Distance of 78xx/79xx Voltage Regulator to Circuitry

[Salar]

To keep temperatures inside a closed cabinet low,
I think about mounting the voltage regulators on
the heatsinked enclosure. Distance to the circuitry will
be around 15 to 20cm, connected via twisted wire.
Any disadvantages to expect?
All the best, Salar

 

[AndrewT]

voltage regulators use feedback to improve accuracy.
The closer the feedback tappings to the circuit the better the regulation.

If you can mount the regulator ON the client circuit with zero mm long leads, then it will perform better.

 

[Salar]

Hi Andrew!
Thanks for the answer! But because Traces have the same length, would´nt this compensate feedback a bit?
But yes, resistance of a 20cm long wire (1sqmm) would be about 3,5 milliohms

 

[vacuphile]

It is inductance, not resistance which is your enemy here, and it only starts to play a role at higher frequencies.

By mounting appropriately sized decoupling caps close to the action part, the effect of this inductance can be negated. I wouldn't worry at all about 15 or 20 cm of twisted wire between the regulator and the circuit, given adequate decoupling.

 

[fabrice63]

In that matter , the closer is obviously the better ;-)

you'd better try to cool them down where they are , you've got plenty of solutions available ;-)

doing that you'll solve easily one problem , without creating many others ;-)




.

 

[Salar]

Having an old scope up to 20mHz, can I measure the effect? In which spectrum should I look for noise?

 

[martin clark]

The 78xx/79xx regulators have their internal gain rolled-off above only 1Khz or so. This means that the worrying about the added inductance of ten cm of wire has little effect on accuracy of regulation, because the regulator already has an intrinsic 2-3 microhenrys in effective output inductance (a side-effect of limited feedback bandwidth)

If you need to mount them a few cm away as described, just don't worry about it. Best-practice in decoupling, as post #4, is very much more important anyway.

 

[vacuphile]

Martin,

Actually, these devices are remarkably fast; I have seen schematics for how to use LM317's as audio amplifiers. Please have a look at tables 10 and 12.

http://www.ti.com/lit/ds/symlink/lm117.pdf

With the appropriate cap, Ro is well under .1 Ohm at all audio frequencies.

Vac

 

[jackinnj]

If you have a choice, use LM317/LM337 instead of 78xx/79xx -- by bypassing the adjust pin with 10uF and you can knock down the noise by an order of magnitude, improve ripple rejection.

 

[Sangram]

If your regulators are expected to run hot, it is either due to large voltage difference or large output current, or both.

A two-stage cascade regulator often works well to reduce the load across a single 78xx or LM3x7. These are reasonably durable devices and can handle 400-500mW continuous dissipation without any heatsink at all, in ambient temperatures of 50 degrees or so. This works out to a current draw of 150mA at 3V drop, which is the minimum for decent regulation.

If you need more current, a TO-220 unit is probably not the best choice without some form of external help or a heatsink. If your voltage headroom is higher, consider a pre-regulator. I've used series resistors when the load is predictable and constant to serve this need, but a second chip really improves PSRR.

I've also used ripple eaters (either series or shunt) close to the load and 3-terminal devices near the power supply when the DC voltage regulation is less of a consideration than noise levels. There are quite a few ways to skin this cat.

 

[Salar]

How would the schematics of a two-stage cascade regulator look like?

 

[jackinnj]

How would the schematics of a two-stage cascade regulator look like?

If you want 5V out -- use a 7809 first and a 7805 second. Wire in series.

If you put regulators in parallel you have to use balancing resistors.

Dig through the application notes on the Texas Inst, Linear Tech and ST websites. All will be revealed.