| soundNERD |
I currently run a 10-inch subwoofer with one LM3886. Not too powerful, so I want to build something bigger. I looked at the Class D LM4651/2 from national, but it seems a little too complicates. Unless somebody has some pcbs using through hole components, I would like to stay away from something that complicated. I then found the Chainclone - http://www.euronet.nl/~mgw/diy/amps...ainclone_1.html - and was wondering if anybody has used one of these before. If so, did anybody use it for a subwoofer? How does it sound? Is it better than the 4651/2?
Another idea I had was to build a stereo amp to drive my B&W DM580 speakers. If I am right, a bridged 3886 can do 220W into 8-ohm, so 440 into 4 and 880 into 2? My speakers are 8-ohm but 220W per channel is still pretty impressive. Or, since my speakers are 8-ohm, would I be better off just using 4 chips instead of 8 and not paralleling them? Can the 3886 drive a 4-ohm load in bridged mode at all?
Any ideas on the sub amp? Would it be worth it to just make a pcb and use the 4651/2? |
|
|
| usekgb |
I'm not familiar with the chainclone, so I can't comment there. As far as using 4 chips, you are going to have to parallel two pairs of them. You then bridge the two paralleled pairs. This will get you the bridged power you are looking for, and will also deliver plent of current to give you the low eon oomph we all love so much.;)
Cheers,
Zach |
|
|
| usekgb |
OK, I just looked at the Chainclone thing. This is a very similar layout to the one that National Semiconductor published. It's basically a bridged/paralled design. It should work great for your sub. It will give you about 220 watts at 8ohms in to your sub.
Cheers,
Zach |
|
|
| soundNERD |
And almost 440W into my 4-ohm sub speaker?
If so, that has some good things. and some bad things. Good, thats a ton of power. Bad news, my subwooofer is a 100WRMS with a 300W handling. Is that too much?
Do I need a larger transformer to get 440W into 4-ohm than to get 220W into 8-ohm? Or would the one on that page work?
How do you bridge them together? Just connect the inputs together and use one output for one side of the speaker and the other for the other side?
If anybody has done this, does anybody have the pcb layout for one channel?
Thanks, Mike |
|
|
| bigparsnip |
Hi, you won't be able to double the power into a smaller load with this design, as the chips become limited by their maximum output current, and thermal dissipation, but you would probably still get around 300WPC, perhaps a little more, which is still quite a lot.
As far as these paticular circuits go, I'm not too sure how they would perform, as the are not totaly even in allrespects. Mainly, you have the inverting amp with Zin of around 1K Ohm, and the non-inveting with an input impedace about equal to that of the chip itself, which will be a lot higher.
Now, I have never built anything like this before, but I am guessig that as the input will drive the two halfs differently, you will have slightly different outputs on each half which could add a large amount of distortion, depending on how ideal your driver is. |
|
|
| soundNERD |
Would I be better off using the schematic from national's datasheet? If so, what is that opamp on the input for?
I would like the second one better since there is less parts, but on that one I really have no idea what that odd opamp is on the input.
heres the schematics:
http://www.diyaudio.com/forums/show...?threadid=22305 |
|
|
| bigparsnip |
Hi, Jojo whent for the second schematic in the end, as it was easier to match everything up than the first one and he seemed to be haappy with it when he was finished.
As far as the op-amp thing at the front of the circuit, that is just a buffer, that will give you a non-inverted and an inverted output (ballanced), so that you can feed the out of phase signals to each half of the bridge when they are both either inverting or non-inverting designs. |
|
|
| Optical |
The opamp at the beginning is just a buffer, and does just that, perhaps not crucial... but id say that would depend on whats driving it
Im personally about to embark on the circuit with the DRV134 at the input as (like bigparsnip says) id rather have each opamp running in the same mode to make it easier to match everything |
|
|
| soundNERD |
First, what do you mean by snap?
second, can I just use a standard audio opamp at the input such as a 386? I am using a crossover based on the lm837 if that would do the same buffereing thing.
And yes, I think i will go with the second one also, it is less parts and less to explode when I short somethig out. :D |
|
|
| bigparsnip |
| no, you will need to use that specific part (or one that does the same job, whihch you could implimnet using two op-amps) as it gives out one + signal, and one - signal where as you will normaly only get one of these from an op-amp. |
|
|
| soundNERD |
| never mind. I only looked quickly at the schematic, and didn't notice it was split. Where do I get that part? |
|
|
| bigparsnip |
| try seeing if you can get a couple of free samples from nationals website. |
|
|
| soundNERD |
OK, I didn't know national made the DRV series. I will try to get some.
Also, I designed a PCB layout. It is on the other computer, so I will post it tonight. Hopefully it will work. |
|
|
| soundNERD |
Nope, national doesn't make them
any other ideas? |
|
|
| tbla |
| the drv-134 is from TI...... |
|
|
| bigparsnip |
| Sorry about that confusion, try here. |
|
|
| sam9 |
Going back to the original post, you can order a PBC for an LM3886 amp from www.marchandelec.com It's the PM21. There are two on the PCB and they can be bridged. You can also look at the schematic and construction guide there.
I think (but could be wrong) Marchand predates 47 Lab and most of the things that call themselves "gainclones".
However, I think it dubious that you can get the power levels you are looking for, especially with a 4-ohms load, out of any LM3886 amp. The reason is heat - the package isquite small which limits heat transfer. It might be OK for a subwoofer amp as those spend most of their time not doing much. I would be generous with the heatsinking and ventilation. Also I would select the non-encapsulated version and use the best rated (lowest thermal resistance) insulating pad I could find. On a posative note, the LM3886 has built in thermal protection so you could give it a try with a reasonable expectation you would not damage the chip.
Oh yes, I built the Marchand amp and found it easy to build and it sounds just fine comparred to your typical ChainStore amp. My personal bias is that no IC based amp (as yet) has the performance potential of a discrete amp. (Note I said potential - IC amps have the advantage of pretty clearly defined decent result where discrete designs run from incredibly bad to incredibly good) |
|
|
| soundNERD |
Yesh, I found that it was made by TI after my last post. Thanks.
Heres the bottom solder side of my PCB (not mirrored) |
|
|
| soundNERD |
And here's the layout
The big pads on the left are power in, and the ones in the center are speaker out.
the 4 unlabeled pads at the top are jumpers and the 3 by the volume control are the power for the DRV134. |
|
|
| soundNERD |
Just wondering who makes standard 1% tolerance resistors? Anything that will fit on my board. I can always mount them sticking up.
Thanks again, Mike |
|
|
| bigparsnip |
Hi, I just thought that you should probably know that your board may need some small alterartions, as if you do go for this design, you will be getting peak currents of over five amps flowing in the circuit, and at the momnet the power tracks are just way too thin to handle that. So, you may want to beef up the tracks you are using for the power lines, output and ground so that they can handle the current you want to put down them. (I would sujest about 0.1 to 0.15 inch for 1 ounce copper board).
As far as the resistors go, for all apart from the output ones, just use standard 1% metal films that you can pick up from radio shack or something seen as this is your first attempt, as you can always change them out at a later date. And as for the output resistors, you will need at least 5W or so rating on them, so I would just go for 5% wire wound ones of rnow too. |
|
|
| soundNERD |
| Heres the revised pcb. I did the original in a trial software program, so it wouldn't let me save and I had to use print screen and paint to save it. So, I had to edit it in paint. Not all the connections are done. Does the + input need a thick wire connecting to ground? I couldn't fit a larger one than I have now. Is the new thickness of the + - and speaker wires thick enough (gnd isn't done and nor is the connections to the otuput resistors)? |
|
|
| bigparsnip |
What you have done is a fair improvement, but it looks like you still have the output resistors as 1/4W sizes, and if you use these you will probably blow them up if you turn the volume up much, so they realy do need to be replaced with higher power ones.
If you can wait for the PCB, I am planing on designing something similar soon (I just changed from using eagle, to Protell, so I am trying to re-learn how to draw a scheematic, then convert it to a PCB design), however, I was going to go for a design with six chips per channel (Bridged, with three chips in parallel so to give better drive below 8 ohm loads). I was also, going to try and put a little more flexibility into the board, so that, it can be used in defferent configurations (two seperate banks to three parallel chips. three lots of two parrallel chips, or six seperate amplifiers.
But as I am comming to the end of term at university at the momnet, I have quite a bit of work on till next week, so I probably won't have a final design until some point near the middle of next week. |
|
|
| JOE DIRT® |
may I suggest a double sided board and having a ground plane on either the top or bottom
DIRT® |
|
|
| soundNERD |
Well, I would like to stick with what I have. I have finished the trace size increases. You did say something about driving lower than 8-ohm speakers and using 6 chips. My speaker is a 4-ohm, will it not work well? Also, I wouldn't use 1/4W resistors. I would use larger ones but just mount them sticking up off the board. I also don't like double sided. I tried one before, on a headphone amp, and yes it worked, but half the parts I had to mount like surface mount because the drilled hole went through two traces.
Are you saying that something isn't good about this design? |
|
|
| bigparsnip |
Hi, the thing about the chips being able to drive a four ohm load can be a bit confusing at times. Basicly if you have it so that you are getting about 50W into an 8 ohm load, the power will not double into a four ohm load, but instead you will get about 60-70W. The same goes for a bridge/paralel configuration with four chips; you will get about 200W (at a fair maximum) into an eight ohm load, but probably only 300W into a four ohm load not the 400w you might be hoping for.
To try and get round this a little you can parallel more chips, so the apparnet load each chip will see is increased and they cna drive closer to their expected maximum rating without having the thermal or current limiting cut in. basiclay, it should get me an extra 50% increase or so in power at lower loads (say around 350-370W into a four ohm load).
But, the design that you have already chosen with the four chips will still happily power a four ohm load, so there is no real need to change it (300W is still about six time the amount of power I have inot my speakers at teh momnet, and I'm pretty sure it would fry them if I did try and run them a 300W). |
|
|
| soundNERD |
| Thanks. Actually, I wasn't complaining about 300W, just from the way that post was worded, it sounded like it wouldn't run stable into 4-ohm. Basically, I just wanted to know how my board would work. I know I didin't connect the mute pin yet, but is the rest ok? |
|
|
| bigparsnip |
I guess so there is notthing glaringly obvious wrong, but it's a little hard to tell just from the drawing.
I do know that you have to conect the mute pin up in a certin way on the LM3668 (even if you aren't using the mute) or else it won't work properly. |
|
|
| soundNERD |
| Yes, I know I didn't have room to attach the mute pin. It won't have a mute switch, so all I do is tie the pins together and connect to ground through 1 resistor, right? Or do I need a resistor for each output? |
|
|
| soundNERD |
| What would be a good transformer? I would like enough current to supply full power at all times, whatever voltage whould be good, and not too expensive ($10 - 20 would be nice). Where can I get such an x-former? |
|
|
| soundNERD |
| any ideas on the transformer? |
|
|
| ddog |
| quote: | Originally posted by soundNERD
Yes, I know I didn't have room to attach the mute pin. It won't have a mute switch, so all I do is tie the pins together and connect to ground through 1 resistor, right? Or do I need a resistor for each output? |
You must connect mute pins to -V and draw 0,5mA from each mute pin to switch mute off. |
|
|
| soundNERD |
| Whoops, my mistake. I did mean to type vee not ground. Does each chip have to have it's own resistor for the connection, or can I connect the 4 pins together and use 1 resistor? |
|
|
| bigparsnip |
I think it would be best to use individual resistors for each chip, in case the voltage at the pin on each chip isn't quite the same.
As far as a transformer goes, if you want to be able to run this happily into a four ohm load, I would sujest going for a 25 - 0 - 25, 600VA asa a minimum per channel, or perhaps a 1kVA if you are going to be running two from the same transformer. |
|
|
| soundNERD |
| OK but who are some manufacturers of transformers? |
|
|
| bigparsnip |
| Well, there are quite a few out there and I guess the ones that would be any good for me would be less so for you as the cost of shipping from one side of the atlantic to the other for such heavy items won't be cheep. However, I have heard a lot of people in the US and canada talking about Piltron and Victoria Magnetics as being two of the more DIY friendly compnaies to deal with, but there is always the option of looking up which standard models someone like digikey stock and seeing if you can use a one of those. |
|
|
| soundNERD |
OK, I am finally getting ready to build it.
I noticed that the feedback resistors are 220K. Normally I use somewhere around 30K. Wouldn't the 220K give it too high of gain?
Thanks, Mike |
|
|
| ddog |
| quote: | Originally posted by soundNERD
OK, I am finally getting ready to build it.
I noticed that the feedback resistors are 220K. Normally I use somewhere around 30K. Wouldn't the 220K give it too high of gain?
Thanks, Mike |
It depend what is value of the other feedback resistor, if it's 10k and amp is inverting, you are going to get gain of 22 (26.8dB), witn NI gain is 23 (27.3dB). that's with 10k, with 1k for example, it will be Inv=220 or NI=221 |
|
|
|
