I agree with the notion that a lot of the bad reputation of chipamps is the stupid "gainclone" philosophy. When treated like any other solid state power amp, ie with a good layout, proper decoupling close to the IC, a good heatsink, good stiff power supply and MOST importantly not trying to exceed the chips rating, they sound very good indeed especially given the simplicity.
I think trying to get the maximum 68W out of the LM3886 is half the trouble. Personally I wouldn't use it past 30W, i.e. 25v rails, giving plenty of margin.
I had the opportunity to compare my LM3386 amp against a relatively decent Arcam Alpha amplifier, and i couldn't really tell any difference. Both amps had 160VA transformers, and the Arcam had 1x6800uF per rail (35v) versus my chipamp's 2x4700uF per rail (25v).
I do suspect that chipamps might be a bit suspect with complex loads, but for driving my Wharfedales and Eltax Liberty 5's, they sound great
I think trying to get the maximum 68W out of the LM3886 is half the trouble. Personally I wouldn't use it past 30W, i.e. 25v rails, giving plenty of margin.
I had the opportunity to compare my LM3386 amp against a relatively decent Arcam Alpha amplifier, and i couldn't really tell any difference. Both amps had 160VA transformers, and the Arcam had 1x6800uF per rail (35v) versus my chipamp's 2x4700uF per rail (25v).
I do suspect that chipamps might be a bit suspect with complex loads, but for driving my Wharfedales and Eltax Liberty 5's, they sound great
"I think trying to get the maximum 68W out of the LM3886 is half the trouble. Personally I wouldn't use it past 30W, i.e. 25v rails, giving plenty of margin."
Yeah, that's probably a very good idea - SOA is a bit challenged and contrary to discrete BJT:s with huge chips these have pretty short thermal time constants. (A typical big BJT like 2SC5200 can take for 10ms double the peak power a LM3886 output transistor can for *0.1ms*)
Yeah, that's probably a very good idea - SOA is a bit challenged and contrary to discrete BJT:s with huge chips these have pretty short thermal time constants. (A typical big BJT like 2SC5200 can take for 10ms double the peak power a LM3886 output transistor can for *0.1ms*)
Attachments
Hi,
I think that the 3886 is suitable for big power into high impedance loads of around 8ohm. eg. 60W into 8r0 using quiescent rails of +-37V sagging to +-34V on full power. This is despite the fact that these chipamps were never intended for this purpose when initially released for consumer electronics.
I have stated repeatedly that if designing for lower impedance loads then the power should be scaled down. I have suggested 30W to 40W maximum for 4ohm loading.
I think that the 3886 is suitable for big power into high impedance loads of around 8ohm. eg. 60W into 8r0 using quiescent rails of +-37V sagging to +-34V on full power. This is despite the fact that these chipamps were never intended for this purpose when initially released for consumer electronics.
I have stated repeatedly that if designing for lower impedance loads then the power should be scaled down. I have suggested 30W to 40W maximum for 4ohm loading.
megajocke said:
I've made very similar plots to yours with BJTs, MOSFETS and several different chip-amps...and I can't agree more. The 3886 doesn't have the "guts" of a discrete amp. National's data sheet is partially responsible, claiming power ratings under conditions that are likely only to happen during development when you can absolutely control the load and rail voltage. To their credit, they do tell you all about the chip, and study the document long enough you'll figure out what it can and cannot do.
I've built them before...for what I wanted them to do (quick, cheap, easy to construct amps) they were great.
When I want more power and better sound quality, I go discrete.
Scott
Technically there is no problem putting out 200 and more watts with chip-amps , its a matter of what one is prepared to spend on the things .
A bridged circuit using two LM3886 is no problem at all and can put out 100 Watts provided that the speaker is not less than 8 ohms. Lower the supply-rails accordingly as the chips will see 4 ohms each and use a very good opamp at unity gain to invert the input signal on one of them is all. Should be enough for sound at home.
So should 50 watts be except for noisy parties.
Michael
A bridged circuit using two LM3886 is no problem at all and can put out 100 Watts provided that the speaker is not less than 8 ohms. Lower the supply-rails accordingly as the chips will see 4 ohms each and use a very good opamp at unity gain to invert the input signal on one of them is all. Should be enough for sound at home.
So should 50 watts be except for noisy parties.
Michael
Unfortunately, there are very few genuine 8R speakers any more. Most domestic loudspeakers are actually closer to 6 or 3R loads now and the old notion that a nominal 8R rating was representative of the speaker's impededance has seldom been the case for at least 25 years, in my experience. Realistically, we need to have power amplifiers with 4R load capability at maximum power for use with typical modern loudspeakers. Bridging is really going in the opposite direction to that required. You surely mean parallel operation, which is perhaps easier with other chipamp families.
I understand that market research and dealer reports have shown that for a given price range and impedance rating, louder speakers sell better than apparently less sensitive ones. The cheapest way to increase apparent sensitivity is to reduce the impedance and that strategy seems to work very well since it's now widely practised.
I understand that market research and dealer reports have shown that for a given price range and impedance rating, louder speakers sell better than apparently less sensitive ones. The cheapest way to increase apparent sensitivity is to reduce the impedance and that strategy seems to work very well since it's now widely practised.
Last edited:
Any solid state amplifier, chip or otherwise, will have good control over the loudspeaker if it is used within its limitations. It is sickeningly easy to use a small amplifier beyond its limitations. Doesn’t take much volume to clip a 40 to 60 watt amp. And when an amp clips, it’s damping factor is determined by the power supply’s output impedance. Face it, no one is going to use a $600 power supply with a $6 chip - you get what you pay for.
If you manage to get it to current limit, you fare far worse. During the time the current limit is active, say when driving a bass reflex speaker at it tuning frequency where the impedance hits 3 ohms, the output impedance goes high. It becomes a current source, which theioretically goes to infinity. Well it does go high enough that the speakers voice coil might as well be open circuited. Any transient, regardless of fundamental frequency, causes the speaker to ring like a bell at its fo. You’re going to get low frequency noise and extra spectral content, even if the offending frequency is higher. A speaker has another impedance minimum as well, and it tends to be in the low midrange where power content is high.
Even if you’re not going into full-on current limit, when the output transistors run out of current gain driving low Z the whole amplifier operates closer to open loop. Damping factor degrades, and so does everything else - distortion, noise, high frequency response..... you’re not going to get the data sheet distortion numbers published for an 8 ohm resistive load with a real speaker. - easoecially one that cheats on impedance to be “louder”.
If you manage to get it to current limit, you fare far worse. During the time the current limit is active, say when driving a bass reflex speaker at it tuning frequency where the impedance hits 3 ohms, the output impedance goes high. It becomes a current source, which theioretically goes to infinity. Well it does go high enough that the speakers voice coil might as well be open circuited. Any transient, regardless of fundamental frequency, causes the speaker to ring like a bell at its fo. You’re going to get low frequency noise and extra spectral content, even if the offending frequency is higher. A speaker has another impedance minimum as well, and it tends to be in the low midrange where power content is high.
Even if you’re not going into full-on current limit, when the output transistors run out of current gain driving low Z the whole amplifier operates closer to open loop. Damping factor degrades, and so does everything else - distortion, noise, high frequency response..... you’re not going to get the data sheet distortion numbers published for an 8 ohm resistive load with a real speaker. - easoecially one that cheats on impedance to be “louder”.
Thanks guys.
this is very useful information , especially for rhythmdiy who started this thread but of course for me as well as it constitutes another pitfall for non engineers so for a lot of people. you could insert it in one or two of me other threads if you want to . Thanks again
this is very useful information , especially for rhythmdiy who started this thread but of course for me as well as it constitutes another pitfall for non engineers so for a lot of people. you could insert it in one or two of me other threads if you want to . Thanks again
Last edited:
Info for rythmdiy and his project, plain speak:
Any solid state amp including chipamps will deliver a good clean bass as long as one does not overdrive it and as long as the ps delivers the necesary juice without seriously dropping rails.
A lot of discrete circuits can deliver a lot more output when clipping because their current capability can be a lot higher than a chip-amps but what they do then is overload the ps which is normally not a problem with music but what this does is draw more current than the ps can deliver so the rail voltages drop.
Since distortion in subwoofers is higher than in any other speaker anyways and since it is less audible than the same distortion in say a midrange-speaker they can be overdriven within reasonable limits. To some it apparently even "sounds good".
BUT : is this really what one wants , ok , then you can do it.
If on the contrary what you want is a clean powerful bass you then will have to invest the necesary material and cash/work to get this result. There is no way around that.
If what you want is listen to good music at home a good chip-amp will do. If you want to throw a party including shaking up your neighbours it will not.
I hope this helps you to decide what you really need to do.
Michael
Any solid state amp including chipamps will deliver a good clean bass as long as one does not overdrive it and as long as the ps delivers the necesary juice without seriously dropping rails.
A lot of discrete circuits can deliver a lot more output when clipping because their current capability can be a lot higher than a chip-amps but what they do then is overload the ps which is normally not a problem with music but what this does is draw more current than the ps can deliver so the rail voltages drop.
Since distortion in subwoofers is higher than in any other speaker anyways and since it is less audible than the same distortion in say a midrange-speaker they can be overdriven within reasonable limits. To some it apparently even "sounds good".
BUT : is this really what one wants , ok , then you can do it.
If on the contrary what you want is a clean powerful bass you then will have to invest the necesary material and cash/work to get this result. There is no way around that.
If what you want is listen to good music at home a good chip-amp will do. If you want to throw a party including shaking up your neighbours it will not.
I hope this helps you to decide what you really need to do.
Michael
The cheapest and best way to get an overdimensioned powersupply with the added advantage of stabilized voltages seems to be a good dual smps, for about 70 to 100 of the little coins you get something capable of up to almost a 1000W, relative soft startup and short protection...............
A lot less weight, as well, and a lot less space. Main trade-off is that a tranformer based ps practically never fails whereas a circuit is more likely to do so. A smps might need adittional filtering as well.
Like everything there are advantages and disadvantages so no matter what one chooses it always is a compromise.
One advice if you want to use smps is to get a speaker-protection circuit (China ones for less than 10 coins) and instead of using it for the speakers use it to dectect the rails and short them to ground in case of a serious imbalance . The circuit needs a separate ps which to keep it cheap can be taken out of a small wallwart for example and is autoprotected as well as if the wallwart fails the rails will never be connected to the amps. The relay must short the amp side of the rails to ground as the relay is likely to arc in case of failure and breaking DC at 30 or 40 volts. The higher the voltage the worse the risk of arcing. The same goes for this as for effective speaker-protection........
A lot less weight, as well, and a lot less space. Main trade-off is that a tranformer based ps practically never fails whereas a circuit is more likely to do so. A smps might need adittional filtering as well.
Like everything there are advantages and disadvantages so no matter what one chooses it always is a compromise.
One advice if you want to use smps is to get a speaker-protection circuit (China ones for less than 10 coins) and instead of using it for the speakers use it to dectect the rails and short them to ground in case of a serious imbalance . The circuit needs a separate ps which to keep it cheap can be taken out of a small wallwart for example and is autoprotected as well as if the wallwart fails the rails will never be connected to the amps. The relay must short the amp side of the rails to ground as the relay is likely to arc in case of failure and breaking DC at 30 or 40 volts. The higher the voltage the worse the risk of arcing. The same goes for this as for effective speaker-protection........
Last edited:
I just finished a SS project with a SMPS stuffed in it. Works actually surprisingly well. Granted, the SMPS spreads the RF 'love' within the case a little bit but nothing serious that could not be gotten under control. Fellows who have experienced problems with the residual RF coming of a SMPS have likely disregarded EMI design rules and perhaps are still using single sided or double sided boards without any plane layers and mindless criss crossing traces everywhere. For those loop area is a foreign concept.
The problem with lower powered amps is the lack of headroom.
The signal quickly hits the power rail and you get horrible clipping.
A lot depends on the music source whether it is highly compressed or not.
Something that isn't compressed much will have very transients that can quickly hit the rails.
My first power amp in 1980 was 50 watts and quickly ran out of headroom with 30 volts rails.
My next amp was 225WRMS with 60 volt rails and was much better.
The signal quickly hits the power rail and you get horrible clipping.
A lot depends on the music source whether it is highly compressed or not.
Something that isn't compressed much will have very transients that can quickly hit the rails.
My first power amp in 1980 was 50 watts and quickly ran out of headroom with 30 volts rails.
My next amp was 225WRMS with 60 volt rails and was much better.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.