lots of lm3886 power amps qs...

[swt]

HI!. although i read past threads, and used the search function, i still have some questions.
I've built the LM3886 power amp from ggg. Changed the 33k feedback res to 22k, the input imp res to 100k and 10uf fb cap to ground to 22uf as in the datasheet.
SO here i go:
do you guys use buffers in front of the power amp?. i think it's useful to use them to adapt the preamp to the power amp. but useful it's not always better.
i've tried a couple of rog circuits, and also the bsiab, and unless i crank the vol, i don't achive gain and volume from the power amp....so: should i use a booster, buffer, or maybe just go up with the feedback resistor to get more gain?
Can i add a couple of diodes ala tube screamer to get soft clipping?. ( maybe through a series resistor to limit them).
Are the single rails preamps enough to excite the power amp?. Because i'm using a bipolar supply of +-25v for the chip.
Thanks a lot for your replies!
will be reporting as i make further advances.
PS: i build tube amps and fxs....so far this is my first transistor amp.

[teemuk]

Some general thoughts...

Buffers: These are not necessarily needed. However, aside some additional complexity, adding a simple OpAmp buffer should do no harm, as it's perfectly transparent as long as you don't overdrive it. The best practice is to use a circuit that can be configured for voltage gain, as there may be a need for it (you definitely seem to have). Usually if the power amplifier is non-inverting and has a reasonably sized resistor to ground in its input then its input impedance is fairly high and no additional buffer is required. Inverting amplifiers are different thing since the source impedance is defining their gain. They benefit from having a buffer that keeps the source impedance constant.

Preamp's output amplitude does not match the input sensitivity of the power amp. Which one to use: A booster, buffer or a tweak of feedback resistor values? Well, the latter solution is definitely the easiest addition but if excessive gain (e.g. doubling the gain) is required then it might be worthwhile to add an additional gain stage (booster) instead. Most of OpAmp based circuits also act as "buffers" since they provide very low source impedance and enough drive current to handle the typical "preamp loads". You most likely won't be loading any preamp stage with impedance that is less than 1 kilo-ohm anyway. Any OpAmp can cope with those loads. Addition of a unity-gain buffer does not solve the initial problem and therefore would make no sense.

Addition of diodes to soften the clipping: Well, I guess the most important question is: Where were you planning to add them? If you put them directly in front of the power amp (that is after the volume pot) they will clip the input signal level to about 600 - 700 mVpeak. The input sensitivity of your power amp is a bit higher than that. At this place using a series resistor can be tricky since there's already a variable series resistor introduced by the volume control pot. Adding the diodes (in shunting configuration) before the volume pot is naturally something you should not do due to reasons explained soon. If you fit an inverting gain stage in front of the power amp you can of course put the diodes into its feedback loop but this configuration will not work either without significant tweaks. (That due to reasons explained soon). Naturally, you can't fit the diodes to the feedback loop of the power amp – at least not in the way done in the tube screamer circuit. Making the power amp clip softly requires some special arrangements but that's a different story.

Do note that if you soft clip the signal so that the power amp can't be overdriven you simultaneously sacrifice the amount of clean headroom, thus you will get less output power with a given amount of distortion. Your 60W (I guess) amp may as well turn into a 30W amp that just has a somewhat smoother clipping. Yet, at 60 watts of output power from the PA the amount of distortion will be quite high as the signal is significantly clipped.

Yes, single rail preamps are enough to "excite the power amp". Problem in this case is something different: The input sensitivity of your power amp is about 1Vpeak. This means you need a 1Vpeak signal to overdrive it. Most preamps, even those powered by low rail voltages, can easily exceed that level. The input sensitivity in question is, however, the sensitivity at the non-inverting input. In practice, the potentiometer in front of the power amp will introduce a significant amount of attenuation (about 9/10 at 50% of the dial in case of a logarithmic potentiometer). This means that when the volume is set to "5" the input of the volume control potentiometer should receive about 10Vpeak signal in order to fully drive your power amp to its maximum (clean) power output. Preamps powered by a single 9V rail (such as most ROG circuit) can only reach about 4V signal peak levels. This is a poor figure in comparison to about 13Vpeak of typical opamp circuits powered by bipolar ±15V rails. It is obvious that 4 volts can still definitely overdrive your power amp, yet that is not the problem: The problem is that the volume control attenuates the signal too much and you need to crank it. The problem becomes worse if the nominal output level of the preamp is significantly less than the 4 Vpeak (which I pretty much guess it is). Now, if you put a diode clipper that limits the signal to about +-600mV directly in front of the volume pot, guess what will happen...?

So, if you feel that the preamp lacks gain you can add an additional gain stage that brings the signal level higher. It might be worthwhile to use higher rail voltages in this stage so that you can feed the potentiometer with a "hotter" signal (that way you don't have to crank it so much). You can also increase the gain of the power amp to decrease its input sensitivity. Realize that these modifications actually do nothing but decrease the clean headroom. Your amp is still overdriven when it receives higher than the 1Vpeak input signal at it's non-inverting input – now it will just receive it sooner in relation to volume potentiometer's dial. Yet this compromise is often seen as a better alternative than a volume control that needs to be cranked all the way up. What is the perfect balance in all this is a highly subjective issue. If you increase the gains too excessively the volume control turns "unresponsive". This time you will have too much gain even at low settings. (For example, think linear volume potentiometers).

[darwindeathcat]

Hi Teemu,

  In your discussion about impedance matching between Pre and Power amps, it seems like you are suggesting that it is good/advantagous to have a preamp that can overdrive the LM3886. Am I correct in this assumption or did I totally miss the point?
  Anyway, I had thought that the reason to use an ROG design is that it is the Preamp that should be creating the desired distortion, and the poweramp should just be cleanly amplifying the signal (regardless of whether it is distorted or clean) that is provided from the preamp. In this case, it seems to me that the output voltage range (3-4 volts) from a stock ROG (or other FET distortion pedal) design should be what you want. ie, they are high enough to be amplified into nicely audible volumes by the poweramp, but not so high as to actually overdrive the power amp itself.
  As for the effect of the "vol" pot: would not it make more sense to wire it as a variable resistor (in series with the output of the pre) rather than as a volume control (shunting variable amounts of signal to ground)? Would you not then have finer control over the output voltages over the entire travel of the pot? i wouldd think a pot value of 500k should give a good amount of contol when used like this...
  What do you think about these issues? BTW, I'm awaiting my lm3886 and j201's mpf102's etc. from futurelec, and I'm planning to build a multi preamp lm3886 amp...

[teemuk]

It's not advantageous to overdrive the LM3886 but it is advantageous to have a preamp that is capable of doing it. Your assumption that the preamp should generate the distortion is correct but, as it was mentioned in my post, the volume potentiometer introduces a significant amount of attenuation to the signal. Its effect must be considered when matching power amp – preamp levels. 4V mentioned was a peak level that you can expect a 9V supply powered circuit can put out. (Theoretically such circuit could swing to ± 4.5V). Note the word peak: Similarly as your guitar can output quite high peak amplitudes when you strum the strings, the nominal signal amplitude is significantly lower.

If you design the power amp so that it cannot be overdriven by the pre, such amp will most likely be very "quiet". This was the problem discussed in my post. One has to turn the potentiometer to a very high setting to reach reasonable loudness levels with nominal signal amplitudes. If the input source happens to have low amplitude it cannot be compensated in any way because there is no reserve gain left for doing that.

The dial in the potentiometer is just numbers. You can arbitrarily choose what these numbers mean. In my opinion, it is perfectly all right that the power amp begins to be overdriven after the volume control is dialled past "6" or "7". At this way you don't have to keep the volume potentiometer all the way up in the highest dial settings. You can always back on volume if it seems that the power amp gets overdriven but getting more volume when the amp is already at "10" is impossible, you see the point? If you are forced to dial in to high volume control settings then it's just as bad as a situation where the only usable range of the pot is at the lowest dial settings.

What's the right range? Well, that is a subjective issue and dependant on many variables such as power amp input sensitivity, preamp output signal level, speaker SPL, softness of power amp clipping, quality of the potentiometer, guitar signal level, preferences of the amp's user etc.


All in all, all this talk may make this issue seem far more complicated than what it really is. If you feel that the amp is not "loud enough" at a certain volume control setting then just increase the gain of the preamp or the power amp a bit to compensate. That's all.

...

I don't understand how the series volume potentiometer you describe could ever work. A shunt configuration works because in it the potentiometer forms an adjustable voltage divider. I don't see that happening in the series configuration. You'd be just increasing the impedance of the circuit but this alone has no effect on signal levels. If you need to control the attenuation then you need the resistive divider somewhere. At this point you're pretty much forced to use the shunting configuration anyway. The only way I see the series configuration could work is when it's a part of an inverting amp (its input resistor to be exact) and as such defining the gain of the stage. I don't recommend trying this idea in a power amp, as it does not sound like a very stabile solution.

[swt]

thanks a lot for the replies!. very useful info. The reason i asked, is because although the power is working fine, i tried to use it in a rehearsal situation, with rog pedals as pre, but the volume needed for, was overdriven the subsequent pedals, such as univibes and delays, and when used at the correct vol, it was too low. I think my soution to this will be: use a fet buffer from, say +-18volts, and a jfet return boost similar config, as a loop. THen run the preamps at normal vol into the loop, and pedals in there. OR might try the tut ultimate fx loop, as the loop, and pre before it.
will be reporting, if anything works. thanks a lot.

[swt]

forgot...the soft clipping power amp i was talking about is implemented in the vox busker, with diodes in the feedback loop of a tda 2040 chipamp. But it has a high resistor fb, and a lower series resistor with the diodes...so i don't think it will have so much headroom.

[darwindeathcat]

Teemu,
    I now understand that you meant that it is advantagous to have a preamp that iscapable of overdriving the amp, but not that the preamp should be overdriving the poweramp. It makes sense to have the extra output from the preamp stage, since, as you say, the volume potentiometer will be shunting significant voltage.

I don't understand how the series volume potentiometer you describe could ever work. A shunt configuration works because in it the potentiometer forms an adjustable voltage divider. I don't see that happening in the series configuration. You'd be just increasing the impedance of the circuit but this alone has no effect on signal levels. If you need to control the attenuation then you need the resistive divider somewhere. At this point you're pretty much forced to use the shunting configuration anyway. The only way I see the series configuration could work is when it's a part of an inverting amp (its input resistor to be exact) and as such defining the gain of the stage. I don't recommend trying this idea in a power amp, as it does not sound like a very stabile solution.

  I think I did not fully explain what I meant by this. If you go to the following link http://www.lwharpamps.com/techmicpot.html you will see three ways of wiring up a logarithmic potentiometer to act as a volume control. The first way is the standard that, as you say, introduces a variable voltage divider, and shunts varying amounts of the signal to ground. This produces a taper that has very fine control at the begining of the pot's travel, but increasingly coarser control as the pot travels (ie less control at higher volumes).
  In the second image the pot is wired "backwards", which still introduces a voltage divider, but offers the "finer" control at the loud end of the pot's travel (which is counter clockwise now)
  The final image shows the situation I was alluding to. It seems that this will, as you say, create a variable output impedance. But won't it also offer finer variation in voltage attenuation? You will never get "total 0" volume with this type of control since there will always be some voltage across the pot, but if you have a significantly high value pot, you should be able to get fine control from very low volumes (pot at max resistance) up to very high volumes (pot at min resistance). I would think that as long as the input impedance of a non-inverting amp is set to be higher than the max output impedance that will be passed from this control, there shouldn't be any problems? Correct?

  I'm not sure that the fine versus coarse issue holds true for linear taper pots. Would it be less advantageous to use a linear pot in this case? (I mean will the "linearness" of the pot's taper make volume adjustments wierd?)

Anyway, thanks for the very informative posts! I just wanted to clarify what I actually meant!