Adding larger capacitors to Audiosector LM3875 kit

[JDawg]

I ordered the LM3875 premium kit from Audiosector a while ago and I just got it finished. Well its not completely finished but good enough to hook it up and see how it sounds. My first impression was that it sounded really good. The sound had really good mids and highs but the bottom end was lacking a little bit. I noticed that the power supply board doesn't use large capacitors like the one sold at Chipamp.com. Would adding larger capacitors make any difference? Is there a specific brand of capacitors I should be using? Is there any difference between using more expensive Jensen caps compared to cheaper ones?

[AndrewGM]

Hi

I use LM 3886 boards that no one would admit to using here on this site - the cheap ones from China.
They suggest 4 x 10,000 uf caps for the power supply board but I had an opportunity to buy bigger ones - so I did - 4 x 22,000 uf.
The bridge rectifier was rated at 35 amps @ 1000 volts so - what the hell.
It's fantastic in the bass, the mid and hf.
You can use whatever quality you can afford - look at the original Gaincard
picture attached to see what that means -
( other one is my first effort with the bigger caps )
You'll know if it was good idea as soon as you listen to it - or not
It's a certainty that someone will explain it more scientifically and warn you
that your rectifier stage will need to cope with the extra capacitance if you go too far - I'm sure that's correct.

Regards

Andrew

Andrew

[Fast Eddie D]

I beefed up the caps in the power supply of a chipamp I built from salvage- I ripped the whole power amp board out of a fubared unit and grafted it into an integrated amp. Doubling the capaticance (to merely 10,000 uF per rail ) made a very significant difference in the bass "punch" delivered by the amp- it made the difference. This was with the original large but inefficient and hot running transformer, and the original rectifiers.

It is a concern that bigger caps can destroy or shorten the service life of the rectifiers, but that is easily rectified too (pun intended). If you look at the data sheet on a typical rectifier, you will see that the peak instantaneous current is at least a magnitude of 10 greater than the service rating. You must take into consideration the transformer used too; if it's a wimpy, old technology transformer, your rectifiers will probably be safe. If it's a really beefy unit, it's a good idea to upgrade the rectifiers anyway.

[JDawg]

If I'm going to add some 10,000 uF capacitors to my power supply board I think I'm going to also snubberize it at the same time. I'm just not sure which value resistor and capacitors I should be using. The snubbered power supply from the Chipamp website uses different values than the ones shown in the schematic for the power supply from Audiosector. It says I should be using a 0.1 ohm resistor and a 3300 pF capacitor but the chipamp power supply has a 1 ohm resistor and 0.1 uF capacitor. Does it matter which ones I use?

[AndrewT]

I think that if you read Audiosector you will find that everything has been tailored to give maximum sound quality in the Mid and Treble ranges when driving high efficiency speakers.

Peter Daniel has omitted the smoothing capacitor stage completely from his implementation. Here he uses the local decoupling caps as the sole transient current supply. I think this seriously compromises the bass performance of the Power Amplifier.

You buy and listen to what you like. Peter's implementation may suit many, but it cannot suit every listener.

My philosophy is quite different and it too cannot suit everyone.

[Praudio]

Originally Posted by AndrewT

My philosophy is quite different and it too cannot suit everyone.

More details please !

[AndrewT]

if you are prepared to research the chipamp topic, you will find I have many opinions that are not considered universal.
eg.
Chipamps don't suit 4ohms speakers.
Chipamps can't perform well if running hot.
Chipamps can produce bass if the PSU is designed to deliver current.
Chipamps don't like low efficiency speakers.
Power Amps in general need to be bandwidth limited at the input.
The NFB and Smoothing and other caps in the amplifier all interact with each other and when chosen badly and/or randomly will interfere with the performance of the amplifier.
A full range amplifier needs a far wider bandwidth than what we are capable of hearing to sound right.

[Fast Eddie D]

Originally Posted by AndrewT

if you are prepared to research the chipamp topic, you will find I have many opinions that are not considered universal.
eg.
Chipamps don't suit 4ohms speakers.

The unit I ripped the power amp unit out of had 6 ohm speakers. It is now driving 6 ohm speakers. I have a pirated chipamp circuit board (don't know how good it could sound, it's sitting in a box) that drove 3 ohm speakers. Both of these amps have "no name" chips with no markings on them.

I have looked at a lot of your posts, Andrew, and I appreciate your perspective. Do you say this because 4 ohm speakers seriously reduce loop gain? Because a 4 ohm load will trigger the overload circuits and not allow full voltage swing into the load? Heat? Inquiring minds want to know.

And I do realise that paralleling them will get around all that. My buddy was bragging about his custom built amp, 100 watts into 8 ohms, and it will deliver full current and full voltage swing into 1 ohm. In fact he has it running an array of speakers in a configuration that provides a load of 1.5 ohms "nominal." Anyway, I went over there and he opened it up, and it is a bridged/ paralleled chipamp, 10 chipamps per amplifier and beefy dual power supplies. It sounds damn clean too. Very clever and has me wanting to make a "little brother" clone of it.

Chipamps can't perform well if running hot.

I can see this. Do the parameters really shift that much with heat? Crossover distortion increase?

Chipamps can produce bass if the PSU is designed to deliver current.

I figured that out when I built my salvage part chip amp. I do have experience designing and building solid state, op amp based, and tube power amps, though. Do you think chipamps are more susceptible to this phenomenon than other amps? Why?

Chipamps don't like low efficiency speakers.

Headroom and nasty clipping?

The NFB and Smoothing and other caps in the amplifier all interact with each other and when chosen badly and/or randomly will interfere with the performance of the amplifier.

More so than other amps?

Did you read my little talk about choosing poles for the RC circuits when designing an amplifier? It's the same concept, isn't it? Because haphazard choice of capacitor values will have the same effect on a solid state amplifier too, and the more poles the amplifier contains, the worse it will be.

A full range amplifier needs a far wider bandwidth than what we are capable of hearing to sound right

No questions about this.

Thanks for considering my questions, Andrew. I understand electronics pretty well, but chipamps are new to me. Evidently they're deceptively simple. Is it because it's so easy to nudge them into instability? Wouldn't conventional "common sense" design avoid this pitfall?

My first chipamp worked real good. Nothing fancy, 100% salvage parts.

[AndrewT]

I'll come back on just one point you picked up, current limiting.

Take the 3886 as an example.
The maximum output current of this chipamp is specified @ Tj=25degC and can vary from a low of 7Apk to a typical 11.5Apk.
Heat up the chipamp and this maximum current will drop.
Heat it a lot and it will drop drastically. I'll guess and suggest that at Tj~90degC, Tc~70degC, Ts~50degC the maximum current could be limited to less than 50% of the cold specification.

Now look at an 8r0 load on a chipamp and let's see if we can predict the currents that will be demanded.
68W into 8r0 is equivalent to 23.3Vac and 2.92Aac to the load.
This is 33Vpk and 4.1Apk.
Now consider a reactive 8ohm speaker load. The maximum current easily reaches double the value expected into a resistive load and in extreme (rare) cases triple and more, than what the resistive load demands.
Our 68W amp will regularly exceed 8Apk into an 8ohm speaker when played loud and occasionally exceed 12Apk. How can a warm or hot chipamp meet these demands when Spike and the other protections are overseeing the health of the Chipamp?
Change the speaker load to 6ohms and the worst case transient currents become >14Apk. The demand for current becomes greater the lower the speaker load impedance. Warm chipamps just cannot meet that transient demand. But we can help them perform. To reduce the current demand use speakers that are higher impedance, use speakers that are less reactive, use speakers that are efficient, use speakers that have a very simple crossover. this is pointing to using a full range 8ohm driver with sensitivity >93dB/W @ 1m.

Finally, of the National Series of Chipamps, the 3886 has the highest output current capability relative to it's heatsink interface thermal resistance. All others are less capable of meeting these transient current demands.

[JDawg]

I just wanted to give an update because I just added some 10,000 uF capacitors to the power supply board and added the snubber components. I sat down and listened to some cd's and I honestly can't tell the difference or if it made any improvement at all. I thought a snubberized power supply with larger caps was supposed to give better bass and a more dynamic sound but it still sounds the same as before I think. Its hard to say since I couldn't do a direct comparison before and after the changes. I think there's a a bit more bass but maybe I'm just hearing things because I was expecting an improvement. Oh well it didn't cost too much for the parts so I guess it was worth trying it to see if it made any difference.