First off, I've been a lurker here for a few years. I've done one lm3875 build a couple of years back. Sounded pretty good. One of my buddies is still using it to this day without complaint. I'm plenty comfortable with an iron. I can read a schematic and get around eagle well enough to compile a PCB if I have to. That being said, time is going to be of the essence here. I have until the end of the month to produce this amp (read I'm slower than crap in eagle and it better be a gloriously kick *** circuit for me to spend that kind of time).
I've been issued a bit of a challenge and have been beating my head against the wall trying to decide which direction to go. I've got a $100 budget excluding power supply and chassis. Basically, I've got a buddy the has stated outright there is no way $100 worth of parts could ever out his quadruple digit amp in true sound reproduction, clarity, and sound stage. So...game on.
Wattage doesn't have to be off the charts...we aren't looking to deafen the neighbors though the ability to rattle the windows in his 1000sq/ft studio would be nice. Test speakers are in the 90db efficiency range and as this amp will be my own after hopefully thoroughly embarrassing him I'll be building my own speakers to suite once it finds its way back home.
As my previous experience was with a gainclone, that's the first thing that popped into my head but would I be better served going a different angle? If more than a decade later the gainclone is still the best option, then removing wattage from the equation (because honestly a well executed lm1875 is enough to rattle some windows) given rather flexible parameters, which chip or combination there of is going to result in the biggest head slap for this guy?
I have a dream...that one day I will make my best friend eat his own words with a nice side of crow. Help me make this dream a reality.
I've been issued a bit of a challenge and have been beating my head against the wall trying to decide which direction to go. I've got a $100 budget excluding power supply and chassis. Basically, I've got a buddy the has stated outright there is no way $100 worth of parts could ever out his quadruple digit amp in true sound reproduction, clarity, and sound stage. So...game on.
Wattage doesn't have to be off the charts...we aren't looking to deafen the neighbors though the ability to rattle the windows in his 1000sq/ft studio would be nice. Test speakers are in the 90db efficiency range and as this amp will be my own after hopefully thoroughly embarrassing him I'll be building my own speakers to suite once it finds its way back home.
As my previous experience was with a gainclone, that's the first thing that popped into my head but would I be better served going a different angle? If more than a decade later the gainclone is still the best option, then removing wattage from the equation (because honestly a well executed lm1875 is enough to rattle some windows) given rather flexible parameters, which chip or combination there of is going to result in the biggest head slap for this guy?
I have a dream...that one day I will make my best friend eat his own words with a nice side of crow. Help me make this dream a reality.
wattage is important - really shouldn't expect a win unless you're close - so how big is your competitor's amp? what is the speaker Z?
bridging and paralleling power audio chip amps can get you up there to the mid 100s of W depending on load
peak current can be an issue with difficult loads, speakers with impedance dips from XO or even just pumping driver resonances and suddenly reversing phase can draw >5x the current than the "name plate" Ohms rating calculation
paralleling helps there too - just need lots more chip amps
Geddes explored some cheap home theater amps - only Pioneer's proprietary labeled chip amps met his crossover linearity requirements
once you are paralleling then you can Class AB bias a few 100 mA per chip each amp's outputs against the others - should help with crossover distortions of "lessor" commodity chip amps
some pursuing low distortion have wrapped a good quality audio op amp feedback loop around the whole chip amp
there are stability concerns that strictly limit how much improvement is possible with a added feedback loop - I don't totally agree with the choices a few such project amps here have made but some improvement in numbers is certainly possible
and I don't think I've seen such with chip amps but Black's Feedforward - used in Quad's power amps can be added as an "improver" even when feedback may not be able to be pushed further
this type feedforward correction uses a 2nd highly linear but lower power amp to correct a higher power amp output - the difficulty is the power combiner has to isolate the feedback inevitably used in both amps from interacting
I simmed a version with a DSL driver CFA op amp corrector and one of Bob Cordell's book example circuits
bridging and paralleling power audio chip amps can get you up there to the mid 100s of W depending on load
peak current can be an issue with difficult loads, speakers with impedance dips from XO or even just pumping driver resonances and suddenly reversing phase can draw >5x the current than the "name plate" Ohms rating calculation
paralleling helps there too - just need lots more chip amps
Geddes explored some cheap home theater amps - only Pioneer's proprietary labeled chip amps met his crossover linearity requirements
once you are paralleling then you can Class AB bias a few 100 mA per chip each amp's outputs against the others - should help with crossover distortions of "lessor" commodity chip amps
some pursuing low distortion have wrapped a good quality audio op amp feedback loop around the whole chip amp
there are stability concerns that strictly limit how much improvement is possible with a added feedback loop - I don't totally agree with the choices a few such project amps here have made but some improvement in numbers is certainly possible
and I don't think I've seen such with chip amps but Black's Feedforward - used in Quad's power amps can be added as an "improver" even when feedback may not be able to be pushed further
this type feedforward correction uses a 2nd highly linear but lower power amp to correct a higher power amp output - the difficulty is the power combiner has to isolate the feedback inevitably used in both amps from interacting
I simmed a version with a DSL driver CFA op amp corrector and one of Bob Cordell's book example circuits
Last edited:
As is truly his style he went and changed the game a bit. Quote "I said there's no way $100 worth of parts could beat A quadruple digit amp. Didn't say which one, so I've ordered a Marantz PM8004 to give ya a decent challenge". As I have heard the PM8004...he may have just technicalitied me out of the game.
So amp is 70wpc
Speakers are an excellent pair he devised. If I can come up with an amp to impress, he's gonna share his recipe with me...so that's whats on the line. 8 ohm 90 db to answer your question.
So amp is 70wpc
Speakers are an excellent pair he devised. If I can come up with an amp to impress, he's gonna share his recipe with me...so that's whats on the line. 8 ohm 90 db to answer your question.
"a chip amp" could easily mean an amp using power audio chip amps for output - no need to restrict yourself to just 1 chip per channel
after all few discrete amps use one transistor per channel on the output
70 W is no problem for bridged and paralleled chip amps - and a few chips can handle the required V in singles but I would still recommend paralleling a few for extra current and power dissipation
after all few discrete amps use one transistor per channel on the output
70 W is no problem for bridged and paralleled chip amps - and a few chips can handle the required V in singles but I would still recommend paralleling a few for extra current and power dissipation
I'd go look for Panson's posts on driving the outputs of a couple TDA7293's (here's your chip amps!) with a LME49811. DC servo (TL072?) and probably a THAT1200 to make it play nice with everything. The parallel pair will allow you to run close to their max voltage.
That's ~$50 in chips, leaving half for PCB's, etc. Should be doable if you're sensible with parts. That's going to mostly be PCB, and a choice few capacitors around the output chips.
That's ~$50 in chips, leaving half for PCB's, etc. Should be doable if you're sensible with parts. That's going to mostly be PCB, and a choice few capacitors around the output chips.
I had actually been kicking around the idea of an opamp driving a pair. I had found a site out in the wild talking about a 49811 driving a pair of toshiba power amps...only to discover just a few minutes ago actually that the guy had discussed it here. http://www.diyaudio.com/forums/chip-amps/191019-lme49811-2sk1529-2sj200.html
I've always loved the idea of Tom's Modulus 86 composite, but that's outside of the scope of my budget for this one.
I've heard great things about some of the tda72xx. Definitely worth a second thought if I could work out a stereo pair within' budget.
I've always loved the idea of Tom's Modulus 86 composite, but that's outside of the scope of my budget for this one.
I've heard great things about some of the tda72xx. Definitely worth a second thought if I could work out a stereo pair within' budget.
Does your design have to be a chip amp based??
Note: I have nothing WHATSOEVER against chip-amps, having designed & built one around a TriPath TA2020.
I ask my question because there are a few beautifully simple, but staggeringly amazing amps from Nelson Pass. The one I am thinking of is basically a MOSFET clamped to a great big mahoosive heatsink.
As for op-amps, well you could do this discretely (and hence many would say therefore better) using cheapo transistors (& maybe JFETs).
Is this going to be an all through-hole based design or are you not adverse to using SMT devices?
Andy
Note: I have nothing WHATSOEVER against chip-amps, having designed & built one around a TriPath TA2020.
I ask my question because there are a few beautifully simple, but staggeringly amazing amps from Nelson Pass. The one I am thinking of is basically a MOSFET clamped to a great big mahoosive heatsink.
As for op-amps, well you could do this discretely (and hence many would say therefore better) using cheapo transistors (& maybe JFETs).
Is this going to be an all through-hole based design or are you not adverse to using SMT devices?
Andy
no point in suggesting Tomchr's implementation the PCBs alone use up the budget.
Then you need nine times as much to finish the project.
Allow 30% for chassis, 30% for psu, 30% for all the hardware and heatsinks and 10% for the amplifier.
Yeah, don't get me wrong I love the idea of Tom's Modulus build, but that's definitely outside of the scope this go around as I couldn't swing two of his PCB's in budget.
The $100 budget is amp only. I've amassed heatsyncs of every size imaginable and I've got spare SMPS's sitting around and a access to a fair number of toroidals. I've done some wood working a little bit of metal work in the past so chassis will be designed and built by myself...already have a fair number of the materials on standby.
Doesn't necessarily have to be chip based...its just all I've done thus far. And once you've replaced a north bridge on a $12000 multi-socket server, no form of soldering is much of a concern
I have been considering attempting a op driving a pair of power's for a while. Found a site that detailed an interesting build of a 49811 pushing a pair of toshibas. Had all but dismissed it until I discovered a few moments ago that the combo had been discussed here as well.
The $100 budget is amp only. I've amassed heatsyncs of every size imaginable and I've got spare SMPS's sitting around and a access to a fair number of toroidals. I've done some wood working a little bit of metal work in the past so chassis will be designed and built by myself...already have a fair number of the materials on standby.
Doesn't necessarily have to be chip based...its just all I've done thus far. And once you've replaced a north bridge on a $12000 multi-socket server, no form of soldering is much of a concern
I have been considering attempting a op driving a pair of power's for a while. Found a site that detailed an interesting build of a 49811 pushing a pair of toshibas. Had all but dismissed it until I discovered a few moments ago that the combo had been discussed here as well.
Maybe this reply will show up...
I like the idea of letting an LME498xx drive and I've heard some great things about the TDA7293. I love Tom's Modulus 98 design but yeah, PCB's alone would break me on this one. Definitely looking into my options for letting an LME498xx chip drive though.
To answer some other questions. Nope, doesn't HAVE to be chip amp...its just all I've done thus far. SMT components don't scare me...once you've scalped and replaced chips on a many 1000's of dollars multi-socketed server board, it takes a lot to worry ya
Thanks everyone for the suggestions thus far. Keep em' comin'!
I like the idea of letting an LME498xx drive and I've heard some great things about the TDA7293. I love Tom's Modulus 98 design but yeah, PCB's alone would break me on this one. Definitely looking into my options for letting an LME498xx chip drive though.
To answer some other questions. Nope, doesn't HAVE to be chip amp...its just all I've done thus far. SMT components don't scare me...once you've scalped and replaced chips on a many 1000's of dollars multi-socketed server board, it takes a lot to worry ya
Thanks everyone for the suggestions thus far. Keep em' comin'!
Ah! Okay--yeah, I'd probably pick out one of your PSU targets that gets you close to that magic 70w (into 8 ohms I'm assuming), and do a high-bias AB output stage (EF2 should be more than enough) around an LME49811. Plenty of decent output stages hanging around. Search up on TMC/TPC (the former being a type of the latter) to get more feedback around the output stage, as that *should* help with higher frequency numbers (whether that will matter or not is another thing).
I still would get a THAT1200 and a DC servo to make sure the amp is playing nice.
Also, someone neutral needs to run this challenge (and gains need matching) otherwise there is a bevy of reasons one amp will likely "win".
I still would get a THAT1200 and a DC servo to make sure the amp is playing nice.
Also, someone neutral needs to run this challenge (and gains need matching) otherwise there is a bevy of reasons one amp will likely "win".
$100 excluding power supply? What's included in the power supply - does it mean transformers, bridge and all supply caps? In which case you can spend as much as you'd like on the caps?
The way to get clarity (and the rest of the desired subjective characteristics) out of a chipamp is improve its rejection of supply noise. So having a low impedance supply is a pre-requisite but I've found running at higher voltage than required and stepping the output down with an output transformer helps a lot. The output trafo doesn't need to be anything expensive - a rewound EI core mains trafo does the trick.
The way to get clarity (and the rest of the desired subjective characteristics) out of a chipamp is improve its rejection of supply noise. So having a low impedance supply is a pre-requisite but I've found running at higher voltage than required and stepping the output down with an output transformer helps a lot. The output trafo doesn't need to be anything expensive - a rewound EI core mains trafo does the trick.
toroids give too much pri-secondary coupling
both direct capacitance from the layered windings and high magnetic coupling bandwidth
most want to isolate amp power from the power line EMI noise
if you aren't trying to minimize material cost, weight and size then you can have better quality in some important dimensions with split bobbin EI mains power transformers as long as you can space them a bit further form the rest of the circuitry and/or add Iron shielding
and more evidence for paralleling: http://www.diyaudio.com/forums/chip-amps/42578-about-linn-power-amps.html - but I would use an even number for the push-pull bias option I mentioned
especially with the TDA7293 - can find many other posts about the current output inadequacies, real need for paralleling them at higher V, output power
both direct capacitance from the layered windings and high magnetic coupling bandwidth
most want to isolate amp power from the power line EMI noise
if you aren't trying to minimize material cost, weight and size then you can have better quality in some important dimensions with split bobbin EI mains power transformers as long as you can space them a bit further form the rest of the circuitry and/or add Iron shielding
and more evidence for paralleling: http://www.diyaudio.com/forums/chip-amps/42578-about-linn-power-amps.html - but I would use an even number for the push-pull bias option I mentioned
especially with the TDA7293 - can find many other posts about the current output inadequacies, real need for paralleling them at higher V, output power
Last edited:
do you understand dynamic headroom? - realistic recordings without loudness war compression have over 20 dB peaks above the average level
even pure acoustic performances at the audience can have >110 dB SPL peaks
with 90 dB/W speakers that is 100 Watts required
clean clipping isn't as audible as some other distortions but I'm pretty sure a 5 W amp won't fare well in completion with 70 W unless you're using headphones
https://web.archive.org/web/20070710004619/http://www.headwize.com/articles/hearing_art.htm is a good read - read it twice in fact, both for protecting hearting and learning real world dynamic peak requirements
even pure acoustic performances at the audience can have >110 dB SPL peaks
with 90 dB/W speakers that is 100 Watts required
clean clipping isn't as audible as some other distortions but I'm pretty sure a 5 W amp won't fare well in completion with 70 W unless you're using headphones
https://web.archive.org/web/20070710004619/http://www.headwize.com/articles/hearing_art.htm is a good read - read it twice in fact, both for protecting hearting and learning real world dynamic peak requirements
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.