Are High quality opamps important in active crossovers?

Hi everyone.

I'm in the process of upgrading my tri-amped system. I started with a pre amp and 3 way active crossover I built using boards I purchased from Rod Elliot (P09 & P97 respectively), and I was/am pleased with the results.

I recently rebuilt my 3 way crossover using plans I downloaded from Silicon Chip's site.

see http://www.siliconchip.com.au/cms/A_30278/article.html

The main reason I did this is everything is on one board, including input and output connectors and power supply. Also the filter stages and output buffers of the circuit are exactly the same as Rods' design. Rods' boards worked just fine,b ut it was a pain to wire all those screened cables and looked messy.

I'm in the middle of building another pre amp, using Rods' circuits for his "High quality Audio Preamp" (P88), and as he suggests I'm using the OPA2134 dual opamps.

It seems silly, though, that after going to the trouble of using good quality chips in the preamp to then use the el cheapo TL074's, from Jaycar, in the crossover. Or is device selection not so important in this role? If it is what would be a good (quad) opamp to use? OPA4134's (same as OPA2134 in a quad) don't seem to be readily available in Oz, not from Farnell or RS Components anyhow. I've spent hours searching thier web sites and I can't find a device with similar specs for less than $25 ea, and I need 6 of them!

Part of my problem, I'm sure, is I don't know enough about what to look for. I've a pretty good idea (I think) of what GBP, THD and slew rate are, but what actual figures are considered "good", and what else should I be looking at?

I read in another thread in this forum that AD8513's might be OK ($18.33 from RS). Data sheet here. However these seem to be surface mount only - can you buy little boards to solder them onto so they will fit the DIP14 sockets I used when building the board?

Thanks all.
Andy
 

funberry

Member
2005-08-17 3:23 pm
Audio General amplification requires GBP of 1MHz min, and slew rate of 0.5 v/uS min.


Active filters place exceptional demands on opamp slew rates, and the greater the Filter's Q (quality factor) the greater the Slew rate needs to be. Since slew rate is closely tied to bandwidth, a high slew rate opamp is likely to have high bandwidths. It's not uncommon in hi-fi active filters, to have to go to opamps with bandwidths well above the audio frequency range (20MHz+)

I would not use any thing with slew rates lower than 15-20 V/uS in an active filter.

For high fidelity, you'll want a noise level of 5nV/rtHz or less. NE5534 has been widely used in hi-fi and is a standard for comparing noise. It sits at 4 - nV/rtHz.

Some newer, exotic parts like INA103 feature noise levels as low as 1nV/rtHz, excellent slew rate and lo distortion. They are often used in pro mixing consoles and studio recording equipment.

The OPA4228 is good value, with a noise figure of 3nV, which is good for a car application, but its slew rate (10v/uS) is borderline for active filters.
NE5534, at 13V/uS, is not ideal either, but in a bind, can do for a low-Q active filter, and is widely available, and cheap.
So, this is what you look for, when deciding on op-amps for your active crossover: noise and slew rate.

Pro equipment generally use single (1-per-package) opamps, to minimize cross-talk, parasitic capacitances, and very importantly- to provide individual power supply decoupling capacitors for each individual op-amp. A dual or quad pack is a convenience, not a pro tool.
 
Hi,
postscript to Funberry on slew rate:- In my opinion highest slew rate only required on treble Hi-pass and if fitted treble Lo-pass.
Mid and bass filters can easily manage with lower slew say 5 to 10 V/uS (3Mhz to 8Mhz bw).
Funberry, any comment?
How do you define Hi Q? above 1? or above 2? or some other figure?
Try an edge on piggy back PCB with one pair of dual opamps into a 14 dip.
 

funberry

Member
2005-08-17 3:23 pm
To me, Hi Q = elliptic, state variable filters.

Higher the Q, the higher the SR required. Higher the ferequency, higher the sr required.

If I were designing a filter now, I'd pull out my college physics book and calculate the required SR now. But I'm just speaking qualitatively, if there's one parameter I really need to pay attention to in an active filter, it's slew rate, more than offset oltage, input current, rail-to rail IO, temperature drift, or anything else.

Because it's easy to say " well, I'm working at audio frequencies, so 1MHz bandwidth is fine" and stick in a 741...

The rationale behind slew rate in an active filter, is a bit counter-intuitive to a beginner.
 

SillyGalah

Member
2005-09-04 6:26 am
AndrewT said:

Try an edge on piggy back PCB with one pair of dual opamps into a 14 dip.

Thanks Andrew, ggreat idea. I'm not sure if there is physicaly enough room, but definately worth investigating.


Originally posted by funberry

Higher the Q, the higher the SR required. Higher the ferequency, higher the sr required.

Q is a concept I never realy quite understood. Something about a ratio of reactance...? I do recall that tuned circuits with a high Q hive a greater propensity to oscilate. Or is it the other way.
 
I too am building Rod's active crossover and would like to know what are the most suitable dual opamps. Not the ones that are not. Slew rate seems important maybe there are a few other considerations.
Short of reading all manufactures' data , this forum should be a excellent source of information.
 

EC8010

diyAudio Moderator Emeritus
2003-01-18 7:57 am
Near London. UK
OPA2134 is a very nice op-amp. You want low noise and low distortion, and it's a generation on from NE5534. Moreover, as it's FET input, it gives you a great deal more freedom about capacitor values and circuit impedances than a pure bipolar op-amp. I see that funberry wants a good slew rate and is alarmed about high values of Q. Most crossovers are based on Sallen & Key topology where the Q is obtained by positive feedback. Positive feedback increases distortion. In June 1978 Peter Billiam published an AES paper about how filters increase the distortion of op-amps ("Harmonic Distortion in a Class of Linear Active Filter Networks")
 

dfdye

Member
2005-12-04 8:57 pm
USA
The safe bet is OPA2134''s. There have been a ton of designs to use these, I'm sure, but the icing on the cake that made me pick them (since I haven't had a ton of experience with a bunch of different opamps) was that these are the parts used in the active crossovers for the Linkwitz Orions.
 
dfdye said:
The safe bet is OPA2134''s. .....


I use OPA4134 (same but quad) and work nice for a 4th order L-R crossover. Measured transfer function with SpectraLab is picture perfect as per simulation, hi and lo phase match exactly.

TI used to provide free samples, only they had only the SOIC package available, which is a pain. Before designing a SMD oriented xover board, I managed by gluing the package *on the underside, legs up* (i.e. the same orientation as if it were a regular DIP on the component side), and soldered thin wiskers from device pads to pcb pads.

Rodolfo