| Alcaid |
I'm going to use the schematic under to build an OPA541 gainclone, and I wondered how to lower the gain to 10 (Because of the low bandwith of the opa541). What is the best combination concerning DC offset and sound quality:
1) 1K (-in to GND) and 10K (feedback),
2) 2K (-in to GND) and (20K feedback),
3) or any other solution?
Do I have to use another resistor (+in to GND) than the 22k used in the schematic?
http://www.diyaudio.com/forums/attachment.php?postid=305699 |
|
|
| Mad_K |
| I would use 1K and 10K for all resistors (or 2K/20K or some other combination if you require higher input Z). 1K in series with the input and 10K to gnd; 10K feedback and 1K to gnd. I've tried it and I had about 1mV DC-offset ;) |
|
|
| Matttcattt |
| what would the safe/sensible minimum and maximum value for the gain be? most people use 22, and some use 10. |
|
|
| macboy |
| Most of National's chip amps are stable only for gains of 10 or higher, so that is the minimum. The maximum depends largely on the gain-bandwidth product. That limits the bandwidth at any specific gain. For the LM3875 and LM3886, this is typcially 8 MHz, but can be as low as 2 MHz. That means that at a gain of 20, the worst-case bandwidth (-3 dB) is 2 MHz / 20 = 100 kHz, but it is typically 8 MHz / 20 = 400 kHz. I would suggest that you try to keep the -3 dB bandwidth to at least 100 kHz in order to avoid any rolloff effects in the upper part of the audible range. A gain of 10 to 25 is probably a good range to stay within. |
|
|
| Mad_K |
| quote: | Originally posted by Matttcattt
what would the safe/sensible minimum and maximum value for the gain be? most people use 22, and some use 10. |
I wouldn't go above 22x. I have run opa541's stable at 2,4x at work... |
|
|
| Shoog |
All the OPA chips are supposedly unity gain stable, but I haven't tested that. I would definately run them at lower Gains than the LM chips because their open loop gain is much lower.
Shoog |
|
|
| Alcaid |
| quote: | Originally posted by Mad_K
I would use 1K and 10K for all resistors (or 2K/20K or some other combination if you require higher input Z). 1K in series with the input and 10K to gnd; 10K feedback and 1K to gnd. I've tried it and I had about 1mV DC-offset ;) |
If I use the 1k/10k combo, What pot do you suggest? Is 50k log ok? |
|
|
| metal |
Hello guys
I have used OPA541 and OPA549 after I read the datasheets thoroughly,and noticed that the full output voltage swing is equal in both directions,the -ve and +ve for the OPA541 .In contrary to OPA549,which suffers from inequal full output voltage swing in the -ve and +ve direction :xeye:
Another thing to mention is that I used 24-0-24 volt 5 Amp transformer for OP541,and it worked very fine with 4 Ohm speaker,when I connected the OP549 to this supply,it shut it self down.This means that I had to use 18-0-18 volt transformer,and I did and connected 4 Ohm speaker also,but the output power was not as high as OPA541,and using a lower impedance speaker in not a practical solution because the distortion will increase,even connecting two 4 Ohm speakers in parallel so as to obtain more power won't be helpful as this will draw more current.
Not forgetting that the protection circuit is much better in OPA549 as it senses the load indirectly.In contrary to OPA541 which uses a power resistor for protection purposes.But I had to consider the +/- 40 volt of the OPA541 and forget about the OPA549 option.
I used 1 K Ohm to ground and 15 K Ohm as the feedback resistor,the input resistor is 15 K Ohm also. |
|
|
| Alcaid |
I thought you said you read the datasheets thoroughly.
The OPA541AP is max +-35V and the OPA549T (and S) is only max +-30V. No wonder it shut down with your dual 24V supply. :xeye: |
|
|
| till |
I have no problems of that kind with an OPA549 amplifier at +-20V.
Post you excat schematic to see whats the mistake. |
|
|
| Alcaid |
Anyone got suggestions for what pot-value to use in the 1k/10k combo?
100k log?
100k lin?
50k log?
50k lin?
20k log?
20k lin?
.......? |
|
|
| Mad_K |
| quote: | Originally posted by Alcaid
If I use the 1k/10k combo, What pot do you suggest? Is 50k log ok? |
are you shure you are going to have enough gain when using it passive?
In this case, I would use a 10K log pot and a 1K in series with the input. |
|
|
| Mad_K |
| None of you are reading the '541 datasheet correctly, as it can take +/-40V ;) |
|
|
| Alcaid |
OPA541AP: +/-35V
OPA541AM: +/-40V |
|
|
| Alcaid |
Actually, that was a bit wrong. I'll try again ;)
OPA541 AP/AM: +/-35V
OPA541 BM/SM: +/-40V |
|
|
| Mad_K |
| Correctomundo!!! ;) |
|
|
| Mad_K |
| It sounds better when you don't max out on the psu voltage though... I found it very nice with +/-27V :yes: |
|
|
| Rookie |
Has anyone experimented with different gain values regarding the sound quality? Lower gain means higher bandwidth, and maybe better sound. Has anyone tried it? I was thinking about building a unity gain chip amp with OPA548 (it is unity gain stable) and drive it with PGA2310. It has enough gain (31.5 dB) and ouptut voltage swing for ~20W/8ohm (it is enough to me).
What do you think about this? |
|
|
| Mad_K |
| Methinks it sounds best with 10x gain... |
|
|
| Rookie |
Mad_K,
Have you tried it with unity gain? |
|
|
| Mad_K |
nope. I guess I will be a little better in the low end, but not as nice higher up in the spectrum..
edit:
Otoh I recon it will be much better than with, say 33x gain (wich I tried) and it distorted/compressed like crazy at high frequencies with high energy. These things one can read from the open-loop frequency graph in opa541 datasheet. The way I see it is that you have to have feedback, but not too much or too little ;) |
|
|
| GregGC |
| quote: | Originally posted by Mad_K
The way I see it is that you have to have feedback, but not too much or too little ;) |
Second that. I think there is an optimum for the NFB. Between 20 and 30 gain is what they say. You don't want the NFB to be too much either (100% at unity gain).
The more feed back the less relaxed the sound and the less distortions too. Less NFB less Distortions but not as relaxed.
I found my GC sounds the best (to me)with gain of 20.
/Greg |
|
|
| Mad_K |
| If you're talking LM chips, I'm with you on 20-30x Greg. If we're talking OPA chips I think 10-20x is more appropriate ;) |
|
|
| macboy |
| quote: | | The more feed back the less relaxed the sound and the less distortions too. Less NFB less Distortions but not as relaxed. |
Uh, you just contrdicted yourself nicely. Which is it? More feedback = less relaxed or less feedback = less relaxed?
| quote: | | If we're talking OPA chips I think 10-20x is more appropriate |
I disagree. With the 1 MHz (or less) gain-bandwidth product, you are pushing the -3 dB point too low to 50 kHz at gain of 20. That will cause measurable and audible amplitude rolloff and phase shift in the upper regions of the audio spectrum. These chips are very different from National's. I would suggest using a gain of 5 to 10 max. |
|
|
| Mad_K |
| Yep, 20x is a little high with the opa chips, but you can get away with it. As said earlier, I prefer 10x. |
|
|
| GregGC |
| quote: | Originally posted by macboy
Uh, you just contrdicted yourself nicely. Which is it? More feedback = less relaxed or less feedback = less relaxed?
|
Apologies: It should say:
More NFB(les gain)=less relaxed, less distortions.
Less NFB(more gain) = more relaxed more distortions.
/Greg |
|
|
|
