A Tale of Three LM1875 amps

[danielwritesbac]

Daniel:

I did not change the value of the Zobel resistor from 1 ohm. Both you and Andrew have remarked that value seems too small. I don't know what effect that might be having on the sound, but I sure like the way these amps sound right now. They are running very cool--the heatsink hardly gets warm, and I can easily hold my finger directly on the chip and feel no heat whatsoever. I believe that is the best indication that the chip is running well within performance limits. Maybe I'm just not pushing them hard enough, but the present volume is adequate if not overwhelming.

Please tell me, in the best way you can, what effect a different value Zobel resistor would have on the sound. I don't care AT ALL about power. It's a little hard to believe you can get 80 watts from a chip rated at 30 max! If you really need that much power, why don't you just use a chip designed for it instead of pushing the 1875 so far outside its optimium parameters? I mean: doesn't that defeat the purpose of a chip designed to provide low levels of quality sound? I am interested ONLY in the quality of the sound. Both NSC and QK50 specify 1 ohm, and I'm not willing to bet against both of them.

Peace,
Tom E

Hi Tom, um, so I don't sound absolutely nuts, I'd better explain that 80 watts. It was with 4 ohm speakers. My meter is slow and I'm sure that it counted more transients than the usual meter. The borderline between fidelity and force is actually somewhere around 60 watts, with gain in the low 40's, and much, much tinkering.

What attracted me to push the LM1875 outside its parameters is that most of its power goes into transients (dynamics), which, to me, is prettier than a high average SPL. For sure, at the same percieved loudness, a highly dynamic sound is less dangerous to an ear. But, really, I just think its thrilling. Ah, so there's why I pushed the LM1875 so hard--I like it, and I like a LOT of it.

The zobel:
The amp is rated a maximum of 4 ohm speakers, which is about 3.3 ohms "reference" load. A 1 ohm resistor, 1.2 ohms of ESR (estimated) is a 2.2 ohm load (your zobel). That's why I say that you need to change that zobel resistor to 2.2 ohms. Then that would be something like 2.2 ohms plus 1.2 ohms (estimated ESR--totally a guess), is 3.4 ohms and possibly, barely, within the "safe" boundries for amplifier design.
Still borderline, that should be at least a 1/2 watt resistor or it will eventually turn brown or make smoke.
To be absolutely sure of safety minimums, you could increase the resistor to the newer factory recommendation of 2.7 ohms resistor or larger.

I would disagree with the factory recommendation for hi-fi use because. . .
In my opinion, the zobel is the last step for the amplifier and shouldn't be installed until then. The zobel, always a masking component, may cause overcompensation (expensive overcompensation) at other areas of the design.
At the last steps, time for zobel install, I would connect a variable resistor temporarily (two connections of a speaker Lpad/rheostat). Then I would interview 0.1uF and 0.22uF caps. When the result is most pleasant (the actual job of a music amplifier), then I would disconnect the L-pad, read off the value with the multimeter and this would be the last word on what resistor to choose (done in minutes) for that particular application. . . except. . . that it must also fit safely to the reference load, because. . .

Hopefully, fine tuning (voicing) the amplifier without the zobel in place has a result of not needing to put the zobel within the audio band--which remains unproven. What has been observed is that the practice allows use of a less powerful zobel (less load, smaller caps, etc. . .)--serving as more of an RF filter and less of a tone control.

Violating the reference load may or may not put a damper on the dynamic abilities of the LM1875 (there's a "quirk" in the safety limiter that can damp transients), so that's why I mention the load violation on that zobel.

My LM1875 is able to make such power, not just because of my tinkering, but also because of Andrew T's advice on improving its efficiency. The little amp does get hot when pumping out the watts, but its like Tripath quality sound and you can't buy an extra-power Tripath kit for $24.
With efficiency assured, transient power can be up to 180 watts (on 4 ohm loudspeakers). Because of a difference in the limiter, the LM3886 cannot do dynamics of that power (and sheer speed) at the same quality of fidelity reproduction.

The sounds of resistors. Oh no, not again! Okay, well, it is a real issue. Its also a really small issue.
So, its one of those "last step" things.
I think that the #1 place to observe a difference is not because of the resistor itself, but rather because carbon can isolate a capacitor better than a metal resistor. It also had more of an effect on source isolation. That would be in your input circuit's in-series resistor, Rb. Yours is marked R1. Practically everywhere else, quality of soldering, the traces, and the distance from the chip, is going to make more difference.

Botique parts:
Given Sonic's little Tripath as an example whereby most users reported glorious results with Blackgate N 4.7uf, then its necessary to buy some because that's an established baseline. However, since you're building an amplifier from scratch, there's good news and bad news. The bad news is that there's no baseline whatever, and the good news is that any brand/price component stands an equal chance of success to carry an audio signal. That's because audio signal uses only a tiny portion of the component's bandwidth.

Botique parts, Part II:
There's an exception, and that's snubberized and/or bypass cap practices in power supplies. If a cap fails to shunt treble noises to ground, and then if a bypass cap trims off only a portion, then you have a (not successfully shunted) big peak going into the amplifier. When input signal + power = output (because that's what amplifiers do) then out comes that peak too. For this reason, the main caps on the power supply do need to be pretty good quality, and not overspend on the bypass cap quality, because of what can happen (egads!) if they shunt more powerfully than the main caps. A similar topic is the "draining" resistor on bottom of a power cap. This is a notch filter. Hopefully it has no effect on the audio band, but if it is used, then I suggest to use the exact same model power caps as the application calls for, or regard that resistor as a potential trouble spot, because of using it without an established baseline.

Hey, sorry for the long post, but I really had some explaining to do. Probably, I'll still sound nuts though. Whichever the case, maybe I gave you some cool things to try out. That's really the point.

EDIT: See page 6 here http://cache.national.com/ds/LM/LM1876.pdf
and note the zobel (Rsn) corresponds to roughly the load of an 8 ohm speaker. The LM1876 is the newer, stereo version of LM1875. Also check the other Overture amplifiers guides and note that the zobels are within the safe parameters for amplifier reference loads.
That's what I was talking about.

EDIT: Hey, if you really like it the way it is, and it doesn't get hot, then I'd say leave that one alone and buy another one to tinker with. Then you can A-B compare. That will tell you more than I ever could. EDIT: Especially because I already covered all that I could.

 

[Zero One]

Hi madisonears

There has been quite a lot of discussion over what feedback resistor to use and I have tried all three, carbon, metal, and SMD which I imagine is a type of metal film resistor.

Long and short of it, the SMD produced the smoothest grain free sound.....but

This could be due to the extremely short feedback loop and the near impossibility of anything causing problems via inductance etc with such a short loop so I won't make any statements that one is better than the other in itself.

Just on the Zobel network I have tried various configurations but I have to say what sounds best is pretty much determined by the speakers you are using, some work great without it at all, but some sound terrible. Probably looking at the impedance curve for a driver would give some good clues as to what might be needed but generally I say just try a few different values an listen to it a while. Once again this gets back to the concept of system synergy.

In short I feel that when someone says this or that is best its really not all that helpful, unless they of course are using the same power supply, speakers, pre-amp and source etc, and that doesn't even take into account the room itself.

Of course a bad amp will sound bad no matter what, but one the whole what we are doing is performing a sonic high wire act and it take a lot of effort to nail that balance, but its fun trying and by all accounts you seem to be doing a great job of it.

 

[danielwritesbac]

Kudos!!

 

[Sparky OR]

Would anyone care to post the actual schematic of BrianGT's LM1875 amp PCB? I know it seems pretty elementary, but I'd like to keep a copy of the schematic with the amp, as I'm getting old and sometimes I forget what the heck I actually built, a couple years later
I have one channel running with stock parts, the other one almost done. It's been sitting there unfinished for months, while I listen to the mono. I didn't load anything into R1, R6, and the zobel parts R5 and C7 are mounted on the speaker terminals, temporarily, til I'm sure I either wanted them at all, or was satisfied with the values. In addition I jumpered C2 instead of using the iffy looking 22uF bipolar cap included with the kit. Overall I'm happy with the unit that is playing, although i can get it to clipping when the 50k volume pot is opened up all the way. I have a nifty looking 20k attenuator bought off epay from some same place in the orient, I forget where now
I've been distracted fussing with a Harbor Freight x2 mill to use in chassis making, it's fun to,but now I need to get back to finishing all the semi-breadboarded amps sitting on the bench.
Thanks a bunch for all the info that got me to this point, mostly lurking.
BTW off topic t amp from a local guy looks pretty cool if you google "ampbox" (not the band)

 

[danielwritesbac]

Would anyone care to post the actual schematic of BrianGT's LM1875 amp PCB? I know it seems pretty elementary, but I'd like to keep a copy of the schematic with the amp, as I'm getting old and sometimes I forget what the heck I actually built, a couple years later

I have one channel running with stock parts, the other one almost done. It's been sitting there unfinished for months, while I listen to the mono. I didn't load anything into R1, R6, and the zobel parts R5 and C7 are mounted on the speaker terminals, temporarily, til I'm sure I either wanted them at all, or was satisfied with the values.

In addition I jumpered C2 instead of using the iffy looking 22uF bipolar cap included with the kit. Overall I'm happy with the unit that is playing, although i can get it to clipping when the 50k volume pot is opened up all the way. I have a nifty looking 20k attenuator bought off epay from some same place in the orient, I forget where now

I've been distracted fussing with a Harbor Freight x2 mill to use in chassis making, it's fun to,but now I need to get back to finishing all the semi-breadboarded amps sitting on the bench.
Thanks a bunch for all the info that got me to this point, mostly lurking.
BTW off topic t amp from a local guy looks pretty cool if you google "ampbox" (not the band)

Hi, you just need LM1875.pdf from National Semiconductor's website.

I also have a compatible design, http://www.diyaudio.com/forums/showthread.php?postid=1513044#post1513044 that is a serious effort. That thread, the audiosector design, the chipamp.com design, and the manufacturer's schematic are all related.

Anyway, my thread does contain many fun and interesting options that just happen to work for your chipamp.com lm1875.

DC offset:
One thing you need to notice/test is that your input impedance needs to be about 10k in order to reduce DC offset (since you removed the NFB cap). One pin of that resistor contacts the chip, and the other contacts the 0v (ground).

However, it can go in this order instead. . . source, dc block cap-20k pot, 20k resistor load. To do that nicely, you may want need two additional (optional) loads, directly upon the RCA jacks, a 220pF and a 1m to block buzzes and RF. All told, this option has the amplifier as a capacitive load, and that does have wider tolerances for impedance matching, as well as support for a wider variety of input caps, such as Nichicon's ES.

Zobel:
Well, the good news is that the speaker terminals are a great spot for the speaker zobel, although the RC components are usually on the inside of the amplifier enclosure. For longevity, a 5w resistor is highly recommendable (and inexpensive too). A great deal of power can run through the speaker zobel, so perhaps resistors of 4R to 10R range is more conservative of the amplifier's power.