Looking for the bad RAM chip(s).

This is a common diagnostic problem for anyone dealing with fixing a board with bad RAM beep code. WHICH RAM chip?!! They are all soldered to the board, they can be very difficult to source and scary and expensive to replace. I have devised this method to hopefully assist us in this most terrible of times. A near-field / EMC probe is used with a spectrum analyser that most of us small repair shop techs don’t have the money for – in proofing new development pre-production electronics for FCC compliance. Some crafty hobbyists / engineers / techs in the past HAVE also used the probe and spectrum analyzer to identify faulty / noisy components. I made my own probe. When I saw that the rtl-sdr Airspy worked so well and spectrumspy worked VERY well with it to make a poor-man’s spectrum analyzer…I put two and two together! A great diagnostic tool (hopefully!) is now within the reach of the rest of us! (I’m poor, so this is exciting to me.)

 

Many RAM chips on a MacBook board. One or more of them are presumably the cause of this MacBook having a “RAM beep code”. Verified all RAM chips have power and signal paths. YES. Check. Bad RAM? Which one? How best, to identify the culprit? I found a lot of 5 replacements for $25. They lookup in their datasheet to be 1600MHz. If they are powered-up properly – and data is doing stuff in them: that data read / write should happen at this frequency. They are made of silicon, plastic, copper. At this frequency – they should behave like little radio stations broadcasting a signal for all to hear. With the right ears.

Near-field probe, 30dB pre-amp, rtl-sdr AirSpy w/ SpyVerter, SpectrumSpy. (An EMC / RFI testing set-up, poor-man’s spectrum Analyzer). My spectrum analyzer tops out at 1.8GHz. I could add a down-converter head-end to work higher frequencies, if need be.

All these components available from Airspy, aliexpress, ebay. – Time for them all to be put to the test.

2nd-order harmonics, resonant frequencies, birdies from all circuitry of Airspy, pre-amp, from my lab equipment’s strong RFI noise, FLUORESCENT LIGHTING: this picture is going to be busy! See if you can spot the RAM data signal…(This is what we find at 1600MHz).

And here’s half-wavelength at 800MHz. If I move the probe an inch or so away from the RAM chip: the signal goes away. Just firing up the board will not guarantee that ALL RAM address spaces have data going to or from them. I plan on feeding proper voltage to RAM chips and pumping a copy of signal data into each section of each chip with my arbitrary waveform generator to test across ALL address spaces equally, controlled. ALL chips DO show activity – intensity varies – shows lower, not as dark of a graph color shift on a couple of them. This may be the bad chips. It may also just show address spaces not used at that time. Beep could be from bad controller, bad DDR3 power supply filtering, brown-out on that supply, a bad resistor on a feedback, all manner of possible candidates! I treat the results as information, data to feed into the equation as yet another “given”, or “known”. Given enough “knowns” – you can deduce and solve for “X”. Beats guessing or not caring and blindly replacing all 16 chips.