The MW102 looks pretty good at the price. Note that spec accuracy is ±0.02 pH according to the data sheet and not the ± 0.01 listed on the website referenced in #3 (given that the thing apparently only resolves 1 mV which is 0.005 pH it would be hard pressed to come up with an accuracy of 0.01). Using standard buffers you aren't going to get accuracy of better than ± 0.02 anyway but that is accurate enough for brewing.
One of the nicest features in this meter is the stability determination capability i.e. as long as the reading is changing the indicator blinks to show you that it is not stable. In many meters you can adjust the stability threshold i.e. specify that the reading must not change by more than x mV in y seconds for the meter to declare the reading valid. This is a very nice feature (though a PITA if you are doing titrations) but can sometimes lead one astray. I always note the reading as soon as the meter declares it stable but then wait another minute or so and then initiate another reading. The electrode will continue to move towards equilibrium in the time between the readings and if the second is awfully close to the first you should wait even longer and then repeat a third time.
If you watched the calibration video on the link in #3 I'd advise doing things somewhat differently. First, you should rinse with DI water from a squirt bottle. Don't rinse the probe repeatedly in the same water and don't use tap water for rinsing. Second, after each rinse shake and preferrably blot the rinse water away from the electrode using a paper towel or something like. Don't touch the bulb directly with the paper towel - its sufficient to get the water off the electrode housing. Third, prepare fresh buffer, open a new packet each calibration or, if you have purchased bulk premade buffer, pour out a fresh bit for each calibration. Don't stick the electrode and RTD in the bottle and then do the same thing next time. Obviously the concern here is contamination and dilution. Even though you have, presumably, rinsed off most of buffer and sample and have done a good job shaking/blotting you haven't gotten rid of 100%.