specifically wondering if bridged operation is worth the extra fuss over peter's single-ended setup, in terms of absolute sound quality. i have heard that some amplifiers sound better bridged, some worse, but much of that lies in the implementation details of course. any consequences to stability? how critical is precise symmetry of the two channels? not that it's hard to achieve with so few parts (can hand-select virtually every critical component - 2 resistors!). yes yes i know, i should just try it myself, but unfortunately i'm a little short on project time these days (hence my being MIA here the last several months) and like anyone else would like to profit from others' experience. 
one thing i definitely do want to incorporate is a transformer-balanced input... yes there are negatives associated with xformers (many of which are abated by use of very high quality design), but i feel it's a much preferrable way to obtain complementary signals over the traditional active solutions (and much easier!), and it has the very important side-benefit of providing RF input filtering... VERY critical in today's high-noise environments with a touchy power op-amp chip IMHO. if you use a good Jensen xformer, you get a low-pass Bessel filter around 100kHz or so for free. it's $$$ (US$100 a pop) but may be worth every penny in this application. i'm sure Kuei Yang would concur.
if i decided to not go bridged, any ideas on how a differential-input single chip gainclone would sound? or perhaps the chip will always sound better in simple singled-ended inverted input mode. aw fudge it, i may just have to stop being such a abum and buy all the parts to try it out myself...
cheers,
dorkus
one thing i definitely do want to incorporate is a transformer-balanced input... yes there are negatives associated with xformers (many of which are abated by use of very high quality design), but i feel it's a much preferrable way to obtain complementary signals over the traditional active solutions (and much easier!), and it has the very important side-benefit of providing RF input filtering... VERY critical in today's high-noise environments with a touchy power op-amp chip IMHO. if you use a good Jensen xformer, you get a low-pass Bessel filter around 100kHz or so for free. it's $$$ (US$100 a pop) but may be worth every penny in this application. i'm sure Kuei Yang would concur.
if i decided to not go bridged, any ideas on how a differential-input single chip gainclone would sound? or perhaps the chip will always sound better in simple singled-ended inverted input mode. aw fudge it, i may just have to stop being such a abum and buy all the parts to try it out myself...
cheers,
dorkus
Hi Dorkus
I pretty much agree with all you say and am also curious to know the answers. My GC uses premium quality components and construction (2 resistors only) and the sound is genuinely good. Though good in the sense that a SET amp is good - colour, tone, microdynamics. A wall of sound it certainly ain't Past experience with bridging is very positive: as there are no ground currents from the load some aspects really improve - one is a killer, very tuneful bass. And of course much improved macrodynamics. My problem is very low impedance speakers - if bridged, then paralleling is a must. About paralleling i am siginficantly less certain - i've never really liked parallel devices - neither at input, nor at output, and parallel +bridge will certainly complicate grounding and PS star-earthing and decoupling. P2P may no longer be an attractive option and don't forget those current-equalising resistors at output - they simply don't sound nice.
If i get around to building it, i'd start with the paralleling and see if i can live with the sound. Then i'll look into the inverting options. No experience with drv134 but BUF634, touted to be really good (and much better than 134) doesn't sound that transparent to me. So, even without trying i'd say the drv134 is not the route for absolute sound. An OPA627 inverter may be a better ticket. And finally, of course, there is the Jensen. It may very well be a good solution, i simply don't know. If you look in the archives, someone built an STK based bridged amp, again using the Jensen and the final result was claimed to be much better than the mickey mouse 3886 Rowland.
If you have highly efficient and easy speakers and listen to 'easy music' - jazz trios, girls with guitars and pianos, small ensemble classical, a good GC, same as good 45 SET is the best solution. If your speakers (and your music taste) are more difficult then it's probably worth to get into bridging. Ultimately there will be price to pay in transparency - bigger parts count, longer signal paths, more to screw up the sound, but it may still be worth it.
cheers
peter
I pretty much agree with all you say and am also curious to know the answers. My GC uses premium quality components and construction (2 resistors only) and the sound is genuinely good. Though good in the sense that a SET amp is good - colour, tone, microdynamics. A wall of sound it certainly ain't
Past experience with bridging is very positive: as there are no ground currents from the load some aspects really improve - one is a killer, very tuneful bass. And of course much improved macrodynamics. My problem is very low impedance speakers - if bridged, then paralleling is a must. About paralleling i am siginficantly less certain - i've never really liked parallel devices - neither at input, nor at output, and parallel +bridge will certainly complicate grounding and PS star-earthing and decoupling. P2P may no longer be an attractive option and don't forget those current-equalising resistors at output - they simply don't sound nice.If i get around to building it, i'd start with the paralleling and see if i can live with the sound. Then i'll look into the inverting options. No experience with drv134 but BUF634, touted to be really good (and much better than 134) doesn't sound that transparent to me. So, even without trying i'd say the drv134 is not the route for absolute sound. An OPA627 inverter may be a better ticket. And finally, of course, there is the Jensen. It may very well be a good solution, i simply don't know. If you look in the archives, someone built an STK based bridged amp, again using the Jensen and the final result was claimed to be much better than the mickey mouse 3886 Rowland.
If you have highly efficient and easy speakers and listen to 'easy music' - jazz trios, girls with guitars and pianos, small ensemble classical, a good GC, same as good 45 SET is the best solution. If your speakers (and your music taste) are more difficult then it's probably worth to get into bridging. Ultimately there will be price to pay in transparency - bigger parts count, longer signal paths, more to screw up the sound, but it may still be worth it.
cheers
peter
dorkus said:specifically wondering if bridged operation is worth the extra fuss over peter's single-ended setup, in terms of absolute sound quality. i have heard that some amplifiers sound better bridged, some worse, but much of that lies in the implementation details of course. any consequences to stability? how critical is precise symmetry of the two channels? not that it's hard to achieve with so few parts (can hand-select virtually every critical component - 2 resistors!). yes yes i know, i should just try it myself, but unfortunately i'm a little short on project time these days (hence my being MIA here the last several months) and like anyone else would like to profit from others' experience.
one thing i definitely do want to incorporate is a transformer-balanced input... yes there are negatives associated with xformers (many of which are abated by use of very high quality design), but i feel it's a much preferrable way to obtain complementary signals over the traditional active solutions (and much easier!), and it has the very important side-benefit of providing RF input filtering... VERY critical in today's high-noise environments with a touchy power op-amp chip IMHO. if you use a good Jensen xformer, you get a low-pass Bessel filter around 100kHz or so for free. it's $$$ (US$100 a pop) but may be worth every penny in this application. i'm sure Kuei Yang would concur.
if i decided to not go bridged, any ideas on how a differential-input single chip gainclone would sound? or perhaps the chip will always sound better in simple singled-ended inverted input mode. aw fudge it, i may just have to stop being such a abum and buy all the parts to try it out myself...
cheers,
dorkus
$100 for each transformer?
the DRV134s were free. there is allways a price to be paid....use effective speakers and you are able to play with small but very transparent amps - but then you have the problems in the speakers, nonlinearities and coloured sound..........or use very "wellbehaved" speakers and you need lots of power = big and sophisticated amplifiers = $$$
hi peter,
thanks for your reply. i have in fact heard a gainclone of sorts in my system, a commercial product (Blue Circle Music Pumps). it uses LM1875 housed in a size 12 lady's high-heel shoe (i kid you not), and while it had some problems sonically, with the right speakers it did some things phenomenally well. while i would hesitate to call the sound "tubey" (it was a bit too lean and bright to be tubey really, though that could have been this particular implementation), you are absolutely right in that it has a rightness of color, tone, and microdynamics. with a small 2-way transmission line, min. impedance about 9 ohms, there was a beauty to intimate musical passages i had never heard before in my system - never has the oboe solo in the 2nd mov't of Rimsky-Korsakov Scherezade been so captivating, or have i gotten so many goosebumps in Ella Fitzgerald's Moonlight in Vermont. you are also correct that macrodynamics are a problem, and with large orchestral music (my primary material) the sound tended to fall apart.
i do indeed remember that STK amp thread, rljones did some fantastic work there... shame the chips he used are no longer available. it was his success that inspired me the most to pursue a bridgeclone, though the LM chips do not have the current capability of the STK's. still, i think my intended speakers are benign enough (particularly if bi-amped) to get by with bridging w/o paralleling (paralell is a whole other can of worms i have no interest in opening).
i think a bridged gainclone using LM3875 on < 24V rails (18-0-18 secondaries) using Jensen input transformer may be a good bet, so perhaps that is what i should try out. because of the need to drive multiple modules, i am considering following the xformer with a simple JFET buffer a la Erno Borbely... i have experimented with one of his designs and found it very promising. i think the sound quality of the buffer may be complementary of the gainclone, and it may help to flesh out the sound a little more by balancing interstage impedances. using a large enough supply xformer with adequate regulation factor, separate rectifiers, and a straightforward pass MOSFET regulator for the buffer, there would be little additional complication of the power supply. AC coupling between the buffer and amp chip would not seem to be too detrimental either, since Peter Daniel has this in place in his circuit anyway. my only questions are with the grounding scheme of the whole thing, what with the multiple modules and bridged topology now.
sound like a worthwhile project? i'll start compiling a parts list.
thanks for your reply. i have in fact heard a gainclone of sorts in my system, a commercial product (Blue Circle Music Pumps). it uses LM1875 housed in a size 12 lady's high-heel shoe (i kid you not), and while it had some problems sonically, with the right speakers it did some things phenomenally well. while i would hesitate to call the sound "tubey" (it was a bit too lean and bright to be tubey really, though that could have been this particular implementation), you are absolutely right in that it has a rightness of color, tone, and microdynamics. with a small 2-way transmission line, min. impedance about 9 ohms, there was a beauty to intimate musical passages i had never heard before in my system - never has the oboe solo in the 2nd mov't of Rimsky-Korsakov Scherezade been so captivating, or have i gotten so many goosebumps in Ella Fitzgerald's Moonlight in Vermont. you are also correct that macrodynamics are a problem, and with large orchestral music (my primary material) the sound tended to fall apart.
i do indeed remember that STK amp thread, rljones did some fantastic work there... shame the chips he used are no longer available. it was his success that inspired me the most to pursue a bridgeclone, though the LM chips do not have the current capability of the STK's. still, i think my intended speakers are benign enough (particularly if bi-amped) to get by with bridging w/o paralleling (paralell is a whole other can of worms i have no interest in opening).
i think a bridged gainclone using LM3875 on < 24V rails (18-0-18 secondaries) using Jensen input transformer may be a good bet, so perhaps that is what i should try out. because of the need to drive multiple modules, i am considering following the xformer with a simple JFET buffer a la Erno Borbely... i have experimented with one of his designs and found it very promising. i think the sound quality of the buffer may be complementary of the gainclone, and it may help to flesh out the sound a little more by balancing interstage impedances. using a large enough supply xformer with adequate regulation factor, separate rectifiers, and a straightforward pass MOSFET regulator for the buffer, there would be little additional complication of the power supply. AC coupling between the buffer and amp chip would not seem to be too detrimental either, since Peter Daniel has this in place in his circuit anyway. my only questions are with the grounding scheme of the whole thing, what with the multiple modules and bridged topology now.
sound like a worthwhile project? i'll start compiling a parts list.
I was also looking into bridging, but didn't have the motivation to try it out yet. It seems to me that althought some advanages are possible, they would be offset by additional complexity of the input stage (transformer, buffers).
I'm not much into "easy" music, and my clone isn't falling apart even on industrial riffs. I believe that if you go for a GC, a right speaker is a ticket (although I didn't find a speaker yet, that would be problematic either).
I'm not much into "easy" music, and my clone isn't falling apart even on industrial riffs
. I believe that if you go for a GC, a right speaker is a ticket (although I didn't find a speaker yet, that would be problematic either).
Matttcattt said:
$100 for each transformer?
yeah, but they aren't jensen transformers, now are they?
if you must use an IC, i would suggest AD8620 dual opamp in complementary configuration. probably the best op-amp out there right now. use it in a tight surface-mount layout and the sound is quite transparent. you won't get the benefits of the transformer solution though.
hi peter (hpotter )
i agree the additional complexity of bridged makes me a bit wary and possibly not worth the investment in time/resources, but i'm wondering if the opportunity is maybe worth the trouble. perhaps the chip sounds really good bridged (like rljones's STK amps), perhaps not. while i am always in favor of simplifying the signal path, if you have to add active and passive components to achieve a means to an ends, Jensen transformers and JFETs are pretty good bets. i hear Jensens are very transparent (at least on par with a really good quality capacitor, maybe better), and i feel the LPF effect of a transformer is truly beneficial in this day and age of RF contamination. the JFET buffer is about as simple a discrete circuit as you can build, and with a little massaging it sounds very transparent. i think even w/o the bridging, some buffering of the input to gainclone is still useful... and if you're driving two modules for bi-amp or what not the 5k input impedance would be too uncomfortable for most preamps. a complementary JFET buffer should be able to drive such a load to 3V RMS with well under .01% THD, while improving the linearity of the input transformer and preceding stages as well.
so, you find the power output of the LM3875 to be quite adequate even with "harder" music, yes? i found the LM1875 to fall short of my needs, but that is a weaker chip and was being fed by only 75VA xformer per channel... i am considering using one big low-noise 500VA Plitron to feed all 4 channels of my gainclone, bridged or not.
btw, does anyone have any further comment on the stability implications of bridged operation? perhaps a RC on the output of each side is necessary to ensure a defined HF path to ground, since the load is no longer referenced to ground? any supply or grounding consequences?
p.s. oh peter, i had one question for you... you use one large supply for both channels yes? you have 4-wire interconnect of supply to each channel, preserving separate grounds for bipolar supply until the star point at the amp chip end. but, since there is a single supply, the + and - grounds are shared between channels. i suppose this could be viewed as a loop of sorts, but it's inconsequential right? i guess since it doesn't cause any detrimental paths for ground currents it's not a big deal.
)i agree the additional complexity of bridged makes me a bit wary and possibly not worth the investment in time/resources, but i'm wondering if the opportunity is maybe worth the trouble. perhaps the chip sounds really good bridged (like rljones's STK amps), perhaps not. while i am always in favor of simplifying the signal path, if you have to add active and passive components to achieve a means to an ends, Jensen transformers and JFETs are pretty good bets. i hear Jensens are very transparent (at least on par with a really good quality capacitor, maybe better), and i feel the LPF effect of a transformer is truly beneficial in this day and age of RF contamination. the JFET buffer is about as simple a discrete circuit as you can build, and with a little massaging it sounds very transparent. i think even w/o the bridging, some buffering of the input to gainclone is still useful... and if you're driving two modules for bi-amp or what not the 5k input impedance would be too uncomfortable for most preamps. a complementary JFET buffer should be able to drive such a load to 3V RMS with well under .01% THD, while improving the linearity of the input transformer and preceding stages as well.
so, you find the power output of the LM3875 to be quite adequate even with "harder" music, yes? i found the LM1875 to fall short of my needs, but that is a weaker chip and was being fed by only 75VA xformer per channel... i am considering using one big low-noise 500VA Plitron to feed all 4 channels of my gainclone, bridged or not.
btw, does anyone have any further comment on the stability implications of bridged operation? perhaps a RC on the output of each side is necessary to ensure a defined HF path to ground, since the load is no longer referenced to ground? any supply or grounding consequences?
p.s. oh peter, i had one question for you... you use one large supply for both channels yes? you have 4-wire interconnect of supply to each channel, preserving separate grounds for bipolar supply until the star point at the amp chip end. but, since there is a single supply, the + and - grounds are shared between channels. i suppose this could be viewed as a loop of sorts, but it's inconsequential right? i guess since it doesn't cause any detrimental paths for ground currents it's not a big deal.
Dorkus
Lucky you. My Sonus Fabers dip to 2.5ohm...
Yes, without the paralleling it will be a worthwhile project. I think you'll be pleasantly surprised with the sound of the 3875, it's a lot more mature and less 'in your face' than the 1875. About the FET buffer, i honestly don't know. It brings extra complexity, power supplies... If i were you i would first build it with no buffer and then try out the multiamping.
cheers
peter
Certainly not. Of course, after years of suffering with SETs i can adjust to almost any power limitations
I wouldn't think there will be any problems with stability of the BC. Not contaminating the ground with high load currents is probably the best feature of bridging in my view.
In fact it will be quite easy for me to try a single bridged channel. I don't have a proper transformer but a Lundahl 1660, while not ideal will probably allow me to experiment.
dorkus said:
It might be a loop of sort, but any way you look at this there is always some loop, when it comes to grounds, as they are usually shared in source equipment and supplies even if totally separate, have same reference to ground.
The grounds in PS are shared as I'm using a single 400VA Plitron. Originally, I run separate wires to ea. channel and the grounds were separate for ea. channel with only one channel's ground connected to chassis. But I found out that in some systems and setups this was causing some interferences in a channel which ground was not connected to chassis. So eventually I ended up with both channel's grounds connected to chassis (from binding post to central point on the chassis.
One other solution would be using mains earth to ground the chassis, but I didn't want to use extra wire required to accomplish that (not enough space in connectors).
If you keep this thread alive, I might try the bridge config. as well
After using Alephs for some time, I don't find GC lacking in any department. BTW, you wouldn't be interested in evaluating one of my amps?
analog_sa said:
yikes, yeah those little bass reflex Fabers can be quite a taxing load... being under the impression that you were more of a SET guy, i would have thought you would have had much kinder speakers!
good to hear the 3875 is more mature-sounding than the 1875, i found 1875 much too "raw," although it did have some definite strenghts.
i will have to try both with and without the JFET buffer of course, though i think if i bi-amp the load w/o buffer will be too low for comfort.
analog_sa said:
ah, very true... that does significantly simplify grounding, not having to worry about load return currents.
let us know if you try a BC, of course... even if i set my mind on doing this, we all know that it takes me a while to get things going, if i ever get to it at all... (*cough* son of dork...)
Peter Daniel said:
true, true... though one good thing about transformer coupling is of course the total isolation (if so implemented) of the source ground. although when i looked at Jensen's app notes (AN-003), i noticed they tend to tie source ground to chassis ground and a leg of the xformer secondary anyway. kind of odd, but i guess this is for maximum compatibility when using the balun as an adapter, rather than internally in equipment...
yuck, yeah i try to avoid using earth ground, though i probably should for safety (even Mr. Pass says so, though if i were building an X1000 i would probably want that beast grounded too)... i dont' think my apartment has properly grounded outlets anyway.
If you keep this thread alive, I might try the bridge config. as well
EVERYBODY KEEP POSTING.
btw, this multiple quoting in replies w/the checkboxes is great! i know it's been around for a while but this is my first time using it.
p.s. i sent you an email about evaluating one of your 'clones.
Konnichiwa,
A few basic cornerstones:
1) High Nickel content core material (Permalloy/Mu-Metal, Radiometal etc).
2) High Common Mode rejection and good secondary winding symmetry
3) Wide bandwidth
4) High Level handling
5) comparably high impedance
A relatively affrodable unit is the Stevens & Billington TX896 from Stevens & Billington which I am using in a new Amplifer Project (Valve based, strictly).
It is billed as 10k "Line Splitter" (1CT:1+1), can handle +12dbu @ 40Hz (THD < 0.5%) offers a +/-0.5db Bandwidth of 15Hz-46KHz and has >65db Common Mode rejection at 10KHz (more the lower the frequency).
http://www.stevens-billington.co.uk/mic_line_splitter.htm
Sayonara
miguel2 said:
A few basic cornerstones:
1) High Nickel content core material (Permalloy/Mu-Metal, Radiometal etc).
2) High Common Mode rejection and good secondary winding symmetry
3) Wide bandwidth
4) High Level handling
5) comparably high impedance
A relatively affrodable unit is the Stevens & Billington TX896 from Stevens & Billington which I am using in a new Amplifer Project (Valve based, strictly).
It is billed as 10k "Line Splitter" (1CT:1+1), can handle +12dbu @ 40Hz (THD < 0.5%) offers a +/-0.5db Bandwidth of 15Hz-46KHz and has >65db Common Mode rejection at 10KHz (more the lower the frequency).
http://www.stevens-billington.co.uk/mic_line_splitter.htm
Sayonara
Stupid Question time!
Does the voltage to power the actual LM3xxx chips come from the first OP-AMP? the schem only shows 15V going to the Opamp, but nothing connected to the LM3's. Is it just implied that the LM3's have a connection? Or is that the complete circuit?
Dang it...how do I make that appear? I did the [img.] [/img.] (without the "." of course), but its not a picture format I guess.
Does the voltage to power the actual LM3xxx chips come from the first OP-AMP? the schem only shows 15V going to the Opamp, but nothing connected to the LM3's. Is it just implied that the LM3's have a connection? Or is that the complete circuit?
Dang it...how do I make that appear? I did the [img.] [/img.] (without the "." of course), but its not a picture format I guess.
Hybrid fourdoor said:Stupid Question time!
Does the voltage to power the actual LM3xxx chips come from the first OP-AMP? the schem only shows 15V going to the Opamp, but nothing connected to the LM3's. Is it just implied that the LM3's have a connection? Or is that the complete circuit?
Dang it...how do I make that appear? I did the [img.] [/img.] (without the "." of course), but its not a picture format I guess.
I think it implies that the supply for the driver is separate from the LM3XXX supply and that the supply to those is present but different than the supply to the driver.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.