Hello
How do you measure such low value 20µV ripple, what is min. required to measure µV ripple?
Are digital oscilloscope possible, Volt/Div is typically higher than 500µV/div.
I saw a video on youtube, EEV and it didn't seem so easy, also active differential probe adds noise.
ps: the older ST STW12NK95Z; newer part substitution is: STW20N95K5
How do you measure such low value 20µV ripple, what is min. required to measure µV ripple?
Are digital oscilloscope possible, Volt/Div is typically higher than 500µV/div.
I saw a video on youtube, EEV and it didn't seem so easy, also active differential probe adds noise.
ps: the older ST STW12NK95Z; newer part substitution is: STW20N95K5
I use an HP/Agilent 34401A 6.5 digit bench top multimeter. Its bandwidth is 300 kHz, so the noise + ripple measured covers the audio band and beyond.
Not really... As you note, the resolution of an oscilloscope generally doesn't cover the µV range. A dynamic signal analyzer or audio spectrum analyzer would be a better option.
Very cool. Thank you.
~Tom
Tom,
Two questions before I "Add to cart".
Those kinds of questions are always welcome...
Yep. Sharing one regulator between two channels is fine as long as the total current draw is within the limits of the regulator. For example, I use one regulator to feed two channels of a stereo 300B amp. Works great.
I highly recommend that you use a thermal pad (SilPad for example) between the transistor package and the heat sink for two reasons:
1) The SilPad fills the voids due to the surface roughness of the heat sink and the transistor itself, thus, providing a better thermal connection between the two. This allows the heatsink to dissipate the heat more efficiently.
2) The drain of the MOS transistor is connected to the metal back side of the package. This means, the heatsink will be connected to the input voltage unless you use a thermal washer. This is a hazard from a serviceability standpoint.
If the heat sink is completely internal and there is no way to touch it from the outside, you can get away with connecting the transistor directly to the heatsink. In that case, you need to use thermal grease, though.
My advice: Spend the $0.50 (NOK 3.00?) for a TO-247 size silicone thermal pad. It's less mess than the thermal grease and then you don't have to worry (as much) about electrocuting yourself.
~Tom
Hello Tom,
Your regulator is really a TOP design, small compact and seems good tested.
Also the test with a light bulb is a very clever test, it's simple to test and a good indicator of a robust design.
Even the LT3080 seems better than the newer LT3081, look at ripple graps, the LT3080 is clearly more stable, look at the ripple rejection at 100k, 50dB for LT3080 and only 20dB for LT3081. Also note the different footprint. I wonder if LT can/want do better than LT3080.
Most companies are designing and creating products that will last for about 10yrs more or less and that's a shame.
I will order tomorrow
ps:
Thank you for your response,
the Agilent 34401A is really nice, Agilent even said MTBF=150000h approximately
Your regulator is really a TOP design, small compact and seems good tested.
Also the test with a light bulb is a very clever test, it's simple to test and a good indicator of a robust design.
Even the LT3080 seems better than the newer LT3081, look at ripple graps, the LT3080 is clearly more stable, look at the ripple rejection at 100k, 50dB for LT3080 and only 20dB for LT3081. Also note the different footprint. I wonder if LT can/want do better than LT3080.
Most companies are designing and creating products that will last for about 10yrs more or less and that's a shame.
I will order tomorrow
ps:
Thank you for your response,
the Agilent 34401A is really nice, Agilent even said MTBF=150000h approximately
Thank you. I do my best to ensure the success of those who build my circuits.
The light bulb load test actually has two layers of cleverness: First off, the inrush current of a cold light bulb tests the peak current capability of the regulator and the stability of the current limiter. In addition, the "wiggle the bulb" test causes some really nasty fast current pulses that can kill a poorly designed regulator.
Awesome. Thanks.
I enjoy using it. I was fortunate enough to pick up a pair of calibrated meters for about 1/3 of the new price during the economic downturn.
~Tom
