P2000

general recent conversations

Discussion on question by 0..: How to…

Imported from a comment discussion on diy.stackexchange.com/qu...

P2000

Jan 12, 2022 18:39

Try electrical engineering (EE). Your question is in my professional field of expertise, but you really have to be more concrete about what you are trying to accomplish. At EE we don't need your drywalling skills, but we do need a lot more context. "I have a ... and I am trying to ... How should I ... so that ... " is a good start.

P2000

Jan 12, 2022 18:39

@Tetsujin I have 37 years experience with force fields and flux capacitors, and once I know what the OP actually needs, I'll know for sure its neither of the two...

P2000

Jan 12, 2022 18:39

So this is not a DIY question for a specific room in or around your home, but a "world building" question. Otherwise, please specify the exact room, its dimensions, doors/windows, how much space you have/need, activity in the room, the types of sound to be kept out/in, your budget and level of skill with drywall/framing etc.. and other constraints. Pictures and drawings would be great too.

Discussion on question by Vivek Mahar…

Imported from a comment discussion on diy.stackexchange.com/qu...

P2000

Dec 21, 2021 15:00

This question has two parts, and both are missing critical details. For a physical answer you'd have to provide a sketch and perhaps some pictures. Also highlight any (shared) utility entry points. For a legal answer you'd have to specify your jurisdiction, and supply pertinent sections of your deed / land registry. In any case, as interesting as it is, it cannot be answered here as asked.

Discussion on question by Fritz: Why …

Imported from a comment discussion on diy.stackexchange.com/qu...

P2000

Nov 19, 2021 14:35

I actually don't think citizenship matters.

Discussion on answer by motosubatsu: …

Imported from a comment discussion on workplace.stackexchange....

P2000

Oct 16, 2021 11:55

@musefan there's a difference between what my standards are for my HR department (or accounting etc..) and how I believe it reflects on my company, compared to what I am willing to put up with when dealing with support departments at another company, whether I do business with them, apply for work there, work for them...

P2000

Oct 16, 2021 11:55

"to resolve conflicts and other employment issues" (musefan)... you're lucky if that's the kind of HR they have at your workplace. There are also places where they are effectively not more than the executive branch of corporate policy. As for the above answer, yes, one experience like the OP's during an administrative interaction indeed says little about the company, the team with the opening, or HR.

Discussion on answer by P2000: Why do…

Imported from a comment discussion on electronics.stackexchang...

P2000

Sep 13, 2021 05:03

@hddh Your point "You don't need a balanced pair to use differential signalling to reject common-mode interference" is not the whole story and it's not 100% correct, but you could ask about in a separate question.

P2000

Sep 13, 2021 05:03

The issue with unbalanced lines is when the shield is the ground, like coax, the shield will form the loop and pollute the signal & amp ground. When there is a separate signal ground, like a shielded twisted pair, the issue is the loop polluting the shield and the pollution coupling into the pair. If the pair is balanced and differential, the common mode pollution can be rejected. If the pair is single ended you have no recourse. The shield must be grounded somewhere to avoid static charge buildup and ensuing high voltage discharge. BTW shields don't do much below 2MHz.

P2000

Sep 13, 2021 05:03

@hddh differential receivers still need a quiet power source and ground plane, and that could be disturbed by noise on the shield. Also, any imbalance of coupling between shield and the two signal lines will still provide a differential noise component. The larger the EMI source the more critical that coupling balance becomes, and it will never be ideal.

P2000

Sep 13, 2021 05:03

@hddh with balanced pairs, if you ground both ends of the shield you again have a ground loop and what it captures can carry to the amp/ground planes etc.. regardless the balanced signals, or you risk grounding one side through the shield and if it has no mains ground the shield can end up carrying ESD back, or fire-starting mains currents (in the event of a ground fault) beyond the rating of the shield

Discussion on question by Jeff R: Wha…

Imported from a comment discussion on diy.stackexchange.com/qu...

P2000

Jun 25, 2021 12:56

@MonkeyZeus yeah, I get your point, and this one remains a bit of a mystery. The OP could perhaps rely on secondary indicators: if they were done poorly you'd maybe see the effects of failure (ineffectiveness, movement, bending...), and if done right you'd see the proper brackets and fasteners.

P2000

Jun 25, 2021 12:56

If they are structural, there would be strong and firm fastening or bracing at both ends of those "rods", perhaps to take a transversal load. So, what is at both ends? The ones along the ceiling serve possibly a different purpose from the ones on the floor. Looks more like best-effort patchwork than calculated carpentry/framing reinforcement.

Discussion on question by sabruri1: S…

Imported from a comment discussion on electronics.stackexchang...

P2000

Feb 24, 2021 05:01

It would help if you could provide data points (plus curve) for reference vs measured RMS (and mean if available) at a few different levels and a few frequencies. I think that would quickly eliminate some suspects and hint at the problem. As pointed out by others, you could have a bias, an incorrect scaling, or maybe even a partial period / transients. Even better if you could dump the input[i] for all n, and plot it for us to see.

Discussion on question by Michael Wea…

Imported from a comment discussion on electronics.stackexchang...

P2000

Dec 21, 2020 19:35

@MathKeepsMeBusy I followed your suggestion and adjusted the R to its max (I think from 2k to 9M or so) and it did not change the minimum GBW of 178kHz or (obviously then) the suggested part with GBW 786kHz. Actually surprising. Perhaps the range does not affect the output impedance/feedback loading enough?

P2000

Dec 21, 2020 16:15

@MathKeepsMeBusy Yes, I don't know what explains the difference in GBW, and it's just an avenue of ideas and counter ideas.

P2000

Dec 21, 2020 06:34

@MathKeepsMeBusy good point, and your earlier 118MHz GWB is indeed perhaps a bit excessive but not entirely facetious, as >50MHz is needed for low distortion audio. Your point on feedback current may be what's really at play.

Discussion between P2000 and Michael …

Imported from a comment discussion on electronics.stackexchang...

P2000

Sep 28, 2020 23:40

@MichaelWeaser The one on the left of the spring would be your R=10K or whatever is specified, and then set your filter to have an Ri = R/10 (e.g. 1K). I wouldn't go for an Ri less than 100Ohm, so if Ri=1K works, go with that. Play with Ri and the filter tool to see what gets you L&C components in a reasonable range (size, price).

P2000

Sep 28, 2020 20:06

@MichaelWeaser If you don't want to do the above math, you could start with a 10K pot and 1K filter input & output. Also add a 10K series before the pot.

P2000

Sep 28, 2020 17:26

@MichaelWeaser Ok, Ri will affect the responsiveness of the "finger control" pot. Simply sketch out the v-divider around the pot plus the filter Ri; curve out the log of the voltage (or its dB) over Ri vs pot position. You probably want -20dB or less if the pot is up vs 0dB when down, and straight-ish in between. A high Ri and high-R pot (10K) load on DelayEq also isolates the channels so pressing one doesn't affect level of others.

P2000

Sep 24, 2020 03:18

@MichaelWeaser If you drive the filter with an op-amp, you can go for 1K at the input. At the output, if you are driving a op-amp you can go with 10K-100K filter output impedance. Since your source is noise to start with, I don't think it's a big deal. This matters more for filters in transmission lines.

P2000

Sep 23, 2020 03:57

@MichaelWeaser the more "off" it is (high or low) the more it will attenuate. Do you recall learning about a source's impedance and the load it feeds? If the load and source match, you get the maximum power transferred into the load. Since you have a passive filter, the power transfer matters, not just the voltage into it. If you are off by 10x or 100x, then you need more gain in the amplifier or opamp which means more noise. Going too low also makes the inductors expensive. Start with one filter.

P2000

Sep 22, 2020 03:54

@MichaelWeaser yes it's going to be very tricky. I don't know how savvy you are with modelling, but you could first model it in Python or Matlab and see what kind of accuracy you really need to be able to recognize speech. If this is not your thing, you could play with a simpler circuit first. But there are many parameters (noise source, pitch & volume control, filter selection - maybe even combine outputs) and the original operators took a year to train...! Anyway, good on you for trying this!!

P2000

Sep 22, 2020 01:52

@MichaelWeaser go with what I wrote re 1K/10K

P2000

Sep 21, 2020 18:57

Remember, you can also build filters around op-amps (Sallen Key). Then you don't have to worry about I & O impedance since you get some gain for free. You'll trade inductors for op-amps. At voice frequencies this should be a better cost tradeoff. And with a voder, a bit of extra noise (from the opamp/power supply) shouldn't be an issue.

P2000

Sep 21, 2020 18:33

What R values are you choosing? The impedance of the LC filter does not affect its tuning, only its attenuation when combined with the driving R and the load R. If your output pots are 100K, then 10K output for the filter is good. If the inputs are driven by 100R then a 1K filter input is ok. If not sure, pick 1K..10K. Issue is that if you are terribly off, the output voltage of the LC filter + pot will be so low that you have to drive-up the pre-amp and you'll have noise.

P2000

Sep 21, 2020 18:28

The impedance is probably not a factor, if your signals comes from a line-output, and you are feeding into a pre-amp anyway. You can choose anything in the 1K to 10K range, and see what L&C values you get, and then check the price for the L. 10K input and 1K output is a good starting point. These values determine the attenuation due to the line-output and line-input impedance of your circuitry around the filter. If you can include a sketch of your circuit setup in your question, that would help: line output, pots, filters, line input etc..

P2000

Sep 21, 2020 18:28

OK that info helps. It seems you know what you are doing. a) Do you you still have a question about the elco? And b), are you able to simulate the filter, with the component values from rf-tools?

P2000

Sep 21, 2020 18:28

I am a bit confused: you wrote "I was on that website calculating values for a 20th order bandpass filters." and then "I am sure I probably don't need a 20th order filter for what I need"? Maybe the youtube you linked would enlighten the reader, but could you perhaps just include your specs (frequencies/bandwidth, impedance, attenuation, Q etc...) in your question, like @Andyaka asked. If you don't know what all that means, then let us know, and we'd have to cover that first.

P2000

Sep 21, 2020 18:28

Wrt its calculations of L&C, you might be ok, as the issue we had was about frequency scaling if t0 group delay was provided instead of f0: electronics.stackexchange.com/questions/519992/… As for the capacitor, if you use elcos then yes the polarity must be correct. So only for biased circuits. Otherwise it's a hazard or at least you won't get the capacitance you thought. Remember you can always use multiple caps to get the value you want

P2000

Sep 21, 2020 18:28

does impedance matching matter in your case? If not, you can use a different value to get the L/C in the range you like better. Also, you could try cascading lower order filters each with L/C in better ranges. 20th order is high, you should really check the pole locations / transfer function with the acutal values returned.

P2000

Sep 21, 2020 18:28

I've tried that tool and found it to provide wrong answers. I don't know why/when, but it's prudent to check it in spice or to derive the transfer function and evaluate it at a few key frequencies. I found this site more accurate: changpuak.ch/electronics

Discussion on question by Leoc: Why i…

Imported from a comment discussion on electronics.stackexchang...

P2000

Sep 23, 2020 21:40

(1) yes you can create your butter-2 from analog, but it disappoints because the fc is so close to the fs. (2) you can manually shift the analog fc to get the digital fc where you want it (3) you can use c2d and hope it does a good job (4) you can transform and prewarp manually to pin it at fc. See also dsp.stackexchange.com/questions/64025/…

P2000

Sep 23, 2020 21:40

I don't think it didn't do anything, rather, you just won't see much benefit from a well damped 2nd order system. If it's flat-ish around fc, a horizontal warp or shift won't do much. Try a higher order filter, and look closely at the fc. Post your plots and we can discuss further. (And yes, my fs comment was for illustration, of course -as you & @AJN mention- fs affects many other aspects of the system)

P2000

Sep 23, 2020 21:40

High Fs is needed to get close to the shape, especially in the stop band. Warping is needed to match it close to some important frequency, usually the cut off. As the filter steepens, matching it matters more. Compare the dB at fcut with and without warp (your plot compares different frequencies). Then repeat with a steeper (higher order) filter, e.g. 7th Cheby: it will illustrate it better.

P2000

Sep 23, 2020 21:40

Can you add the updated comparison plots in your question (and with axis down to maybe -80dB not -300dB)? Without warping they should match close with high fs, but not exactly. They will diverge especially in the stop band.

P2000

Sep 23, 2020 21:40

good point, it's always 2pi somewhere. well I have not specifically used c2d but I do recall needing fs > 4*f0 for the warping to work. Try a higher fs.

P2000

Sep 23, 2020 21:40

Where did you apply the pre-warp frequency? I read 125.6KHz

Discussion on question by Harper - Re…

Imported from a comment discussion on diy.stackexchange.com/qu...

P2000

Aug 13, 2020 02:20

I believe an issue may be that everything has to be "UL listed" for US: plugs, fixtures, extension cords ... Could be a matter of code and home insurance.

Discussion on question by crasic: Is …

Imported from a comment discussion on electronics.stackexchang...

P2000

Aug 1, 2020 23:15

Crasic, does it matter what type of noise it is? You could use a spectrum analyzer to identify it (or its components). Also, your application could filter the noisy measurement (analog/digital), but I'm guessing you've thought of that, and you are focusing on the noise source here.

Discussion on question by equave: Why…

Imported from a comment discussion on electronics.stackexchang...

P2000

Jul 21, 2020 14:47

Did you cap the power pins of the flop (and opamp)? Perhaps show us a picture of your breadboard/pcb.