very specific question about dc paths

[caress]

i'm building a sixth-order active LPF to add onto a toy keyboard of mine.  i'm using fig.6-20:  flattest amp values (pg.144) from don lancaster's active filter cookbook, for those who have it...  (great book, btw)

i've built this up and tested it and it works really well for my needs except that there is a buzz like a florescent light (none here) when i tested it...?  hmmm.  i tried adding a dc-blocking cap at the output/input but the sound persists - maybe that's not the problem?  my question is about the little asterisk at the input where it states, "must return to ground via low-impedance dc path."  asny help on this statement or my problem in general?

[brett]

Hi
buzzes are often ground loops, which are AC, so it's not surprising that that a cap won't solve things.
If the input requires a DC path to ground, why not put in a low value resistor (e.g. 1 k) between your input and star ground (you have star-grounded everything, right?).  Of couse, you'd then need something with low output impedance to drive such a low input impedance thing.

But someone else will probably have better ideas, so wait for a few replies.
cheers

[George Giblet]

I don't have that book.  An it might be of some help if you mentioned if the filter sections are non-inverting (ie. Sallen and Key) or inverting (MFB).  eg. see

http://www.ele.uri.edu/~diceccoj/Final_w_Pic.pdf

and how you have done the power rails.

Some possibilities:
- you haven't used shielded cables enough.
- the circuit isn't close enough to ground
- the wiring is too close to the transformer (if there is one)
- A power supply issue. Try adding a cap (single supply), or caps (dual supply) across the power rail(s).
- A bias network issue (single supply)

I don't know what filter frequency you are using but if you are hearing buzz above the cut-off frequency of the filter then I would be loking at power supply issues or the output cabling first.

[caress]

i triple checked the ground and i don't think there's a loop there because it's a bit higher frequency than a ground hum.

i forgot about the dual supply, though, so that's probably my problem...  i still need to add some caps across the rails, too.
i'll work on it.

[caress]

update:

* i used this 555 voltage inverter to get the negative supply.  http://www.aaroncake.net/circuits/vinvertr.asp
* added power supply filtering
* added a .33uf input cap

it works!  i'm getting some mild oscillation, probably from the 555, but it's not shielded and i'm using alligator clips everywhere and the dual supply is practically sitting on top of the filter...  it will probably get ironed out when i get it boxed up, though.  this filter boosts the output quite a bit and is sensitive to input volume.  my silverface fender twin was set on 4 without the filter, then when engaged i needed to turn down to 2.  pretty huge boost in volume.  also ... i really should have made the filter frequency 1.5-2khz instead of 1, but it's ok... 

edit:  this turned out to be not very specific to dc paths... sorry

[Paul Perry (Frostwave)]

Temporarily running a circuit on batteries, can often help work out what is going on with mysterious whines & hums.

[George Giblet]

Good idea Paul.

I'm fairly sure it's the 555 though.  My rough calculations show it's oscillating at about 3khz which is going to produce a whine.  Dropping the 3300pF cap to 1000pF (1nF) would push the frequency up to 10kHz.  With normal caps you probably don't want to run them over about 10kHz (but you can try your luck).  Low ESR's and tants will go higher.  With that circuit it's important to keep C2, C3, and C4 all close to the 555.  And you don't want the circuit to be too close to audio signals!

[caress]

Good idea Paul.

I'm fairly sure it's the 555 though.  My rough calculations show it's oscillating at about 3khz which is going to produce a whine.  Dropping the 3300pF cap to 1000pF (1nF) would push the frequency up to 10kHz.  With normal caps you probably don't want to run them over about 10kHz (but you can try your luck).  Low ESR's and tants will go higher.  With that circuit it's important to keep C2, C3, and C4 all close to the 555.  And you don't want the circuit to be too close to audio signals!

i was using a battery...  but then again, i'll only be using a battery with this. 

i'll try dropping a different cap in there to see if it cleans up, but why wouldn't i want to run normal caps over 10khz?  it's also really really close to the audio signals right now since i'm just testing; it will be nicely arranged/shielded when i drop it in the keyboard.  the caps, though, are as close as possible to the 555... pretty much touching it.

[George Giblet]

At low levels there is a peak in your ears frequency response around 3kHz, so the 3kHz tone is most easily heard.  Pushing it up to 10kHz makes it hard to hear.  16kHz would be better, that's like the whistle from CRT TV's which most people can't hear (I usually can) and 20kHz + is ideal because it is beyond the audio range.   *If* some interference is getting in before the LPFs then the LPF will remove some of it, again the higher the oscillator frequency the more the LPF will remove it.