The usual ULBs have a run time of 30 days. Certainly in the case of AF447 that run time was not adequate. In the case of MH370 there was probably no satisfactory search area to even begin looking for an ULB.
While ULB technology may seem archaic - how often does the 30 day run time and short detection range really make any difference?
After AF447 there were proposals to require 90 days of run time instead of 30. Would that have helped find any aircraft lost at sea in the last 10 years?
https://www.easa.europa.eu/system/fi...%202012-16.pdf
With the searches of recent years I would have tried for the following modest improvements in ULBs:
Very accurately controlled frequency: Not 36.5-38.5 KHz but something more like 37.5 +/- 3 Hz. Easy to attain with current technology. And this would allow use of serious digital signal processing methods to extend the range by a factor of 5 to 10 (a guess without serious calculations)
Reduced on-time after the battery gets below something like 40% capacity
Further reduction at 20% with a goal of 6 months run time.
Frequency shift modulation might be nice - encoding the aircraft ID.
Easy to do with a very slow rate and very small deviation.
Dukane has been offering a 3 month run time ULB for quite awhile.
As a design engineer I could list 5 or 10 more wonderful features... but my view is that what we have now is very nearly adequate - if not satisfying all of our wishes for easy recover of the FDR and CVR.
We have come a long way from the FAA proposed specs of 1968 - which was when I was working on avionics! Among other things: 200 foot depth range
http://www.fire.tc.faa.gov/pdf/na68-7.pdf
edited ELB/ULB error, hat tip to Coagie!