I've been throwing varieties of 7805s at voltage regulation designs for so long that I've gotten behind the curve and would like to know what is best for 21st voltage regulation.
Requirements - cheap, through hole that fits on breadboards for prototyping, cheap, is more efficient than the 7805, and is cheap.
Ideas?
Smiley
I like the PT5101 switchers. Same 3 pins. Dont get hot. But they cost something.
I do not know which application you use, but for long time those were very good. For low voltage applications I personally use LM1117.
For generic, not too high load current, non-LDO, applications that can tolerate a few mA of device current, a 780x is still fine.
I tend to shift away if the load current is under 10mA, if the dissipated power would be over roughly 200mW, or the minimum input-output differential is under about 3V.
What I shift over to depends on which of those thresholds was crossed.
Jim
I like the PT5101 switchers. Same 3 pins. Dont get hot. But they cost something.
Is cost important? :twisted:
As hinted at above, "it depends".
If you are creating e.g. 5V from a 9V battery and aren't concerned with LDO and modest current draw--then any cheapest would probably work well.
But if you are starting from a raw +24V or raw 36V, then you are dumping a lot of heat. Those heating considerations and solutions (bigger track; heat sinks; ...) might outweigh the cost savings for the regulator itself.
A quick trip to DigiKey and wading through the many thousands of "hits", the lowest-cost shown appears to be Diodes Incorporated AS78L05Z family. $0.40 qty. 1; $0.20 qty. 100; less than $0.10 in full-reel. 100mA; up to 30V raw.
I have no idea how that might compare to what you typically use.
I like LM2930 for automotive applications.
P.S.
I was digging in my parts today, I should have said LM2937. It has better specs for temoerature and load dunp. I use it, not the LM2930, could only remember 3 numbers, and made an error. I am discovering old age.
If you do not need a LDO solution then I like the lm317, you can adjust the voltage with 2 resistors (pro is that you can set the output voltage precisely, con is additional parts) the feedback pin is exposed so you can externally decouple it for better noise performance. It is dirt cheap, and widely available from multiple sources, it is also quite a versatile chip so you can use it for things other than voltage regulation.
For switching, I like the microchip stuff, but they are only good for about 500mA.
Anyways if the 7805 is working well for you then really there is no reason to switch.
I like LM2930 for automotive applications.
Perhaps I am wrong but the LM2930 is not automotive qualified and only rated for -40C to 85C.
I think I still have a couple LM309s in a junk drawer
The largest known prime number: 282589933-1
It's easy to stop breaking the 10th commandment! Break the 8th instead. 
I also prefer LM317. Especially the TO-92 are nice for breadboarding. The voltage drop is about 1.7V at 100mA when hot.
If you need lower drop then there is xx2950/51 which has it at ~380mV.
Depending on the manufacturer those come with thermal shutdown and over-current protection.
6ee98
I have used SR10S05 but again not cheap. there are 0.5 Amps versions aswell.
APP VIN range and typical Load Current ?
well compared to LM78XX NOT cheap but very reliable and a drop in replacement for most apps...been using Recom by the bucket load to replace almost all previous LM78xx apps with >9 vin to 5 or 3.3 vout and 100ma to 500ma current and especially those that required a heat sink which you can eliminate
Recom even started shipping a lower cost version of their previous version!
http://www.digikey.com/product-detail/en/R-78E5.0-0.5/945-1648-5-ND/2834904
There is another part number for >500ma.
murata also has started to try to compete with Recom
Have a box of NOS PT5101 gathering dust since the Recom replaced them some time ago.
That's a cute part and about half the price of the one I was looking at.
Curious: The price break is at 42 pieces.
The largest known prime number: 282589933-1
It's easy to stop breaking the 10th commandment! Break the 8th instead.
2N3055 and a zener....
Seriously, as has been said, it depends. I've seen a few circuits with two regulators, one for when the micro is in power down or sleep mode, which is low current, but high efficiency, and one that is high current and not so efficient, switched on by the micro. In fact, the project I'm currently programming for has just such an arrangement, I haven't really studied how it's all going to hang together yet.
I have huge fun with regulators; I frequently need to get 3.3v from 5v at low current with very low standby current.
Most regulators include high-temperature protection which is of limited use for me since I need to work up to 150C; often the protection includes hysteresis which is even more annoying since it won't turn the thing on again until the temperature drops ten or fifteen degrees... and down the hole, that just ain't gonna happen! Also, the protection temperature is often not well specified.
I have found though that the LM2936 is rather good; the thermal protect has no hysteresis and I haven't found a chip where it comes in below 155C, usually 160C. Operating current is a few microamps down the middle leg and it's happy with 1.7v across it, up to about 25mA - I rarely use more than 3mA.
Curious: The price break is at 42 pieces.
which is of limited use for me since I need to work up to 150C;
2N3055 and a zener....
often the protection includes hysteresis
since it won't turn the thing on again until the temperature drops ten or fifteen degrees...
Also, the protection temperature is often not well specified.
Requirements - cheap, through hole that fits on breadboards for prototyping, cheap, is more efficient than the 7805, and is cheap.
If you go to Digikey, and sort linear regulators 100mA-1.5A by price (500+, removes 1+ variations) you find the biggest separator is Vin.
eg if 6V Vin is ok, buys you a cheaper part.
Also beware that Lower Iq regulators, often have quite strict load Cap specs. (uF and mOhms caveats )
The 7805 series shows up on first page as L78M05CDT-TR
More expensive, and ok to 100-150mA is LP2951, which includes a PGOOD output, and has 30V rating.
This does need higher decoupling caps.
just today I noticed on the front page of digikey this device
R-78E5.0-0.5 and it's "only" $2.5 and you save the heat sink. but max 28V in and 0.5A
By using a switching regulator like MC34063 or LM2574 Family and a few passive components you can save much cost. The solution is not as small as 7805, but very efficient and very easy to built.
By using a switching regulator like MC34063
Forget about tight current loops or shielding on a breadboard. And prepare for the fact everything is jumping around, multimeter shows same values on AC and DC range..
So, IMHO for analog the only viable solution on a breadboard is to use a linear regulator powered with a battery. Alternatively an SMPS power supply + some nice passive LC filters.
but very efficient
The problem is that an SMPS is totally unsuitable for analog circuits.
If a switching power is well designed, it makes not much more noise than the linear. But you are right: it should not be built on a breadboard. Just built a pcb in a size that can be used as an adapter for your breadboard, perhaps some LC-filtering in the Output.
Analog? Even digital circuits are internally built from analog components :)
I have seen pretty strict power supply noise requirements with chips receiving or transmitting 10Gbit/s video links over 100m length. Other video transmission chip datasheet suggested a linear regulator with LC filter as the power source so noise at 1MHz is virtually zero, but it is not simple as the chips eat multiple power nodes with multiple voltages and current requirement can easily exceed 500mA per supply voltage, so basically it was required to drive an LDO with switching power supply, and arrange correct power turn on sequencing. All the extra LDOs and coils eat up a lot of PCB area.
Of course this is far from sensitive analog circuits.
With sensitive analog I follow the switcher with some bulk capacitance and a linear LDO. And pause the switcher during measurements. Quiet as a mouse.
If the switcher is without a shutdown pin, a diode to pullup the FB pin works as shutdown, just be careful with layout.
???I have made VHF receivers
If a switching power is well designed, it makes not much more noise than the linear.
I have seen pretty strict power supply noise requirements with chips receiving or transmitting 10Gbit/s video links over 100m length.
Guys, it is obvious one can make a ripple arbitrarily small with any SMPS and there is no point discussing if 1uV is possible or not.
But the topic is not SMPS design but, quoting OP:
Requirements - cheap, through hole that fits on breadboards for prototyping, cheap, is more efficient than the 7805, and is cheap.
IMHO, as I said earlier, the only justification of an SMPS (be it integrated or discrete) in such breadboarding circumstances is the 4A blinking LED project.
With sensitive analog I follow the switcher with some bulk capacitance and a linear LDO.
