Electronic Design

Bob's Mailbox

Any big trips to exotic spots planed this year? We’re headed for Newfoundland and Prince Edward Island for a change of pace. (I may go to Scotland in September. /rap) My question: Do you have some circuitry I could use for an electronic bagpipe simulator? It would need nine notes selected by removing fingers from some form of contact that would reasonably simulate a finger hole. I don’t need too many specifics, just a broad idea and maybe a part suggestion. I believe the reference frequency (low A) is around 46 Hz. The notes run from low G to high A on a Mixolydian scale. There are also three drones: two tenors tuned to low A and a bass tuned to an octive below low A. The intent here is to “pipe” the noise directly to earplugs so as not to annoy my dog, and of course my wife. I do have a practice chanter that is nowhere nearly as loud as the pipes, but it’s still loud enough to cause irritation to some members of the household. There is debate on whether the bagpipes create music or just organized noise.


I’ve heard that you can buy a “practice” bagpipe. Don’t any of your friends who have bagpipes have a simulator? I don’t know much about circuits for music simulators, or even noise. I’d hate to have to reinvent that wheel. Aren’t there any books on how to make synthesizers? Good luck.


What do you know about op-amp nonlinearity? (I just happen to know everything about op-amp linearity. I’ve almost finsished writing a 25-page app note on linearity, and I have measured dozens of good op amps. Will 0.1 ppm be good enough? I think the LM4562 will be about the best. Go to www.national.com/rap and look for the LM4562 datasheet. /rap) I have an application that requires extreme dc linearity, like sub-ppm. I need positive gain in the range of 4 to 10 with input between –0.4 and +0.4 V, multi- GO input impedance, and minimal noise with source resistances from 100 O to 1 kO or so and bandwidth from about 0.1 to 10 Hz. It seems that so-called “crossover distortion” (For many good op amps, the crossover distortion is quite negligible. /rap) may be a dominant nonlinearity and can be alleviated by drawing a constant current from the output using a CRD to the negative supply, or something like that. Are there other types of nonlinearity I should watch for, and how might I deal with them? Thanks! Keep up the great work.


Ask me another question after you read the app note.


I’d like to short out a resistor. Well, not exactly short it out, but reduce its resistance to the channel resistance of MOSFETs. The problem is one end of the resistor is at 5 V, and the other end can be as low as 0 V or as high as +5000 V. Is there a way to connect MOSFETs in series to reliably do this? There are FETs on the market with breakdowns as high as 1500 V. I can’t use a relay for a variety of reasons.


I am not an expert on this. I know that people with high-V switching to do sometimes stack up several high-V FETs and turn them on with photo-pulses with a photodetector at each gate. Call up the people who make the 1500-V FETs and ask them how to turn on a stack of four of them with simultaneous photo-pulses. They will know better than I do, as there may be some tricks.


Silly question: When I look at a noise spec in nV/Hz, do I use equivalent bandwidth times spec or square root of equivalent bandwidth?


It is not nV/Hz but nV/vHz. Let’s say you have an amplifier with 20 nV/vHz in the flat band. Let’s also say you have an audio bandwidth of 10 Hz to 20 kHz for the –3-dB points. If the rolloff is a smooth 6 dB per octave above 20 kHz (simple single break, not a lot more rolloffs), the noise bandwidth will be p/2 × 20 kHz, or 31,416 Hz. A good book on noise will remind you about that factor of p/2.

After you subtract the 10 Hz also × p/2 (about 16), the effective noise BW will be about 31,400 Hz. The sqrt (31,400) is about 177.4, and then you multiply that × 20 nV, which will be 3.55 µV rms, referred to input. So the output noise will be about (gain) times 3.5 µV. If the gain is +10, you would have 35 µV of output noise. That’s how you use the sqrt. (If you have a gain of –10, the noise gain will be 11, so you would have 39 µV at the output. This all assumes the resistors are low enough to not contribute to noise. RIN = 1k or lower will contribute less than 4 nV/vHz, which is negligible, as 4 + 20 = 20.8...)


Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.