Does position matter?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Yes, if frequency distribution of levels is unequal (and it is heavy towards bass in most audio), you may get a headroom benefit from cutting off high-level unwanted stuff before further processing.

So low cut first as a rule-of-thumb.

At least I always believed so....

Jakob E.

 
It’s a really interesting question. I’d always assumed you’d ideally want HPF first to maximise headroom as Jakob says, followed by your high frequency band - so that if you boost, you’re boosting HF noise from as few previous stages as possible. Then mid, low, and LPF (again, to cut noise from all the previous stages).

That said, I’ve never actually seen that topology in any schematics so I may well be overthinking it, or missing something elsewhere! Or maybe it would just make for a really awful front panel layout...  ;D
 
pucho812 said:
Which comes first hi or lo pass filters?  Does it make a difference ?
Filter before eq? Does it make a difference?
My rule of (old) thump is (passive) LPF filter first to prevent slew overload...(rectification, RFI, etc).

While early HPF can help preserve headroom, doing it too early generally involves non-ideal capacitor dielectrics. Some manufacturers (not me) would undersize the electrolytic cap in series with typical gain resistor in mic preamps to cheat and roll off more LF at high gain (reducing audible 1/F noise etc. ), making the preamp appear quieter than it actually is. My preference was to tune early HP poles very low with large electrolytic capacitors to reduce the AC voltage across them in use, that can cause non-ideal behavior, and as soon as practical perfect a proper HPF pole to define the audio bandpass using a film capacitor (that doesn't suck) at higher impedance to keep it practical.

In theory it doesn't matter, in practice it depends on what are the weakest links of your audio chain.  It also depends on what you input source is. Line level sources are generally not fast enough to slew limit modern electronics. Mic inputs are still the wild wild west (probably worse now for RF than back in my day). 

JR
 
well I am thinking this way.  as far as signal flow.  preamp to filters to eq.  so by the time I am at the filters I am at my line level without question.    But which filter first hi or lo?  We can expand the question to eq bands.  which bands first?  the last bench gig had the following in our designs, LMF HMF, LF, HF as signal...  I never asked why we went that way.  I figured they would be in order Low to high or high to low.
 
I would agree with pre-amp => filters => EQ.  Not sure the order of hpf, lpf,  and the EQ bands really matters all that much.  Mostly it would just be dictated  pcb layout and the order you want the front panel controls.
 
john12ax7 said:
I would agree with pre-amp => filters => EQ.  Not sure the order of hpf, lpf,  and the EQ bands really matters all that much.  Mostly it would just be dictated  pcb layout and the order you want the front panel controls.

true... 
 
In a true linear system, it does not matter.

If wind-rumble is overloading the system, you need to cut bass BEFORE  it distorts something.

Unless mad mosquitoes are swarming the mike, and your levels are too hot anyway, there is probably no rush to filter audio highs. (RF should be reduced ASAP.)
 
PRR said:
In a true linear system, it does not matter.
Indeed if mic preamps were linear up to light I wouldn't use a passive LPF in front of them.

An active LPF needs to be fast enough to filter, I learned that the hard way designing filters around BBD delay lines back in the 70s using slow op amps of the day. They weren't fast enough, but using a passive real pole at the high edge rate node worked.

Input transformers are generally bandpass filters all by themselves, so do not suffer slew rate problems.
If wind-rumble is overloading the system, you need to cut bass BEFORE  it distorts something.
or reduce gain, but if wind noise is a common problem and you can't use a wind screen on the mic by all means put your HPF early.
Unless mad mosquitoes are swarming the mike, and your levels are too hot anyway, there is probably no rush to filter audio highs. (RF should be reduced ASAP.)
As I shared I prefer to set the audio bandpass HPF pole once with a high quality film cap... setting the other coupling caps in series octaves lower to be out of the way.

Most NF networks have LPF poles across them and these roll-offs are cumulative so multiple cascaded 200 kHz LPF could eventually add up to -3dB @ 20kHz but it would take a bunch of them. The input RF filter can be higher than 200kHz unless bothered by AM radio (as low as 500kHz).

A/Ds or sampled circuitry (like BBDs) will generally use their own dedicated LPF.

JR
 
JohnRoberts said:
Indeed if mic preamps were linear up to light I wouldn't use a passive LPF in front of them.

Missed the point completely.

"Linear" as in "no distortion". Sine waves come out sine, drum wave comes out the same drum wave, and wind-noise in the microphone does not overload.

I've done gigs where the unwanted sub-sonics were near as loud as the talent. Filtering at 20 or 150Hz gains ~~~6dB headroom before the harpsichord or organ is intermodulated with blower or rumble.
 
I would say it depends somewhat on the topology of said filters.
One very common issue with Sallen & Key variable HPF is the need for dual pots with one wafer being twice the value of the other. A common workaround consists in using a gain of about 4dB instead of unity to the voltage amp. Since this gain must be compensated for, it can be done before the filter, adding an attenuator and a voltage follower (S&K filter wants to be driven from low Z), which increases noise, or after the filter, which reduces headroom by 4dB.
I would say all sorts of permutations have been tried (some designers even advocate putting dynamics before the filters, which is silly IMO).
It seems some kind of consensus has been achieved with filters being before EQ, the order of filters making no significant difference in operation, as long as they are decently designed.
BTW I fully agree with JR's recommendation of preventing any VLF or VHF right from the start, using passive filters.
 
abbey road d enfer said:
I would say it depends somewhat on the topology of said filters.
One very common issue with Sallen & Key variable HPF is the need for dual pots with one wafer being twice the value of the other. A common workaround consists in using a gain of about 4dB instead of unity to the voltage amp. Since this gain must be compensated for, it can be done before the filter, adding an attenuator and a voltage follower (S&K filter wants to be driven from low Z), which increases noise, or after the filter, which reduces headroom by 4dB.
I would say all sorts of permutations have been tried (some designers even advocate putting dynamics before the filters, which is silly IMO).
It seems some kind of consensus has been achieved with filters being before EQ, the order of filters making no significant difference in operation, as long as they are decently designed.
BTW I fully agree with JR's recommendation of preventing any VLF or VHF right from the start, using passive filters.
I did a lot of cascaded multistage active LPF back in my BBD delay line days.  ;)

Topology also matters wrt limited op amp GBW or slew rate capability. One active 2 pole LPF topology seems to tolerate out of band HF signals well. The "multiple feedback" topology 
256px-MFB_Topology.png

If we look at signal flow, the first thing the signal hits is a passive C to ground, so the signal is slowed before it reaches any active gain stage.

I even played around with a variant on MFB topology to incorporate and use the natural bus capacitance of a long console bus run, putting that bus C in parallel with C2 using the above topology.  I did not end up using this in production. The only time I had a RF rectification problem in a sum bus amp was inside a small 6x4 mixer. I fixed that one with an on the fly engineering change to a bifet op amp in the sum bus stage. Bifet opamps are less susceptible to rectification from RF due to higher (Vth?).

JR
 
JohnRoberts said:
256px-MFB_Topology.png


I even played around with a variant on MFB topology to incorporate and use the natural bus capacitance of a long console bus run, putting that bus C in parallel with C2 using the above topology.
As a summing amp, it seems replacing R3 with an inductor may improve HF response and offer more gain margin. It's just a hint, but confirmed by simulation.
Purely speculative since analog summing improvements are akin to improving the petrol lamp or the pioneer cart.
 
Back
Top