diyAudio (http://www.diyaudio.com/forums/index.php)
-   Chip Amps (http://www.diyaudio.com/forums/chip-amps/)

 noddy55 28th August 2012 12:39 AM

Dear Sir,

I am really confused about certain issues with chip amp design. I am sure you could help me.

1.How we can determine the exact bandwidth of a particular design like LM1875 K50 Kit circuit diagram?

2.If we want to design a bandwidth of 20Hz to 20Khz, then lower -3db cut off should be one decade below the required frequency i.e 2Hz and better if we can go up to 0.6Hz. What about the higher frequencies -3db cut of? Do we need to go one decade or more with them also? ie. 20Khz*10 = 200Khz?

3.We have two high pass filters (C1/R3 and R4/C3) in k50 circuit diagram. Why two? What if we set both of them to different -3db cut of? Which one of them will be effective? How do they both affects each other?

4.There is no Low pass filter in K50 diagram.How we limit the higher frequency to say 20Khz(required) or 200Khz (need to be set to get the actual required)? If we do not set the higher frequency limit, by any means what will be the effective higher frequency, in such a scenario and how we will determine this effective frequency?

5.We determine the value of c3(22uf) in conjunction with R4(10k) with a formula:
fl = 1/2 * pie * F *C
Is the determination and final selection of this capacitor value is also affected by R5(180K) or R3(22K) in any way?

6.Please, recall the relation of R3(22K) with R5(180K).

7.R1(1K) with R2(1M) forms the asymmetric L-Pad atteunator as well as impedance matcher.How we will determine the actual atteunation performed by these two resistors in decibel(db)? and, with what direction (higher side/or lower impedance side) the impedance is matched with these two resistors in k50 diagram?
I want to know the way if and how i could change these values for a different configuration but with the same effect?

8. GBP is a gain and bandwidth product which always remains constant, no matter we change the gain or the bandwidth. If we increase bandwidth, gain will decrease and vice-versa.

We know that: GBP = Gain*Bandwidth, or Bandwidth = GBP / Gain.
Also, Bandwidth / 1000 = Bandwidth in Khz. GBP for LM1875 is 5.5Mhz.

So, i found certain values (assuming GBP=5Mhz) based on the above formula and k50 diagram:

R5(Feedback resistor)/R4(Feedback Shunt Resistor)

150k/10k (gain 16.00V/V) (bandwidth 313Khz)
47k/2.7k (gain 18.41V/V) (bandwidth 272Khz)
180k/10k (gain 19.00V/V) (bandwidth 263Khz)
47k/2.2k (gain 22.36V/V) (bandwidth 224Khz)
68k/2.7k (gain 26.19V/V) (bandwidth 191Khz)
56k/2.2k (gain 26.45V/V) (bandwidth 189Khz)
100k/3.3k (gain 31.30V/V) (bandwidth 160Khz)
47k/1.5k (gain 32.33V/V) (bandwidth 155Khz)
100k/2.7k (gain 38.04V/V) (bandwidth 131Khz)

You can see the product of gain and bandwidth is always near about 5Mhz or constant. Now my questions or confusions are:

i). Do this 5Mhz GBP is always automatically itself attainable all the time? Can't we have a lower GBP at a particular time?

ii). If we decrease the gain by Rf/Ri ratio, will it automatically(due to GBP contant rule) increase the bandwidth, without touching any high pass or low pass filters in the circuit? or,

iii). If we insert a low pass filter at the input of the amplifier, thus only intending to change/limit the bandwidth, will it also automatically(due to GBP constant rule) change the gain set by Rf/Ri?

Thanks.

 sreten 28th August 2012 01:23 AM

Hi,

Loaded questions seeking confirmation aren't really questions at all.

You haven't provided a link to the circuit in question.

rgds, sreten.

 abraxalito 28th August 2012 01:59 AM

Just cherry picking the more interesting questions (ones not requiring the schematic details) here...

Quote:
 Originally Posted by noddy55 (http://www.diyaudio.com/forums/chip-amps/218649-hi-please-help-issues-lm1875-chipamp-design-post3142421.html#post3142421) 2.If we want to design a bandwidth of 20Hz to 20Khz, then lower -3db cut off should be one decade below the required frequency i.e 2Hz and better if we can go up to 0.6Hz.
I can't see anything better by going down to 0.6Hz.

Quote:
 What about the higher frequencies -3db cut of? Do we need to go one decade or more with them also? ie. 20Khz*10 = 200Khz?
If you want 0.1dB flatness to 20kHz then yes.

Quote:
 3.We have two high pass filters (C1/R3 and R4/C3) in k50 circuit diagram. Why two? What if we set both of them to different -3db cut of? Which one of them will be effective? How do they both affects each other?
They're in series I presume (no k50 schematic provided) so the lower one will be the most effective.

Quote:
 4.There is no Low pass filter in K50 diagram.How we limit the higher frequency to say 20Khz(required) or 200Khz (need to be set to get the actual required)? If we do not set the higher frequency limit, by any means what will be the effective higher frequency, in such a scenario and how we will determine this effective frequency?
Its good engineering practice to limit the bandwidth at the input of the amplifier - for RF protection at the very least. Normally this is done with a single RC filter, though inductors are employed on occasion.

Quote:
 8. GBP is a gain and bandwidth product which always remains constant, no matter we change the gain or the bandwidth. If we increase bandwidth, gain will decrease and vice-versa.
Yes - but note there's a production spread in GBW - min, max and typ.

Quote:
 i). Do this 5Mhz GBP is always automatically itself attainable all the time? Can't we have a lower GBP at a particular time?
You answered this question yourself above 'always remains constant'.

Quote:
 ii). If we decrease the gain by Rf/Ri ratio, will it automatically(due to GBP contant rule) increase the bandwidth, without touching any high pass or low pass filters in the circuit? or,
The bandwidth is the bandwidth of the amplifier part, not including the low pass filters.

Quote:
 iii). If we insert a low pass filter at the input of the amplifier, thus only intending to change/limit the bandwidth, will it also automatically(due to GBP constant rule) change the gain set by Rf/Ri?
No, the low pass filter is not included in the GBW of the amplifier - its quite separate, being your circuit and not the chip manufacturer's and therefore has no effect on that figure.

 noddy55 28th August 2012 10:00 PM

1 Attachment(s)
Sorry! Mr sretan, my apology,

Here is the circuit diagram for K50 LM1875 Audio Kit.

Thanks.

 noddy55 28th August 2012 10:34 PM

Thanks! Mr. abraxalito for some of the answers.

Quote:
 The bandwidth is the bandwidth of the amplifier part, not including the low pass filters. No, the low pass filter is not included in the GBW of the amplifier - its quite separate, being your circuit and not the chip manufacturer's and therefore has no effect on that figure.
So, does this means that we can have a different GWP for the overall circuit (which can be lower than or equal to the internal chip GWP (never higher than this)) as compared to the chip GWP, which is always constant? The overall circuit design can limit its overall GWP to a lesser value, say 2MHz as compared to the LM1875 chip's internal GWP(5.5Mhz)?

Quote:
 If you want 0.1dB flatness to 20kHz then yes.
What value we should consider proper for a true Hi-Fidelity amplifier? or what should be the proper cut of frequency in case of LM1875 or any other Hi-Fi audio amplifier?

Quote:
 but note there's a production spread in GBW - min, max and typ
So, what does this production spread of min, max and typ actually means?

Thanks.

 abraxalito 29th August 2012 04:42 AM

Quote:
 Originally Posted by noddy55 (http://www.diyaudio.com/forums/chip-amps/218649-hi-please-help-issues-lm1875-chipamp-design-post3143560.html#post3143560) So, does this means that we can have a different GWP for the overall circuit (which can be lower than or equal to the internal chip GWP (never higher than this)) as compared to the chip GWP, which is always constant?
The term 'GWP' only applies to the LM1875, not the whole circuit. The circuit can have a different bandwidth, a different gain but I can't see a meaning to a whole circuit having a gain-bandwidth product. I am not sure I understand your question.

Quote:
 The overall circuit design can limit its overall GWP to a lesser value, say 2MHz as compared to the LM1875 chip's internal GWP(5.5Mhz)?
Its quite possible to reduce the gain of the circuit and keep its bandwidth constant - you do this by adding what I call 'gain stealing networks'. In your case the network would be an RC (in series) between the + and - pins of the LM chip.

Quote:
 What value we should consider proper for a true Hi-Fidelity amplifier? or what should be the proper cut of frequency in case of LM1875 or any other Hi-Fi audio amplifier?
People have different views on this - my own view is less than 0.1dB droop at 20kHz is preferred. Though for an amp I design solely for myself I'd relax this to 0.1dB at 17kHz as my hearing doesn't extend to 20kHz.

Quote:
 So, what does this production spread of min, max and typ actually means?
It means there's a spread of values in practice, a statistical distribution. Most will be near the typical.

 noddy55 29th August 2012 11:42 PM

Quote:
 The term 'GWP' only applies to the LM1875, not the whole circuit. The circuit can have a different bandwidth, a different gain but I can't see a meaning to a whole circuit having a gain-bandwidth product. I am not sure I understand your question.
Thanks! You cleared my confusion about this for ever. Yes, i did not actually means the whole circuit's GWP. I just wanted to confirm that a circuit can have any bandwidth and can be set at any gain value and the product of both (which is not known/Called/Identified as GWP) can be different from that of the chip's internal product of bandwidth and gain (which is in actual a GWP) , but it's maximum range is always limited by the max. value of the chips GWP which is 5.5Mhz.
In short, we can set the gain and bandwidth product of overall circuit to any value, so far as it lies in the range of 5.5Mhz.

Quote:
 Though for an amp I design solely for myself I'd relax this to 0.1dB at 17kHz as my hearing doesn't extend to 20kHz.
And how you do that actually? By choosing, one decade above 17Khz or two decades above?

Thanks.

 abraxalito 29th August 2012 11:59 PM

Quote:
 Originally Posted by noddy55 (http://www.diyaudio.com/forums/chip-amps/218649-hi-please-help-issues-lm1875-chipamp-design-post3144947.html#post3144947) In short, we can set the gain and bandwidth product of overall circuit to any value, so far as it lies in the range of 5.5Mhz.
Delete the word 'product' there and yes, I agree. In practice the bandwidth will always be lower than 5.5MHz divided by the minimum stable noise gain of the chip which is normally around 26dB.

Quote:
 And how you do that actually? By choosing, one decade above 17Khz or two decades above?
I use the rule of thumb that the 0.1dB frequency is a factor of ten from the 3dB frequency. So to get 17kHz at 0.1dB I set the 3dB at 170kHz.

Quote:
Have I answered your question now? Or do you misunderstand about dBs? More questions always welcome if you're not clear yet.

 AudioLapDance 30th August 2012 01:31 AM

Check the chipamp forum for recent posts about the 3886.

There is a good discussion on the proper sizing of the input cap and its relationship with the capacitor in the feedback network. As well as comments regarding a high freq cap you can put accross the inputs.

I have the K50 but our very own

Chipamp.com

has a much tighter PCB. You get 2 compact 1875 boards and a power supply board.

Good luck and keep us posted!

Cheers,
Jeff

PS the chipamp site is a bit rusty but they seem to be stirring again.

 noddy55 3rd September 2012 10:14 PM

Thanks, Mr. abraxalito,

Point 3, 4th last part, 5, 6 and 7 above are still not answered.

If anyone can? Then Please. Can't go for more questions until.

Thanks.

All times are GMT. The time now is 05:43 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.