I am planning to order BrianGT's excellent Kit but I notice that there is still a little room for customization in the way of transformer VA rating and secondary winding voltage, or adding a stiffening capacitor to the power supply. One thing I'd like to know about, and I think it would be of value to discuss it here, is what things limit an amplifier's dynamic capability (GCs in particular, or NIGCs specifically, if anything can be said specifically about them), causing it to clip. I know in my own design, I'd like to provide the amp chip with the best possible operating conditions to minimize clipping as much as I can.
Are there things that I can do to minimize the possibility of amp clipping or is it pretty much out of my control (meaning the limiting factor is generally in a component whose value cannot be better optimized without impairing the circuit's function, or is on the amp chip itself)?
Are there things that I can do to minimize the possibility of amp clipping or is it pretty much out of my control (meaning the limiting factor is generally in a component whose value cannot be better optimized without impairing the circuit's function, or is on the amp chip itself)?
There are a number of aspects that limit an amps output: voltage capability, current capability, thermal considerations, tolerance of the components, and about a thousand other things. For an amp to be theoretically "clip-proof", it would have to have access to infinite voltage and current and be able to operate at amazing efficiency. As it is, having a good power supply can allow the amp to sustain higher output for extended periods of time, but if the power supply begins to run out of gas despite good design you get clipping. Adding more or larger filter capacitors can help, but I believe those come in to play only when the amp is approaching its peaks, not to mention there is a point of diminishing returns.
An example I use to demonstrate clipping (as opposed to distortion) is the kick-drum test. I have a CD track that starts with several kick-drum hits in quick succession and have an amp attached to an oscilloscope. The first few hits show a clean signal, but as it progresses and the filter capacitors are used up, the output becomes less in intensity and the pattern is more "tilted". A "good" amp will usually have greater stamina and the clean patterns will last longer. An "average" amp will take two or three hits then begin to degrade. The "Best Buy tax-return-special" receivers will usually have one clean hit, then degrade.
This is something of a layman's explanation (as I am a layman when it comes to the specifics), but I hope I have explained it intelligently none the less.
An example I use to demonstrate clipping (as opposed to distortion) is the kick-drum test. I have a CD track that starts with several kick-drum hits in quick succession and have an amp attached to an oscilloscope. The first few hits show a clean signal, but as it progresses and the filter capacitors are used up, the output becomes less in intensity and the pattern is more "tilted". A "good" amp will usually have greater stamina and the clean patterns will last longer. An "average" amp will take two or three hits then begin to degrade. The "Best Buy tax-return-special" receivers will usually have one clean hit, then degrade.
This is something of a layman's explanation (as I am a layman when it comes to the specifics), but I hope I have explained it intelligently none the less.
mjarve's example is a good one to show the importance of a good power supply.
An amp clips when the input signal requires the output signal to exceed the voltage that can be delivered by the amp.
First, with a regulated power supply, provide the maximum voltage that the chip can safely operate with, at a higher current than will be required by the load. Next, mount the chip on a BIG heat sink. The chip will get hot at high power output running at higher voltage so good heat sinking is a must.
To avoid clipping you need to have high power avalible for the peaks. High voltage and plenty of current is how you get the higher power.
Any amp can be driven into clipping. Matching speaker efficency and amp power is the way to avoid the problem in the first place. If your speakers need 200 watts to get rockin' then you need a 200 watt or larger amp, not a 60 watt amp running it's guts out.
I'm not trying to talk you out of building a chip amp, I'm just trying to give you an idea of what you need to think about. If 1 or 2 watts gets the guy next door yelling at you then a chip amp will work just fine for you.
An amp clips when the input signal requires the output signal to exceed the voltage that can be delivered by the amp.
First, with a regulated power supply, provide the maximum voltage that the chip can safely operate with, at a higher current than will be required by the load. Next, mount the chip on a BIG heat sink. The chip will get hot at high power output running at higher voltage so good heat sinking is a must.
To avoid clipping you need to have high power avalible for the peaks. High voltage and plenty of current is how you get the higher power.
Any amp can be driven into clipping. Matching speaker efficency and amp power is the way to avoid the problem in the first place. If your speakers need 200 watts to get rockin' then you need a 200 watt or larger amp, not a 60 watt amp running it's guts out.
I'm not trying to talk you out of building a chip amp, I'm just trying to give you an idea of what you need to think about. If 1 or 2 watts gets the guy next door yelling at you then a chip amp will work just fine for you.
More good reasons to biamp or tri-amp! Even if the demands on the bass amp can't be kept up with, it won't affect the higher frequencies. 
Since I began DIY speakers, I have been using a "nothing" amplifier, a 60w/channel home theater reciever (probably cheaper than Sony). So no matter what I do my GC will probably be a better amplifier than the one I have now. I'm just trying to think about all the considerations that I can. How much capacitance per channel would you recommend? I have been thinking about getting a piece of circuit-board and making it just for filter capacitors. I discovered that I can have 5000uF of capacitance per channel (10000uF total) for about $12 extra in parts. Is that enough or should I go for 10000uF per channel (doubling the number of capacitors)?
I've seen 25WPC amps with dual power supplies and 9200uFd capacitance per channel, 65 WPC amps with 25,000uFd per supply rail, and 120WPC brutes that sounded good with as little as 8600uFd per supply rail. I always believe in over engineering myself, so if I were doing it I would use the larger caps. But you have to make sure that the rest of the system is up to the task or your asking for premature failure elsewhere.
I say start with the 1000 (or 1500) uF on each pin and see what it sounds like. Then you can add more and see if you like it (or not).
Again, if you bi-amp, you can tailor each amp to it's respective frequency range with more capacitance on the bass amp.
(Sorry if I sound like a salesman for Gainclones 🙁 )
Again, if you bi-amp, you can tailor each amp to it's respective frequency range with more capacitance on the bass amp.
(Sorry if I sound like a salesman for Gainclones 🙁 )
An important factor not mentioned yet is the load you plan to drive. As an example, if you speaker is 4 ohms, and you amp can deliver a max current of 6 amps, the max output voltage is 24 v peak. Trying to pump out more will clip the amp on current. So, even with 50 V supplies, tgere will not be more than 24 V peak at the output.
There are many related issues as mentioned above. The max current may not be available if the chip is hot, and the max current will definitely not be available if the supply is high, because the internal chip's protection circuitry will (hopefully) watch the Safe Operating Area.
If you want max power output before clipping, see what the max output voltage is with the planned load and the max available current. Then take your supply a few volts above that (the amp needs a few volts to function above the output signal). Next, provide for good heatsinking to keep the amp cool to avoid premature SOA limiting.
Finally, make a stiff supply so thast the supply voltage doesn't sag during prolongued high power demand, which would cause voltage clipping.
A regulated supply would be a good ides in this case, although you get more dissipation in the supply, your AMP can be tweaked for max output power.
Jan Didden
There are many related issues as mentioned above. The max current may not be available if the chip is hot, and the max current will definitely not be available if the supply is high, because the internal chip's protection circuitry will (hopefully) watch the Safe Operating Area.
If you want max power output before clipping, see what the max output voltage is with the planned load and the max available current. Then take your supply a few volts above that (the amp needs a few volts to function above the output signal). Next, provide for good heatsinking to keep the amp cool to avoid premature SOA limiting.
Finally, make a stiff supply so thast the supply voltage doesn't sag during prolongued high power demand, which would cause voltage clipping.
A regulated supply would be a good ides in this case, although you get more dissipation in the supply, your AMP can be tweaked for max output power.
Jan Didden
Too many speakers today have 4 ohm woofers and 8 ohm tweeters...
Manufacturers call these 6 ohms, some even call it 8 ohms.😱
And on these days of bass reflex, speakers have big impedance dips, some 8 ohm speakers fall to as low as 3 ohms in the bass frequencies.
B&W 602 S3 does this, those guys are speciallists.
So, to make the amp more "universal", an amp that can drive this, doesn't get too hot, doesn't go into protection, or doesn't clip too much, you would be better to target the PSU voltage for a 4, or max. 6 ohm load.
If you look to the LM3875 or LM3886 datasheets (averaging these two), you'll find out that the ideal voltage for these chips is at around 30v.😉
Oh, these chips don't like capacitance at all!
I wouldn't pass 1500uf per rail, per chip.
Sound gets worse as you add capacitance.
Try it, and decice.
You can always regulate, before the regs you can put the big caps, no problem.😀
Manufacturers call these 6 ohms, some even call it 8 ohms.😱
And on these days of bass reflex, speakers have big impedance dips, some 8 ohm speakers fall to as low as 3 ohms in the bass frequencies.
B&W 602 S3 does this, those guys are speciallists.
So, to make the amp more "universal", an amp that can drive this, doesn't get too hot, doesn't go into protection, or doesn't clip too much, you would be better to target the PSU voltage for a 4, or max. 6 ohm load.
If you look to the LM3875 or LM3886 datasheets (averaging these two), you'll find out that the ideal voltage for these chips is at around 30v.😉
Oh, these chips don't like capacitance at all!
I wouldn't pass 1500uf per rail, per chip.
Sound gets worse as you add capacitance.
Try it, and decice.
You can always regulate, before the regs you can put the big caps, no problem.😀
- Status
- Not open for further replies.