[USRP-users] 12V to 6V Converter for USRP Supply

Jeff Lambert jeffl at k1vzx.com
Sun Feb 17 20:31:46 EST 2013


You could reduce the dissipation by using some zener diodes to drop the 
voltage before the regulators.  And you can parallel regulators to 
provide higher output.  At the end of the day, a linear regulator is 
likely to be less noisy than a switch-based DC to DC converter.  More 
expensive does not mean better performance.  There is already a linear 
regulator on the board, so you really don't need the best power supply 
anyway, unless for some reason you need better noise performance than it 
already offers.

-Jeff


On 13/02/17 19:16, Radio User wrote:

> HHH wrote:
>
> > Would just find an adjustable regulator from ebay cost only several 
> USD more cost effective?
>
> Good question.    Answer: Not likely.
>
> The regulators I've seen on eBay are all based either on the LM317 or 
> the LT1764.
> These are linear regulators.
>
> 1. The LM317 is good to about 1.5A with careful attention to thermal 
> engineering.
> (This includes fiddly stuff like the kind of thermal compound, careful 
> application,
> the right surface finish on the heat sink, and the right fasteners and 
> torque.
> You'll no doubt find folks who say "it works fine for me, and I use 
> discarded
> bandaids and paperclips to hold it to the heat sink."  Good for them.)
>
> 2. The specified drain for the N200 and WBX together is about 2.6 
> amps.  That's
> 3 in round numbers.  3 > 1.5.
>
> 3. The LT1764 is a little beefier, but the thermal challenge remains. 
> Max spec'd
> output current is 3A.  Too close to the requirement for my tastes.  The
> low drop out is a nice feature (and the compelling one if your real 
> problem is
> a battery voltage that is close to the output requirement). But it is 
> still a
> linear regulator.
>
> 4. Linear regulators have three major problems in this application
>
>      I) Given a 12V input and a 6V output at (let's say) 2.5A, we'll 
> dissipate
>         6*2.5 = 15W in the series pass device (the regulator 
> package).  In still
>         air (no fan) with a really good contact to the heat sink, 
> that's going to produce
>         a temperature rise at the heat sink surface of T = 5000 deg C 
> / S  where S is the
>         exposed surface area of the heat sink in square centimeters.  
> (Source --
>         Reference Data for Radio Engineers converted from the constant 
> 55 deg C per watt per in^2.)
>         The sample I looked at seems to have about 40 cm^2 of heat 
> sink (charitable)
>         so the still air temp rise at 2.5A is going to be about 125C.  
> So this solution
>         requires a fan, where the temp rise is much much smaller than 
> 5K C/W/cm^2.
>         (And if you think I'm being conservative, note that real 
> models can predict
>         substantially higher temp rises for this load and heat sink. 
> 5K deg C/cm^2
>         may be quite optimistic...)
>
>         I don't want a fan.
>
>       II) At 2.5A a linear will waste one watt for every watt 
> delivered to the radio.  That's additional
>         drain on the batteries.  Compare this to the switcher 
> efficiency at 90 to 95% where
>         a switcher will waste 100mW or less for each watt to the radio.
>
>       III) Linear regulators tend to produce lots of broadband noise, 
> especially when
>         you push them. The decoupling on these supplies is not 
> adequate for a radio
>         application.  Assuming all the other problems could be 
> addressed, careful users
>         will want to add significant decoupling/bypass caps to the 
> output of any linear
>         regulator.  A bazillion uF cap is not sufficient, it ceases to 
> behave as a capacitor
>         well below frequencies of interest.  And most cheap power supply
>         manufacturers aren't using the highest quality caps in any case.
>
> So, yes there are inexpensive linear regulator solutions here.  They 
> require
> moving air past them to stay within thermal limits, they are running 
> close to (or over)
> their capacity, they may produce broadband noise that needs to be 
> knocked down,
> and they are only 50% efficient on a good day.
>
> Given the cost of a USRP, saving $30 or even $50 by buying a cheap 
> power supply
> seems like a false economy.
>
>
> _______________________________________________
> USRP-users mailing list
> USRP-users at lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20130217/13232d3c/attachment-0002.html>


More information about the USRP-users mailing list