# Brian park

's profile

## Brian park

's contributions
• 01.22.2018
• Teardown: A19 LED bulb
• What is the forward voltage of the LED array when it is running? Each package must have multiple junctions to run at such high voltage (forward voltage of LED array must be greater than highest DC voltage available). Is this lamp "world-wide power compatible"?
• 04.12.2007
• Real-world power tests model FPGA’s thermal characteristics
• Why not just measure the current going into the FPGA while it is operating "conventionally", and multiply that by the operating voltage to get power? The frequency to the FPGA could be varied to adjust the power to anything desired.
• 03.01.2007
• Low-cost current monitor tracks high dc currents
• You are injecting the oscillation onto the DC line! There is a simple solution, and you need not worry about loading of the oscillator. Put two cores in the circuit instead of one. On both are equal number of turns for the oscillator winding. The windings are connected in series and phased so that when the cores are stacked, the flux in the two cores go in opposite directions. Winding carrying the DC goes through both of the cores. Magnetic arrangement is similar to that used for magnetic amplifiers. If the cores are matched and the sense windings matched number of turns and phased properly, there is no AC coupling between the DC winding and the sense winding.
• 11.28.2017
• Cancel PWM DAC ripple with analog subtraction
• There is a simpler way! (following explanation is for 8-bit PWM). The secret is to manipulate the order of the (PWM period x 2^8) so the power output of the "PWM" is pushed higher in frequency. You do this by writing the PWM generator in software (routine in assembly requires 9-10 instructions per iteration and 256 byte table). This requires a 256 byte bit-reverse table. In software, you have 8-bit counter running at PWM frequency x 2^8). This value is put in index to read from table. This result is bitwise ANDed with 8-bit amplitude. If result is zero, you write zero to port, else write 1. This goes to conventional RC filter, but highest power is at PWM period x 128. (Each bit lower has frequency half, but also power 1/4th the previous). In comment below, some do this in FPGA. Why not put the "rate multiplier" circuit in FPGA?
• 12.23.1999
• Simple tester checks Christmas-tree lights
• All of these schemes require contact to the string, which is both troublesome and dangerous. I set an (analog) line-powered and line-grounded oscilloscope to the highest gain and touch the probe to either the lamp glass or the wire insulation. Rest of strategy is as you describe, except plug is reversed to analyse "other side of broken circuit. Unfortunately, this scheme (and the one you describe) does not work well for LED strings, especially rectified ones. If you use portable scope, you must ground it to get readings. Circuits can be built which are high-gain high-impedance (such as unterminated CMOS input) can be used instead.
• 08.21.2008
• It’s an electromagnetic-mechanical world
• De-gausser won't work. It will "solve" the problem for awhile, until the parts re-magnetize from the earth's or other field. Better is to have some strong magnets waved over to ferret out parts that will magnetize. Even the head itself can magnetize (this was persistent problem in tape recorders), but in this case, it is no problem.
• 04.27.2016
• Control an FPGA bus without using the processor
• About USB dongles on PCs: they are great except when they don't work, which is almost always! They fail to follow the PC protocol totally, and by Murphy's law, your application will demand it. Am example is the "break" code, which is staying in the "space" state for longer then 10 x baud period. Most dongles indicate "zero" for the byte so sent. Debuggers often use this code in the initialization as a preamble to the "security code" that must be sent. Go to the flea markets and scoop up those old PCs! Other dongles choose their baud rate seemingly randomly, including not even standard baud rates. Most often baud rate will be 2x or 1/2x rate set in PC (set by some random number table). This caused much consternations in expensive military program I worked on.