About make an audio power amplifier.

[Tony Stewart]

".Use 5A L and Plastic C instead of huge e-caps.". referred to my suggestion to use cheaper, better SMPS with lower ripple using inexpensive small LC.


In your case, you need, huge e-cap, depending on your tolerance to muddy distorted bass with 100 Hz hum at high power. Without the added complexity of cascode current sources, your supply ripple rejection will be poor in PA at high power.

What C value?

 

[spec]

I have thought it is almost like unity gain. As you said, gain≈1, why not use unity gain, 1k worth?

The 1K resistor has little effect on the voltage gain of the opamp stage in the audio band, it just gives isolation of the input from stray capacitances etc at high frequencies. Remember that the OPA 2134 has a higher frequency response than normal opamps. The 100 Ohms on the output of the opamps is to isolate the output of the OPA2134.

spec

 

[spec]

Here is an outline circuit of the Elliot speaker protector with AC monitoring and relay drive. This basically has the functionality of your circuit, as discussed, but simplified and rationalised. There is always a danger with monitoring circuits that they will not turn on. I have not investigated this aspect. The time constants R12/C10 and R13/C9 will give around one second before AC monitoring starts.

spec

ERRATA
(1) C9 should read 22uF

NOTES
(1) R10 is not strictly necessary but is just good practice.

 

[Nikolai Petrenko]

What C value?

I already have 10000uF old Epcos cap (extract from old amplifier).

 

[Nikolai Petrenko]

My active crossover, certain has many error:

 

[spec]

My active crossover, certain has many error:
View attachment 100066

Impressive and well executed. A good schematic layout helps a lot to understand, check, analyze, modify, and fault-find a circuit.

Suggest these minor amendments to reduce cross-talk and possibly improve PCB layout:
IC2A change to IC1B
IC3A change to IC2B
IC2B change to IC3A

IC1A and IC2B need a 47 Ohm resistor between their outputs and the high pass filter inputs.

Are you planning on combining the filter function and preamp all in one to reduce the number of gain stages that the audio signal goes through?

One thing I did notice is that you have two low pass filters where one could do the job. This can be implemented by combining the left and right audio channels, for the bass channel, before the low pass filter. On reflection I am not sure if it is better as the signal overhead may be less with a single low pass filter- something to think about.

In an amplifier chain there is always a trade off between distortion, dynamic range, and noise and active filters are notorious headroom thieves so this is an area to be investigated.

Why have you suddenly changed from an OPA2134 to an NE5534 for the bass sub output?

spec

 

[Nikolai Petrenko]

Thank, I have fixed the error, I have wrong at two low-pass filter, gain=2 is replace by unity gain

 

[spec]

Hy Nicolai,

I have had a more detailed look at the filter and preamp designs.

This is what I suggest:
(1) Combine the preamp (mine) with your filter circuit.
(2) Replace the two filter input opamps with the two output opamps from the preamp (left and right channels) in the preamp.
(2) Delete IC4
(3) Delete IC3B
(4) Connect R15 top to IC1A output
(5) Connect R16 lower to IC1B output
(6) The output of the IC3A is then the output to the sub amplifier

That will leave one spare OPA2134 opamp (I think).

spec

 

[spec]

About earthing and star points: While it is good that you are attending to the earths and hopefully supply lines, it is a little more complicated than just connecting every earth line to one point. The important thing is to visualize what the opamp sees at its non inverting and inverting inputs and also practical implications. Tony Stewart mentioned similar.

As a result, you may have a few earth star points. Don't let this put you off though. The earthing can easily be sorted. The main thing is to get the topology and functionality correct. The next stage will be fine-tuning and optimization.

It is quite handy that you have posted the filter design because it will be useful for me on another project. I hope you don't mind me stealing it?

spec

 

[spec]

If it is not too much hassle, operate the OPA2134 at 17.5V. They sound better with high voltage supply lines and also you get more overhead.

spec

 

[spec]

This is what I was thinking about for the combined preamp and filters:

ERRATA
(1) Change R16 to 10K
(2) Change R24 to 10K
(3) Change R25 to 10K
(4) Change pin identity from, 'LEFT MID OUTPUT' to 'LEFT MID/HIGH OUTPUT'
(5) Change R10 to 47R
(6) Change R1 to 47R
(7) Change R19 to 47R
(8) Change R22 to 47R

DATA SHEETS
(1) OPA2134
https://www.ti.com.cn/cn/lit/ds/symlink/opa2134.pdf

 

[Nikolai Petrenko]

EDITED

It is quite handy that you have posted the filter design because it will be useful for me on another project. I hope you don't mind me stealing it?

How many schematic and design I have and will steal from you?

If it is not too much hassle, operate the OPA2134 at 17.5V. They sound better with high voltage supply lines and also you get more overhead.

As you know, NE5532 are my choice now.

This is what I was thinking about for the combined preamp and filters:

Why call output of high-pass filter is "mid"? I think should be "mid+treb.", may I am too stupid fastidious?
And only need one opamp in low-pass filter section, so go for NE5534?

 

[Nikolai Petrenko]

Oh, your design not only help me to save money, it also reduce the duty on power supply, also save more PCB and space too. One dual opamp has been kicked out . What will happen if 12K5 arenot available in the store? Any answer if it happen?
With your lastest schematic and plugged NE5532, is it neccessary to use coupling caps for the power amp input?
The Elliot amp has voltage gain=23 (was I wrong), and Elliot also said DC offset< 100mV, so maxium DC in the input without capacitor will be <4mV. <4mV DC offset, is it possible? Or any wrong in my think?

 

[Nikolai Petrenko]

About earthing and star points: While it is good that you are attending to the earths and hopefully supply lines, it is a little more complicated than just connecting every earth line to one point. The important thing is to visualize what the opamp sees at its non inverting and inverting inputs and also practical implications. Tony Stewart mentioned similar.

As a result, you may have a few earth star points. Don't let this put you off though. The earthing can easily be sorted. The main thing is to get the topology and functionality correct. The next stage will be fine-tuning and optimization.

And each star point will be connected to main star point?

 

[spec]

EDITED

How many schematic and design I have and will steal from you?

As you know, NE5532 are my choice now.

Not possible without decoupling capacitors, so they would lead to more expensive overall.

Why call output of high-pass filter is "mid"? I think should be "mid+treb.", may I am too stupid fastidious?

Correct- fastidious is good.

And only need one opamp in low-pass filter section, so go for NE5534?

Got it

spec

 

[spec]

Oh, your design not only help me to save money, it also reduce the duty on power supply, also save more PCB and space too. One dual opamp has been kicked out . What will happen if 12K5 are not available in the store? Any answer if it happen?

Those are only notional resistor values at the moment and not worth worrying about. You can make any resistor you like by parallel and series standard resistors.

The formula is, (R16/2) + R25= 15K

With your latest schematic and plugged NE5532, is it necessary to use coupling caps for the power amp input?

Yes- but other complications arise because of the high NE5532 input current and offset voltage.

The Elliot amp has voltage gain=23 (was I wrong),

Not wrong. The Gain of the Elliot amp is 1+ R5/R4= 1+22K/1K = 23
**broken link removed**. But due to C3 the gain is 1 at DC.

and Elliot also said DC offset< 100mV, so maxium DC in the input without capacitor will be <4mV. <4mV DC offset, is it possible? Or any wrong in my think?

You are on the right track but detail not right. At a first approximation though, yes you can DC couple throughout, providing you use the OPA2134.

spec

 

[spec]

Just for reference, Rod Elliot audio power amplifier, Nikolai version:

(1) OUTPUT TRANSISTORS (Q7/Q8)
ON, NJW3281/NJW1302 https://www.onsemi.com/pub_link/Collateral/MJW0281A DATASHEET.PDF
(2) DRIVER TRANSISTORS (Q5/Q6)
ON, MJE15028/MJE15029 https://www.onsemi.com/pub_link/Collateral/MJE15028-D.PDF
(3) VAS TRANSISTOR (Q4)
ON, (MJE243G)/MJE253G https://www.mouser.com/ds/2/308/MJE243-D-113102.pdf
(4) LONG TAILED PAIR TRANSISTORS (Q1/Q2)
xxx/xxx
(5) LED (D1)
xxx
(6) TAIL CURRENT (Q3)
xxx
(7) VBE MULTIPLIER (Q9)
xxx

VOLTAGE GAIN: 23
SENSITIVITY: 1.414V peak (1V RMS)

**broken link removed**​

 

[Nikolai Petrenko]

Not wrong. The Gain of the Elliot amp is 1+ R5/R4= 1+22K/1K = 23
**broken link removed**. But due to C3 the gain is 1 at DC.

Well, I forgot the C3 , so input and output DC will be almost the same.

 

[spec]

Well, I forgot the C3 , so input and output DC will be almost the same.

The voltage difference between the amplifier output and zero volts (which is what counts) will be dependent on.
(1) The difference between the VBEs of the two long tail pair (LTP) transistors
(2) The difference between the LTP hFEs
(3) The difference between the LTP transistor collector currents, which are not necessarily equal in the Elliot amplifier.

spec

 

[Nikolai Petrenko]

Those are only notional resistor values at the moment and not worth worrying about. You can make any resistor you like by parallel and series standard resistors.

The formula is, (R16/2) + R25= 15K

What about (10k/2) + 10k =15k???