Amplifier Design Configurations: So simple yet so powerful aspects of design.....

[Ravindra Desai]

Hello all,

Amplifier design has so many variations.
Even a chip amp like LM3886 or LM1875 can be configured to give much superior performance than that claimed in the chip manufacturers datasheet.
Not that the chip manufacturer does not want you to get the most out of his chip, but the application note provided by the manufacturer, more often than not, provides only some basic things to get you started and then its up to your imagination, design skills and experience that decides what you can do.

e.g. LM1875 is a 20W audio amplifier IC that delivers 20W into 4E or 8E @ 0.1% distortion at any frequency within the power bandwidth.

Can LM1875 be used to deliver 120W into 4E? Yes, with some additional circuitry, it is possible.
Can LM1875 be used as an audiophile amplifier with vanishingly low distortion? Yes, with some additional circuitry, it is possible.

Can the LM1875 be designed so that it is a mismatch with most pre-amplifiers but a perfect match with a 'specially recommended' pre-amplifier?
Can the LM1875 be designed so that it can respond to changes in cable?
The answer to the above two questions is also yes!
Why would anyone want to do such a thing can only be best imagined.

Lets take a look at what parameters decide 'universal acceptance' of an amplifier with any small signal pre-amplifier stage:

Refer to fig. (a) in the attachment:
It is a non-inverting amplifier. Simple, robust and it is what most people use as a design topology.

Refer to fig. (b) in the attachment:
It is an inverting amplifier that provides better common mode performance than the one in fig. (a).

However, the amplifier in figure (b) has an input impedance of only 560E as against the more standard audio amplifier input impedance of 22k offered by the one in fig (a)
To improve the input impedance while retaining the better common mode performance, the inverting amplifier is always used with a buffer as shown in fig. (c)

Now, the gain of the amplifier in fig. (b) and (c) is exactly the same and it is 22k/560E = 39.286.

In fig. (c), If I reduce the resistor 560E to 56E and 22k to 2k2, the gain will still be the same and the circuit would work just fine.

What if we do the same to the resistors in fig (b)? The gain would be the same, however the input impedance will now fall to 56E!
Such a circuit will have trouble finding a proper pre-amplifier and would exhibit different gains with different pre-amplifiers or for that matter different cables!
Now, needless to say that if I build a preamplifier with the last stage of which is a buffer, then it would be a perfect match with the 'tailored' inverting amplifier stage.

It is a general practice (called as best practice) that an inverting amplifier should have a buffer preceding it and in the same product.
If not you will have an amplifier that would work only with a 'specially recommended' pre-amplifier else will give different results with different pre-amplifiers and even cables.

Stay away from very sensitive electronics equipment and buy a product that is universal in nature and which does not 'crib' to work at its best.

I could not make this less technical but it is something that an electronics student who has completed his 2nd year will know and agree to.

I hope that this helps you understand some 'inside details' of the audio gear.

Regards,

Ravindra.

P.S. Kindly note that the diagram in the attachment omits a lot of design details such as de-coupling capacitors, the Boucherot cell and should only be used as reference to the text in this post.

 

[Aniket]

Good initiative Ravindra.


speaking of LM1875, it is basically a high power op-amp inside, with quasi complementary output stage and internal V-I and current limiter. I would consider it as a 10-15W amp, it has amazing performance until the limiters kick in.

I have some left over PCB's of a 2.1 project based on LM1875. if any one needs for experimenting, i am happy to share.


Regards,
Aniket

 

[Ravindra Desai]

Good initiative Ravindra.


speaking of LM1875, it is basically a high power op-amp inside, with quasi complementary output stage and internal V-I and current limiter. I would consider it as a 10-15W amp, it has amazing performance until the limiters kick in.

I have some left over PCB's of a 2.1 project based on LM1875. if any one needs for experimenting, i am happy to share.


Regards,
Aniket

Hello Aniket,

The LM1875 is a superior version of the now obsolete TDA2030 and TDA2030A.
It is one of the beauties manufactured by National Semiconductors, now taken over by Texas Instruments.

I like chip amps as they can be modified externally for superior performance.
As a bonus, you get precision bias current setting and all the comprehensive built-in protection for free, making the application circuit compact and simple.

Have you tried any variations to the basic application circuit?
Using LM1875 inside the feedback loop of an audiophile grade op-amps tremendously improves the noise performance and slew-rate.
You can actually HEAR the BASS beating in correct rhythm as against falling back in most amplifiers, making it a very fast amplifier.
One song that you easily demo this feature is Love you Zindagi from the movie Dear Zindagi.

Regards,

Ravindra.

 

[Ravindra Desai]

What is your opinion on the oldy LA4440 chip ? I had great fun with it about 25 years back.

I know the LA444X series of amplifiers from Sanyo. Those were hugely popular in the time frame that you mentioned.
That along with the BA5214 tone controller was a great combo.

In fact, the BA5214 was used in most BPL TVs. It was unique as it had DC controlled tone/ vol controls.

Regards,

Ravindra.

 

[Aniket]

Hello Aniket,

The LM1875 is a superior version of the now obsolete TDA2030 and TDA2030A.
It is one of the beauties manufactured by National Semiconductors, now taken over by Texas Instruments.

I like chip amps as they can be modified externally for superior performance.
As a bonus, you get precision bias current setting and all the comprehensive built-in protection for free, making the application circuit compact and simple.

Have you tried any variations to the basic application circuit?
Using LM1875 inside the feedback loop of an audiophile grade op-amps tremendously improves the noise performance and slew-rate.
You can actually HEAR the BASS beating in correct rhythm as against falling back in most amplifiers, making it a very fast amplifier.
One song that you easily demo this feature is Love you Zindagi from the movie Dear Zindagi.

Regards,

Ravindra.

Yes true, or consider it a direct replacement of 2040, with much better frequency response and overall performance.

I tried subtle modifications to the application circuit. with improved PCB layout design. These chips are very unforgiving regarding PCB design, and can oscillate badly if not properly designed. feedback path and supply decoupling being the most critical. I did put a low pass and high pass filter at the input to limit its bandwidth and improve stability.

While chips amps can offer great performance in a small package, with inbuilt protection and bias mechanism, they can no where match the outright performance and dynamics of discrete designs. in one of my recent designs, slew rate >50V/us and SNR>120dB and peak current capability exceeding 150A was achieved. thanks to the excellent discrete devices available today, with excellent gain characteristics, Ft, safe operating area and current capabilities.
In fact we can design a compact discrete amplifier in range of 30-50W, having the same form factor as chip amps with outstanding performance. if members are interested we can start off a new amplifier thread.

Regards,
Aniket

 

[surfatwork]

What is your opinion on the oldy LA4440 chip ? I had great fun with it about 25 years back.

Ah the good old la4440. Cheap as chips (pun intended...), ubiquitous and easy to build. Remember wiring up many of those about 3 decades back!

 

[Ravindra Desai]

Did not know about BA5214. I used my double deck Sanyo tape player with 5 band equalizer to drive them. I build my first LA4440 amp back in 11th Grade. I remember redesigning the stock PCB with better ground, using photoresist and drilling holes. I made mono block amps with some great British Transformers. The amp played my wooden 2 Way Pentagonals and we had countless late night B'day parties in college. I made some 5 setups for friends making a (declared) profit of Rs 500 for each setup including speakers. It was a very musical sounding setup. Back then I even did not know if a setup could sound bright or fatiguing for long listening..... All goodies were taken for granted and well received.

About chip amps, is this one of the best bet and musical sounding ?
LME49810/11 with bipolar output & LME49830 with FETs. Consensus is that the bipolar version sounds better and Sanken are most preferred.

More here:
http://www.audioxpress.com/article/You-Can-DIY-The-TERESA-Amplifier

regards,
Samir.

Unortunately TI has shelved the LME49810/11 and LME49830 probably because they wanted to force market their class-d ICs.

Regards,

Ravindra.

 

[Kannan]

@Ravindra Desai Sir, this is more of a technical thread started by you beyond my understanding.

However as an end user, how would you characterize the sound signature among the various class of amps (Tube, SS - Class A, AB, D etc) keeping apart the technicalities of design.
Also how does the design affect the signature?
I have seen you mention that amps made by you are engineered to sound like tube as is claimed by many other high end manufacturers.

From my personal experience, I prefer the sound of Class A as it has the most open mids when driving passive setups.
I have used different classes in my both home and car (3-way and 4-way) active setups. I find that when one takes the active route, other Classes in well designed amps match Class A with added advantage of size, power consumption and far lesser energy loss.
The only time I have run Class A in my car was to the tweeters crossed at 11khz, but at that frequency it is hard to decipher.
Now I am running full Class D in car with great results.

Other members can also bring in their perspectives.

 

[Tmr_suresh]

single ended Tube amplifiers always sound best for all audiophiles. the drawback of tube amplifier is that drive current . so it requires high efficiency speakers . apart from tube amplifiers class A SS amplifier sounds almost near tube . why class A ? to avoid cross over distortion ! then an amplifier with even harmonic profile sound good . monotonically decreasing harmonic profile with 2nd,3rd,4th order is good for ear . anything from 5th will cause ear fatigue . all this i learned from articles of great designers like nelson pass, john curl, and many more.

i liked chip amplifiers and i have built many . they are easy to use . but when our demand for hi end comes discrete amplifiers are needed. just take a look at parasound halo a21 spec. it has 16 beta matched output transistors and can provide instantaneous current peak of 60 amps.

http://www.parasound.com/a21.php

i fully agree with Mr. Kannan opinion on passive and active route.

 

[Ravindra Desai]

single ended Tube amplifiers always sound best for all audiophiles. the drawback of tube amplifier is that drive current . so it requires high efficiency speakers . apart from tube amplifiers class A SS amplifier sounds almost near tube . why class A ? to avoid cross over distortion ! then an amplifier with even harmonic profile sound good . monotonically decreasing harmonic profile with 2nd,3rd,4th order is good for ear . anything from 5th will cause ear fatigue . all this i learned from articles of great designers like nelson pass, john curl, and many more.

i liked chip amplifiers and i have built many . they are easy to use . but when our demand for hi end comes discrete amplifiers are needed. just take a look at parasound halo a21 spec. it has 16 beta matched output transistors and can provide instantaneous current peak of 60 amps.

http://www.parasound.com/a21.php

i fully agree with Mr. Kannan opinion on passive and active route.

Thank you Suresh for chipping in!

@Ravindra Desai Sir, this is more of a technical thread started by you beyond my understanding.

However as an end user, how would you characterize the sound signature among the various class of amps (Tube, SS - Class A, AB, D etc) keeping apart the technicalities of design.
Also how does the design affect the signature?
I have seen you mention that amps made by you are engineered to sound like tube as is claimed by many other high end manufacturers.

From my personal experience, I prefer the sound of Class A as it has the most open mids when driving passive setups.
I have used different classes in my both home and car (3-way and 4-way) active setups. I find that when one takes the active route, other Classes in well designed amps match Class A with added advantage of size, power consumption and far lesser energy loss.
The only time I have run Class A in my car was to the tweeters crossed at 11khz, but at that frequency it is hard to decipher.
Now I am running full Class D in car with great results.

Other members can also bring in their perspectives.

@Kannan ,

Suresh more or less answered to your point of sound signature getting its characteristics from its amplifier design.

The technical points in post # 1 were to show how amplifiers can be made to respond to changes in cable or how they can be made to that it works best with its recommended counterpart pre-amplifier.

Coming to your points:

My designs are based around two ICs: LM1875 and TDA7294. The first has a BJT output stage and the other a MOSFET.
My designs for both ICs uses same modifications to the feedback loop which is a deviation to the manufacturers datasheet.
With everything the same, if I do a blind A/B test, you will not be able to make out one from the other, forget about noticing the differences.
No one, in a blind test can confirm BJT sound from a MOSFET. Period!
So that takes care of MOSFET vs BJT. This is my personal opinion. So treat my conclusions accordingly.

Class A/B/AB can all share the exact same design, theoretically. The difference is in the bias current setting. Bias current is current that flows through the output stage when there is no input signal present. It does not flow into the speaker. So you cannot hear it. Its like a scooter that can be adjusted for different idleing settings. Ideling keeps your scooter in start position but does not actually move it

Class-B: No idleing. Hence, you have to raise the accelerator to get started and there after once you are in motion, its a smooth ride. But near zero speed, you struggle. It is fuel efficient.

Class-A: Idleing is so high that you can take off to maximum speed the moment you release the brakes after starting. It is fuel inefficient.

Class-AB: Ideling is in-between class-A and class-B, but nearer to class-B, most of the time. Fuel efficiency is in between class-a and class-b

Similar to the scooter, the amplifier in class b struggles when the input signal, which is a sine wave, crosses zero. This results in cross-over distortion.
Out brain responds to it as fatigue causing.
With Class-A there is no crossover distortion. Hence, no fatigue.
With Class-AB, the crossover distortion can me small enough to not cause any fatigue.

So that takes care of Classes of the amplifier.

You answered the part of active vs passive yourself.
In passive, it is important to have a well calibrated crossover. Something that you cannot DIY without measuring instruments. So, more often than not you are inviting disaster with a home made multi-way system.
With active crossover, you get a choice to hear, fiddle and set and you can end up happy.

All of you forget one important point with amplifiers, the one I have tried to bring up so many times: How an amplifier responds to a complex load like loudspeaker since its impedance can change -200% to + 400% about its nominal value
That is what sets it apart.
Tubes handle this part best. Semiconductor amplifiers can be tailored to handle this as nicely or even better.

So that takes care of what really matters with amplifiers, besides basic design.

I hope that things are more clear now.

Regards,

Ravindra.

 

[Aniket]

" All of you forget one important point with amplifiers, the one I have tried to bring up so many times: How an amplifier responds to a complex load like loudspeaker since its impedance can change -200% to + 400% about its nominal value
That is what sets it apart.
Tubes handle this part best. Semiconductor amplifiers can be tailored to handle this as nicely or even better. "

well IMHO that's totally wrong, tube amps are the worst at handling reactive loads, thanks to their considerably high output impedance.
While driving a complex/reactive load, which in this case is a loudspeaker, a low sensitivity 8 ohm loudspeaker can have impedance as low as 3 ohms at certain frequencies, it can in fact damage the tubes and the output transformer also, when driven into clipping or below clipping level. The current demands increases drastically and can even cause arcing in tubes. Probably that's the reason most listeners prefer high efficiency speakers paired with a tube amp, as it presents a easy load to the amp.
on the other hand, a properly designed solid state amp which is stable at 4 ohms can handle 8 ohm reactive loads with ease. even it can drive impedance as low as 1.5 ohms during transients and produce more than twice the rated power in bursts.
While all the theories explained in the above posts can be a very good starting for a DIY enthusiast, they are all still very preliminary. a lot more goes in the designing of a working amplifier. One has to consider different stages in an amplifier design and their working and circuitry and trade offs.

Regards,
Aniket

 

[surfatwork]

well IMHO that's totally wrong, tube amps are the worst at handling reactive loads, thanks to their considerably high output impedance.

Correct. This is why tube amps often have different terminals for 4/8/16 ohm speakers, which of course correspond to different taps on the output transformer.

 

[Hari Iyer]

I started my hifi journey with a Sony TA-AX44 amp from the early 80s which was built in India by a company called "ORSON". It has a Sanyo hybrid chip IC - STK2044 which is a 45 + 45 watt amplifier. I liked the sound of this amplifier very much and have used this amp for 20+ year from 1987 to 2007. I still use this amplifier for break-in speakers and my other small project and testing and is a very robust built amplifier i have ever seen. Now the amp has completed 31+ years with me with no sign of breaking down.

 

[Kannan]

i had some good vintage gear (nothing exotic) including Dynaco SS and tube pres and amps, an old Marantz receiver (1960s I think), a 1955 Fisher tube gear, a telefunken 4-in-1, couple of JBL monitors, some large Russian and Japanese alnico drivers, three NAD 3020 and some autographed paraphernalia.
They were all in decent to very good condition except the drivers which had all original voice coils but the cones had seen better days.
One fine year, when my audio journey was in absolute limbo, I jumped the band wagon and gave off the entire stuff retaining one NAD 3020. .Only the Dynaco was sold and it is still being enjoyed in a home in Coimbatore.
The last NAD 3020 also I sold recently after recapping with Mr. Syed to a Chennai FM.

A friend of my father who was into repairing amps had collected over 400 amps and various speakers over the years, but suddenly passed away one fine day when I had relocated from Chennai for work.
I came to know couple years later that the entire stuff was sold to a reseller for pittance by his family who did not have any idea about the value of the gears.

 

[Ravindra Desai]

" All of you forget one important point with amplifiers, the one I have tried to bring up so many times: How an amplifier responds to a complex load like loudspeaker since its impedance can change -200% to + 400% about its nominal value
That is what sets it apart.
Tubes handle this part best. Semiconductor amplifiers can be tailored to handle this as nicely or even better. "

well IMHO that's totally wrong, tube amps are the worst at handling reactive loads, thanks to their considerably high output impedance.
While driving a complex/reactive load, which in this case is a loudspeaker, a low sensitivity 8 ohm loudspeaker can have impedance as low as 3 ohms at certain frequencies, it can in fact damage the tubes and the output transformer also, when driven into clipping or below clipping level. The current demands increases drastically and can even cause arcing in tubes. Probably that's the reason most listeners prefer high efficiency speakers paired with a tube amp, as it presents a easy load to the amp.
on the other hand, a properly designed solid state amp which is stable at 4 ohms can handle 8 ohm reactive loads with ease. even it can drive impedance as low as 1.5 ohms during transients and produce more than twice the rated power in bursts.
While all the theories explained in the above posts can be a very good starting for a DIY enthusiast, they are all still very preliminary. a lot more goes in the designing of a working amplifier. One has to consider different stages in an amplifier design and their working and circuitry and trade offs.

Regards,
Aniket

@Aniket :

I think you missed the point of my post # 13.
I think you overlooked @Kannan 's post # 10 where he wanted me to relate sound signatures to amplifier design topologies and in a simple and non technical way

Also, you again missed the point I was making about an amplifier handling complex load.
My statement was about ability of an amplifier in handling variation in load impedance and not about power handling capacity and ability to handle heavy load currents at low load impedance or transient current capability.

I know a thing or two about design but in this forum there are not very many who understand circuit design and for them I have to give analogies and make things simple and interesting.
Nothing wrong about that, right?

Regards,

Ravindra.

 

[diyaudio]

Sound reproduction is about accuracy and faithfullness where good designed semiconductor amps excel. An amp should add nothing and subtract nothing. A good designed class AB (chpiamp or discrete) will have better characteristics almost in all parameters. There is a very small place for even class A. All the talk of tubes, class A etc is very subjective. Amps are just a gain stage. They are not intended to impart a character which is why all good amps will sound the same. Tube etc impart a signature by way of adding lower order distortion, while it may be preferred, its not accurate.
Class D has now come of age and will soon replace other types in this regard. The Bruno Putzey designs have nicely put the filter in feedback loop and matured the Class D technology. Even the TI designs are very good even though they have filter outside the loop.

Distortion is very less understood by most amp designers, a 10% distortion distortion may sound good, and 1% may sound very bad, so many switch to Class A heaters. Its not the higher distortion thats the culprit, its the order of distortion. MP3s have 80%-90% of information removed yet sound good. A 300Kbps MP3 is very hard to distinguish from the original, because MP3 compression removes the stuff not percieved by the human ear/brain.

Also, an amp is a voltage source, it will swamp any impedance variation of the load, it must do it, to do it, it has low impedance in its intended bandwidth. Any amp that has a higher impedance is a bad idea, its good for tweakers who like to keep tweaking endlessly. Faithfullness and accuracy will get lost.

 

[Aniket]

@Aniket :

I think you missed the point of my post # 13.
I think you overlooked @Kannan 's post # 10 where he wanted me to relate sound signatures to amplifier design topologies and in a simple and non technical way

Also, you again missed the point I was making about an amplifier handling complex load.
My statement was about ability of an amplifier in handling variation in load impedance and not about power handling capacity and ability to handle heavy load currents at low load impedance or transient current capability.

I know a thing or two about design but in this forum there are not very many who understand circuit design and for them I have to give analogies and make things simple and interesting.
Nothing wrong about that, right?

Regards,

Ravindra.

Hello Ravindra,

I think you missed my point completely, i referred to tube amps handling reactive/complex loads which also includes any loudspeaker on the planet and of course it's impedance will vary with frequency(read it twice again), which as stated by you was wrong.
Also referring to our other forum members and all your above posts, your rationale was absolutely correct, but, i think you proactively considered my post wrong.
And again i would state the same thing, that all the above posts are very preliminary to understand how linear amplifiers work. Lot of other things have to be taken into account to have a working amplifier at your bench. also, its not just the amplifier class (class A,B,AB,D,G,H ....) but also its topology and how various stages in an amplifier are configured, biased, driven and implemented to give any amp its sound signature.
We all are here in this forum to learn, share and help each other and correct if any wrong info is passed. perhaps, i can also write up on amplifier design and classes, but, again having a practically working design, and its details are beyond the scope of this thread. That's why I said earlier, that if members are interested, we can start of with a simple 40-50W amplifier design, and discuss its working and circuitry as a learning for all.

Regards,
Aniket

 

[sadik]

Designing is not my piece of cake, I can only execute the work after being designed. So I will just keep on reading this thread as for now.

However i would like to mention that the Class A Jean Hiraga Amplifier (30 Watts version by Dann) which i made had sounded the best to me till date.

 

[Kannan]

@sadik your construction techniques, whether speakers or amps are beyond the DIY realms. They are highly professional accomplishments.
You may not design amps, but you are an expert with tools

 

[sadik]

thankx @Kannan