LM3875 Kit Finished - Could use some more bass

[Peter Daniel]

I didn't try it with 10R but I listened to Gaincard and it sounded pretty good.

The DC offset from the preamp isn't bad - maybe 3mV max. Maybe I'll just bypass the DC blocking cap to see if that makes it any better.

If it's 3mV than it's not much indeed, but the amp has DC gain and 3mV will be amplified.

Try it without cap, but before connecting speakers measure DC offset: anything below 100mV should be fine.

 

[dfdye]

This is actually cork floor panels, they worked better than acrylic I used later in production models.

Well now! I was rather off! Still I do like the look of the cork, but the picture makes it look like a much "fancier" wood.

Also, I had a little giggle when you posted the link to the gainclone sporting the large burl blocks--flashbacks of the bronze/brass heatsink argument suddenly popped back into my head. (Just for the record, let's NOT rehash that!)

 

[danielwritesbac]

Do you have a part number or model number? And maybe a range of capacitances to look for? I'm worried about using too small a capacitance. The RC circuit formed by that input cap has a resistance equal to the output resistance of the preamp plus the input resistance of the gain clone (around 22k due to the shunt resistor, yes?). So that's about 0.5 + 22k = 22k. Once the cap's capacitance hits about 0.2uF, we have a filter that does -3dB at 30Hz, and we don't want to move that highpass filter's corner frequency up any further...

Unless the cap gets put between the shunt resistor and the chip input instead of before both?

The DC offset from the preamp isn't bad - maybe 3mV max. Maybe I'll just bypass the DC blocking cap to see if that makes it any better.

For Wimas (poly types)? 100v and ranges nearby 0.7uF (as in 1.2uF, 1uF, 0.9uF, 0.8uF, 0.7uF, 0.6uF, 0.5uF).

For electrolytic? 100v, sometimes 250v, sometimes 63v, and certainly not 50v. In ranges of 4.7uF, 4uF, 3.3uF, and 3uF.
Some people like BPT (speaker type) electrolytics.
Its popular to add a small polyester bubble or ceramic disc bypass cap (parallel cap) to electrolytics in order to achieve an "extra air" effect.
Popular sizes of bypass caps centerpoint at 47nF, meaning anywhere between 100nf and 22nf can help out an electrolytic, depending on application. Its unnecessary to use bypass caps with electrolytics that contain ceramic powder, like Elna's Tonerex and Cerefine; however, the finer electrolytics sometimes need to operate at smaller capacitance sizes than standard models.
The only thing certain with electrolytic caps is that you'll get a lot of variety and that the one you really like will be discontinued immediately.

P.S. The phrase "-3 by 30hz" means that its active at 60hz. So, instead of using 0.25uF to start rolling off the bass at 60hz, try out 0.5uF to start rolling off the bass at 30hz.
Remember to give your speakers at least an octave of room so that their "-3" sounds aren't reduced to -6.
Easy availability of a 1uF film cap is the whopping big blue thing at the local radio shack, its decent, made for decoupling audio, and it would position somewhere just barely below the audio band. Its great for Tripaths, but I haven't tried it with chipamps.

P.P.S. For going without an input filter cap, then you'd need a really nice electrolytic (or junk+bypass) at position Ci (NFB) to prevent amplification of DC. That does help make for zero DC offset, although finding a lovely sounding ecap is quite the adventure. EDIT: Same adventure as input caps, except that the values are larger, like 47uF, 68uF or so.

 

[BWRX]

I would imagine that 2R7 and 0.1uF would mess up the sonics quite a bit - that would short a lot of audio frequencies through the zobel. The 10R and 0.1uF setup shouldn't, though. It would only get rid of stuff up around 150kHz and above.

I see a number of you suggesting small changes to values of zobel components based on speculation alone. Will 2.7 ohms in series with a 0.1uF cap be that much different from 10 ohms in series with a 0.1uF cap?

The impedance of an ideal cap is 1/(2*pi*f*C). For a 0.1uF cap at 20khz that equates to 79.6 ohms. As the frequency decreases the impedance clearly increases proportionally, whch means that a zobel network has even less of an effect at lower frequencies. 79.6 + 2.7 = 82.3 ohms and 79.6 + 10 = 89.6 ohms - either of these combinations in parallel with any normal speaker load will have a VERY minimal effect at 20khz and even less so at lower frequencies.

 

[johan851]

I see a number of you suggesting small changes to values of zobel components based on speculation alone. Will 2.7 ohms in series with a 0.1uF cap be that much different from 10 ohms in series with a 0.1uF cap?

The impedance of an ideal cap is 1/(2*pi*f*C). For a 0.1uF cap at 20khz that equates to 79.6 ohms. As the frequency decreases the impedance clearly increases proportionally, whch means that a zobel network has even less of an effect at lower frequencies. 79.6 + 2.7 = 82.3 ohms and 79.6 + 10 = 89.6 ohms - either of these combinations in parallel with any normal speaker load will have a VERY minimal effect at 20khz and even less so at lower frequencies.

Exactly. You want it to have low impedance at high frequencies so that it 'shorts' high frequency oscillation or RF, but doesn't have an effect on signals in the audio range. I read 2R7 wrong because I'm not used to that notation, but there's probably not a huge difference between 1R and 2R7.

 

[BWRX]

Of course the purpose of a zobel is to terminate the output of the amp at higher frequencies, but you said that 2R7 and 0.1uF would mess us the sonics quite a bit because it would short a lot of audio frequencies... it doesn't. The difference between 2.7ohm+0.1uF and 10ohm+0.1uF is rather negligible.

 

[johan851]

P.S. The phrase "-3 by 30hz" means that its active at 60hz. So, instead of using 0.25uF to start rolling off the bass at 60hz, try out 0.5uF to start rolling off the bass at 30hz.
Remember to give your speakers at least an octave of room so that their "-3" sounds aren't reduced to -6.
Easy availability of a 1uF film cap is the whopping big blue thing at the local radio shack, its decent, made for decoupling audio, and it would position somewhere just barely below the audio band. Its great for Tripaths, but I haven't tried it with chipamps.

P.P.S. For going without an input filter cap, then you'd need a really nice electrolytic (or junk+bypass) at position Ci (NFB) to prevent amplification of DC. That does help make for zero DC offset, although finding a lovely sounding ecap is quite the adventure. EDIT: Same adventure as input caps, except that the values are larger, like 47uF, 68uF or so.

You've got me confused. A certain value of capacitance by itself shouldn't position the signal anywhere - it's all dependent on the equivalent resistance of the RC network it creates.

I can't seem to find what I'm looking for at Mouser. You keep saying "poly"...do you mean polypropylene or polyester?

 

[johan851]

Of course the purpose of a zobel is to terminate the output of the amp at higher frequencies, but you said that 2R7 and 0.1uF would mess us the sonics quite a bit because it would short a lot of audio frequencies... it doesn't. The difference between 2.7ohm+0.1uF and 10ohm+0.1uF is rather negligible.

Right, that's just what I said in my latest post (correcting myself). I thought 2R7 was 2e7 ohms. It's not. I'm used to writing out my values on paper, not on forums.

 

[BWRX]

Ah, I see. Your other post said there's probably not a huge difference between 1R and 2R7. I guess you meant to say 10R and 2R7?

 

[johan851]

Right.

 

[BWRX]

I can't seem to find what I'm looking for at Mouser. You keep saying "poly"...do you mean polypropylene or polyester?

Either will work fine. Polyester types will be physically smaller (and cheaper) than polypropylene types of the same capacitance value and voltage rating.

 

[johan851]

Ok. Because I already have a polypropylene as the input cap...

Wima part #: MKP4-1.5/400/10P27
http://www.wima.com/EN/mkp4.htm

I was under the impression that this was a decent cap.

 

[danielwritesbac]

Input Filter (Cin)
All caps are theoretically decent for some application. Your application defines the bandwidth (portion of the cap's bandwidth) by the 22k (20k?) resistor.
So, the very decent Wima brand could work even better when not 300% off the mark for size. Sorry, my apologies, I know that its rude to say it like that, but I'm having trouble explaining.

Zobel
An overlooked possibility is using a 0.047uF (47nf, bar473) ceramic cap as part of the speaker zobel. That shouldn't harm the audio band, but still ought to be able to contribute to stability and block RF. I believe that Radio Shack has a 100v model available locally. Its for radio projects, so it could be great to shunt (block) radio signals from coming into your amplifier over speaker cable.
They, and Parts Express, also have extreme high ESR Tracon green polyester bubble caps in 0.047uF, 0.1uF, and 0.22uF. This is useful if you have already purchased the expensive 2.7 ohm resistors for zobel. The 0.047uF 100v polyester caps are also frequently used for bypass cap at input and aboard power supplies.

Groundthing (I forgot the term!!)
Its a 10w, 10 ohm resistor, 1 pair diodes and a 100nF (0.1uF, bar104) ceramic cap. You turn one of the diodes backwards polarity to the other, and put all 4 items in parallel with each other as a unit to make the. . . Groundthing? Oh heck! Can somebody mention the actual term?
Earth ground (3 prong's round prong), goes to one end of this unit, and chassis ground of the amp goes to the other end.
Not only does it help prevent noises, but it also helps prevent power on/off thumps.

 

[johan851]

Input Filter (Cin)
All caps are theoretically decent for some application. Your application defines the bandwidth (portion of the cap's bandwidth) by the 22k (20k?) resistor.
So, the very decent Wima brand could work even better when not 300% off the mark for size. Sorry, my apologies, I know that its rude to say it like that, but I'm having trouble explaining.

But it having a higher capacitance than it needs to will only move the corner frequency of the filter lower... I must be missing something.

Groundthing = ground loop breaker. I don't know if it's necessary because I don't have 50/60Hz hum, but I might give it a try for the turn off pop. Turn on is silent. I don't have resistors attached to the capacitors to drain them after turn off - could that cause a problem?

 

[danielwritesbac]

Input circuit, and noise. . .

Well, the product you purchased is an excellent power amplifier. However, that's not the same as an integrated amplifier.

When you add a decoupling cap, then the power amp has become partially an integrated amp, and I think that you need to follow the guidelines on input circuit for integrated amplifiers.

The caveat is that its not well published, so here's a quickie:

Anyway, integrated amp input circuit is a source, load, pass, load, pass, load, poweramp--wherein all of the load are resistors.
The two "pass" are usually in-series resistance and input filter cap.
One or more of the "load" can be 1 megaohm or 500k resistors.
The "load" at source can be either a potentiometer or an equivilent resistor.
The load that is located at the power amplifier module is vital at attracting the signal into the amplifier.

Example: Source, 22k, 1k, 1M, 800nF, 22k amplifier.
Example2: Source, 50k, 1.5k, 1M, 400nF, 50k, amplifier.
Example3: Source, 10k, 500R, 1M, 2uF, 10k, amplifier.
Example4: Source, 100k, 2.2k, 1m, 200nF, 100k, amplifier.
notes
Format is Source, load, pass, load, pass, load, amplifier.
Specified caps are maximum size for poly (multiply by 6 for electrolytic sizes).
RF filter may parallel the load at the amplifier.

P.S. When input filter cap is overlarge, it can make loss of speaker driver control via allowing naturally present Theta (7.14hz) to enter the amplifier, become amplified and slow down the woofers. Of course, its easier just to say "use the right size" rather than try for an explanation.

 

[johan851]

P.S. When input filter cap is overlarge, it can make loss of speaker driver control via allowing naturally present Theta (7.14hz) to enter the amplifier, become amplified and slow down the woofers. Of course, its easier just to say "use the right size" rather than try for an explanation.

That makes sense. Thanks! I'll try to get some smaller Wima caps, although I can't find the blue ones at Mouser. Is MKP or MKS ok?

 

[danielwritesbac]

That makes sense. Thanks! I'll try to get some smaller Wima caps, although I can't find the blue ones at Mouser. Is MKP or MKS ok?

Polypropylene is MKP.
EDIT2: Except for Wima, my impression of other polypro caps is a "hard" (angry women) sound, so I do keep a small collection of polyester film on-hand.

My apologies again. I use 100k inputs, so I haven't tested input filter caps for 22k. Thus, I haven't tested a specific model for you, so I can't say "okay."

Peter recommends Blackgate "N" electrolytic at 4.7uF for his amplifiers. If they're unavailable, amp builders previously used Elna Cerefine for the same applications, and they are current production models.

I would try for a small collection of the right capacitance sizes at 100v and 250v. I would not purchase 50v rated ecaps (except for those above).
Bandwidth (usually undocumented) is much more important than price.
On searching for documented caps (here at diyaudio.com) for 22k loads, then the buzzwords for laid back sound could be "euphonic" "depth" and "soundstage" as some good clues.

If you build speakers, check to see if you might already own a 3.3uF BP electrolytic (BP means that both ends are crimped).
This can be used along with 0.047uf 100v polyester bubble cap (parallel), or nearby (smallish) values like that--whatever is small enough to boost the air without making voicetracks shout--is the right size bypass cap (IMO).

EDIT: Except for specifically documented applications, price of capacitors doesn't relate to performance. Wima, BlackGate and Elna, have quite a few exceptions to this rule in their lineups. Still, success varies by individual model's accomodation to your application. For success in applying cash as a solution, purchase an inexpensive collection with a lot of variety.

EDIT3: Shopping "cheat sheet"
Electrolytics most copesetic at 100v and higher ratings, from 3uF to 4.7uF
Poly most copesetic at 100v and lower ratings (if you want "euphonic" sound), in values neighboring 0.8uF (800nf).
EDIT4: This includes the big blue 1uF film cap at the Radio Shack. I think it prefers something like 10k to 15k loads though.

 

[BWRX]

When you add a decoupling cap, then the power amp has become partially an integrated amp, and I think that you need to follow the guidelines on input circuit for integrated amplifiers.

Where do you get this incorrect information from Daniel? And why do you continually post it on the forum as if it were true?! The definition of an integrated amplifier, as has been commonly accepted, is a unit that includes a preamp and a power amp. A preamp can be anything from a simple passive volume control to a full active preamp with multiple source selection. Adding a coupling cap to a power amp does NOT make a power amp a partial integrated amp.

P.S. When input filter cap is overlarge, it can make loss of speaker driver control via allowing naturally present Theta (7.14hz) to enter the amplifier, become amplified and slow down the woofers. Of course, its easier just to say "use the right size" rather than try for an explanation.

Just more BS folks!

 

[johan851]

Just more BS folks!

Hmm. Yeah, when Daniel said that I figured that if there was some kind of low frequency like that, then it would be good to block it. I googled Theta though, and the only thing that shows up at 7.14Hz are brain waves. So...given that Daniel said I should block such a thing, and also that he said no input cap would be just fine (which would pass them) I'm just going to go ahead and drop the whole input cap discussion. Input caps provide DC blocking, and as long as they're big enough to have a corner frequency mostly outside of the audio range, they should be good. I've double checked my input caps, and they're Wima MKP4 1.5uF caps. I don't see what would be wrong with them.

Thanks for speaking up BWRX - I was trying to apply my knowledge of engineering to those ideas, and I was rather confused. It's nice to know it doesn't make sense to you either...

 

[BWRX]

Input caps provide DC blocking, and as long as they're big enough to have a corner frequency mostly outside of the audio range, they should be good.

I would agree with that general statement

I've double checked my input caps, and they're Wima MKP4 1.5uF caps. I don't see what would be wrong with them.

They should be just fine. In my personal experience no cap is the best cap, but I have had good results with polyester and polypropylene film caps used as coupling caps.