part of something everyone should do is test the quiescent current draw of your battery powered expensive tools, particularly ones off ebay, because the soft switch can be fucked up. A good soft switch should be around 5uA draw, slightly more with a RTC.
Any alkaline battery will get messed up if the quiescent draw is too high and you deep discharge it (and maybe even nimh too). Playing battery bingo will not save your expensive stuff if it has a defect in the soft start transistor. And some stuff is not engineered properly also.
Doing this simple test with a microamp meter can save you thousands if you like portable equipment. Once you get into the C/D batteries, its less important, but with 9V, AA, AAA.. its critical. Scopemeter on 4 C's is probobly OK, but fluke 289 or expensive LCR meter on AA's is gonna trash pretty much anything if its leaking. We are at greater risk because if ebay purchases.
The way to do it is to take a piece of paper and glue two conductors to it and stick it in the battery cavity between the battery and a contact and hook it up to a microamp compatible meter and test the draw when its off. Now microamps are easy to measure. Just make sure the meter turns on with the current meter in the circuit, because its easy to have it not connected.
Also, mechanical switches could be contaminated from previous battery leaks and present a leakage. Don't trust it because it has a mechanical switch.
And keep in mind how bad soft switches were, some people never learned, i.e. printers stealing energy. Imagine those engineers were tasked with making a soft switch for a multimeter.
And, you can also have a low impedance fault while the meter is turned on, which can easily hammer the battery by drawing 200mA to heat a transistor, whereas normally it should draw 20mA, but still be functional. Deadly to intermittent use 9V batteries, which are common on multimeters. Perfectly fine when you draw the expected 10mA, but when you turn it on, it ends up drawing half an amp unknown to you, it will cause the battery to explode after a few cycles.
But, since current is not measured, and you have it hooked up to some weird 9V meter (i.e. DE-5000) that you power on once a season, you end up killing the batteries WHILE THE METER IS ON and it looks like the battery died because 'it sucks', because you drew a heavy load from a 9V for 30 minutes a month because of a undetectable internal fault (particularly with heavily insulated rubber coated meters where the heat will take forever to escape and will be unnoticeable because you will be long done with the measurement and put the meter into an insulated pouch when you are done measuring the odd inductor). Deciding to whack in 18650's to power a 10mA load is NOT the correct engineering solution, you are potentially hiding a problem!! It's similar to a bad AC repair job, where you do not find the leak or purge the system from moisture or follow proper procedures, if you ever worked on a car air conditioner.
Restoring leaked equipment is a difficult process, you need a ultrasonic to do a good job IMO, + I recommend dremel rotary brush tools, vinegar, Distilled water wash, etc. And you typically need to replate contacts. There is a good chance if you did a 'scrub the contact with a screw driver' repair job on a leaked battery that there is a impedance shift in the unit because of corrosive mist, vapors, etc. And you can have a humidity and temperature dependent leakage problem once the circuit is soiled. You can't trust your eyes to do this repair, it needs to be measured for leakage. Thankfully its simple once you understand the procedure.
I don't like how people want to shift the industry to change battery chemistry and capacity and force changes to accommodate for broken equipment and poor quiescent current and unnecessarily high current draw designs. Wrong tool for the wrong job and being and idiot IMO, in the end you are accommodating for bad efficiency. Do we need the same problem as MS-WORD in the micropower electronics world? Where you need a 2GHz processor to run a word processor program?? It's not M$ windows! Think high efficiency ASM code, not high level inefficient crap thats poorly coded. Lazy and dumb engineers will use higher capacity insensitive batteries and rechargeable batteries to cover their tracks. Eventually you will need a nuclear reactor and a liquid cooling system to run a 3.5 digit meter.
I would like to see people measure the draw of the devices that are making everyone start to hate primary cells. I don't want a pouch cell soldered into my DMM. I want to be able to go to the store and buy a pack of batteries in 5 minutes if the need arises and forget about it. The art of electronics 3rd edition has a chapter on soft start switches everyone needs to read.
Fluke 187 - 9V battery. Fluke 289 - 6AA cells. WTF. I am guessing at this rate the future fluke 387 will have a 30Wh LIPO in there.
Also, try setting a primary lithium on fire some time, cut it open and spit in it. It burns like a road flare. I don't like this. It is not necessary to put that in a fucking wall clock.