Transistor Amplifiers continued 10

Why these LM3875 gizmo's are wrong


Below the schematic diagram of the National Semiconductor LM3875 (TM). It is an IC with a nice LF power amplifier in it, that is capable of delevering approx. 50 Watts to a speaker. One instantly notices, looking at the below internal schematic diagram, that this is a standard audio amplifier. One of the few minor set backs is the simple semi-complementary output stage and the protection circuit, that makes this amplifier unfit for High Quality (HQ) audio purposes. NS never intended this IC to be for the HQ audio section.
So far so good, NS has made a good and usefull IC with this LM3875.

The internal basic diagram by NS.


The basic test-application of the LM3875. You see that it is very easy to construct a simple power amplifier with it of quite good performance characteristics.


Typical test aplication by NS.


Now we have below a practical application of one of the audio gizmo's of today a certain mister Thorsten Loesch who hides behind the chinese name Kuei Yang Wang. 
Why is this not a good amplifier?

    The guru's "wisdom".


Well look at that power supply, 1000 uF??? What electronic principle is that fundamented on??? Clearly it is not knowledge. Lows suck this way!!! IM distortion gets up to high levels. Not to mention other disadvantages. You will not find any qualified audio/electronic engineer advising to built a power supply THIS way intended for the use in audio gear. This is NOT the way to build a power supply intended for the use in HQ audio gear. NS suggests another approach! The gizmo approach is wrong and the reason why these gizmos do this is to suggest that this amp is faster than it really is.
My advise is: put on each power rail at least 10000 uF and put at least an extra >100 uF close to the IC pins. So on the + side one big capacitor and also on the - side, making 20000 uF for each amp. And >200 uF close to the power connections of the IC in total, per amp. Put on all power electrolytic capacitors parallel one 0.05 uF ceramic 200V capacitor. With this you can keep possible HF out of your amp.
Now the lows rock. Many tried it and confirmed the difference. So you see with some basic knowledge you can make a good amplifier with these IC's of NS after all.

This is how a power supply is done.

This .22 Ohm resistor in the speaker output? Used to kill tendencies of oscillation of the amp? Well its a good way to kill the  (average) damping factor of the amp: get rid of this ugly thing. Kill tendencies of oscillations of amps the appropriate way, the way NS suggests: by HF decoupling. Nothing against the 1 uF capacitor, but why leave the caps away while suggested in the NS application, read the datasheet of NS, the NS application is better, do it the NS way, NS knows about electronics. If you really need a compensation network in the output circuit use something like the below in stead of the silly resistor. 

This is how hf decoupling is done.


          This circuit is to replace the .22 resistor.

Leave away the 100k volume control, volume control is in your pre-amp (unless its used as a preset).
Do you need shottkey diodes in the power section? No you don't.
Is this a low cost amp? No it isn't. Low cost amps are 2nd hand amps and the reuse of amps is better for the environment.
Is this an easy to built amp? No, basicly it isn't easier as any other amp, if you want the job done right. Take a look at the factory made LM3875 amp as an example of bad building practices.
What is the advantage of this amp then? Well its a small one, you can put 5 amps for home cinema use into one housing formerly used for a stereo amp, it really can be made small with this IC.
Is this a recommendable amplifier for your HQ home HiFi? No it isn't really.
Is this amp only suitable for high efficiency speakers? No it isn't, all speakers can be used.
Do you need to put any parts in your kitchens oven for 10 hours to get them function right?
No do not believe this gizmo nonsense, your kitchen equipment is not fit to put electronic parts in.
Do you need inverting amps? No you do not, do not mix non-inverting and inverting amps in multi-channel and surround configurations, you can get into phasing problems. In audio non-inverting amps are the commonly used ones.
Be aware of the differences between inverting and non-inverting op-amps as the above amp is an op-amp. Be aware that feedback affects output impedance and harmonic distortion equally for both types, but their characteristics themselves are different: input impedance is high for non-inverting and low for inverting configurations. The noise gain differ, that will become significant in low gain applications, in the unity gain case an inverting amp has twice the noise gain of a non-inverting amp. Bandwidth is similarly related: for the unity gain case a non-inverting amp will have twice the bandwidth of the inverting amp. So one must have a reason to make the choice for the inverting amp in this case. But there are obvious reasons to make the choice for the non-inverting amp and not for the inverting counterpart. The reasons of the gizmo to make the choice for the inverting amp are at least contradictory.

Does this amplifier have a slightly romantic, warm and soft tone, with strong dynamics and good clarity? No not at all, the sound of this amplifier built this way is not good and may not even meet modern HiFi standards.Follow NS suggestions for better results.
Do you need to use metal film resistors and cermet track volume control for a brighter tone? No, metal film resistors are generally used to lower the noise in preamps. Try to get a better tone first by building this amplifier following traditional electronic standards as i.e. suggested by NS.
Is this IC no good? This IC is a high quality product of NS and applicable in most audio situations where we need a Watt or 50.
But it is not for HQ HiFi or so called "High End" home audio. Its very nice for in your car or tv set or dvd surround set! Use it bridged for 100W party & disco PA, use 5 of them in your diy multi channel home cinema set. Hundreds of nice applications can be thought of. NS did their job well...... In my opinion the gizmo did not understand the NS application datasheet very well.

This is another example not to believe these todays audio-guru's (or what they call themselves) that sputter all kinds of fake scientific audio stories around the net  about audio equipment and related stuff that make no sense at all, and the only thing they show off is their limited electronic knowledge and experience: I have seen one of these LM 3875 stereo amps of one of these gizmo's, he actually used UTP cat 5 network cable as power rail and even as speaker cable completed with four pages of arguments (about ie. the disadvantages of stranded cable!!!) why this cable is the best there is for audio, can you believe that??
Some other people came with other far more interesting ideas involving this IC and this kind of amplifiers: for instance click at some of the links at the bottom of this page or:
Look around on the net concerning this issue and make your own conclusions. 
And below you can find interesting links to LM3875 related sites.
Of course this story is only our opinion, it is not to offend any person.


Now here is the experience of Fabio Chiozza from Argentina. He experiences the same as the described in this article, but let him do the talking below:

I happened to get hooked by the GC fever, and since I am the "seeing is believing" kind of guy, I put hands to work and built a GC. I am certain I made no gross mistakes. I got DC a offset of 14 mV, and it is cold as ice when quiet, so it is not HF oscillating either.

I agree 100% with your point of view, I have certain skills in electronics (I'm 44, electric engineer, tech and sensibly oriented approach towards audio, I mean don't believe at all in "directional cables", rocks on top of your CD transport, esoteric designs, green markers, etc). It is very noticeable that these nonsense practices don't reach our 3rd world. Maybe they are consequence of the money excedents available that motorize marketers appetites to apart people from their money (or viceversa). This kind of hypes would have little or no effects here, since there is much less cream to skim if you get the picture. In fact, nobody knows about green markers or isolating damping feets, etc here.

Anyway, so I built the "inverted" GC, and ... it is surely a good sounding amp, but it is WAY behind the old faithful 3020i for instance, which, btw, you can get on ebay for 50 USD approx.

I agree with you that it is hard to believe that this fancy design would sound any better than the manufacturers' application circuit depicted on the datasheet. This is not a technical statement, rather it is a matter of common sense: Could they be so dumb as to depict a poor application of their own opamp?? Could it be so easily bettered by a non-skilled community? Also as you say, I believe them at National KNOW SOME electronics for sure...

I have a Marantz 2252 (call me a vintage man) which also sweeps this little -not-so-wonder out of sight (or hearing). It has NO bass, and surely those 1000 uF caps are to blame.
Just to see what happens, I bought a couple of 2200 uF to add in there.

I believe all this hype is motivated by mostly honest people, but without any knowledge on basic electronics, so to most of them this may have been their first DIY amp. Only advantage is that it is really easy to build. Being a teenager I tried several amps and even an awful preamp but abandoned the activity upon finishing tech college (age 18). Some 26 years later, grabbed the solder again, only to discover this is not what is said to be. It is fun, though, but not much more than that. Seriously, can anyone expect that this little circuit would beat an amp from a serious maker? Hardly so ... the real world just is not that way.

On the pictures there is only one channel shown. This is not the definitive disposition, just the "testing stage". The big transformers are +9V 0 -9V each, the middle connection is not used, I connected in series the secondaries to get +18V 0 -18V as this seems to be the "recommended" supply voltage at several DIY pages.
I used no PCB, only veroboard for the PSU rectifier, the rest of external components are soldered directly to the LM3875. Also as recommended.

The only tweak I did until now is, just to be sure, I added a modest couple of 2200 uF caps to the power supply and OBVIOUSLY as one would expect, it does improve the sound noticeably, and not only the bass, clarity is also improved probably because IMD goes down.
But, still behind my NAD3020i of course on the clarity and dynamics departments.
The GC reminds me the Yamaha RX-V395 receiver I commited the mistake to purchase, only because it was on discount. It is very good for watching a Rambo DVD full of explosions and still get the dialogues, but not for serious music listening. I mean, it lacks so much dynamics that everything happens on the last 5 dB of the scale. 

In conclusion:
1.- National does a very good job indeed making this op-amp. Cheap, good quality and reliable stuff.
2- Yes, you can easily get a fine amp from this, and
3.- You donīt need more skills apart from knowing which end you should hold the soldering iron once it's hot -hmm,   
     OK, actually you should know a little little more than that, but not much.
4.- It is fun to build as a beginner because it works right away, no extra special cares or final adjustments needed.
5.- But ... sorry, no, it does not sound like hi-end gear to me, maybe just hi-fi. Sounds no more than OK.
6.- And ... entry-level GOOD gear you can easily get for USD 50 or so, sounds inquestionably better.

What can I say, last weekend I left it on for several hours, just to leave out any "burn-in" period sound anomalies (which I donīt believe to exist, anyway)
Everything remains the same. Only improvement I accomplished was quite predictable: the adding of 2200 uF capacitors per rail fixed the bass a lot, without this addition I definitely do not like at all the GC sound. I listen it through a pair of old Acoustic Research AR38s. Still behind my old creatures (NAD 3020i, Marantz 2252, etc)

My modest opinion is that the majority of tweaks one can find over the www, have the same effect as drinking water when you are ill, (does not cure you, but don't do no harm either). 



Now the continuing story of a fake Chinese plonker:

Here below the proof that this fake Chinese "guru" is not learning, although he told on several diy-forums and emails that he only "copied" the psu from the "original" amp, whatever that may be. And he said he "forgot" about the psu, how can you do that? 
Look at that amp and psu of his recent "design" (in fact a stolen together copy of some older designs): again the same stupidities. Now this mister Kuei Yang Wang / Thorsten Loesch proved himself to be a plonker. This is really a shame. Stop this fraud! The fellow is now in the MFB business (where he shows off his ignorance) and some other fields of audio, beware of his stupid comments and agression. Many have the same reaction, like the following guy who was attacked by him, as happens in most audio discussion groups he writes:
<<Thank you very much for your negative feedback Thorsten, very helpful indeed (not)  there is no need to show open doors and ask questions which have obvious answers in such a bashing way. Maybe you better read again my post and get what I intended to say, or are you really that ignorant>>.



Made in Japan "Final-labs" brand power IC based amplifier inside, what do you think: is this wiring well?

Or how do you like another spagetti? Does not look tasty not? And where they left the ketchup?

Do you want to know how to set up a good (chip) amplifier ?

Then you just have to click on this link, enjoy !





To complete this story: below a link to the major data and diagrams of the LM3875.


National Semiconductor made recently a new audio stereo poweramp chip: the LM 4780, 
take a look at this nice design here:


Links to related sites:


 (LM3875 is a trademark of NS)

Copyrights Hans Hilberink PE1MMK & Fabio Chiozza and associated editors and many others who sent their comments, last update 27-12-2005.Va.