Thoughts on DSP multiamp/attenuator setup?

Administrator
Joined 2004
Paid Member
I have been meaning to build a flexible passive XO box for a long time-
I'm a fan of good passive crossovers, so I would never say no. :) Passive can take a long time to dial in, tho. And coils aren't cheap.
But at least you have target crossover points, 250 Hz and 1.25 kHz. No idea how those were bent with DSP, but it gives you a good place to start.

How's your measurement setup?
 
I have located the original thread, the design changes quite a bit since the first post and the final board is on page 5:

CS3318 PCB Layout

See what you reckon, it's more work than an off-the-shelf controller but there may be benefits to you (mine has remote control, it can fit in aesthetically with your 19" units, you can level match your different drivers in analogue domain etc.) but I would recommend a bal/unbal stage I think, partially because the IC costs £20 so using two adds quite some cost!

The cost of an additional chip/board is not really an issue, but assembly might be. I don't think I'd be able to solder the chip in place- I just can't do the tiny stuff (even basic full size opamps are an issue for my hands)- I'm not 100% going down the path of needing an attenuator, but if that's where I land, I'll certainly think hard about seeing if your project's adaptable (it certainly seems to be). Thank you!
 
I'm a fan of good passive crossovers, so I would never say no. :) Passive can take a long time to dial in, tho. And coils aren't cheap.
But at least you have target crossover points, 250 Hz and 1.25 kHz. No idea how those were bent with DSP, but it gives you a good place to start.

How's your measurement setup?

My measurement is capable enough- I have an Omnimic setup (Dayton) which does enough for me for FR/Phase, and an impedance tester in the WT3 device. I don't have a good option for far-field measurement, however, as my space isn't conducive to it (and I haven't prioritized it as I can get to very clean results without that data).

If I do go passive, I have a few values in 10 awg aircore (Solen) ranging up to 5mH or so, as well as a plethora of smaller units in 18 awg. I guess it's time to start measuring as that will inform what I can "get away with" as the most simple/affordable endpoint to this project. I've done a fair bit of work in passives, so that's not really a big issue. If the 14"s aren't naughty, the 250 could be pretty close to textbook, and could even be active or line level- just because I'm (in this scenario) not triamping with FIR filters, doesn't mean I have to use single-amp setups.
 
Just a few thoughts as I work through this project:

PC based- too much complexity and UI issues for systems the family needs to use. They put up with enough. Also, the multichannel dacs of any reasonable quality are very pricey.

FIR filtering - Delays are too much and mess with source compatibility, my current/relevant systems are both video-integrated and without a PC managing video buffering (see above why no PC). Standalone processors that do FIR in my budget are operating at 48khz, have limited taps, AND have severe latency with no PC to buffer video. I really prefer to keep to 96k+.

Level setting in dbx- there are input and output levels available, so I can tailor the output from +4dBu to +22dBu (I'd almost certainly use 4), and input from 14-28dBu so dynamic range and noise control should be pretty manageable so long as I can throw a solid (read- higher than standard LL preamp) volume controlled input at it (up to 24dB higher in than out is pretty solid)

https://3e7777c294b9bcaa5486-bc9563...iveRack_VENU360_Manual_5058681-C_original.pdf Page 94/96 (Page 96 of document, 94 labeled) is where I got the info on levels.

Big honking passive XOs- so much darned work but I should do it anyway as I have plenty of systems floating around to test and play with. The Evoi specifically appears to be a 2.5way, even though they describe it as a three way. There are certainly three filters, but given the midbasses are in a single vented volume, it would be very odd to only use one through the lower octaves, though the amp would basically short out an actively high-passed woofer below 250hz.

Just putting these thoughts down for feedback and posterity, and to try to keep myself honest as I work this through. Before I invest in gear I should probably keep my marriage together by fundraising with some of my existing, unused toys. Who wants an SE-OTL?
 
Last edited:
Found my old thread. A few lifestyle changes, but I found a very good price on used 360s, and apparently with factory warrantee (per their customer service)- I suspect refurb, since the 360 seems to have had a problem with the PSU connector, but will see the unit when I get home tonight.

For the time being this will primarily be fed from a McIntosh Mx-121 Pre-Pro, and feed mostly NCore 400 and 500. It'll get used from a variety of sources over time, likely I'll just keep it on the +4dBu setting to optimize for maximum input voltage (and thus resolution).

When I want to go purist, I'll probably run direct from a Squeezebox touch with a transformer for 75:110 ohm. Link to seller below.


DBX DriveRack VENU360
 
The reason for the standalone unit was to simplify the gear configuration- trying to avoid adding devices!

I do need to figure out some post-DSP attenuation for gain staging- as it stands, I typically need a net voltage gain of 0dB or less throughout the system (high power amps and sensitive speakers in a home setting). The best pre-built options I've seen come from Naiant, properly designed, and with variants for line level, mic level, etc. The 8 position balanced attenuators are one heck of a nice way to add input trimming to an amp, but there are adaptor and cable built-in attenuation options available too.

https://naiant.com/custom_audio_reproduction_equipment/inline-devices/

I'll probably just build my own as it's just 3 resistors per cable and I have the parts for the most part, but the naiant stuff seems well made and is very reasonable.
 
Last edited:
I recently came across a cheap used Driverack 260, and am going to use it to figure some stuff out and see where it takes me. I am coming from an Ashly XR2000 analog crossover.

My biggest hang up, which may be similar to something OP is dealing with, is that my power amps do not all have the same gain, and therefore when set up at one volume, once I turn it up, the balance is affected, no?

Is attenuating the input levels on the power amps going to allow gain matching between the amps? My impression was Gain was a bigger concern than wattage. Ya?

Apologies if I'm hijacking or distracting. If so, let me know and I'll start a different thread.
 
My biggest hang up, which may be similar to something OP is dealing with, is that my power amps do not all have the same gain, and therefore when set up at one volume, once I turn it up, the balance is affected, no?

Unless they have nonlinear gain (IOW, high distortion), once you balance the input levels and gain for parity, you'll be in good shape. This is absolutely a good place for this conversation!

The DSP I am using (symetrix symnet 8x8) has software controlled gain/attenuators on the input and outputs. With 100dB/1W speakers I don't notice any hum.

The 360 has clipping range controls on input and output, as well as individual output channel balancing levels and a master control in software.

The issue for my application is not expected to be with noise, though naturally that's to be avoided, but I want to be able to ensure that I can use a pre-DSP volume control without causing the A/D converters to have insufficient dynamic range available (Or use a digital volume control pre-digital input). The software attenuation has a decent range to it because of the wordlength headroom in the processing chain, but I tend to listen at a wide range of volume levels. Rarely super high but often super low.

It's really about keeping the preferred, simple UI from the processor or digital source as my volume control interface, and the system flexibility that goes along with that. If I can have that flexibility without performance costs by throwing some resistors on the amp inputs (I could modify them for unity gain but that's a pain if I want to use them in other setups), then that makes good sense to me.

If it matters, I ordered several of the naiant products. Too affordable a quality solution to pass up. I picked up a quad of the 24dB attenuator cables, and a pair of their SPD variable attenuators (it's a small switched resitor network built into a little adaptor, very convenient way to add variable input attenuation). Between those, I should be able to keep my ducks in order, even when working with a variety of amplifier/speaker efficiency combinations. I won't use any output attenuation on the subwoofers, for example, but might need more on the tweeter than the 24dB that the midranges need. The goal is to have as little gain balancing needed within the DSP as possible, as that's digital headroom that's better allocated to input sampling.

I haven't gotten them yet, but expect them sometime in the next week or so, at which point I'll start integrating the Venu and see where it gets me.
 
You know, I keep a pair of these around for whenever i have too much speaker sensitivity and amp gain. The switchable attenuation has always let me find a range of digital/analog level control that maintains good gain staging.
Amazon.com: Hosa ATT-448 Input Attenuator, XLR3F to XLR3M: Gateway

I think mine are -10, -20, -30dB....which i like...

Those are a great solution! However, for my use, my concern with those is that they're very low impedance, which isn't ideal for the amp input attenuation. The other issue with some attenuators (not necessarily these) is that the attenuators aren't symmetrical- they use an L-pad on just one leg, unbalancing the system and potentially bringing common mode noise along for the ride.

I'm going to use these for cases where 24dB isn't a good match:

https://naiant.com/studio-electronics-products/inline-devices/vpd-variable-inline-attenuator/

The steps offer quite a bit of range, at a more appropriate impedance, and are known to be properly balanced.
 
Last edited:
I'm going to use these for cases where 24dB isn't a good match:

VPD variable inline attenuator

The steps offer quite a bit of range, at a more appropriate impedance, and are known to be properly balanced.

Those look really cool...bound to work.

fwiw, and not to be challenging at all ........I think too much attention is given to gain staging / impedance matching / etc ...these days.....those things strike me as finishing details with very little significance ....compared to even basic simple tuning ...
 
IDIOT'S GUIDE TO MOVING TO DSP ALMOST PAINLESSLY

There are many Behringer DCX2496 on eBay cheap. A million in use. All-singing, all-dancing, does everything. Stand-alone with its own display panel.

Cheap too is a USB to AES box, maybe $35, thereby by-passing the analog DCX input stage that some people think is faulty (without any evidence I've ever seen). But not applicable to the new more modest DCX variant.

If you are a purist, you might want to run you digital section all-bits-on-deck. In that case, you need a cheap ganged volume control at the far-side of the DCX. Cheap.

Works great. You can do all kinds REW tests and fine-tune your EQ in minutes. You'll wonder why you didn't do it a decade earlier.

Passive crossovers make no sense in theory or in practice, unless you are crossing over pure resistances and have no sympathy for amp loads.

B.
 
Last edited:
Those look really cool...bound to work.

fwiw, and not to be challenging at all ........I think too much attention is given to gain staging / impedance matching / etc ...these days.....those things strike me as finishing details with very little significance ....compared to even basic simple tuning ...

That's perfectly fair. As a counterpoint, however- I care about the details, and the little stuff does matter, but perhaps also important is that it's additive. Better amps of similar power don't matter as much as a better crossover, but their improvements stack on top of the better crossover improvements. I have invested enough in the small stuff to put the effort into the OTHER small stuff. Low diffraction cabinets, tweaked drivers, etc. I don't want to make overstated golden ear claims like a schmuck but the little stuff matters to me.