3-way active XO ... can amp input coupling caps be tailored for the frequency band?

[andyr]

My theoretical EE knowledge is limited so I am asking for some assistance from the tech-savvy here.

I have built 3-way active XOs for my Maggies using Rod Elliott's P09 PCBs; XO points are approx 330Hz & 2,600Hz. Thus I have 3 stereo amps covering the frequency spectrum.

The question which I would be grateful for an answer to, is based on the following assumptions (and please correct me if I'm wrong! ):
* The value of the amp's input coupling cap imparts a certain phase shift to the signal?
* Good coupling caps like Mundorf/SoniCap etc are expensive ... but smaller values are cheaper!

So my question is ... should the input coupling cap for bass, mid & treble amps be the same? Or can I use, say:
- a 2uF input cap on the bass amp, for maximum low-end output,
- a 0.47uF input cap on the mid amp (which which doesn't have to deliver much below 300Hz), and
- a 0.22uF input cap on the ribbon amp, which doesn't have to deliver much below 2KHz?

Or will doing this introduce undesirable phase problems?

Amps all have a 43Kohm input impedance.

Thanks,

Andy

 

[planet10]

Best is to use the input coupling cap as one of the poles of the XO...if you have already built the XO up, you'd have to mod it a bit

dave

 

[andyr]

Best is to use the input coupling cap as one of the poles of the XO...if you have already built the XO up, you'd have to mod it a bit

dave

Sorry, Dave ... you need to speak English (not EE tech-speak)!

Regards,

Andy

 

[Eva]

You can safely forget about input coupling capacitors. The speakers themselves produce far more phase shift in the middle of the audio band than any coupling capacitor, and crossover has to be designed to cope with that and obtain suitable acoustical phase shift.

For example, linkwitz-riley electrical crossovers are useless, but you can get LR acoustical response with proper electrical filters and phase shifters. In general, generic active crossovers are useless (like generic passive ones).

 

[AndrewT]

So my question is ... should the input coupling cap for bass, mid & treble amps be the same? Or can I use, say:
- a 2uF input cap on the bass amp, for maximum low-end output,
- a 0.47uF input cap on the mid amp (which which doesn't have to deliver much below 300Hz), and
- a 0.22uF input cap on the ribbon amp, which doesn't have to deliver much below 2KHz?

if the amplifier can pass 10times the bandwidth of the crossover+speaker, then the phase effect of the amplifier is negligible compared to the speaker.
For bass amplifier your 2u2F and 43k has F-3dB = 1.6Hz. It will pass a 16Hz signal to the bass speaker with negligible phase alteration.
Similarly a mid amplifier with an F-3dB = 33Hz will be OK. This needs C>=0.12uF & 43k.
The treble amplifier must pass 2600Hz and so select it's F-3dB <=260Hz. for Rin=43k select C>=15nF

There is an alternative as suggested by Zen and alluded to by Eva.
Use the amplifier C as a single pole part of the overall crossover. Q=0.7
For LR xo you now need a further 3pole Q=0.7 filter. Part of this could be a C feeding/protecting the driver and the remaining 2pole filter with Q=1 would give that 4pole LR. Now you have to add EQ to get the driver to do what you really want from it. This is the difficult bit. Try using a DCX2496 to determine the EQ required to get the driver working properly. Then imitate that DCXfilter discretely in your active crossover EQ.

 

[andyr]

You can safely forget about input coupling capacitors. The speakers themselves produce far more phase shift in the middle of the audio band than any coupling capacitor, and crossover has to be designed to cope with that and obtain suitable acoustical phase shift.

For example, linkwitz-riley electrical crossovers are useless, but you can get LR acoustical response with proper electrical filters and phase shifters. In general, generic active crossovers are useless (like generic passive ones).

Mmmm, thankyou, Eva. Let me see if I can paraphrase what you said.

1. "L-R electrical XOs are useless?" OK, so what is better to use?

2. "Generic passive XOs are also useless?" OK, what I was trying to do with my active XOs was mimic the passive Maggie XOs in that they have the same order and slope ... but slightly different "@frequencies", to smooth out the (electrical) frequency response. And have the amps in direct control of the drivers ... vs. the stock passive setup, where inductors and caps are "in the way"! .

So I guess I would have to say that my active XOs are not "generic"? They were modelled using lspCAD.

3. You can safely forget about input coupling capacitors. The speakers themselves produce far more phase shift than any coupling capacitor.

So are you saying it will have negligible (hearable) effect if I tailor the amp's input coupling cap to the frequency band it is dealing with ... vs. using the same (2uF) cap on all 3 amps?

Thanks,

Andy

 

[andyr]

For bass amplifier your 2uF with 43k has F-3dB = 1.6Hz.
It will pass a 16Hz signal to the bass speaker with negligible phase alteration.

Similarly a mid amplifier passing 330Hz with an F-3dB = 33Hz will be OK.
This needs C>=0.12uF with 43k.

The treble amplifier must pass 2,600Hz and so select its F-3dB <=260Hz. for Rin=43k select C>=15nF

Thank you very much, Andrew.

OK, let me see if I have understood you. Taking a decade down (with my 43K amplifiers):
* 2uF on the bass will pass >16Hz with neglegible phase alteration.
* 220nF will be more than enough for the mid amp which needs to have an F-3dB of 33Hz, in order to deliver >330Hz with neglegible phase alteration.
* 22nF will be more than enough for the ribbon amp, to deliver >2,600Hz with neglegible phase alteration.

And if I use these values (thus saving myself an awful lot of money, compared to using 2uF throughout! ), I will get neglegible phase difference between the 3 amps in their relative frequency bands?

Regards,

Andy

 

[AndrewT]

yes, check the F-3dB frequency of the high pass filter that each of these caps will produce.
You will also find that the NFB DC blocking cap can be scaled lower value with increasing frequency. This in turn allows lower value PSU capacitance requirements.

You save in the input filter, in the NFB filter and in the PSU filter.
The NFB filter for both the mid and treble amplifiers could be 50V/63V metalised plastic film. MKS/MKT/PES would be near ideal here.

Polypropylene capacitors don't cost that much. I hope your not thinking of buying the exotic high retail cost caps that have little science to why they might/mightnot improve the sound quality.

 

[Eva]

Mmmm, thankyou, Eva. Let me see if I can paraphrase what you said.

1. "L-R electrical XOs are useless?" OK, so what is better to use?

2. "Generic passive XOs are also useless?" OK, what I was trying to do with my active XOs was mimic the passive Maggie XOs in that they have the same order and slope ... but slightly different "@frequencies", to smooth out the (electrical) frequency response. And have the amps in direct control of the drivers ... vs. the stock passive setup, where inductors and caps are "in the way"! .

So I guess I would have to say that my active XOs are not "generic"? They were modelled using lspCAD.

3. You can safely forget about input coupling capacitors. The speakers themselves produce far more phase shift than any coupling capacitor.

So are you saying it will have negligible (hearable) effect if I tailor the amp's input coupling cap to the frequency band it is dealing with ... vs. using the same (2uF) cap on all 3 amps?

Thanks,

Andy

1: The LR concept is excellent from an acoustical point of view, but the electrical filters required to achieve acoustical LR responses are quite different from LR and may be quite complex sometimes. For example, to approach a true LR24 alignment, the electrical filters required are going to be 12dB/oct or 18dB/oct and may have to include some EQ and phase shifters (worth the effort, though).

2: Drivers have inherent built-in 12dB/oct filtering, both for the lower and upper roll-off. Lower roll-off is 24dB/oct for bass reflex, and there is a small portion of 24dB/oct followed by 12dB/oct on horns. This inherent filtering produces substantianl phase shift and group delay in the passband preventing textbook electrical filters to work properly.

3: There is no magic in capacitors. 2.2uF 50V film is widely available and not expensive at all. The huge higher voltage parts have nothing but drawbacks. Electrolytics are also fine, provided that capacitance is high enough to simulate a short circuit in the audio band (like 22uF to replace a 2.2uF film). Esoteric stuff is a true waste of money. A good crossover design (involving measuring or estimating the phase response of the drivers fed directly from an amplifier and making a set of electrical filters to convert that into LR12 or LR24 acoustical output, or at least, to achieve decent acoustical phase matching) does have a real impact on perceived sound, particularly on bass

 

[andyr]

Thanks, Andrew & Eva.