Another killer on the loose is ,ehem, forgive me, but Peavey VSX 26.
Beats me, but I like it better than BSS Omni. And its' dirt cheap.
Just looked inside of it, and its output stage is just as crappy as it should be. How would it sound with a reasonable analogue stage...
Near-impossibe to tweak, anyhow, with its plethora of SMD components.
Boys there has anyway got something right with this unit. Is it that Blackfin or what, I dunno..
Back to the on-topic: Those powerful shelving features above and below XO freqvencies has tended my choise to I-O Xovers, altough I otherwise prefer analogue reproduction. It's just a trade-off,
and I know my gain might very well be pain of some one else
Beats me, but I like it better than BSS Omni. And its' dirt cheap.
Just looked inside of it, and its output stage is just as crappy as it should be. How would it sound with a reasonable analogue stage...
Near-impossibe to tweak, anyhow, with its plethora of SMD components.
Boys there has anyway got something right with this unit. Is it that Blackfin or what, I dunno..
Back to the on-topic: Those powerful shelving features above and below XO freqvencies has tended my choise to I-O Xovers, altough I otherwise prefer analogue reproduction. It's just a trade-off,
and I know my gain might very well be pain of some one else
The quality of the Marchand will be set by the National Semiconductor LF353 opamps (price = 42 cents each in 95 qty.) and any associated electrolytic coupling caps - there are probably several in series before the signal makes it all the way through the crossover. Specs of the LF353 are average, nothing too amazing, but a long way from the worst. The JFET input helps reduce DC offsets, which makes it possible to use fewer DC-blocking electrolytic caps. Upgrading the electrolytics is probably worthwhile, at least those in the direct signal path.
As for tube gear, stay away from both the Marchand and Audio Research units - high-distortion circuits run at weird operating points. I'm not aware of any commercial tube crossovers that are well-designed. At least the solid-state stuff is harder to mess up, thus the generally better sonics.
As for tube gear, stay away from both the Marchand and Audio Research units - high-distortion circuits run at weird operating points. I'm not aware of any commercial tube crossovers that are well-designed. At least the solid-state stuff is harder to mess up, thus the generally better sonics.
Some interesting recommendations here by Chucko of Silicon Valley. I'm a "video op amp fetishist", so the LM6172 or LT1358 look good to me - I like really fast transistor circuits. The downside can be high input bias currents, which in turn can require the nuisance of electrolytic coupling caps on the output to get rid of the DC offsets.
There's a nice recommendation for the JFET-input OPA2134, which has pretty impressive specifications and can probably be DC-coupled throughout.
There's a nice recommendation for the JFET-input OPA2134, which has pretty impressive specifications and can probably be DC-coupled throughout.
Lynn Olson said:
The LF353 is OK - so is the OPA2134PA.
I use the superior LM4562 opamps - WOW
I use Teflon caps and film caps everywhere including the power supply except the little ceramics on the 4562s
I also don't use metal film resistors - I like the sound of matched carbons
For attenuators I use Daven stepped attenuators.
My circuit guru Uncle built the power supply, it's in a separate box
The pinout is standard so that part is easy. What may not be compatible is the power supply bypassing since the LM4562 is a faster chip. You'll probaby need to add some good quality caps bypassing the rails to ground right at each chip to avoid oscillation issues. Keep the leads as short as possible.
Borbely high end cross over
The best discrete cross over I have ever seen is Borbely Audio's
cross over kits.
JFET designs!
http://www.borbelyaudio.com/products.asp
There are three different types of kits:
EB-802/214"Bass Extender Board"
EB-1102/215"Basic Crossover Board"
EB-1102/216"Extension Board"
Here's the block diagram for the discrete 215:
Can be used with balanced or SE inputs.
Power supply could be Borbely's super shunt reg for ex:
Sigurd
The best discrete cross over I have ever seen is Borbely Audio's
cross over kits.
JFET designs!
http://www.borbelyaudio.com/products.asp
There are three different types of kits:
EB-802/214"Bass Extender Board"
EB-1102/215"Basic Crossover Board"
EB-1102/216"Extension Board"
Here's the block diagram for the discrete 215:
An externally hosted image should be here but it was not working when we last tested it.
Can be used with balanced or SE inputs.
Power supply could be Borbely's super shunt reg for ex:
An externally hosted image should be here but it was not working when we last tested it.
Sigurd
Hi
As for opamps I wouldn't go with JFet's ( and for sure not with the LF353 ! ) as all of them I was auditioning shared a very specific soft-focus characteristic in sound .
Beside the never ending discussion about "what is the best sounding opamp" currently, I'd like to draw the attention also towards the general architecture .
Given a tweeter / mid / bass / sub arrangement we are talking about 8 channels in total.
One of the main culprits in active XO design is about grounding technique and power supply consideration.
Beside NAIM I am not aware of any other manufacturer nowadays that consequently considered all impacts to grounding throughout an asymmetric operated chain ( the one-ground DIN connectors are a direct consequence from this, actually ).
In fact what makes the main difference is that you split into several branches ( eight in our example ) at the XO unit rather than just dealing with a daisy chain device.
Lets assume at least 3 ground returns for power supply and IN and another 7-8 ground returns ( filters / OUT / bypass) for each branch. Even IF you manage to keep the filters that simple this easily makes 50+ returns !
To bring them all to a centre-star-point with SHORT leads is almost impossible. This applies equally to discrete ss, opamp and valve designs .
If you decide to go the symmetric route with tubes, you end up with 8 - 10 transformers that will cost a fortune and limit harmonics to about –60 dB ( though I admit this must not necessarily turn out to be that bad for the mere sound ) along with some pickyness regarding source and / or load impedance matching ( and cable selection on longer runs ).
At least ONE thing the "B" does better than most designs I have seen so far is that they use INTERNAL symmetric signal transfer at the point where they jump from the DA-chip to the separate filter / output board . A quite unusual feature I haven't seen to be recognised and emphasised much .
Interstage cross talk via the power supply is also not easy to handle in such a multistage device and may cause some further head ache to get it right . This easily mud's the sound unbearable – much more than simple distortion due to imperfect circuits or devices . Especially as there are more or less heavy complex loads to be driven at certain stages rather than simple resistive ones .
Lynn, thanks for your great summary about the specific " cinema sound "in the other thread. This isn't pointed out clearly such often.
If you seriously intend to design an active crossover from scratch that meets your over the top requirements I have only little hope your OB project ever gets finished (maybe after another several thousand posts or so in this thread ).
Greetings
Michael
As for opamps I wouldn't go with JFet's ( and for sure not with the LF353 ! ) as all of them I was auditioning shared a very specific soft-focus characteristic in sound .
Beside the never ending discussion about "what is the best sounding opamp" currently, I'd like to draw the attention also towards the general architecture .
Given a tweeter / mid / bass / sub arrangement we are talking about 8 channels in total.
One of the main culprits in active XO design is about grounding technique and power supply consideration.
Beside NAIM I am not aware of any other manufacturer nowadays that consequently considered all impacts to grounding throughout an asymmetric operated chain ( the one-ground DIN connectors are a direct consequence from this, actually ).
In fact what makes the main difference is that you split into several branches ( eight in our example ) at the XO unit rather than just dealing with a daisy chain device.
Lets assume at least 3 ground returns for power supply and IN and another 7-8 ground returns ( filters / OUT / bypass) for each branch. Even IF you manage to keep the filters that simple this easily makes 50+ returns !
To bring them all to a centre-star-point with SHORT leads is almost impossible. This applies equally to discrete ss, opamp and valve designs .
If you decide to go the symmetric route with tubes, you end up with 8 - 10 transformers that will cost a fortune and limit harmonics to about –60 dB ( though I admit this must not necessarily turn out to be that bad for the mere sound ) along with some pickyness regarding source and / or load impedance matching ( and cable selection on longer runs ).
At least ONE thing the "B" does better than most designs I have seen so far is that they use INTERNAL symmetric signal transfer at the point where they jump from the DA-chip to the separate filter / output board . A quite unusual feature I haven't seen to be recognised and emphasised much .
Interstage cross talk via the power supply is also not easy to handle in such a multistage device and may cause some further head ache to get it right . This easily mud's the sound unbearable – much more than simple distortion due to imperfect circuits or devices . Especially as there are more or less heavy complex loads to be driven at certain stages rather than simple resistive ones .
Lynn, thanks for your great summary about the specific " cinema sound "in the other thread. This isn't pointed out clearly such often.
If you seriously intend to design an active crossover from scratch that meets your over the top requirements I have only little hope your OB project ever gets finished (maybe after another several thousand posts or so in this thread
).Greetings
Michael
Well, it doesn't use any bias through the load. The DC current is just the remaining offset voltage that the two two stages with Vce of opposite polarity actually are intended to cancel out.
The source of this is most probably the different bias points of the two stages since Vce is dependant on base-current which by itself is dependant on collector current.
A slightly modified circuit (equal bias current and own emitter follower for Q1's collector) showed better behaviour in this respect. I Will post it later on.
Regards
Charles
Lynn,
I am wondering why do you need a crossover at all?
Maybe all you need is a pre-amp with two cathode followers using two output transformers for each channel.
Your para eq will handle the low pass, and a tube amp with a tiny capacitor (somewhere in the chain of signal) will handle the high pass.
I am wondering why do you need a crossover at all?
Maybe all you need is a pre-amp with two cathode followers using two output transformers for each channel.
Your para eq will handle the low pass, and a tube amp with a tiny capacitor (somewhere in the chain of signal) will handle the high pass.
mige0 said:Hi
Lynn, thanks for your great summary about the specific " cinema sound "in the other thread. This isn't pointed out clearly such often.
If you seriously intend to design an active crossover from scratch that meets your over the top requirements I have only little hope your OB project ever gets finished (maybe after another several thousand posts or so in this thread
Greetings
Michael
I'm not as much a fan of multi-amping as others. Amps like the Karna as so over-the-top, and take so long to design (several years), I have no desire to build several of these things.
I'm be splitting the low-level audio spectrum no more than two ways - one path for a prosound crossover/parametric EQ unit, which will drive something like a QSC pro amp, and the other path a passive HP filter, a minimalistic buffer, and the all-triode Karna, which uses input transformers for complete galvanic isolation from anything else. The dividing frequency is going to be somewhere between 300 and 600 Hz, depending on the choice of midrange driver.
By the way, that -60 dB 2nd-harmonic distortion figure most likely means you have a bad sample, or perhaps your test setup had some DC leakage going through the primary. 2nd harmonic in a transformer indicates DC somewhere, or a gross defect in the core material.
There are transformers out there, with no DC on primary, that have less than -120 dB distortion at 1 kHz and above - some candidates are Jensen professional, Lundahl, Dave Slagle, and Tribute. The Audio Precision system has always used a Jensen studio transformer on the input section of the analyzer - you don't see any -60 dB 2nd harmonic on an AP test system, do you?
Others of the all-transistor multi-amp persuasion can build their direct-coupled EQ and electronics as complex as they want - that's an area where I diverge from Siegfried Linkwitz, who is still firmly in the all-solid-state camp, opamps and all. I'd rather work on the OB system, that's a more fun and interesting challenge.
agent.5 said:
My amplifiers are not conventional RC-coupled tube amps - far from it. They are transformer-coupled throughout, and there aren't any good locations in the circuit to add RC-coupling without a major re-design. A re-design I don't want, since I don't care for the sound of RC-coupled circuits in the first place, and spent nearly ten years on the Amity, Raven, Aurora, and Karna getting rid of those miserable phase splitters and equally miserable RC-coupling. Get rid of those two things, and just about all of the traditional "PP tube amp sound" disappears.
See circuit here. As you can see, both over-the-top and no good place to add RC-coupling. Despite the archaic appearance of the circuit, the intrinsic forward-path distortion is one-tenth to one-twentieth that of an equivalent-power Williamson or Dynaco circuit.
Also not a fan of conventional cathode-followers, which to me, sound murky and have a lot more distortion than I'd like. A cascoded current-sourced cathode or emitter-follower, though, is another matter, with 100X less distortion than a simple resistor-loaded circuit. (I've measured the distortion reduction, it's real.)
The cascoded current-sourced follower is still the leading candidate for the highest-quality signal path - thanks again to joe, our friend from Oz, for the modern circuits - much appreciated!
Like the recording/mixing/mastering chain responsible for maybe 90% of recorded music?Lynn Olson said:
It would be fascinating to see the recorded media collection of those with SS/Op-amp free reproduction systems. I would love to hear all those recordings free of imbedded glare/grain played back on glare/grain free reproduction systems. How clear that must sound played through loudspeaker/room X!
I imagine Earl Geddes only used SS/Op-amp smeared recordings when using that Pioneer receiver at RMAF?
Perhaps if Earl reads this he can enlighten us.
cheers,
AJ
Lynn Olson said:
Anyone mentioned these, yet?
http://www.diyaudio.com/forums/showthread.php?s=&threadid=58757
http://www.diyaudio.com/forums/showthread.php?s=&threadid=59244
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64220
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.