it has very little to do with noise at these impedance and voltage levels.johan851 said:
It is much more related to what range of Rs the power amp will tolerate and what the Rs will do with the RF filter that is likely to be built into the power amp front end.
Some power amp manufacturers specify the optimum Rs.
Ah, I think I get it now...we're worried about frequency response as a result of the filter created at the input of the gainclone? Or is it just that the power amp is expecting a preamp with very low output impedance, and we're messing with that? Is there an optimum rs for the gainclone?
*Edit* Specifications indicate that my preamp has an output impedance of less than 0.05Ω. That's pretty low...
*Edit* Specifications indicate that my preamp has an output impedance of less than 0.05Ω. That's pretty low...
I don't recall National suggesting an optimum Rs for the basic chipamp.
Most builders report improved results from using a low source impedance.
The range of optimum Rs will depend on the external components around the chipamp but mostly it will be determined by the Input stage of the chipamp and what current National have chosen to pass that stage.
I suggest you add a second output to your headphone amp and insert a 10r in the outlet line.
Use this to feed any power amp with Zin>=600ohm.
Most builders report improved results from using a low source impedance.
The range of optimum Rs will depend on the external components around the chipamp but mostly it will be determined by the Input stage of the chipamp and what current National have chosen to pass that stage.
I suggest you add a second output to your headphone amp and insert a 10r in the outlet line.
Use this to feed any power amp with Zin>=600ohm.
johan851 said:
Since your preamp/headphone amp puts out a relatively high voltage for a preamp, you may want to set the chipamp's gain to somewhere in the range of 13--16 (22.5--24dB), which is typical of smaller amps anyway. At this gain, only 2v in will give you over 60w into 8 ohms with 36-volt rails. The chip sounds fine at this gain level, and there's plenty of feedback margin, and the M3 is probably more comfortable at this level anyway.
Craig
Well, the amp is supposed to sound better at a gain of around 30, and the kit comes with parts to set the gain at 33, so I'll just stick with that. As I look at the numbers, I'm going to need some kind of input attenuator no matter what I do.
Someone can double check my math here in case I've screwed it all up...
13Vpp (from preamp) = 6.5VAC = 4.6V RMS
Using a transformer with 18VAC secondaries, I get about +/- 25V volt rails. I don't know quite where this guy will start clipping, but the gain of the supplied parts in the kit is about 33. Even if I adjust the gain down to 13, my output signal with 4.6V RMS input will be 59.8V RMS.
59.8V RMS = 84.6VAC = ~170Vpp. That's obviously not going to work.
So if I leave the gain at 33, and attenuate the input signal by a factor of 10:
460mV RMS in gives me 15.2V RMS out, which is 43Vpp.
43Vpp is about right, and corresponds to an output power of about 30w RMS into 8 ohms, 58w RMS into 4 ohms.
So I'll just stick an attenuator pot in there and dial it in until I have it about where I want.
Someone can double check my math here in case I've screwed it all up...
13Vpp (from preamp) = 6.5VAC = 4.6V RMS
Using a transformer with 18VAC secondaries, I get about +/- 25V volt rails. I don't know quite where this guy will start clipping, but the gain of the supplied parts in the kit is about 33. Even if I adjust the gain down to 13, my output signal with 4.6V RMS input will be 59.8V RMS.
59.8V RMS = 84.6VAC = ~170Vpp. That's obviously not going to work.
So if I leave the gain at 33, and attenuate the input signal by a factor of 10:
460mV RMS in gives me 15.2V RMS out, which is 43Vpp.
43Vpp is about right, and corresponds to an output power of about 30w RMS into 8 ohms, 58w RMS into 4 ohms.
So I'll just stick an attenuator pot in there and dial it in until I have it about where I want.
Gainclone gain
At a gain of 16, you'll get 32v out (64w into 8R) from 2 volts in.
Suggestion: Download the design spreadsheet from http://www.national.com/appinfo/audio/files/Overture_Design_Guide15.xls
(it works fine in the free OpenOffice, by the way) and play with parameters until you find something you like. This also amounts to an interesting basic course in op-amp design.
Remember also, as someone pointed out above, that the National chips are not guaranteed to be stable at gains less than 10.
At a gain of 16, you'll get 32v out (64w into 8R) from 2 volts in.
Suggestion: Download the design spreadsheet from http://www.national.com/appinfo/audio/files/Overture_Design_Guide15.xls
(it works fine in the free OpenOffice, by the way) and play with parameters until you find something you like. This also amounts to an interesting basic course in op-amp design.
Remember also, as someone pointed out above, that the National chips are not guaranteed to be stable at gains less than 10.
johan851 said:
I think you've multiplied something twice. The 84.6v is already peak-to-peak (each way). And it's way more than the National chip can deliver; as I say above, 2v in will be very close to maxing you out, no matter what PS you use. Remember that nearly all discussion is in terms of either RMS AC voltage or peak-to-peak AC voltage. A 50v rms AC transformer gives you a single-ended 70v Vcc, and opamp specs are in terms of |Vcc|+|Vee| -- it's easy to get lost in all this... Just keep one foot on the floor and your eye on the square root of 2 ...
I've taken several courses in opamp design, all the way down to the gate level (analog EE major) and I think I know the basics work. (emphasis on "I think", not "I know" 
) The one thing I'm messing up is what figures we're throwing out that are voltage peak to peak and voltage in terms of wave amplitude. One is half the other. I'm not a power guy.
Looking at Wikipedia, I think that whenever I said VAC, I meant Vpeak, and when I said my transformer voltages were VAC, they should be VRMS. I'll edit it and fix it up. I think the results are similar, though, if not the same.
http://en.wikipedia.org/wiki/Alternating_current
I think it should be this:
) The one thing I'm messing up is what figures we're throwing out that are voltage peak to peak and voltage in terms of wave amplitude. One is half the other. I'm not a power guy. Looking at Wikipedia, I think that whenever I said VAC, I meant Vpeak, and when I said my transformer voltages were VAC, they should be VRMS. I'll edit it and fix it up. I think the results are similar, though, if not the same.
http://en.wikipedia.org/wiki/Alternating_current
I think it should be this:
So aside from my labeling I'm pretty sure that's correct. And I realize that 84.6Vpeak or 170Vpp is way (way) more than the chip could deliver...that's why I'm saying that even with a gain of 13, I'm going to need some kind of input attenuation.
Hi,
let's start at the speaker end.
60W into 4r0 requires sqrt[60*4]=15.5Vac=21.9Vpk=43.8Vpp
Gain is set to x33 (+30.4db).
Input sensitivity=15.5/33=470mVac
I suggest the pre-amp maximum output overhead should be between 10db and 20db. This requires a maximum output of between 0.47V *3 to 0.47*10 i.e between 1.4Vac and 4.7Vac.
Your maximum peak output of 13Vpp (=6.5Vpk=4.6Vac) is very close to the +20db overhead figure. Perfect!!!!!
Next, how quiet is the preamp stage?
If the noise output from the pre-amp after being amplified by your x33(+30.4db) gain power-amp still leaves the hiss from the speaker as inaudible in a quiet room, then again, perfect.
But it may be audible if you go close to the speaker. Assess this one after you have built it. Tell us the result.
Now, back to the beginning and look at PSU requirement based on that 60W into 4r0.
National seems to suggest that 4.1V is lost between Chipamp supply input and Chipamp output.
Your target output is ~20.9Vpk.
PSU should be about 20.9+4.1~=+-25Vdc. Perfect
Think it through in little bites, it does start to make sense and soon becomes the big picture.
To complete the big picture you need an attenuator (volume control) as part of your pre-amp input. This is what stops your system playing too loud or distorting due to clipping.
let's start at the speaker end.
60W into 4r0 requires sqrt[60*4]=15.5Vac=21.9Vpk=43.8Vpp
Gain is set to x33 (+30.4db).
Input sensitivity=15.5/33=470mVac
I suggest the pre-amp maximum output overhead should be between 10db and 20db. This requires a maximum output of between 0.47V *3 to 0.47*10 i.e between 1.4Vac and 4.7Vac.
Your maximum peak output of 13Vpp (=6.5Vpk=4.6Vac) is very close to the +20db overhead figure. Perfect!!!!!
Next, how quiet is the preamp stage?
If the noise output from the pre-amp after being amplified by your x33(+30.4db) gain power-amp still leaves the hiss from the speaker as inaudible in a quiet room, then again, perfect.
But it may be audible if you go close to the speaker. Assess this one after you have built it. Tell us the result.
Now, back to the beginning and look at PSU requirement based on that 60W into 4r0.
National seems to suggest that 4.1V is lost between Chipamp supply input and Chipamp output.
Your target output is ~20.9Vpk.
PSU should be about 20.9+4.1~=+-25Vdc. Perfect
Think it through in little bites, it does start to make sense and soon becomes the big picture.
To complete the big picture you need an attenuator (volume control) as part of your pre-amp input. This is what stops your system playing too loud or distorting due to clipping.
johan851 said:
If your kit has NFB (gain loop) of 22k with 680R, then you could change that 680R for 1K (Ri) without harming its sound.
In fact, its usually lower DC output that way.
In addition, you could change its input-side in-series resistor up to 2.2k (Rb), which is also a tiny little change.
Then, you should be able to set input loader (Rin) to 18k for a bit stronger load (corresponds to 17% stronger load than NFB).
Those are the only three "gain drop" changes that I have documented of benefit.
P.S. The amp that sounds better at a gain of 30 is the one with the two very large caps on the amplifier board. But, this general advice of 30 gain doesn't apply to the chipamp.com kit, so you may use lower gain.
Thanks for looking over that AndrewT. It looks like my numbers were right, but the thing I missed was this:
I believe you, though, so I'm just going to build it. But I'd like to understand that part of your calculation.
It seems to me that at max preamp volume (it does have a volume control) we'd be getting 4.7VAC out, resulting in an attempted 155VAC chipamp out, which would cause all kinds of clipping. So I don't understand how you got the preamp maximum output overhead' figure.
I believe you, though, so I'm just going to build it. But I'd like to understand that part of your calculation.
Hi,
if a pre-amp can only just manage to pass the full signal voltage then it is in the region of high distortion since many are single ended ClassA. Many bad recordings are outputting near maximum all the time.
It is usual to restrict the normal maximum signal to well below the clipping limit of the pre-amp and similarly in any preceding stages.
There is a further benefit in having this overhead. If any very fast transient interference signals contaminate your wanted signal, these will be slightly attenuated by the RF filters built into the amp stages. But some gets through. If these fast transients are clipped they are more audible.
I have seen many design their pre-amps to try to achieve that +20db (x10) simply to get the alledged better sound that ensues.
Think back to the days of vinyl replay.
~20db of overhead was theoretically sufficient to pass the highest recorded velocities on the recording. It sounded terrible. +30db sounded better. Some strove for +35db and I believe that exceptionally +40db was achieved. There is that spare 10 to 20db range again to lessen the audibility of the snap, crackle and pop that folk thought they were getting rid of when they adopted CD.
if a pre-amp can only just manage to pass the full signal voltage then it is in the region of high distortion since many are single ended ClassA. Many bad recordings are outputting near maximum all the time.
It is usual to restrict the normal maximum signal to well below the clipping limit of the pre-amp and similarly in any preceding stages.
There is a further benefit in having this overhead. If any very fast transient interference signals contaminate your wanted signal, these will be slightly attenuated by the RF filters built into the amp stages. But some gets through. If these fast transients are clipped they are more audible.
I have seen many design their pre-amps to try to achieve that +20db (x10) simply to get the alledged better sound that ensues.
Think back to the days of vinyl replay.
~20db of overhead was theoretically sufficient to pass the highest recorded velocities on the recording. It sounded terrible. +30db sounded better. Some strove for +35db and I believe that exceptionally +40db was achieved. There is that spare 10 to 20db range again to lessen the audibility of the snap, crackle and pop that folk thought they were getting rid of when they adopted CD.
you never intentionally use the overhead. It is there to ensure the signal is never clipped and to keep the wanted signal in the most linear amplification range but above the noise.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.