Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
203.12.194.242
I've been using using Rod Elliott's opamp-based active XO boards (P09) for my tri-amped 'Frankenpans' (and before them, my IIIas) for 15 years or so. These XOs boards are very flexible - they're nominally for 24dB slopes but I was able to modified them to deliver 18dB bass LP and 12dB slopes elsewhere that are comparable to the stock IIIa slopes. Plus they have gain control on each band.However, I had been looking around for the last 5 years, for an alternative which used discrete transistors, not opamps - and, late last year, I discovered these articles by Grey Rollins, on TNT Audio:
Part 1: http://www.enjoythemusic.com/diy/1109/xenover_2.htm
and
Part 2: http://www.enjoythemusic.com/diy/0310/xenover_part2.htm
So I am about to replace my Rod Elliott active XOs with these - which I believe offer 2 significant advantages:
1. Using discrete components, they will operate with far less feedback than an opamp design,
and - and far more importantly
2. Output DC offset can be adjusted to zero - which means there don't need to be any caps between the final XO buffer and the following power amp. (Whereas I need a cap on each amp, with the Rod Elliott XOs.)
So I thought I would share these links, in case anyone is interested in building some active XOs.
And yes, I thought about using PLLXOs but adding level controls made this problematical with my preamp.
Regards,Andy
Edits: 03/30/14 03/30/14Follow Ups:
I'm going to print out the schematics and see if I can make full-sense of all this.
The 6db / 12db versions would go perfectly to attempt to duplicate my 1.6 crossovers…..
Too much is never enough
Excellent - be sure to PM me if you decide to go ahead, as I'm a fair way down the track with developing my own version and am working on some improvements. This will have:
1. 18dB bass LP - which has the following "block diagram":
input buffer with pot for level control
---> 6dB lowpass filter
---> intermediate buffer
---> 12dB LP filter
---> final buffer with pot for zero DC offset.
2. 12dB mid bandpass:
---> 12dB LP filter
---> intermediate buffer
---> 12dB HP filter
---> final buffer with pot for zero DC offset.
3. ribbon buffer, to provide:
* level control, and
* zero DC offset.
Regards,
Andy
andyr wrote:
"2. Output DC offset can be adjusted to zero - which means there don't need to be any caps between the final XO buffer and the following power amp. (Whereas I need a cap on each amp, with the Rod Elliott XOs.)"
Most power amps have a cap at the input. Not a must to have one at the output of the crossover.
Correct, Roger - but, with this new JFET-based active XO, I can remove them from the new amp modules I am about to implement. No cap is the best cap! :-)) At least where coupling caps are concerned.
Regards,Andy
Edits: 03/31/14
Sometimes an input capacitor blocks input stage biasing (DC) from appearing at the input jack. Have you confirmed this will not be an issue with your new amps?
Dave.
andyr,
Are you bulding the new crossovers on PBCs or on prototype breadbords?
There is no PCB for these active XO circuits - so I will use my (patented!! :-)) ) construction method! :-)) This is to use a 3mm thick sheet of teflon to hold the components in place ... and do point-to-point wiring on the underneath. Teflon is easily drilled with a miniature hand-drill and withstands the heat from a soldering iron. Also, I suggest it degrades the signal much less than breadboard.
Regards,
Andy
Aha, but without a layer of copper it is difficult to have the components stay in place. Sure, I would like to see pictures when you are finished!
Here is a pic of the JFET-based phono stage that I make:
Yes, this does have a copper ground plane on the under-side ... but other components that I have used this method for - like PSes - have not ... and nothing untoward has happened.
Regards,
Andy
Andy,
What are these large film caps? Are they part of the frequency setting or supply bypass?
That is an early picture - v2 of the 'Muse'. The circuit is currently at v4, so I am using different caps now.
Regards,
Andy
I like DC coupling in principle. I like to see only one cap in the signal path if possible. If not, then use a coupling or other DC filtering cap as part of the XO if you can.
Surely if you are building a minimalist active XO you would be better off moving part of the XO (at least one pole from high pass filters) to the other side of your buffer or gain stage and not bother with the DC offset.
In any case, I am with you on not having Salen-Key feedback loops in your XO. It is why I sold my Marchand to a happy horn guy who got tons of foil caps for nearly nothing.
I am still trying to figure out how to do it like Manley in the Massive Passive parametric equalizer. Get lots of gain (and buffering) at the input and then do the filters. Perhaps output buffer with transformers?
Andy are you using laser trimmed op-amps such as the OPA2134? Your xo's are old (just like us) and I'm thinking you may have TL072's which Rod Elliott says are fine for this application but I don't think they are laser trimmed. Laser trimming minimizes DC offset. Just a thought.
Edits: 03/30/14
But R U sure 'laser trimming' removes DC offset? :-))
I measured some DC offset ... which was not a problem with my present amps bcoz they have an input coupling cap.
Regards,
Andy
Well, I thought the purpose of laser trimming was to eliminate the DC offset but more knowledgeable people than I (that means the rest of you guys) could correct me. In any case, the offset should only be a few mv ( <10) without trimming and almost all the newer opamps are trimmed.Have you discussed it with Rod Elliott to see why the problem exists?
BTW I have toyed with the idea of changing the OPA2134s but have never gotten around to it. Can you really hear a difference between the AD826 and OPA2134. At this point in my life I have reached audio nirvana as I can barely hear my wife in the background when I am listening. I think that is what is meant by a "blacker background" because when I get done with my listening sessions, there seems to be a dark cloud in the room :).
BTW since you seem to be headed toward a JFET DIY have you given thought to making it balanced rather than single- ended?
Edits: 04/01/14 04/01/14
No, neo. :-((
The rest of my system is not balanced and I do not have the design knowledge to do anything more than just use Grey Rollin's schematics and calculate the component values appropriate for my slopes & knee frequencies.
Regards,
Andy
Pots are terrible sounding. Yes, you get zero offset but you are listening to a a very non linear thing (tiny pots.....or any pot). What you want to do is to put a large value resistor across one of the source resistors till you get zero offset. Eliminate the pots. You can use a larger value pot to do the trimming and then replace the pot with a fixed resistor. This will sound way better. This basic circuit will sound better than any xover you can buy including the expensive Pass Labs thing. You are in for a treat!
I am not sure I follow you -not disagreeing just trying to wrap my head around it. Using a JFET buffer will require a pot to balance out the DC offset where as laser trimmed opamps don't need pots, yes? Using a pot and then replacing with fixed resistors seems like a very good idea -not only eliminates the pot but also eliminates re-tuning when the pot drifts. However since resistors come in fixed values, can a suitable value always be found to give zero DC?
I don't think that should be a problem, neo. People like Holco make resistors in all sorts of strange values and, if you create a "parallel resistors calc" spreadsheet, you can play with using several values to achieve the desired end-value.
Alternatively, if you need a value which is not quite the same what you end up with, using several available parallel resistors ... then you can make up a variable resistor combination which should be able to give you the final, fine level of adjustment you need.
Simply parallel:
* a fixed resistor which is around 10% greater than the value you need to end up with, and
* another fixed resistor plus a pot.
EG. if the value of the pot across the 22.1 ohm Source resistor in the schematic turned out to be 2,690 ohm, to get zero DC offset, you would use:
* a 3,160 ohm Holco, in parallel with
* a 10K resistor plus a 10K pot (set to 8,086 ohms).
I use this method for 7 crucial resistors in my 'Muse' phono stage - yes, you still have a pot in the mix but it has a resistor paralleling it which takes 70 or 80% of the current ... so the degradation associated with a pot is, IMO, substantially reduced (compared to just using a pot). And I use Vishay 1280G pots for this purpose, anyway.
(Again, using a "parallel resistors calc" spreadsheet is essential, here.)
Regards,
Andy
I use the same thing with all my level controls resistor (usually a mills wirewound or metal oxide film) is parallel to a pot, and if 0 is not a required resistance option, then I will add a series resistor to the pot and resistor parallel circuit to avoid a 0 resistance.
Do I understand you correctly:* I temporarily attach a large-value pot against each of the 2 x 22.1ohm Source resistors - on the +25v side, say (which will reduce these values) and adjust till I get zero DC offset?
Q: What is "large value", in this situation? 20K? 50K?
* If I can't (get zero DC offset), then it's the -ve side Source resistors which need to be reduced, to end up with zero DC offset.
* Then I replace these pots with fixed resistors of the same value?
Thanks,Andy
PS: I went to your web-site and read that you encountered Trev Lees, back in the 70s! He's even more eccentric now than he was then!! :-))
Edits: 03/30/14
Yes, try a 1K pot across one of the resistors and see which way the offset goes. If putting the pot on one resistor lowers the offset then turn it down....if it raises the offset then put it on the other resistor and then turn it down to zero it and then replace the pot with fixed resistor. The offset will drift with some time so let it burn for a while with the pot in there and then re-adjust the pot a little and then replace with fixed resistor.
.
Tks! I am bookmarking it. There's real promise there. I am not really fully against using opamps but I prefer discrete. Not that I need this for now, with the PLLXO working so well. However, I've kept my eyes open in case a pair of very cheap larger (3-way) Maggies suddenly fall on my lap. Something like this xover could then be the proper approach.
(BTW, I am going through some reading of recent findings on our malady. As soon as I finish I'll be sending you some email notes to see if the wallabies agree : - ))
.
Andy,You're essentially taking a step backward with those designs. You already have a superior design implemented in your system currently.
Discrete components have no inherent advantage regarding feedback usage or lack thereof. Much depends upon the topology utilized and not the packaging of the devices.
DC offset correction could easily be added to your existing op-amp designs to avoid the use of blocking capacitors if you have heartburn with those. A servo can employed or a static offset correction circuit.Cheers,
Dave.
Edits: 03/30/14
Are you saying that there is no feedback in Rod's opamp-based XO?
And re. your statement " DC offset correction could easily be added to your existing " ... it would be hard to add extra components to the P09 PCB.
Regards,
Andy
No, I'm not saying that. I'm saying that there is actually feedback employed in those JFET buffer stages. (It's not obvious, but it's there.) Most folks just analyze schematics looking for global feedback loops and if they don't see one assume it's a "no-feedback" design.
Of course, all this presumes that feedback is a bad thing. Many good audio design engineers would disagree with that. :)
"Easily" is not the correct word. Sorry. However, you could scab some daughtered breadboard type things into the P09 pcb and achieve. Probably though, the best approach would be to use different active crossover boards that have servo correction built on already. I've seen some nice DIY boards on DIYaudio.com a few years back. There was even a group buy IIRC.
Regardless, you should achieve a good result using those buffered passive circuits and it will be an excellent learning experience.
Cheers,
Dave.
Completely disagree. Op amps sound very compromised compared to zero feedback fet buffers. This is based on years and years of listening to opamps and other circuits.
Edits: 03/30/14
Completely disagree with your disagreement. I think we'll probably have to agree to disagree. :)
I accept your subjective evaluations. By definition they are incontrovertible.
Cheers,
Dave.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: