Attenuator BEFORE or AFTER active crossover?

[Michael Chua]

Originally posted by SmellOfPoo
I already have a dact-style 50k 6-ganged stepped attentuator from goldpt.com.

Can't get better than this.

Cheers
Mike

 

[Michael Chua]

no I did not scope the HP output. But Jens' active filter boards are pretty good and I don't think you can fault the grounding there. BTW, there was no hum or buzz in either the satellites or the bass speakers.

Hi Andrew

Here's a sample of my crossover HP. Its just a basic 12dB/oct Unity Gain HP with NE5532. Nothing fancy. Taken with Input shorted to ground at 5mV resolution. You will of course get lower noise with the new generation of opamps. Hope it helps.

Regards
Mike

An externally hosted image should be here but it was not working when we last tested it.

 

[AndrewT]

5534 and others of that ilk will give less opamp noise.
I think the problem is likely to be a combination of resistor noise and use of inherently noisier FET input opamps.
That was why I chose opa275 since it claims lower noise than many of the similar priced real FET input opamps. I think my experiment with reducing the Rs in the RCs confirmed that resistor noise is partly to blame.

It's just that I have never read that active crossovers have a noise problem that appears to be much worse than using passive solutions.

 

[BobEllis]

In my experience, pot on the input end of a crossover is usually fine from a noise perspective. Of course attenuating the outputs will reduce the noise, but I usually am satisfied with the results.

I had a noise issue with the pot before the crossover when using some really deep (40 dB) notch filters for a Cauer-Elliptic filter. Not ready to buy a 6 gang attenuator, I placed a fixed attenuator (15 dB IIRC) between the last filter stage and the buffer. Not perfect, but driving the filters harder improved the signal to noise ratio.

 

[Dr_EM]

I fully plan to implement 6 channels of PGA2310 after my 3 way active crossover for volume. It just seems more sensible to have a constant, high level through the crossover. A passive speaker crossover doesn't but it's another aspect that can be improved apon when going active!

 

[Michael Chua]

Originally posted by AndrewT
It's just that I have never read that active crossovers have a noise problem that appears to be much worse than using passive solutions.

Hi Andrew

Noise and hum are very problematic in active crossovers. I eventually decided on Unity Gain for my crossover because I figured with CDs' line level output, there's really no reason for gain.

5534 and others of that ilk will give less opamp noise.

I do agree. Trouble is on a personal level, I don't like the sound of the 5534. I base my selection on subjective qualities like Tone, Transparency and Dynamics. It doesn't bother me the number of "zeros" in THD or noise figure. As long as I can't hear it, its fine.

Frankly, I have not used opamps for many years because of my disdain for them but I must admit I was stunned by the samples I received from NS. I am really impressed by their new LME series.

Regards
Mike

 

[Michael Chua]

Originally posted by BobEllis
Not ready to buy a 6 gang attenuator, I placed a fixed attenuator (15 dB IIRC) between the last filter stage and the buffer.

Hi

You may want to try out VCAs. Personally, I think it is the next best option to 6-ganged stepped attentuators.

Excellent tracking, very easy to use and best of all, don't sound like opamps.

Regards
Mike

 

[lineup]

Yes, but in practice it's not that bad to have the volume control before the active crossover.

I think so too in general, but how bad is 'not that bad'?

Think there is no general answer.
Both ways are quite possible.
Can depend what circuit and needs you have.

 

[HK26147]

As Michael Chua pointed out:
The purpose on the input is functionally a sensitivity control; to adjust the voltage to the device ( in this case a xover ).
The voltage level standard for pro sound and commercial line level are different.
With a system of devices each providing gain, it easy to overload the input stage of a device in a chain.
A sensitivity adjustment is fairly easy and relatively easy to implement.
IMVHO: I would include it - even the very inexpensive Beringer includes input atten. and gain control.

 

[audionut]

I have a passive line level crossover using C and L. The train of componants is DAC to stepped attenuator (10K) to BOSOZ pre-amp to C and L filters to Aleph Amps...

Do I worry about noise in this situation? I don't think I hear any. Is this just an issue with active crossovers using chip based op amps??

 

[AndrewT]

about noise in this situation? I don't think I hear any. Is this just an issue with active crossovers using chip based op amps??

I think it's a combination of resistor values (usually recommended ~10k) and inherent noise in IC opamps.

 

[Michael Chua]

I think this is what SmellOfPoo has in mind.

All source outputs need to be adjusted for the same sensitivity. The level will depend on whether crossover has gain or is at unity. The idea is to maintain the strongest signal possible within the crossover.

With opamps, perhaps not exceeding +/- 10V peak to peak?

Cheers
Mike

 

[dzzmiller]

Re: Maximize bits

The active crossover (if it is using DSP, eg Behringer etc) has a A-D conversion. The bits are assigned according to the signal (voltage). If the signal is not hot enough then you are giving up bits. This will increase your quantization error.

Back in the days of PCM (before sigma delta mod was popular), we worried about this. With the newer converters (sigma delta), the problem/consequence is not as severe or audible. However it is still there. BTW, the signal can also be too hot for the ADC and you will get clipping.

So you really need to examine the gain structure, dynamic range you are expecting, and the ADC of your crossover (if it is DSP). It really is about maximizing bits. That is why the attenuator is best located after the crossover's DAC. It is more fool-proof that way.

Better still, stay digital if from a digital source. That means coming into the crossover on spdif. But I'm changing my triamp computer fed system to stay digital until it leaves the crossover to the amp. That means there's no A/D and only one D/A conversion.

I'm playing vinyl again, but that is on a separate one amp system. I'm not going to run vinyl through an A/D and D/A process.

 

[AndrewT]

The idea is to maintain the strongest signal possible within the crossover.

With opamps, perhaps not exceeding +/- 10V peak to peak?

There's the problem.

If the amp has a sensitivity of 1Vac for maximum power, it needs just 0.1Vac for normal maximum SPL listening.
If you want to listen to music that is not loud, you will more likely feed it a signal in the range 10mVac to 50mVac.

I would expect the preamp and crossover to have a minimum headroom of 10dB above the maximum signal the power amps expect.

For that 1Vac the peak voltage is 3Vpp and +10dB is 10Vpp.
If the amps need 2Vac the Vpp becomes >=20Vpp
If you want +16dB headroom then it is 40Vpp. That is just about possible with +-21V opamp supplies. But, you have to take the noise that comes with ICs and that voltage driving ability.
5534 can do it, but they are BJT at the front end, not FET, which is better in filters due to their inherent high input impedance.