UBM Tech
UBM Tech

Samuel Kerem

's profile
Senior Staff Member

Biography has not been added


's contributions
  • 05.08.2013
  • Connect a 4×3 matrix keyboard to a microcontroller using two I/O pins
  • For the circuit: to misbehave,- the resistors must drift 49% to produce erroneous result. A penny resistor is guaranteed 1% over temp and life time. (if you think that 555 can drift more than couple %% @ temp, time, Vcc combined--it is a new chapter in electronics)
  • 06.20.2012
  • Adjust power-efficient LED switch to any light intensity
  • First: CD4011 is 4xNAND gate and CD4001 is 4xNOR, but 4xAND is shown here for all 3 choices. Now let’s consider “no light” situation. BC459 and BC177 are off because there is no base current. What we have now is floating CMOS input. This situation is no-no for any design that is considered to be reliable. BC177 will look just as diode with anode connected through 90M-100M to +5V (look at the emitter direction). If you think there is no current going through the gate input – that is wrong, there is always current – it can be very small, but the data sheet (i.e. for TI part shows it can reach +/-0.1uA even @ room t. The input current as low as 0.05uA will make 5V across the resistors changing your assumed logical 1 to logical 0. Any CMOS with floating input is extremely sensitive to any noise or gate charging current and sooner or later starts to oscillate. Your only hope is BJTs leakage current , but it is difficult to imagine a reliable circuit which relies on stray parameters. You are driving LED from 5V w/o any limiting resistor. The habit to rely on internal resistance of CD4011 is the bad one, again this parameter is not controlled and you only ASSUME (probably based on your experience) that it is safe to do that. If you are striving for MINIMAL parts count, then you don’t even need 10K resistor – it is difficult to envision the situation when a REVERSE current through illuminated LED reaches the dangerous for BC459 level (many 100s of mA). Not like I doubt that such illumination can be achieved, but the green LED will be destroyed before BC459 even feels slightest heat. This circuit can be better reproduced with a single comparator (a very inaccurate one is shown here built on 4 gates and 2 BJTs) and will be smaller, cheaper and more reliable. even for such approach it is not innovative at all.
  • 04.21.2011
  • Offline supply drives LEDs
  • Let's talk about safety. The input is 110V or 220V AC. The Q2 is off when input voltage is over ~40V. At that moment the load (LED chain) becomes floating, electric potential at any point in that chain FOLLOWS the input. The author is only vaguely right in the last paragraph,- the more appropriate statement is:-"you will be "DIRECTLY connected to the LIVE electrical line",- if it is 220V-- you will get 220 (minus diodes drop). And it is only part of the problem. A touch in vicinity of R5, C2, Q2 will create the charging path for C2 (63V rating) to charge to 110V or 220V and then to blast spectacularly,- or at least it will be the interesting race who can hold on longer to/in this circuit: you or the capacitor.
  • 09.05.2012
  • Reconstruct the input current in a grounded-impedance current sensor
  • Wow, this DI just re-invented Kirchhoff's circuit law. It also teaches: if you have 2 circuits and your goal to have I(t)_1 x Z(t)_1 = I(t)_2 x Z(t)_2 ( I(t) and Z(t) are time (or frequency) dependent current and impedance) and you have managed to make I(t)_1 = I(t)_2, - you'd better have Z(t)_1 = Z(t)_2 also (or vice versa).
  • 08.17.2012
  • Circuit automatically switches off DMM
  • Really? This circuit shows the most basic "common source FET configuration" and it did cut into the "Design idea" section? ? If I wanted to be a prick I would claim such inaccuracies, in this idea, coming from the variation in gate threshold voltage for 2n7000 which is 0.8V to 3V @1mA – yes fairly fixed for one particular FET at fixed temperature but varies greatly from a sample to a sample and temp. That variation gives the “On- time” from ~33sec to ~63sec for 9.5V fresh battery. If I want to be a prick I would claim the “On- time” depends on the battery condition (yes-that is mentioned in the article), if the battery discharges to still usable7.5V the above time range will shift to ~25sec to ~55 sec (the numbers are approximate just to show the trend). If I wanted to be a prick I would even claim the capacitor, which is shown as polarized, and as polarized i.e. tantalum (or worse -- Aluminum) may (worst case) have "significant" ~1+uA leakage which gives an equivalent parallel resistor in vicinity of 2.7Meg shown. If I wanted to be a prick i would ask the benefit of the circuit, as the majority of even very cheap DMMs have internal shut off mechanism often based not on total time as in this idea, but on the absence of activity, and you dont need to push another button and to remember to push it again f you need to make few sequential measurements. But I dont want to be a prick, I just want to know what is the "creativity threshold" set by EDN to assume that an idea is worthy of publication
  • 05.26.2011
  • Logic gates form high-impedance voltmeter
  • It is very well known that a transfer characteristic of MOSFET varies greatly from sample to sample. That is why CMOS input logical level is defined as below 30% of VCC is 0 and above 70% of VCC as '1". These numbers can vary slightly, but the significant "gray area" (in my example 40% of VCc) does exist where the output is not guaranteed. It is also well known fact that logic gate will oscillate when "slow" signal (that is why there is requirement in DS for rise/fall) pass the mention above "gray area". That is why Schmidt trigger is used -- to prevent an oscillation during slow transition. (CD4011 shown in the article is a regular gate not Schmidt). The “0 to 1” threshold is not only IC to IC dependent but also varies greatly with temperature (input current temp dependent as well). It looks the author never heard about a comparator and a reference
  • 05.10.2012
  • Power-supply decoupler protects your UUT
  • What is this idea is promoting? That it is possible to build a rudimentary voltage comparator (based on Q1) which is neither accurate and temperature stable?