Bridge mode Ref-C
I will not bore you too much on this subject again. Realised last night that Microcap 8's model of the opamp is the wrong way round, I have corrected it just in case somebody else wants to try it, and before I get condemned to the Looney bin and start writing articles for positive feedback while in confinement
Here is my last suggestion with correction and with a 100k input Z. BTW I know that BelCanto uses this technique for their digital amps. It is just unfortunate that I do not have there schematic.
I will not bore you too much on this subject again. Realised last night that Microcap 8's model of the opamp is the wrong way round, I have corrected it just in case somebody else wants to try it, and before I get condemned to the Looney bin and start writing articles for positive feedback while in confinement
Here is my last suggestion with correction and with a 100k input Z. BTW I know that BelCanto uses this technique for their digital amps. It is just unfortunate that I do not have there schematic.
Attachments
Rudy,
I doubt anyone would really need more than the ~40 good Watts RMS that "My_Ref" can give (for home use, that is). Any decent speaker already produces its maximum SPL with a few Watts, and only create unacceptable distorsions with anything more than that. What would all those extra watts be needed for?
Thus all this story about bridging is kind of silly... but, if you really want to try and get more, you should consider a few things:
1) a couple of bridged "My_Ref"s will NOT behave the same as a single one, neither technically nor sonically. It would be a different beast, and without careful investigations and though re-design the results are unpredictable.
2) if you allow for more voltage (by bridging), you also need to allow for more current capability, or you'll only get less control and other nasty side effects. That is, you would need to parallel, too.
3) if you carefully look at Mauro's schematic and really understand it, than you'll see that there are other ways to obtain a bridged circuit, without the need of using an external phase splitter in front of it.
Hint: the "current pump" in Mauro's design can be used in either inverting or non-inverting mode.
You can obtain both bridged and parallel operation "simply" by using more current pumps (half inverting, half not), all driven by a single "SE" signal from the LM318 (well, you'll probably either have to use separate LM318s for each pump or use some proper buffer after it).
Of course you'll have to change the way global (voltage) NFB is applied accordingly, take care of the influence of the changes on (i.e. re-design) the compensation, etc.
That is, you'll have to create a brand new design, though based on the very same principle of operation as of Mauro's one. BUT that's the way to go if you really want to get a more powerful amplifier based on Mauro's concept.
It's not trivial, but it can be done, if you know what you're doing.
Of course Mauro himself would be the best one to try to do such a though re-design, if he would be interested... which I doubt, unfortunately.
On the other end, if you just want to do a simple experiment, than I would rather use an audio transformer.
Just get a line transformer with 1:1+1 (that is, 1:2 with center tap) or any similar ratio (as long as the secondary is center-tapped) and you're done. You'll also get the extra benefit of galvanic isolation of the input.
RS sells some suitable audio transformers, designed for the "pro" field. They are small and dirty cheap, and should be good enough (BTW: have anyone tried 'em?).
BTW: actually, what I believe may be more useful instead would be to try the possibility of increasing the current capabilities by using more "blocks" in parallel...
As a "Quick & Dirty" simple experiment, just try to connect two (identical!) channels in parallel connecting together not only their inputs and outputs but also the pin 3 of their LM318s.
With some "nasty" loads (e.g. Thiels, electrostatics, etc) it may be quite useful to double the available output current...
I doubt anyone would really need more than the ~40 good Watts RMS that "My_Ref" can give (for home use, that is). Any decent speaker already produces its maximum SPL with a few Watts, and only create unacceptable distorsions with anything more than that. What would all those extra watts be needed for?
Thus all this story about bridging is kind of silly... but, if you really want to try and get more, you should consider a few things:
1) a couple of bridged "My_Ref"s will NOT behave the same as a single one, neither technically nor sonically. It would be a different beast, and without careful investigations and though re-design the results are unpredictable.
2) if you allow for more voltage (by bridging), you also need to allow for more current capability, or you'll only get less control and other nasty side effects. That is, you would need to parallel, too.
3) if you carefully look at Mauro's schematic and really understand it, than you'll see that there are other ways to obtain a bridged circuit, without the need of using an external phase splitter in front of it.
Hint: the "current pump" in Mauro's design can be used in either inverting or non-inverting mode.
You can obtain both bridged and parallel operation "simply" by using more current pumps (half inverting, half not), all driven by a single "SE" signal from the LM318 (well, you'll probably either have to use separate LM318s for each pump or use some proper buffer after it).
Of course you'll have to change the way global (voltage) NFB is applied accordingly, take care of the influence of the changes on (i.e. re-design) the compensation, etc.
That is, you'll have to create a brand new design, though based on the very same principle of operation as of Mauro's one. BUT that's the way to go if you really want to get a more powerful amplifier based on Mauro's concept.
It's not trivial, but it can be done, if you know what you're doing.
Of course Mauro himself would be the best one to try to do such a though re-design, if he would be interested... which I doubt, unfortunately.
On the other end, if you just want to do a simple experiment, than I would rather use an audio transformer.
Just get a line transformer with 1:1+1 (that is, 1:2 with center tap) or any similar ratio (as long as the secondary is center-tapped) and you're done. You'll also get the extra benefit of galvanic isolation of the input.
RS sells some suitable audio transformers, designed for the "pro" field. They are small and dirty cheap, and should be good enough (BTW: have anyone tried 'em?).
BTW: actually, what I believe may be more useful instead would be to try the possibility of increasing the current capabilities by using more "blocks" in parallel...
As a "Quick & Dirty" simple experiment, just try to connect two (identical!) channels in parallel connecting together not only their inputs and outputs but also the pin 3 of their LM318s.
With some "nasty" loads (e.g. Thiels, electrostatics, etc) it may be quite useful to double the available output current...
Hi Paulo
You have very valid points and for myself, I would never want to employ parallel devices wherever I can help it. And the amp in its current configuration work very well for me.
The person I want to build these for have very inefficient speakers and at full volume it does not make much of an impact even with the 31db gain of the amp.
The person that supplies me with ribbon tweeters also has a few newly developed ribbons that have an impedance of 1.5ohms (no transformer). The Ref C drove them at low levels but at higher levels they were clipping rather viscously. For him this would also work
thanks
Rudi
You have very valid points and for myself, I would never want to employ parallel devices wherever I can help it. And the amp in its current configuration work very well for me.
The person I want to build these for have very inefficient speakers and at full volume it does not make much of an impact even with the 31db gain of the amp.
The person that supplies me with ribbon tweeters also has a few newly developed ribbons that have an impedance of 1.5ohms (no transformer). The Ref C drove them at low levels but at higher levels they were clipping rather viscously. For him this would also work
thanks
Rudi
rudi said:
Troubles arise if you try to connect voltage sources in parallel, or current sources in series. Paralleling current sources, just like connecting voltage sources in series, creates no problems at all.
If you connect two or more Mauro's in parallel inside the global voltage NFB loop (that's what the q&d trick of connecting pin 3 together was for) basically you are connecting current sources in parallel. No problem!
check the actual power output... just measure the voltage on the speaker terminals with an analog DVM. If it's really hitting the ~33V limit, than you're right, they are so inefficient they would need more power... BUT...
in such a case, those would be really "black hole" type speakers. From my humble experience, you'll NEVER ever get anything "sound" out of such crap. Tell him to trash 'em, and get some decent ones instead!
Otherwise, most likely it is a CURRENT limit problem.
A single 3886 can not provide more than some 10A (specs says 7 to 11.5).
Thus, to get full power @35V without clipping (even worse: internal protection kicking in...), the load should never fall below ~4.5ohm at any frequency!!
With many "hard load" speakers (way more than you would guess...), you may need 4 (four!!!) parallel 3886s just to get the rated full power @ 35V !
the bridging? oh no, that would make things even WORSE!!!!
As said above, it will be the parallel connection that would help. And with such a low impedance he would probably need at least 3 or 4 parallel modules!!!
This reminds me of some ads... it said "power is nothing without control", or something like that... in this case, I would insist that:
Power means nothing without enough output current capability! 
UnixMan said:
Hi Unixman, I have been having a short email conversation with Mauro about a modular appoach to "My Ref" which he is working on. We should hear more soon, but if I understand Mauro correctly it will do just as you say. He is calling it "REV E".
Cheers!
Russ
[edit] fixed typo[/edit]
Me personally - for revT - it already sounds interesting but i like to jump on at a more 'mature' stage
UnixMan has some very valid points about bridging. In lesser designs it usually brings sonic improvements as it simplifies ground return currents.
If you still want to pursue a transformer phase-splitter i can only say good things about the Lundahl 1544 - cheap and very decent.
analog_sa said:
Me personally - for revT - it already sounds interesting but i like to jump on at a more 'mature' stage
I am not sure if you are being serious or not.
It's not at all a question of maturity, the changes made to "My REF" are really to accomodate different approaches to the same basic circuit, the same can be said for REV A vs REV C. Rev C is not really more mature or necessarily "better" than REV A unless you like the way it behaves for you. They are simply different, but based on the same core "current pump" driver. REV E will not really be any major shift as far as I know.
Cheers!
Russ
analog_sa said:
do perhaps know were i can get hold of two of these locally. would be interresting to do the experiment.
But i am also not convinced that it will give the best sound (that is the sytem in bridge mode compared to single ended) so personally i am staying to the current single ended design.
if you are interrested i still have a spare board till we get to ref-T