Preventing a 3875 chipamp from blowing tweeters

bcdzt5

Member
2007-10-27 6:22 pm
It would appear that the 3875 chipamp is very capable of amplifying signals wanted and otherwise.

My preamp, a PC attached to a external USB Soundlbaster, generates a series of DC voltage spikes when initiating. The chipamp happily amplifies these signals and fully extends my speaker's drivers to their full reach for a fraction of a second, followed by a series of annoying/worrysome pops. Up until yesterday, I just shrugged these off, since I seldomly reset my computer, and the whole process of pops is over in about 10 seconds


Then I blew a tweeter. :xeye:


It is also possible that I am underpowering my speakers, causing problems. They are "max rated at 250 watts/channel" my system is 56Watts/channel. I have never been able to detect any clipping...
http://www.jbl.com/home/products/product_detail.aspx?prod=ND310II&CheckProduct=Y


Now heres the question:
If I hookup some 10k volume attenuating pots, and normally keep the volume low, I should be significantly safer. Right?

As of now, the amp is operating at full volume 24/7 and that seems risky to me...

I understand this is mostly a preamp problem, but I don't think I have funds to address that directly. That is unless someone volunteers rewrite the soundcard's shotty initialization code :D .
 
For a simple system, you could just use an Alpa A20K dual gang stereo potentiometer. And you'll need a pair of caps--one for left and one for right. That goes between the potentiometer and the amplifier. It seems that your amplifier has no DC decoupling (yet). Many sources do output DC, and X-Fi is one of them.

I'm imagining these caps to be Blackgate (or Nichicon ES) 4.7uf with Vishay Orange Drop 0.0068uf in parallel with the Blackgates (the bypass cap practice for more "air" since the Blackgates have more bass).

For polarized caps, the + faces the amplifier, and the - faces the potentiometer. If you get a coloration, try a physically larger cap and/or a little bypass cap.

EDIT: Its a wonder that the woofers have survived, since they have no defense against DC. There's another "soundcard puts out DC" and you can check it out too. ;)
 

bcdzt5

Member
2007-10-27 6:22 pm
dark_avenger:
Yup, thats what I've done so far, but I have occasionally forgotten to turn off the amp before restarting the PC/preamp.

Leolabs:
Fully coupled? Well, I've made no provisions otherwise, so I guess it is. But, I would think big caps across the signal path would be bad, and I haven't heard of many others using them (I think?). Do you think the reduction in audio quality would be an issue for a mid-entry level gainclone system?


danielwritesbac:
The soundcard isn't an X-FI, its a "SoundBlaster MP3+". Probably a similar DAC though.

DC coupling from what I understand is a high-pass filter on the input. I should shoot for something like 5hz for the filter? What are the caps you suggested for? Shouldn't they be much bigger, like hundreds or thosands of uf?
 
bcdzt5 said:
dark_avenger:
Yup, thats what I've done so far, but I have occasionally forgotten to turn off the amp before restarting the PC/preamp.

Leolabs:
Fully coupled? Well, I've made no provisions otherwise, so I guess it is. But, I would think big caps across the signal path would be bad, and I haven't heard of many others using them (I think?). Do you think the reduction in audio quality would be an issue for a mid-entry level gainclone system?


danielwritesbac:
The soundcard isn't an X-FI, its a "SoundBlaster MP3+". Probably a similar DAC though.

DC coupling from what I understand is a high-pass filter on the input. I should shoot for something like 5hz for the filter? What are the caps you suggested for? Shouldn't they be much bigger, like hundreds or thosands of uf?


Many of the widely available kits have the option of a DC blocking cap on the signal input to the board. I have installed the blocking caps at the output of my preamp, to prevent any DC issues with my sources and they work just dandy..... now if I could just solve my ground switching problem, life would be peachy! ;)
 
well, it can not be DC for the simple reason that you have a high pass filter in series with the tweeter, that blocks everything below 3-5000 Hz (crossover freq 5000 Hz)

typically this includes one or two capacitors in series with the tweeter.

What can happen is that if you have a large DC spike , this contains a lot of higher frequencies up and about , causing high power load into the tweeter.

Same as when clipping (very simplified ) an amplifier.
(see Rod Elliots webpage , he has a very good description)

However that has to give a LOT of power to blow(burn) the tweeter, more likely there is a mechanical damage due to large cone excursions.

In this case applying caps wont help because they just pass these frequencies anyhow. But they would protect the woofer from sub audio frequencies which can cause extreme cone excursions. But it still will snap, crackle and pop ...

The volume pot helps of course by reducing the power to the speaker.

Otherwise , use a mute relay on the output of the soundcard , driven by the power of the PC . The delay has to be as you say ca 10 s...
 
bcdzt5 said:
danielwritesbac:
The soundcard isn't an X-FI, its a "SoundBlaster MP3+". Probably a similar DAC though.

DC coupling from what I understand is a high-pass filter on the input. I should shoot for something like 5hz for the filter? What are the caps you suggested for? Shouldn't they be much bigger, like hundreds or thosands of uf?

An A20K potentiometer will attract some excess voltage and ground it (with a 20k "load"). Unless it is turned "all the way" up, there will also be much (protective) in-series resistance as well.

Compare a 20k ohm load (not much) to an 8 ohm load.
The capacitor for "all audio" high-pass will also be a much different size.

Its probable that 4.7uf on the input side of the amplifier, is slightly too large. But, at least its big enough. So, that's a great place to start.

If you did DC decoupling on the output side, the capacitor might be in the 1500uf to 2200uf range; however, this approach isn't recommended (except for use when nothing else works).

What they're saying above, is that your tweeter should have already been DC decoupled because its crossover has a capacitor in its path. The cause of its demise remains a mystery.
So,
Perhaps your amplifier isn't equipped with a zobel? Many people leave them off because they can have a too-strong effect. Well, they are adjustable.
My favorite is a little green polyester film cap, 0.22uf, with either 2.2 ohms or 2.7 ohms resistor. That's pretty gentle and still effective at wasting sounds "above" the audio range that your tweeter doesn't need anyway.

There you have it. ;) Suggest shopping list comes out to:
Input:
A20K dual gang potentiometer
Pair of "made for audio" 4.7uf electrolytic capacitors
Output:
(zobel)
pair of 0.22uf green jacket polyester film caps
(Intercap, Radio Shack, Parts Express)
pair of 2.2 ohm (or 2.7 ohm) resistors

P.S. The "sound of" capacitors relates greatly to their physical size, which somewhat relates to their voltage ratings. So, while you're shopping for ecaps, get both 50v and 100v sizes, and for polyester, try both 100v and 250v sizes. That will get you some variety. I don't know which is better because that's all about personal preferences.