESC. No resistors.

[lazy-b]

Guys, I've seen a lot of Commercially made products use Parallel......it look like it will Eventually equalize, cause as Semiconductor heats up, the Voltage Drop increase, therefore some of the current will be diverted to the other Transistor connected in Parallel.

[darkith]

IMHO, I'd think you'd be fine with no load balancing resistors (or using the minimal resistance in traces/connections) as long as the load doesn't spike so quickly that the Vreg bearing the brunt of the load shuts down before any idling Vregs can ramp up (causing cascading failure).

Similarly, I don't know why everyone recommends pulling the positive pin on one ESC when putting multiple identical ESCs in an multi-engine plane. I'd think that the resistance of the servo connection would be sufficient to cause some degree of load balancing, and it would be better to have the redundancy of multiple BECs rather than just one BEC.

D.

[rich smith]

IMHO, I'd think you'd be fine with no load balancing resistors.

Similarly, I don't know why everyone recommends pulling the positive pin on one ESC when putting multiple identical ESCs in an multi-engine plane.

Same reason they build steel and concrete bunkers in the back yard to charge lipos. Mindless internet urban myth. More entertaining than American Idol. :)

[Chippie]

Maybe the small value resistors are there because of good engineering practice- as recommended by the manufactures, but the fact that the resistors are missing in speedos is down to economics...?

I agree though about pulling the Red lead...waste of time imho!

[Dan Baldwin]

they won't start hunting if their output voltage is close?

I don't really mind if one regulator does all the work and the others are just hanging around waiting for the occasional load spike. Can I leave the resistors off with no ill effects in this case?

Thanks, Hans

I haven't tried running multiple regulators in parallel, but I don't think that they would tend to "hunt".

The resistors only help to make sure that the load is fairly evenly shared. If you don't mind that one regulator is taking most or all of the load, and runs hotter than the rest, then the resistors are not needed. On an LM317 the output voltage does drop as the junction temperature rises above 25C, so as long as the voltage difference isn't extreme, they would end up sharing the load eventually.

Dan

[Tomapowa]

Directly from Linear.com:

Re: Linear regulators:

"Can I parallel two LM78xx linear regulators?
Paralleling of 3-terminal regulators is generally not recommended because the devices will not share current equally. If, for instance, you try to make a 2 Amp regulator using two LM7812s or two LM7815s, the device with the higher output voltage could be carrying more load than the other. Or even worse, the second regulator may be totally off. The reliability of such a system is poor because of the combination of high temperature and high current in the first regulator.

A simple way to improve sharing is to insert a low value resistor (about 0.1 Ω ) in series with each output. The problem with this approach is that load regulation is poor, because the voltage drop across each resistor will vary as the load current varies, in this case 0.1V error for full load of 1A per regulator.

A better solution is to use either a linear regulator with a sufficiently high current limit, or a linear regulator controller with an appropriate pass transistor. For this example, the LM1085 quasi-lowdropout (QLDO) linear regulator would be a good integrated solution, or the LP2975 controller could be used for a more customized design. See the online design environment of the Power WEBENCH for recommended regulators to meet your system specifications. "


Re: Switching regulators:

"Can two of the LM2621 switching regulators be put in parallel to boost the output current?
It is easy to parallel two of the LM2621. All that is needed is to tie the two outputs together. They will most likely not have the exact same bandgap and the resistors used will not be exactly equal, but whichever one ends up with a slightly higher output voltage will pump current more often. When it is at its current limit and the output starts dropping slightly, the other one will start switching to supply the extra current needed. This will cause many strange frequencies on the output. Therefore an LC filter may be a good idea if noise becomes an issue.

If paralleling two LM2621s, a current sharing resistor could be used, but is not needed. If connected to the right resistor, it could divide the supply current more equally, but because we could never know which will have a slightly higher bandgap or hysteresis window, adding a resistor to the output may not help and therefore is not required. Consequently, one of the ICs will do most of the work at lower load currents, which is okay."

[Hell-e-Guy]

Same reason they build steel and concrete bunkers in the back yard to charge lipos. Mindless internet urban myth. More entertaining than American Idol. :)

I guess you haven't had the fun of a LiPo fire inside your home... :p

You can parallel two Regs without resistors, it all depends on the quantity you manufacture and the environment you work in.
When you build one or two circuits at home you have good control on the end product and are able to check 100% of your circuit.
When you mass produce at great quantities than you start to really feel the tolerance between silicone batches.

Take out most decoupling capacitors on your circuit it will probably work the same, but put it once in a noisier environment and you start to see the difference.

Hell-e-Guy

[rich smith]

Actually I was hired to intentionally "abuse" hundreds of lipos and did manage to get a couple to flame. Far more difficult than most here think but it can happen. And it WAS huge fun.

As far as parallel regs my company has shipped hundreds and Turnigy, Hextronics, etc. have shipped millions without a single failure traced to that practice. At the risk of sounding like a stuck record I suggest you look at the back of any ESC. Some parallel 8 or 16. No resistors.

Silicon tolerance and bypass caps have nothing to do with it. Contrary to Tomapowas mfg links they do not have to be closely matched. The real issue is regulators do not sink current, they source only. As mentioned above one or two may get warm at first but as current increases they quickly equalize and share the load.

I guess you haven't had the fun of a LiPo fire inside your home... :p

You can parallel two Regs without resistors, it all depends on the quantity you manufacture and the environment you work in.
When you build one or two circuits at home you have good control on the end product and are able to check 100% of your circuit.
When you mass produce at great quantities than you start to really feel the tolerance between silicone batches.

Take out most decoupling capacitors on your circuit it will probably work the same, but put it once in a noisier environment and you start to see the difference.

Hell-e-Guy

[jeffs555]

As far as parallel regs my company has shipped hundreds and Turnigy, Hextronics, etc. have shipped millions without a single failure traced to that practice. At the risk of sounding like a stuck record I suggest you look at the back of any ESC. Some parallel 8 or 16. No resistors.

Sixteen linear regulators in parallel? Haven't seen any ESC like that. Which ESC are you talking about?

[rich smith]

Sixteen linear regulators in parallel? Haven't seen any ESC like that. Which ESC are you talking about?

I have two of them from few years back. I think they said HWE but not sure who made them or capacity. Shrink is gone. At first I thought FETs but after removing heat sink saw they were 7805. There's a jumper on the end and I have no idea what that's for or how to program. Very non-linear throttle and crummy controller in general so gathering dust.