Understanding and redesigning the input stage of LM1875

[noddy55]

........can be replaced with a 5.6K to 56K......

Hi, Hope you are fine,

What if i wanna go beyond 56K? I want to have 100k as the input impedance so as for sure always that i can use any of the input sources which can have 10K output impedance at max. Offcourse, no problem if the output impedance of the source is lower than 10k. Source's output impedance allowable up to 10K is the safest and peace of mind value for me.

1. Will using the 100K fix input impedance resistor have any bad effect on the circuit performance or sound quality, except, other than being acting as the high pass filter along with the input capacitor(1uF) (whose -3db cut off frequency i will manage accordingly by changing the capacitor value and keeping the 100K constant) ?

2. Using and fixing 56K maximum range is just because of the reason that hardly any source's output impedance can be greater than 5.6K (56K is 10 times of 5.6K) or does it implies other important reasons also?

3. Input resistance(1K) along with 1M resistor also forms an input voltage divider network. Will it be reasonable to omit either or both of them? Will either of these resistors have any effect on the high pass filter (1uf+22K)?

Thanks.

[danielwritesbac]

LM1875's have a broad range of tolerances and this makes them excellent for exploring amplifier fine tuning. What we can do with the chip amp is change the design by altering the support circuit. We can't really change a design that is sealed shut inside a chip, so that is much different from a discrete parts amplifier. Even so, many different designs can be done with the chip and we can personalize it for system symmetry.

The prospect needs adjustable prototypes. And we would need to know about the ranges of tolerances and what audible effects they can do.

Explanations are seriously hindered by the fact that all of audio is interrelated compromises that you balance towards best effect and thus any document will both help and hinder. It is the extent that matters. Please constrain problems by double-checking with other sources.

[noddy55]

Thanks! for such a long response.

1. Yes, I forgot to mention, R5(180K) will offcourse can go upto 100K with R3 being 100k as they both are related to each other.

2. I am targeting an initial gain of 16 and would like to go upto a gain of 25 if the sound quality still remains preserved. Mr.madisonears has already build one such project with k50 kit and he changed the R5(180k) to 150K to jump from a gain of 19 to a gain of 16. He felt some harshness in high frequency tones with a gain of 19 (K50 kit), which diappeared with a gain of 16 (can we think of going beyond the gain of 16 in such a scenario?). So, it seems that a gain of 16 is just fine. Or, may be it is due to the kind of source he is using with his project. I would rather initially like to test the amplifier with a sony dvd player line out with original cd soundtracks and also with an hp laptop audio output with winamp being a player for a .wav and .mp3 files.

However, i seek your advise regarding the preferred gain setting value for general playback (but never at the cost of sound quality compromise) so i can accomodate most of the input sources without changing or setting the gain each time for every source.

My possible input sources would be dvd player, laptop audio, desktop audio, apple ipod and my collection of vintage stereo boomboxes, cassette decks line out along with a vinyl record player. The amplifier will reside on a desktop computer table, where any source could be attached.So, the desktop PC and a dvd player will always for sure be a alltime source input.
Speakers would be either the bookself or a floor standing with a tweeter and a twoway full range speakers with 8ohm impeadance (i am also working on this project) so, that i can later shift to a parallel build of LM1875, if i found the sound quality interesting with a stereo build.

For the LM1875, we know that 22k and 100k feedback resistors work well but we also know that somewhere between those two values may be better performance. The mathematical average of 60k appears to work well across a wide variety of chip amplifiers; but the tolerance is broad, so 47k could work just as well.

3. So, many k50 kits are sold and build with a feedback resistor of 180k. They are working fine .Even the LM1875 datasheet has a feedback resistor of 20K(split supply) and 200K(single supply). So, it seems that the range is from 20k to 180K. The average of 20K and 180K comes up to be 100K and not 56K or 68K. So, i see the large sound field and possibilities with 100k. Your opinions are still reserved. Do you know the exact and adequate value of negative feedback current that should be passed through the NFB resitor for LM1875? Did you measure anything? I think best cannot be only our ears and predictions. Best should possibly be calculations and measurements. Super clean treble seems soothing for sometime and a deep bass could excite our senses for a party, but the midranger audio with limited bass and little clean treble, could be relaxing in a long run. Plz, correct me, if i am wrong.

The LM1875's bought at the fruit stand in the market and the LM1875's at $1 from the bargain online vendor are all fake and they cannot be pushed to LM1875's 27 watts without corruption that leads to chip burnout sooner or later.

4. This is good for us. This way we would know that we are dealing with a fake. We don't wanna live with a fake and with a wrong impression that the LM1875 has a very poor performance for our entire life. Let the fakes burn and die soon for its fate and destiny. Let's not allow fakes to bring bad image and poor impression to the original ones.

Midrange tone:
This is the job of sizing the capacitance of the power caps located on the amplifier board. Too small is too forwards, slightly too small is slightly forwards and lively. Too large is recessed, slightly too large is laid back and relaxing.

5. I will comeup to you again for more on this later when i will actually build the power supply section for LM1875 or before bying any components for power supply section.

Important Point!

6. In the temptation to achieve good sound quality and wide middle soundfield we indulge in setting the feedback or shunt resistor values to be those in mid (neither too higher nor too lower) range, individually. What i also see is the fact that, it is not the values of these resistances, individually that matters. It is the ratio of these resistance which matters. No matter what values we choose, if the ratio is same, the effect will be same. Ratio of these two resistance is the desciding factor for the amplifier gain, only (see them isolatedly). 56k feedback and 5.6K shunt gives the gain of 10. 68K feedback and 6.8k shunt also gives the gain of 10 and again 100K feedback with 10K shunt gives the gain of 10. Is there any difference between the three combinations? Will the feedback current be different in three cases when the gain is same? I don't think so. Also, we know that GBP is the Product of Gain and Bandwidth which is always constant no matter what values u choose and what gain you set.
This clearly means that if we increase the gain, bandwidth will automatically be decreased and vice-versa. So, the sound quality seems to be the function of the bandwidth which is the function of the gain, which in turn is the function of the ratio of two (feedback and shunt) resistors (Not there values). If the ratio is same we will achieve the same gain and hence the same bandwidth and the same general sound quality (as far as other conditions remains the same). Now, this bandwidth can also be manipulated further by external, high/low pass filters and capacitor values (increasing or decreasing -3db points) which in turn again changes the bandwidth and hence the gain and hence the sound quality. It is not until we change either the bandwidth or the gain we get the different sound quality (with all other things being the same in the setup). So, why should we so worry only about changing/using the mid range values for shunt and feedback resistors rather then about there ratio? Infact, We should be worry about the ratio of the gain and the bandwidth rather then being worry about the single feedback or shunt resistor values. It is the overall setting of this ratio (gain/bandwidth) which gives us the different sound qualities, GBP being always remaining constant. There should be some perfect ratio (of gain and bandwidth) and different for everyone (because we all have different setups like speakers and sources) no matter how or with what values we achieve this ratio. GBP for LM1875 is 5.5Mhz. Please, anyone, correct me to lift/support/help, for anything or if i am confused and wrong (i am still in the learning phase) with any point or words above in this last and important point.

Thanks.

[danielwritesbac]

Originally Posted by noddy55

Mr.madisonears has already build one such project with k50 kit and he changed the R5(180k) to 150K to jump from a gain of 19 to a gain of 16. He felt some harshness in high frequency tones with a gain of 19 (K50 kit), which disappeared with a gain of 16 (can we think of going beyond the gain of 16 in such a scenario?). So, it seems that a gain of 16 is just fine.

The gain didn't cause the tone/noise change. Actually, the increased feedback current helped the amp work better. The better treble is because of less treble noise. Too bad he didn't try an even smaller value feedback resistor and simply set the gain conveniently.

It is possible to boost either treble or bass dynamic by the noise added with extreme feedback resistor values like the K50 kits have a treble effect at detriment to resolution, but I don't suggest doing that.

However, I seek your advise regarding the preferred gain setting. . . My possible input sources would be dvd player, laptop audio, desktop audio, apple ipod and my collection of vintage stereo boomboxes, cassette decks line out along with a vinyl record player.

LM1875 is ideal for this broad range of use. The remaining task is to choose the gain setting that will allow the weakest source to push a 27 watt amplifier until it barely begins to clip. Your gain will be somewhere in the range of 35 to 40 because many of the music source devices listed will sound better if not straining or clipping themselves.
This time we can look at feedback resistor (FB) versus feedback-shunt-resistor (FBS) as a voltage divider.
FB, FBS, Gain
25K, 680R, 37
27K, 750R, 37
33K, 910R, 37
39K, 1K , 40
47K, 1.2K, 40
51K, 1.5K, 35
56K, 1.5K, 38
60K, 1.6K, 39
62K, 1.8K, 35
68K, 1.8K, 39
75K, 2.2K, 35
82K, 2.2K, 38

In the selections above, notice how the feedback current (the current/force of negative feedback) changes over a fairly broad range but the voltage output (gain) can remain nearly the same.

68K/1.8K will give you enough gain and it will give you some relief on NFB cap size due to the 1.8K feedback-shunt resistor. That selection is the most convenient for quick success.

The record player will need a Riaa preamp or Turntable preamp added because that is a frequency response leveling necessity for turntable response curves, and that is a need of the turntable not related to any given power amp.

All of your other sources have one thing in common and that's a 10k line level load (those headphone sources go from ~16R to ~10K, those line level sources go from ~9k to 100k), so the input of your amp will be 10k and that's perfectly acceptable since non-inverting LM1875 can correct its 10k input load versus 68k feedback resistor without No audible problem whatsoever.

P.S.
There is one difficulty and it is from high gain. A high gain amplifier is necessarily more sensitive to all of its inputs, every pin is an input and that includes power. Instead of relying on the excuse of power noise rejection specs, we'll need a bit of assurance. You'll need some high efficiency 330uF's (level) or 470uF's (attractively laid back) for amplifier board power caps. You'll also need a CRC power supply, but the good news is that it is very easy to do.

[noddy55]

Thanks a Lot! You convinced me for the higher feedback or shunt resistor values are in no way better for good performance.

1.What about lower values(below 56K for feedback)? I still have a doubt about them. Offcourse, a higher feedback current will flow with too lower values as explained by you. What problem do we face with higher feedback current? Just check out the LM4780 datasheet (Page 17, bottom left corner).
It says:

For best performance, lower values of resistors are used. A value of 1K is commonly used for Ri and then setting the value of
Rf for the desired gain.

Accordingly, this way if i want a gain of 28, best combination of Rf would be 27K and Shunt resistor would be 1K. Please, comment on this.

68K/1.8K will give you enough gain and it will give you some relief on NFB cap size due to the 1.8K feedback-shunt resistor. That selection is the most convenient for quick success.

All of your other sources have one thing in common and that's a 10k line level load (those headphone sources go from ~16R to ~10K, those line level sources go from ~9k to 100k), so the input of your amp will be 10k and that's perfectly acceptable since non-inverting LM1875 can correct its 10k input load versus 68k feedback resistor without No audible problem whatsoever.

2. What value of input impedance resistor (R3 22K) you suggest with the above conditions?

The remaining task is to choose the gain setting that will allow the weakest source to push a 27 watt amplifier until it barely begins to clip. Your gain will be somewhere in the range of 35 to 40 because many of the music source devices listed will sound better if not straining or clipping themselves.

No, i don't want to run amplifier at its max volume and source at its lowest. And, i also assume that even at its max or nearly 80% volume, none of my source will clip. So, instead i want to share the load between both (source and the Amplifier). Also, when source at its lowest volume and amplifier at its max, we experience somewhat lower Signal/Noise ratio, and the sound quality seems dull with lower S/N ratio.Hence i need a lower gain with LM1875 as compared. I am not sure but i think 20 to 25 (V/V) gain would be enough. I think this is purely a matter of experimentation.

3. Well, what lower or middle range gain you found the most exciting, perfect and giving the best, widest, sweet soundfield with most of the general input sources, during your experimentations with LM1875?

You'll need some high efficiency 330uF's (level) or 470uF's (attractively laid back) for amplifier board power caps.

4. Can u suggest some best names for high efficiency and online sources to buy from? We only have "PHILIPS" or "KELTRON" here.

You'll also need a CRC power supply

5. What is CRC power supply?

Thanks.

[danielwritesbac]

Originally Posted by noddy55

Thanks a Lot! You convinced me for the higher feedback or shunt resistor values are in no way better for good performance.

This can vary slightly by operating voltage since that also changes the current used.

What about lower values (below 56K for feedback)? I still have a doubt about them. Of course, a higher feedback current will flow with too lower values as explained by you. What problem do we face with higher feedback current? Just check out the LM4780 datasheet. Accordingly, this way if i want a gain of 28, best combination of Rf would be 27K and Shunt resistor would be 1K. Please, comment on this.

Pretty good, rather basic, more intense negative feedback than some of my suggestions, insufficient gain for many digiplayers. Not exactly my preference, but I've used it that way and it works. LM1875s are quite malleable due to wide tolerances. It is hard to go wrong except that low gain doesn't promote ease of use.
I've seen some of your calculations. Apparently, your skills are greater than mine in this area. Perhaps I should be asking you the questions?
Notes:
In general, higher gain is somewhat easier for stability, so if you're flying blind (building without scope) then keep the gain a bit high.
A bigger transformer for your power amp does not make your iphone, digiplayer, or computer more powerful; therefore the bigger the amplifier, the more gain you'll need.

No, i don't want to run amplifier at its max volume and source at its lowest.

The majority of MP3 players, digiplayers and portables compress, blare, distort if pushed to max. I suggest that you don't cause the source to distort.

Well, what lower or middle range gain you found the most exciting, perfect and giving the best, widest, sweet soundfield with most of the general input sources, during your experimentation with LM1875?

We're not faced with the narrow sweet spot problems of a miller comp amp (TDA7294) that has only one ideal gain setting. Instead, LM1875's have a broad range of tolerances with many similar results across the range. The possible differences are so tiny that you'd probably rather focus on something more important, like power circuit and layout.

What value of input impedance resistor (R3 22K) you suggest with the above conditions?

You did quote the answer already. But again the amplifier input load resistor is 10k.
For a non inverting amplifier, a strong input load like 10k is excellent for noise reduction and high resolution results.
One of your music sources is a computer. Anyone who thinks a computer power circuit is clean may vote "yes" by doing a headstand. See how important it is to have a strong input load on the amplifier to block noise so we don't have to do unnecessary headstands?

Can u suggest some best names for high efficiency and online sources to buy from? We only have "PHILIPS" or "KELTRON" here.

Panasonic FC is the #1 most mentioned at diyaudio.com.
Low ESR types are also useful on amplifier boards.
However, ordinary standard types can be more useful on power supply boards.

What is CRC power supply?

It is a noise filter that fixes a problem. Over-reliance on power noise rejection advertisements is the same thing as unnecessarily polluting the negative feedback circuit with power noise. The Fix: Instead of inserting noise into the power amplifier, we can put a simple filter into the power supply to block noise. The most popular noise filter is the CRC.
The Diyaudio.com store has a CRC power board. Check it out: P-PSU-1V20 - PSU Cap Diode Board (NEW STOCK EXPECTED MID AUGUST)
Also scroll down the page and click on schematics and other documents.
The next logical accessories (filters) to add are RC filters across the transformer primary and secondary windings to block HF pollution.

If you want a more technical description, see this page on the Lenard Audio Institute. Scroll down the topic of ripple and you'll see a CRC has fixed it.
I propose to use power noise rejection as a defense against sporadic noise, but use power supply filtering as a defense against nonstop noise. See the difference?

P.S.
Your questions that endlessly go around in circles directly relate to this: Everything is interrelated and what we need to do is simply avoid any sort of large one fell swoop effect from any one given area.
P.P.S.
The real job of fine tuning an amplifier into balance with itself is indeed like your questions and going around in circles for at least a day really does happen when building a new amplifier. For safety during fine tuning I suggest bleeder resistors and LED's on your power board. Got light, don't touch.

[noddy55]

I've seen some of your calculations. Apparently, your skills are greater than mine in this area. Perhaps I should be asking you the questions?

Thanks! But this is not my field. Still, i wanna learn more about amplifiers and audio electronic circuits. Audio is my hobby. I destroyed many audio circuits and perfectly working boomboxes initially, at my childhood. My father scolded me for this several times. I am still learning. I cannot replace a teacher and a helper like you and others at diyaudio!

You did quote the answer already. But again the amplifier input load resistor is 10k.
For a non inverting amplifier, a strong input load like 10k is excellent for noise reduction and high resolution results.
One of your music sources is a computer. Anyone who thinks a computer power circuit is clean may vote "yes" by doing a headstand. See how important it is to have a strong input load on the amplifier to block noise so we don't have to do unnecessary headstands?

Nearly, all of my boomboxes and cassette decks hava a line out impedance of 2.2k(2200ohms) to 4.7K(4700ohms). Its a rule that the input impedance of an amplifier should be at least 10 times greater than the output impedance of the load. So, i cannot think to go below 47K (ten times of 4.7K)? How can i use 10K input impedance (R3) for LM1875? I am still really confused about this.

About using CRC Filter

Yes! i will use CRC filter in my power supply. Should i go with .47R/5W pie filter resistor or some other value? How many filter capacitor per rail and with what value you suggest (2/rail are minimum for pie filter) for LM1875?
Be sure that the similar power supply may later be used for a parrallel build, if required.

Thanks.

[danielwritesbac]

Thank you for the compliment.

Originally Posted by noddy55

Nearly, all of my boomboxes and cassette decks hava a line out impedance of 2.2k(2200ohms) to 4.7K(4700ohms). Its a rule that the input impedance of an amplifier should be at least 10 times greater than the output impedance of the load. So, i cannot think to go below 47K (ten times of 4.7K)? How can i use 10K input impedance (R3) for LM1875? I am still really confused about this.

I must assume that all of the boom boxes could drive a 16 ohm or 32 ohm headphone from their headphone jacks. They're not necessarily perfect for it but yet they are all designed to do it reasonably and therefore 10k or even 5k is certainly not a challenge for any device that could drive a headphone. This also applies to the majority of computers, iphone, digiplayer, mp3 player. . . all of which may be seriously bad performers if driving an insufficient load, mainly lackluster dynamics, noise, and inferior bass. See why we wouldn't want a really light load like 47K?

While considering buffers preamps, etc. . . I just hatched an alternative idea: Have the 47K resistor (and RF filter) at the amp board. Rig a switch on the front panel to engage a parallel resistor of 12k because 47k//12k=10k. The amplifier is now switchable between 47k input load versus 10k input load. In this case, your input cap also mounts near the front panel. The switch at the 10k position will act as a noise damper, but if source current noise is too much from driving that load, you could flip the switch for 47k for the choice of an easier load to drive into a slightly more noisy amplifier. You will see that the audio differences in practice are very tiny except that the 10k load gets you better bass from your computer.

I find it very weird that a cassette deck is made incompatible with 10k line level spec, but the cure for such poor design is a buffer.
Here is a JLH buffer that you can add to sources that need a buffer when/if the manufacturer was too cheap to install a buffer.

I believe that you could use OPA2132 or OPA627 op amp for buffer, powered by OnSemi MC78**, MC79** regulators, and good quality caps, since fine parts run from clean power is a recipe for fast success. There's also some nice BC560C/BC550C and jFet Diamond Buffer projects documented here at diyaudio.com--just use the search feature and you can find many.
You actually don't need to add active stages when using LM1875 power amp. But if you have a personal preference to have everything accommodate theoretically better despite a zoo variety of sources, then you will want a buffer. If you had LM675, TDA7294, I would say that you want a buffered preamp; however, LM1875's have plentiful gain capacity so in this case you want a simple buffer.

The next step up in complexity is to add either a buffer or a buffered preamp to every source, individualized for the needs of each particular source. But, do you really need such complexity? I cannot answer this one for you. However, your turntable does require its own turntable-specific preamp. There are also many Computer preamp projects, and again do please make use of the search feature of this forum.

Yes! i will use CRC filter in my power supply. Should i go with .47R/5W pie filter resistor or some other value? How many filter capacitor per rail and with what value you suggest (2/rail are minimum for pie filter) for LM1875? Be sure that the similar power supply may later be used for a parrallel build, if required.

Voltage: A "36VCT" is 18,0,18 center tap. . . or an 18+18 dual secondaries
Current: Watts times 1.5 = VA transformer requirement
Question: Are you going to design a Stereo (or VDM) Parallel LM1875 to drive 4 ohm speakers (implying the necessity of both large transient delivery and overcurrent protection in one power supply design)? That would result in the maximum size power supply. However, due to the wattage difference, a Monobloc (or Dual Mono) for an 8 ohm speaker will result in the smallest size power supplies.
I need some more input prior to estimating your power supply needs.
Are you going to use one power supply or two?
Is transformer availability Center Tap or Dual Secondaries?
Are you going to use 8 ohm speakers or 4 ohm speakers?

[AndrewT]

The 0r47 value depends on how much voltage you can afford to drop.
With low voltage rails used in chipamp implementations expect to drop an absolute maximum of 1V, but typically this resistor will drop far less when playing music. Maybe just 50mV to 100mV.
If you guess the value too high, just add another resistor in parallel. Start with a 1r0 600mW. Keep adding 1r0 until it seems about right. eg. 3off 1r0 600mW in parallel is equivalent to 0r33 1.8W. Able to pass a continuous current of 1.5A all day (small ClassA amplifier territory). The cap before the R is loaded very heavily with ripple. You must buy a cap or bank of small caps that can accept this ripple without overheating.
Read TNT and ESP for more detail.
If you adopt 10mF (10,000uF) for the second C then about 5mF is about right for the first C. Use 5off 1mF caps in parallel for high ripple capacity.
If you want your amp to reproduce the lowest of deep bass well, then expect to double all these capacitance values.

Where have all your other questions gone? Or have you found the answers elsewhere?

[danielwritesbac]

That explanation is so nicely concise. Thank you sir.

Originally Posted by AndrewT

The cap before the R is loaded very heavily with ripple. You must buy a cap or bank of small caps that can accept this ripple without overheating.

I do really like to heatsink the bridge rectifier(s) so that the diode heat can go out of the caps instead of into the caps. And the R is a concern since the more voltage it drops the hotter it gets.

Catch diodes for the R can assure the voltage drop doesn't go beyond 1v, and a series pair of TO220 fast silicon is ~1v total and can be heatsinked. But I think the catch diode concept, costs and caveats, is unnecessary for lower power amplifiers.

So, is there an easier way to heatsink the resistor? Do we just rely on the PCB to heatsink the resistor? Or what ways are there to deal with that heat and keep the caps cool?