Hello, I have been viewing this forum for a few weeks but this is my first post. I find that people here are very helpful and have found answers to many of my questions but i have a few left. I'm building a gainclone using two LM3886 (one per ch.) @8ohms with an AS-3224 transformer http://www.antekinc.com/pdf/AS-3224.pdf like this http://www.chinabaike.com/uploads/allimg/120413/204G3L91-0.png but with 10uF film caps not .1uf as one user recommended also I may use more smaller main caps in series in place of one big cap per rail because several people have recommended so. My questions are: 1. I plan to place fuse(s) somewhere i know it should be slo-blo but where should that somewhere be and what current should the ratting be? 2. I might go with 2x 500ish uF per rail is this good? i plan to push this amp to the max from time to time if that means i need more uF so be it 3. What brand of caps and resistors? Im at loss with all the options and im kinda on a buget so something middle of the road or less would be nice but i still what it to sound good 4. Any good websites to order from? I live in the USA Thanks in advance for any feedback.
Looks like you have your bases covered OK here.
The fuse (which is a slo-blow) belongs on the primary side of your TX, along with the switch (DPDT) . Yes you can parallel out a pair of 500uF caps instead of one big 1000uF cap.......fact is you can pile on any values trying to get to 1000uF....in parallel of course. The 0.1uF are there to take out any remaining high frequency switching noise. You could get by just fine with some small caps across each diode(snubbers) to quiet down things.....and then you could delete the second set of 0.1uF caps.
Have you run your proposed PS thru any software? I would recommend "Duncan amps" free PS program....so you can simulate the performance.....and see if your design is up to snuff. Be very generous with voltages & amps......If you caps will see only 24V....use one rated for 63V.....If you diodes will see 2A....get one rated twice or more. Factory designs will run components right to the edge.......as a DIYer we all should be generous......besides, it's just pennies more to "overbuild".
_____________________________________________________Rick......
The fuse (which is a slo-blow) belongs on the primary side of your TX, along with the switch (DPDT) . Yes you can parallel out a pair of 500uF caps instead of one big 1000uF cap.......fact is you can pile on any values trying to get to 1000uF....in parallel of course. The 0.1uF are there to take out any remaining high frequency switching noise. You could get by just fine with some small caps across each diode(snubbers) to quiet down things.....and then you could delete the second set of 0.1uF caps.
Have you run your proposed PS thru any software? I would recommend "Duncan amps" free PS program....so you can simulate the performance.....and see if your design is up to snuff. Be very generous with voltages & amps......If you caps will see only 24V....use one rated for 63V.....If you diodes will see 2A....get one rated twice or more. Factory designs will run components right to the edge.......as a DIYer we all should be generous......besides, it's just pennies more to "overbuild".
_____________________________________________________Rick......
Thanks Richard for your input. So i should go with .1uF not 10 uF as another user said? I don't really want to get into snubbers as im trying to keep it basic at least for now. Im going with four 560uf main caps which of these Capacitors | Aluminum Capacitors | DigiKey would be the best for me (other than the ones that have a min. quantity in the 100's) longer life would be nice but not a must.
ECQ-P1H104FZW Panasonic Electronic Components | P3872-ND | DigiKey sorry i gave the wrong link would this fit the bill? on the specs. it says "Features High Frequency and High Stability" that what what made it stand out to me
Last edited:
Hi megadeth,
If i were you I would use both a 0.1uF and 10uF in parallel, this will help to smooth out a lot more noise than just a single cap. I would double what rick says about 'overbuilding', the one massive advantage of DIY'ing is you can build to much higher specs at 10% of the cost. A few extra capacitors will cost you pennies and go a long way.
Also note that it was pointed out these should be in parallel, capacitors in series are an exercise in futility.
If i were you I would use both a 0.1uF and 10uF in parallel, this will help to smooth out a lot more noise than just a single cap. I would double what rick says about 'overbuilding', the one massive advantage of DIY'ing is you can build to much higher specs at 10% of the cost. A few extra capacitors will cost you pennies and go a long way.
Also note that it was pointed out these should be in parallel, capacitors in series are an exercise in futility.
Thanks for posting. Yeah nannoo your right lol I got my wording backwards parallel not series. So going in order from wall to amp (left to right in the schematic im basing it on from my first post) trans, MRU860's, 2.2kOhm, 560uF, 560uF, 10uF, 1Ohm+.1uF. Is this the right order? If anyone would do anything different or if anyone feels like this is correct feel free to say so.
Your schematic showed 10000uF, not 1000uF. But I would recommend more, probably at least 22000uF per rail. And I would recommend many more but smaller caps in parallel. Use at least ten caps per rail, to get lower inductance and ESR, from paralleling. Use any brand. Nichicon are perfectly fine, for example (actually, they are quite good). Check mouser.com, too, not just digikey.
The music signal that makes the sound that comes out of the speakers is CURRENT that comes DIRECTLY from the capacitors. See the image at: http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-38.html#post3117390
Use cap-bank PCBs, if possible:
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-103.html#post3155293
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-121.html#post3166558
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-101.html#post3154202
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-102.html#post3155158
If you can, just put all of the caps as close as possible to the chipamp power pins. If you can't, then at least pack a bunch of smaller caps around the pins, like ten 220 uF or more. The ONLY thing that really matters is what impedance is seen BY THE POWER PINS. The output impedance of the power supply means nothing, unless the output is right AT the power pins of the chipamp.
Decoupling cap calculations: http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-169.html#post3320547
By the way, those small-value caps on your schematic will do no good at all, and could do some harm (forming resonant LC tanks with stray inductances), unless they are within less than about 2 mm of the chipamp power pins.
A resistor, with an optional small cap in series, COULD do some good, as a snubber to damp the spiking and/or ringing from the diode turn-off. But that would need to be across the transformer secondary, or the bridge input terminals (or across each diode). But there would be NO possible way to know, yet, what the resistance or the capacitor value should be. It could be done like this: http://www.diyaudio.com/forums/powe...lm-caps-electrolytic-caps-30.html#post2828689
Last edited:
Thanks your post was very informative. So your recommending something like 20x UKW1J222MHD Nichicon | Mouser 10 per rail? From what I understand the caps supply the current to the speakers which varies depending on the signal in. This is best achived by having a large enough bank so that it never runs empty under high demand. More small caps in parallel offer access to more current per given time due to lower impedence because what good is a large bank that can't be easily accessed. I was planing on housing the psu in a separate but close by case. the only components that should be in the psu case are the transformer, 8x Digi-Key Part Search to convert to DC and the 20 caps linked above. What about the 2.2k resistor does it serve a purpose?
Please do not quote the entire post just above yours. I have fixed it. tnxUsing 10 caps per rail would be fine, I'm sure. I am going to use 100 per rail, with two proper arrays on a 2-sided pcb, mostly as described in the links, except using two halves of one PCB. (But I am pretty far out there, sometimes. I mainly just want to measure the results. It might be way into overkill but I want to find out.) I am also going to use the LM3886 board as a sort of daughterboard, just above where the two cap array power planes almost touch, so that the power and ground pins for everything can go straight down into the power and ground planes of the cap arrays, which should only be about 3/16ths of an inch away, or maybe 1/4 inch (about 4.76 to 6.35 mm).
And without any onboard caps, the 3886 pcb can be made to be pretty small. I'm still figuring out how to mount everything, with heatsinks, while still getting good current diffusion through the cap arrays and lowest impedance at the chip pins, etc etc.
Yeah, it sounds you have the right ideas about the caps. But you also have to consider the high-frequency needs, which probably weren't covered in my linked post about calculating the needed cap values. Look at this one:
http://www.diyaudio.com/forums/powe...lm-caps-electrolytic-caps-23.html#post2806854
And most of the rest are linked from here:
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-3.html
It turns out that you need HF currents, too. When quantified, it enables calculation of a maximum inductance that can be tolerated, as seen by the chip's power and ground. That usually equates to a maximum LENGTH of conductors that can be used, from tghe caps rto the pins, and/or how many of what value caps to parallel.
You could probably just add a few smaller caps near the pins, actually; maybe three 68 uF per power pin. But the layout is more difficult than it seems. See links.
Cheers,
Tom
And without any onboard caps, the 3886 pcb can be made to be pretty small. I'm still figuring out how to mount everything, with heatsinks, while still getting good current diffusion through the cap arrays and lowest impedance at the chip pins, etc etc.
Yeah, it sounds you have the right ideas about the caps. But you also have to consider the high-frequency needs, which probably weren't covered in my linked post about calculating the needed cap values. Look at this one:
http://www.diyaudio.com/forums/powe...lm-caps-electrolytic-caps-23.html#post2806854
And most of the rest are linked from here:
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-3.html
It turns out that you need HF currents, too. When quantified, it enables calculation of a maximum inductance that can be tolerated, as seen by the chip's power and ground. That usually equates to a maximum LENGTH of conductors that can be used, from tghe caps rto the pins, and/or how many of what value caps to parallel.
You could probably just add a few smaller caps near the pins, actually; maybe three 68 uF per power pin. But the layout is more difficult than it seems. See links.
Cheers,
Tom
If its really your first build and you've done nothing simpler before, then consider going for a much simpler chipamp like TDA8561. Yes its lower power but boy is it easier to get going 
The decoupling is much easier with small caps nearby the pins because those smart boys at Philips knew what they were doing and put the power pins right next to each other on the package. Thirdly its bridged, so no tedious power supply grounding issues - it runs on a single supply (I'm using a laptop SMPSU). A fourth point is the pin spacing fits straight in to 0.1" prototyping board so you don't need to get a PCB specially made for it.
My final point to recommend it is the awesome sound - miles ahead of LM3886 when done right - meaning with the power supply really low impedance like gootee is suggesting.
The decoupling is much easier with small caps nearby the pins because those smart boys at Philips knew what they were doing and put the power pins right next to each other on the package. Thirdly its bridged, so no tedious power supply grounding issues - it runs on a single supply (I'm using a laptop SMPSU). A fourth point is the pin spacing fits straight in to 0.1" prototyping board so you don't need to get a PCB specially made for it.My final point to recommend it is the awesome sound - miles ahead of LM3886 when done right - meaning with the power supply really low impedance like gootee is suggesting.
Thanks for the links ill look into them more. As for the TDA8561 i already ordered the LM3886 but my dad wants also want to build an amp too so ill look into that for him. As for the 2.2k resistor in the link of my first post does it do anything besides constantly draw power should i include it or not?
Alright thanks for the quick reply that makes sense I will be sure to include them.
Last question hopefully. I have made changes in caps thanks to peoples advice but the psu is still like the link in my first post: two bridges, with two separate rails one with +PGND and the other with -PGND at the end and what looks like a connector between the +PGND and -PGND. What I am confused about is how to wire it. Do I connect +PGND, -PGND, the purple wire of the transformer (link to is also in first post), and the chassis all directly to earth (3rd round prong in the US)?
Now I am confused, "connect +PGND, -PGND, the purple wire of the transformer ...and the chassis all directly to earth"
You can EARTH, yes. Connect chassis to AC third prong ONLY. Make darn sure it is a REAL Earth! Yes, you can buy a cheap tester at your local DIY shop.
Are you saying I was correct and should connect all four of them to earth? The case and the purple wire which is a shield go to earth that much I know. Its the +PGND and -PGND that confuse me do i connect them to the earth ( 3rd round prong in the US) or are they something entirely different?
Wired with two rectifiers as per the diagram in your first post http://www.chinabaike.com/uploads/allimg/120413/204G3L91-0.png , NONE of the transformer secondary wires should be connected to your chassis earth! Yes you should connect the +PGND and -PGND to the chassis earth (otherwise known as safefty earth). This point is your zero volts reference.
Ah I see now. The purple wire on the transformer is a static sheild, in this case yes I would say it is fine to connect it back to the Chassis earth point.
Tony.
Ah I see now. The purple wire on the transformer is a static sheild, in this case yes I would say it is fine to connect it back to the Chassis earth point.
Tony.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.