Hello,
Hope I'm not repeating the thread. I'm looking for some help in choice of parts for an active crossover.
The crossover design I selected is a fixed frequency electronic crossover. It's kindly presented on sound-au.com and it just so happens I'm after the same crossover frequency and slopes.
Linkwitz-Riley Electronic Crossover
The problem is my lack of knowledge in this field:
1. There are seven op amp sockets. Prices range from very affordable to 80 euros a piece for a double discrete op amp. Is it worth investing in discrete op amps?
2. Resistors? I've seen mill's to be popular for passive crossovers, there are quite a lot od them inside here
3. Capacitors - polypropylene? To look for some high end mundorf?
Do any of these parts actually affect the sound?
Hope I'm not repeating the thread. I'm looking for some help in choice of parts for an active crossover.
The crossover design I selected is a fixed frequency electronic crossover. It's kindly presented on sound-au.com and it just so happens I'm after the same crossover frequency and slopes.
Linkwitz-Riley Electronic Crossover
The problem is my lack of knowledge in this field:
1. There are seven op amp sockets. Prices range from very affordable to 80 euros a piece for a double discrete op amp. Is it worth investing in discrete op amps?
2. Resistors? I've seen mill's to be popular for passive crossovers, there are quite a lot od them inside here
3. Capacitors - polypropylene? To look for some high end mundorf?
Do any of these parts actually affect the sound?
Mills are power resistors for passive crossovers. You want 1/4W metal film resistors that can mount
on a pcb for an active crossover.
Mundorf are physically large capacitors for passive crossovers. You want small, pcb mounting
polypropylene capacitors for an active crossover.
Be sure to use the ESP board for this circuit, it is complex and would not work well otherwise.
on a pcb for an active crossover.
Mundorf are physically large capacitors for passive crossovers. You want small, pcb mounting
polypropylene capacitors for an active crossover.
Be sure to use the ESP board for this circuit, it is complex and would not work well otherwise.
Last edited:
Hello,
Just some pointers from my building and use of a Marchand XM-9 and a Pete Millett Tube Electronic Crossovers.
The resistors and capacitors used in the op-amp feedback loops that perform the frequency divider functions need to be tight tolerance and closely matched. Use 1% film caps, order extra, and hand match them, if you can.
Use, at least, 1% metal film resistors. Vishay, TECon., KOA, YAGEO, pick your poison. I like Vishay but you may find supplies are getting short on some values at most vendors.
Don't get all fancy with op-amp rolling until you get your troubleshooting done and the unit is up, running, and stable. Start with a cheap and cheerful op-amp that is unity gain stable like the OPA2134.
Order and use Rod's PCB. His PCBs are very nice and make building a breeze. Read everything on the circuit on Rod's site, twice!
I started using electronic crossovers over 20 years ago, and have never looked back!
Have fun!
Dan
Just some pointers from my building and use of a Marchand XM-9 and a Pete Millett Tube Electronic Crossovers.
The resistors and capacitors used in the op-amp feedback loops that perform the frequency divider functions need to be tight tolerance and closely matched. Use 1% film caps, order extra, and hand match them, if you can.
Use, at least, 1% metal film resistors. Vishay, TECon., KOA, YAGEO, pick your poison. I like Vishay but you may find supplies are getting short on some values at most vendors.
Don't get all fancy with op-amp rolling until you get your troubleshooting done and the unit is up, running, and stable. Start with a cheap and cheerful op-amp that is unity gain stable like the OPA2134.
Order and use Rod's PCB. His PCBs are very nice and make building a breeze. Read everything on the circuit on Rod's site, twice!
I started using electronic crossovers over 20 years ago, and have never looked back!
Have fun!
Dan
Hi,
I like Tyco´s R-series (YR1xxxx in 0207 casing) for throughhole and Susmu´s Rxx-series for SMD.
The Tycos are high precision, very stable NiCr thinfilm resistors at a very competitive price, especially when bought at numbers of >10.
The NiCr layer is sputtered on a ceramic element, end caps are welded on and tinned copper wire is used for its legs.
The Susumus are basically the same without legs but typically a bit costier.
Both types don´t cost much more than standard thickfilm resistors .. well at least when You are used to low parts count numbers and HighEnd prices
They excel in applications where low tolerance, stability and precision are mandatory, such as xover filters, RIAA, etc.
For caps I like to use low tolerance silver mica or PPs (see Kemets F461-464 series) or PEN and PPT dielectrica for SMDs, as well as COG/NPO ceramics.
Two points that imho almost guarantee for off-optimal sonic results in xovers are standard textbook filter curves and the use of OPAmps.
All speaker drivers require some means of equalization to achieve the desired acoustical xover shape.
It is standard good design with passive filters to design the equing and the filter function at the same to reduce the parts number count and with the driver´s own response.
Hence a good passive filter will electrically not have a textbook filter response.
But amost all active filters aim at perfect textbook response and fail to include the equalizers, or they add dedicated equing stages, increasing complexity and killing sonic performance.
Designing the active filters just as it is typical with passive filters and the driver response in mind will result in electrical responses off of textbook but less complex circuitry and better sonics.
Replacing OPAmps by simple discrete buffers -which is perfectly possible with unity-gain Sallen-Key filter designs- and you may actually be able to listen to music instead of technical playback ;-)
jauu
Calvin
I like Tyco´s R-series (YR1xxxx in 0207 casing) for throughhole and Susmu´s Rxx-series for SMD.
The Tycos are high precision, very stable NiCr thinfilm resistors at a very competitive price, especially when bought at numbers of >10.
The NiCr layer is sputtered on a ceramic element, end caps are welded on and tinned copper wire is used for its legs.
The Susumus are basically the same without legs but typically a bit costier.
Both types don´t cost much more than standard thickfilm resistors .. well at least when You are used to low parts count numbers and HighEnd prices
They excel in applications where low tolerance, stability and precision are mandatory, such as xover filters, RIAA, etc.
For caps I like to use low tolerance silver mica or PPs (see Kemets F461-464 series) or PEN and PPT dielectrica for SMDs, as well as COG/NPO ceramics.
Two points that imho almost guarantee for off-optimal sonic results in xovers are standard textbook filter curves and the use of OPAmps.
All speaker drivers require some means of equalization to achieve the desired acoustical xover shape.
It is standard good design with passive filters to design the equing and the filter function at the same to reduce the parts number count and with the driver´s own response.
Hence a good passive filter will electrically not have a textbook filter response.
But amost all active filters aim at perfect textbook response and fail to include the equalizers, or they add dedicated equing stages, increasing complexity and killing sonic performance.
Designing the active filters just as it is typical with passive filters and the driver response in mind will result in electrical responses off of textbook but less complex circuitry and better sonics.
Replacing OPAmps by simple discrete buffers -which is perfectly possible with unity-gain Sallen-Key filter designs- and you may actually be able to listen to music instead of technical playback ;-)
jauu
Calvin
The filter circuit impedances are around 10k or so, indicating low current noise is an advantage, so NE5532 or good JFET dual opamp would be good. Most discrete opamps have no proper datasheet to judge them by, and are hugely expensive. No point wasting the money - NE5532's are cheaper than most film capacitors, start there I'd suggest, you can try some high-performing JFET ones like the OPA1652 later (they need SOIC8 -> DIP breakout boards).
Use standard metal film, pay attention to the tolerance requirements in a filter circuit, especially higher order filters. If using surface mount components you need "thin-film" resistors, this is SMT speak for "metal film".
PP or PS or PPS or PTFE, whatever is cheapest at that tolerance. You could use PET but it has slightly more distortion. Definitely avoid ceramic unless you are _sure_ its C0G or NP0 (which are usually extremely good).
For all components go for reputable brands (and avoid anything labelled as "audiophile", that's a keyword that means "charge me 20 times the price for the same spec component and fill my head with nonsense")
Apart from ceramic caps you wouldn't be able to tell in a double blind test, but a good distortion analyzer should be able to tell PET from PP for instance. Metal film resistors are basically perfect compared to anything else in the circuit (as metals obey Ohm's law to very high precision).
You may be able to hear the noise floor with sensitive speakers, so make sure the opamps are low current noise(*) (less than 1pA/√Hz is good). For voltage noise less than 10nV/√Hz seems sensible (to match the 1pA, assuming circuit impedances are around 10k (10nV = 10kohms x 1pA)
(*) Very low voltage noise opamps like the AD797 are usually poor for current noise, so avoid these in this kind of circuit - they are intended for very low impedance mic and MC cartridge preamps.
Last edited:
Thanks for suggestions, just the info I was looking for. Considering all the parts, it shouldn't cost more than a 100 euros (PSU and PCB excluded), even when using opa2134.
On the other hand, how would you comment on an active crossover, that uses tubes??
I've seen two variants in marchandler's lineup, both costing significantly more.
Does replacing OPamps with tube circuitry produce better sound?
On the other hand, how would you comment on an active crossover, that uses tubes??
I've seen two variants in marchandler's lineup, both costing significantly more.
Does replacing OPamps with tube circuitry produce better sound?
You mean high tolerance? High tolerance means more precise (yes, its counter-intuitive, but that's how the word is used in precision engineering, and yes many people get this wrong because its fairly misleading at face value). You'll probably have heard the phrase "manufactured to the highest tolerances" - that doesn't mean "made with a stone tool"!!. (Unless its a granite surface plate)
So perhaps "tight tolerance" or "high precision" are better terms. Though usually all you need to do is order a 1% part or better if a 1% tolerance was specified.
After all 1% may be precision in one context and 0.01% may be hopelessly crude in another.
Last edited:
As an example, I use :
Resistors -- Vishay Dale RN55 tempco 25ppm/K
Capacitors -- WIMA FKP2 (preferred) or MKP2, depending on size / value
https://www.diyaudio.com/forums/ana...ve-crossover-filter-solution.html#post5591077
For Sallen Key buffers, I use discrete JFET source followers for load above 10k :
B1-Turbo on a Chip
For loads < 10k, I use discrete Sziklai JFET/BJT source followers instead :
B1-Turbo on a Chip
If gain is necessary, then discrete opamp with JFET input :
One Last Attempt at Discrete Opamp in DIP8 Footprint
If I have to use commercial opamp, then 2x ADA4627 on DIP8 adaptor.
For lower cost OPA2140 on DIP8 adaptor.
And low profile sockets Mill-Max 115-43-308-41-003000.
Patrick
Resistors -- Vishay Dale RN55 tempco 25ppm/K
Capacitors -- WIMA FKP2 (preferred) or MKP2, depending on size / value
https://www.diyaudio.com/forums/ana...ve-crossover-filter-solution.html#post5591077
For Sallen Key buffers, I use discrete JFET source followers for load above 10k :
B1-Turbo on a Chip
For loads < 10k, I use discrete Sziklai JFET/BJT source followers instead :
B1-Turbo on a Chip
If gain is necessary, then discrete opamp with JFET input :
One Last Attempt at Discrete Opamp in DIP8 Footprint
If I have to use commercial opamp, then 2x ADA4627 on DIP8 adaptor.
For lower cost OPA2140 on DIP8 adaptor.
And low profile sockets Mill-Max 115-43-308-41-003000.
Patrick
I would _strongly_ suggest for chains of multi-pole analog filters then opamps are very much better than discretes as each stage needs solid precise performance and ideal buffering to get reliable placing of the poles and zeros for accurate filter response.
A high Q Sallen/Key stage needs accurate x1 voltage gain for instance, not easily achievable with discrete followers driving frequency-dependent load impedances.
A high Q Sallen/Key stage needs accurate x1 voltage gain for instance, not easily achievable with discrete followers driving frequency-dependent load impedances.
Small update: I have ordered Rod's PCBs - one for 2way crossover and two for linear power supplies. I'd like to thank everyone for posting, it helped a lot.
Further dilemas:
1. Rod states that a metal box should be used and he put several advices for grounding. Would it be possible to put all those components in a wooden box, with a metal rear panel? Provided that rear panel is grounded and all other components (two PSUs, a crossover and a raspberry) are all connected to the same metal plate for ground. Wooden box would give me additional piece of mind, since only the rear panel can conduct electricity. It could be tucked away into a shelf and made unaccessible, just in case.
2. Is it possible for a Rod's P05 power supply to provide 3 amps at 5 volts (requirement of a raspberry pie)? I plan to order larger toroidal transformer of around 80VA with outputs of 15-0-15 (for crossover) and 9-0-9v (for raspberry), so the two PSUs would share the transformer.
3. Rod states that PSU PCB can be powered both by 15V and 15-0-15V transformer. Is there any advantages by using 15-0-15V over just a single 15v winding?
Further dilemas:
1. Rod states that a metal box should be used and he put several advices for grounding. Would it be possible to put all those components in a wooden box, with a metal rear panel? Provided that rear panel is grounded and all other components (two PSUs, a crossover and a raspberry) are all connected to the same metal plate for ground. Wooden box would give me additional piece of mind, since only the rear panel can conduct electricity. It could be tucked away into a shelf and made unaccessible, just in case.
2. Is it possible for a Rod's P05 power supply to provide 3 amps at 5 volts (requirement of a raspberry pie)? I plan to order larger toroidal transformer of around 80VA with outputs of 15-0-15 (for crossover) and 9-0-9v (for raspberry), so the two PSUs would share the transformer.
3. Rod states that PSU PCB can be powered both by 15V and 15-0-15V transformer. Is there any advantages by using 15-0-15V over just a single 15v winding?
Not really- only UHF and above frequencies can really sail through such slots, and will fields reduce with distance from the slot exponentially, at audio frequencies that's close to a perfect Faraday cage so long as the components aren't pressed up against the slots. If you're worried (I would worry about small metal objects (like coins) being able to drop through), just add a wire mesh layer under the slots.
Components are all mounted on the lower panel, but some of them (high capacitor and the top of a transformer) reach relatively close to cutouts - around 2-3 cm (~1 inch).
Could those cutouts boost a wi-fi signal in the case? There will be a small raspberry inside. I left a somewhat bigger gap around it's PCB antenna.