Good work but!!
If you read the whole of this thread you will see that the op is taking some very good advice and applying it .
I have no doubt that your amplifier is very quite but the layout has issues. in fact i have built many lm3886 amplifiers the first one i did had a terrible layout but it was still very quiet. In the 18 months i have been building these types of amplifier. I have found the layout has quite a profound effect on the sound of theses fickle I.C's.
Best Regards Ian
Belay those mods!
Everything I mentioned about "Pin 7" is wrong. It is not the "load ground".
So connect the speaker return ground wire directly between the filter cap grounds, near the left end of the chip.
The small decoupling caps should still connect as directly as possible between that load-ground return point and the power pins. And larger decoupling caps should too, but the distance isn't as critical for them. By the way, instead of 0.1 uF X7R caps, I would use the largest capacitance value you can get that still comes in the smallest case size.
The Zobel network's ground also needs to connect at the speaker return ground, by the filter caps.
By the way, I believe that you would get better decoupling performance if you used two 1000 uF electrolytics per rail (instead of one 2200 uF), especially if you could make it so that there was a total of two inches or less from chip power pin to cap pin and other cap pin to load ground point, for each 1000 uF cap. I believe that you would also want to have four to ten paralleled 2200 uF smoothing caps per rail, in the power supply. If you can then connect the rails from power supply to amp board with four inches or less of wire, you will have a nice-looking very low impedance for the power distribution system, over the entire frequency band that the chip can use, which should give excellent transient response.
Everything I mentioned about "Pin 7" is wrong. It is not the "load ground".
So connect the speaker return ground wire directly between the filter cap grounds, near the left end of the chip.
The small decoupling caps should still connect as directly as possible between that load-ground return point and the power pins. And larger decoupling caps should too, but the distance isn't as critical for them. By the way, instead of 0.1 uF X7R caps, I would use the largest capacitance value you can get that still comes in the smallest case size.
The Zobel network's ground also needs to connect at the speaker return ground, by the filter caps.
By the way, I believe that you would get better decoupling performance if you used two 1000 uF electrolytics per rail (instead of one 2200 uF), especially if you could make it so that there was a total of two inches or less from chip power pin to cap pin and other cap pin to load ground point, for each 1000 uF cap. I believe that you would also want to have four to ten paralleled 2200 uF smoothing caps per rail, in the power supply. If you can then connect the rails from power supply to amp board with four inches or less of wire, you will have a nice-looking very low impedance for the power distribution system, over the entire frequency band that the chip can use, which should give excellent transient response.
Last edited:
Hi Tom.
Yes i have made a double sided one . it makes the board real simple and seems to aid stability so much so that because the smoothing caps are so close to the I.C im not using any 100nF at the chip pins. In fact it is a very minimalist version Inverting version. I dont have to worry about the problems with a volume contorl as i use a dac.. 220k for feedback 10K on input. non inverting input straight to ground. 2x1400mF very high ripple rating on board . only 25mm of wire between trafo and rectifier and another 25mm from rectifier to the board only 10 mm of track from smoothing caps to chip pins. No zobel or any other form of filtering apart from the 10Uf film cap on the input. has been driving a 4ohm load with no problems and very litle heat for last 12 months . The speakers are however are a well behaved good load and do not dip too low also they are 91bd sensitivity so play lloud with not much power. Has a very magical sounding midrange although a bit forward sounding treble is very nice too. Bass is tight and detailed but can seem a little lacking somtimes due to the forwardness of the mid range.
Regards Ian
Wow, that sounds really great, Ian.
I'm paranoid about RF so I would want the 220 pF cap, for sure. And I would want to have both a Zobel AND a parallel series output resistor and inductor. You never know when the RF environment might change drastically for the worse, or some "odd" speaker cables might show up.
Anyway, about the 0.1 uF (100 nF) caps not being used: Coincidentally (and I believe that Ivor Katt might agree), I have been running simulations, lately, where any small cap (< 0.5 uF, say) in parallel makes everything worse, and it looks like they simply aren't needed if the larger caps are close-enough to the pins and the power and grounding layout is done right. My simulations were aimed at investigating decoupling cap configurations, so they included basically ALL of the parasitic inductances and resistances of the pcb traces and wiring, and the capacitors' geometries' inductances and internal series resistances, etc. And it turns out that large caps do high frequencies fairly well if there isn't much trace and lead inductance. Apparently smaller caps are used for very high frequencies just because they are physically smaller and so there is much less inductance due to the lengths of the conductive paths involved. (And at really high frequencies, all of these caps are only inductances, anyway.)
It turns out that if you have at least 200 uF close to the pins (although 470 uF is better and 1000 uF is better still, etc), then the large reservoir capacitances can be up to at least 4 inches away and the chip won't know the difference. Of course, there, too, the closer the better.
Anyway, if it were me, I would try temporarily tacking on some larger reservoir caps (probably 22000 uF per rail), with an inch or two of wire (or up to four inches), just to see if it affected the bass. Your 1400 uF can handle the transients but I wonder if strong bass could just simply deplete its charge, because it's not large enough. I doubt it but it might be worth a test. There, too, (i.e. for the 22000 uF test), it is better to use smaller caps in parallel rather than one larger one. And it's even better if they each have their own wires, instead of being "daisy-chained", because that way their inductances combine like parallel resistors and get smaller, which won't happen nearly as well, if at all, if they share mutual inductance, which would be the case if they were all paralleled on the same two conductors.
I believe that ideally you DO want to have SOME kind of transient-handling caps that have no more than two inches, TOTAL, for the paths from cap output to power pin plus cap ground to load ground return. And the less total distance the better.
So I recommend also adding at least SOME kind of somewhat-smaller caps that are very close to both the power pins and the load ground return, literally just because smaller ones will fit closer, where bigger ones can't, and the inductance of the longer conductors and larger cap geometries is very significant at the higher frequencies. Something like a 10 uF (but probably as big as you can fit) would probably help a lot. Even a large-ish polypropylene would probably be OK, too, like 2.2 uF or more, or two of them in parallel. But read on.
By the way, I have done some math that showed that 200 uF, with 20 nH total between its pins and the power pin and load ground point (probably about 0.33 inch of trace on each side would make 20 nH), is enough to provide all of the current needed, and exactly when needed, for a worst-case LM3886 transient, i.e. driving a 4-Ohm load from 0 Volts to 40 Volts in 4 microseconds (which is at or slightly above the LM3886's typical maximum slew rate), while pulling the power rail down by only 0.1 Volt. And yes, that included the cap geometry's inductance and the ESR. (But to be safe, use more capacitance, probably 470 uF if you can squeeze it in there, or, better yet, parallel smaller caps there.)
Last edited:
Hi Tom.
I can understand what you are saying about using larger resevoir caps and the chance that my relativley small 1400Uf would not be able to keep up with the demands of the amplifier. now 1400uF has plenty of ability to be able to supply all the power needed when it is playing at very low levels say a watt or so. I have tried larger resevoir caps many times and they really do affect the mid range with no help in the bass untill higer levels are required. I used to have a 220pF across the input pins to stop rf . i also used to have 100nf at each power input to the ic i thought these would be good to help the instant short term power requirments too especially for the higher frequencys. All the caps i have used are decent quality i have tried ceramics ,flm types and electrolytics.. They all took somthing away from the midrange and seemed to slow the amplifier affecting the attack of snare drums and the leading edge of symbols seemed to slurr. I think peter daniells has had the same problem and he too uses very little uF in the psu. Also i thought maybe there would be a very good chance of the amplifier becomming unstable with long input cables as this is an inverting amp and input is connected to feedback loop. i also concidered longer speaker cables so i tried that too i can report no problems . I then let a friend use the amplifier for a few months he is sympathetic to nothing and if it would go unstable and blow or missbehave in some way then it would be while he was abusing it . It survived and he nearly cried the day i took it back. I nearly forgot to mention if i use eq to pull the mid range back slightly the overall ballance is fine and the psu caps seem to cope with the bass no problems and remmember this amplifier is wide open apart from the input cap and can produce deep controlled bass. This little amplifier sounds awsome on jaz and very very nice on female voices. While messing with this little amp it has frustrated me to hell and then rewarded me with some wonderfull sound too. I havent experimented in a good few months as i needed to learn some other things to do with amplifier design. However all this talking has given me an hunger to start experimenting with the chips again . I have a few ideas to try . I just have a few jobs to do on an old cambridge p35 you mayhave guessed removing inductance and resistance from the supply lol so i have somthing to listen to . I cant wait . Thanks for re kidleing my intrest in the lm Chips again gootee. And a last word . inductance can be a bad thing in psu wiring i have noticed lots just like to ignore it and throw caps at it to cancel it out . I will be trying the ideas you gave me with the caps and will let you know how my ears percive the sound i love this hobby..
Best Regards Ian
.
I can understand what you are saying about using larger resevoir caps and the chance that my relativley small 1400Uf would not be able to keep up with the demands of the amplifier. now 1400uF has plenty of ability to be able to supply all the power needed when it is playing at very low levels say a watt or so. I have tried larger resevoir caps many times and they really do affect the mid range with no help in the bass untill higer levels are required. I used to have a 220pF across the input pins to stop rf . i also used to have 100nf at each power input to the ic i thought these would be good to help the instant short term power requirments too especially for the higher frequencys. All the caps i have used are decent quality i have tried ceramics ,flm types and electrolytics.. They all took somthing away from the midrange and seemed to slow the amplifier affecting the attack of snare drums and the leading edge of symbols seemed to slurr. I think peter daniells has had the same problem and he too uses very little uF in the psu. Also i thought maybe there would be a very good chance of the amplifier becomming unstable with long input cables as this is an inverting amp and input is connected to feedback loop. i also concidered longer speaker cables so i tried that too i can report no problems . I then let a friend use the amplifier for a few months he is sympathetic to nothing and if it would go unstable and blow or missbehave in some way then it would be while he was abusing it . It survived and he nearly cried the day i took it back. I nearly forgot to mention if i use eq to pull the mid range back slightly the overall ballance is fine and the psu caps seem to cope with the bass no problems and remmember this amplifier is wide open apart from the input cap and can produce deep controlled bass. This little amplifier sounds awsome on jaz and very very nice on female voices. While messing with this little amp it has frustrated me to hell and then rewarded me with some wonderfull sound too. I havent experimented in a good few months as i needed to learn some other things to do with amplifier design. However all this talking has given me an hunger to start experimenting with the chips again . I have a few ideas to try . I just have a few jobs to do on an old cambridge p35 you mayhave guessed removing inductance and resistance from the supply lol so i have somthing to listen to . I cant wait . Thanks for re kidleing my intrest in the lm Chips again gootee. And a last word . inductance can be a bad thing in psu wiring i have noticed lots just like to ignore it and throw caps at it to cancel it out . I will be trying the ideas you gave me with the caps and will let you know how my ears percive the sound i love this hobby..
Best Regards Ian
.
Last edited:
Ian,
You might want to check out Post 252 (and related material), at:
http://www.diyaudio.com/forums/power-supplies/106648-paralleling-film-caps-electrolytic-caps-26.html
and Posts 221-230 here:
http://www.diyaudio.com/forums/power-supplies/106648-paralleling-film-caps-electrolytic-caps-23.html
Actually, you should probably go back to at least Post 202, and go forward from there.
Cheers,
Tom
You might want to check out Post 252 (and related material), at:
http://www.diyaudio.com/forums/power-supplies/106648-paralleling-film-caps-electrolytic-caps-26.html
and Posts 221-230 here:
http://www.diyaudio.com/forums/power-supplies/106648-paralleling-film-caps-electrolytic-caps-23.html
Actually, you should probably go back to at least Post 202, and go forward from there.
Cheers,
Tom
Sorry for revive an old post, but i modify the quadtech PCB and i want to know what do u think about it. Here it is:
I change some components to smd like some decoupling caps and some resistor.
Any opinion?
An externally hosted image should be here but it was not working when we last tested it.
I change some components to smd like some decoupling caps and some resistor.
Any opinion?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.