How important is the preamp?

Status
Not open for further replies.
From what I learn here so far, the main source of distortion is on the OPS in the signal chain. Even the IPS is not quite in the picture. How important is the preamp? By preamp, I mean the one that do the switching between AUX1, AUX2, VIDEO etc. And the volume control.

from my understanding, you really don't need to have gain, no frequency compensation. You literally can get away with taking the signal from different sources, go through a selector switch and drive the volume pot. Then just go straight to the power amp!!!

Attached is my design. It's just relays switching the 3 sources. Video is just straight pass through with no repeaters. The audio is buffered before driving to the volume pot, then to the output RCA jacks.

I put trim pots to enhance the highs and lows just a little. But it can by adjust to bypass everything and have flat frequency response. Do I really need to put in this, or should I just get rid of all the trim pots and make it just a straight buffer?

What else do I want to add or look out before I sent out the pcb?

Thanks

Alan
 

Attachments

  • Preamp2.jpg
    Preamp2.jpg
    249 KB · Views: 347
I use the bandwidth of the power amplifier as my guide to the required bandwidth of ancilliary equipment.

Let's suppose my PA is flat (F-1dB) from 4Hz to 100kHz, then I aim for a pre-amp or Buffer or Source to have an extra octave at each end of the audio spectrum, i.e. F-1dB of ~2Hz to ~200kHz and this results in an F-3dB bandwidth of 1Hz to 400kHz.

So I DC block at the input @ 1Hz (160ms for passive RC) and RF attenuate @ 400kHz (0.4us for passive RC).

You don't have RF attenuation.
Your DC block is set to 440ms.
Your DC bias is rather coarse.
Using a pair of diodes (1n4148/914) across the adjustment pot limits the adjustment to +-600mV. This should be fed from the two supplies and very well smoothed to remove both ripple and noise.

Are you using a "pull up resistor" to bias the opamp output stage into single ended ClassA?
 
Last edited:
Pull-down resistors at the input are much recommended, else switching sources may not be noise-free. Commercial designs also tend to have 330-470 ohms in series and 220-330 pF to ground at each input, presumably for better RF suppression.

The circuit wouldn't be an exercise in lowest noise, and the tone controls look rather crude. Anything wrong with a classic inverting Baxandall?

Are there any specific reasons for Class A bias current coming from the positive rail? I don't know what sort of preferences the LM4562 has - I suppose both pull-up and pull-down would work, but if in doubt I'd prefer pull-down, with npn output transistors usually being a bit more robust.
 
I propose the other way.
+0dB (Buffer, or straight through) used when the music is already loud enough with the source in current use.

Then add in +6dB, or +12dB, only when the Source cannot give sufficient volume with the vol pot at maximum.

The vol pot at maximum leaves no resistance in series between the Source and the Power Amplifier and thus introduces no added noise to the speaker output.
 
I use the bandwidth of the power amplifier as my guide to the required bandwidth of ancilliary equipment.

Let's suppose my PA is flat (F-1dB) from 4Hz to 100kHz, then I aim for a pre-amp or Buffer or Source to have an extra octave at each end of the audio spectrum, i.e. F-1dB of ~2Hz to ~200kHz and this results in an F-3dB bandwidth of 1Hz to 400kHz.

So I DC block at the input @ 1Hz (160ms for passive RC) and RF attenuate @ 400kHz (0.4us for passive RC).

You don't have RF attenuation.
What is the effect of not having RF attenuation?
Your DC block is set to 440ms.
I thought if I dip below 1Hz pass band, it's even better. I can easily change to a smaller non polar electrolytic (10uF) to go back to 1Hz.
Your DC bias is rather coarse.
What DC bias? I don't know what that is.
Using a pair of diodes (1n4148/914) across the adjustment pot limits the adjustment to +-600mV. This should be fed from the two supplies and very well smoothed to remove both ripple and noise.
Do you mean a diode across the two ends of the pot?I found out the value of the pots are wrong. I meant to use 500ohm pot that only adjust very little only. I never meant to make it like a tone pot, just a very slight (10%) adjustment on both high and low ends. It's like the cheap amps that have a button you push to give you slightly more highs and lows when you play at low volume. I don't even know whether it is necessary. Do you think I should remove all these and go for true fidelity?
Are you using a "pull up resistor" to bias the opamp output stage into single ended ClassA?

Yes, I use the pull up resistor to force the opamp to work in Class A to eliminate crossover distortion however low it may be. I only use 5K to pull 3 mA because the output voltage of the opamp should never exceed 1V peak. I plan to use a 5K(2K if needed) dual gang potentiometer for volume control. So 3mA should force the opamp to run at class A even if it swing to 6V peak.

Thanks for you comment, one thing I never understand is what is the importance of RF input filter.

Thanks

Alan
 
Pull-down resistors at the input are much recommended, else switching sources may not be noise-free. Commercial designs also tend to have 330-470 ohms in series and 220-330 pF to ground at each input, presumably for better RF suppression.
Do you mean pull down resistor at the relay input so nothing at the input relay signal path is left floating when not selected?

The circuit wouldn't be an exercise in lowest noise, and the tone controls look rather crude. Anything wrong with a classic inverting Baxandall?
I put the wrong value for the pots. I meant all the pots to be 500ohm pots. I never meant to have a tone control, just an experimental boost of the high and low ends to see whether it gives fuller sound at low volume. It is like a button in most of the cheaper amps that I even forgot the name, something called the "loudness" or something. My question is whether I should get rid of all these adjustment and make it flat pass through.

Also, why is the circuit not the lowest noise? there is nothing that is not low noise. The only thing I can see is not having a feedback capacitor across the feedback resistor(including the pot) to band limit the opamp.

Are there any specific reasons for Class A bias current coming from the positive rail? I don't know what sort of preferences the LM4562 has - I suppose both pull-up and pull-down would work, but if in doubt I'd prefer pull-down, with npn output transistors usually being a bit more robust.

Thanks for the reply. I pull up ONLY because it's easier to pull to +V on the layout, it's a two layers board, AND I want a ground plane on the top layer. So I really have single layer for trace and avoid putting any trace on the top layer that cut up the ground plane as much as possible.

Thanks

Alan
 
Last edited:
Thanks guys for the detail reply. My questions still:

1) Should I eliminate the pots and make it straight pass through?

2) If the preamp mainly to attenuate the signal, why do I even need the opamp? Why not straight pass through driving a 25K volume pot that most sources can drive. No active device in the signal chain always give the lowest distortion ( try no distortion)!!!

3) Any down side of pulling up at the output to force the opamp to run in Class A?

4) What is the sonic effect of not having RF input filter?
 
I modified the schematic.
1) I added 1M pull down resistor to every signal path that can be an open circuit when the path is not selected to avoid "pop" when switched due to charging up.

2) I corrected the value of the trim pots to the correct value of 500ohm instead of 5K. The part symbol was created for 5K and I just simply forgot to change the value after pulling in the part symbol.

3) I changed the connections of the trim pot a little so it is always add highs and lows, not subtracting.

4) I added 680ohm and 330pF RF filter to filter at 710KHz.

Let me know what else should I look for.

Thanks
 

Attachments

  • Preamp2.jpg
    Preamp2.jpg
    314.1 KB · Views: 201
1) good, could be even higher.
2) are these gain trim pots? The 0.1 is 100nF and seems very large and will have an audio frequency effect.
3) intended as frequency response alterations. I would not.
4) high enough that no one could accuse these of attenuating any of the treble content.
I would add some small capacitance at each input socket. try 10pF to 100pF at EVERY Hot to Cold. You can also add some from every Cold to Chassis.
These are to attenuate the interference getting re-radiated from the short length of cable after the socket/s.

Any chance you could look at the high pass filters/DC blocking and rescale the values so that small affordable film capacitors could be used instead of bi-polar electrolytics?
Personally I feel happier using 2u2F and 100k rather than 22uF+22uF (=11uF) and 20k, even though theoretically an electrolytic used as a coupling cap has no audio signal across it, if it is selected as "big enough".
 
Last edited:
Thanks guys for the detail reply. My questions still:

1) Should I eliminate the pots and make it straight pass through?

2) If the preamp mainly to attenuate the signal, why do I even need the opamp? Why not straight pass through driving a 25K volume pot that most sources can drive. No active device in the signal chain always give the lowest distortion ( try no distortion)!!!

3) Any down side of pulling up at the output to force the opamp to run in Class A?

4) What is the sonic effect of not having RF input filter?
1) I would eliminate the filtering of the caps across the trimmers. Add frequency manipulation elsewhere when and IF it is needed.
The gain effect could be useful, but should not be needed twice.
Either at the input for signals that are too high and risk extra distortion, or at the end to trim outputs to suit different sensitivities of receiving equipment.

2) read the Forum article
http://www.diyaudio.com/forums/diyaudio-com-articles/186018-what-gain-structure.html
and select your "normal system gain" such that you can turn the vol pot all the way to maximum without clipping the power amplifiers.
Then add in some optional gain for low output sources. Switchable +6dB and +12db gain stage could allow inputs to vary from 500mVac to 2Vac and all would be able to just reach clipping of the power amplifiers.

3) increased heat in the opamp may affect performance, particularly duals and quads. You need to find out which works better: +ve or -ve current sink, resistor or simple CCS or complex CCS, low current sink, or high current sink.
You need all these answers before you build the pre-amp.

4) increased interference can ruin audio performance, ranging from mobile phones & wifi repeatedly bleeping in your speakers, to thermostats switching inductive loads and giving annoying spikes in the audio, to increased "mush" due to fold back of intermodulation distortion, to TV & Radio channels coming through from strong transmitters.
 
Last edited:
2) are these gain trim pots? The 0.1 is 100nF and seems very large and will have an audio frequency effect.
3) intended as frequency response alterations. I would not.


Any chance you could look at the high pass filters/DC blocking and rescale the values so that small affordable film capacitors could be used instead of bi-polar electrolytics?
Personally I feel happier using 2u2F and 100k rather than 22uF+22uF (=11uF) and 20k, even though theoretically an electrolytic used as a coupling cap has no audio signal across it, if it is selected as "big enough".

the trimpots does not alter the mid band gain, just raise the gain of below 300Hz and about 5KHz by 10% max. If adjust to the minimum, all the frequency alteration is gone and become a flat pass through.

Actually the DC block is 22uF with 20K resistor. The break frequency is 0.3Hz, so there should not be any voltage across the cap between 10Hz to 100KHz.

Thanks
 
2) read the Forum article
http://www.diyaudio.com/forums/diyaudio-com-articles/186018-what-gain-structure.html
and select your "normal system gain" such that you can turn the vol pot all the way to maximum without clipping the power amplifiers.
Then add in some optional gain for low output sources. Switchable +6dB and +12db gain stage could allow inputs to vary from 500mVac to 2Vac and all would be able to just reach clipping of the power amplifiers.
I only have DVD recorder to play back, no phono or radio. So the output should be 2Vpp max. My power amp has gain of 21. If I use +/-40V, the gain match will be perfect if I don't even use opamp and use the output of the DVD recorder driving the volume control which drive the power amp.
3) increased heat in the opamp may affect performance, particularly duals and quads. You need to find out which works better: +ve or -ve current sink, resistor or simple CCS or complex CCS, low current sink, or high current sink.
You need all these answers before you build the pre-amp.
Pulling 3mA with 30V across the opamp is 0.09W per amp. The total power dissipation of the pull up will be 0.18W on a Dip8, I think that's safe.

My gut feeling is even if I have the opamp with gain of 2, the gain matching won't be too bad. Nobody said you have to ever crank the volume to 10!!!

The main reason I even put the opamp is to prevent loading of the output of the DVD recorder, and also I want to use a lower value volume pot like 2K or 5K. The low impedance will make it less likely to pick up noise when connect to the power amp.

I know better to put the volume control pot after the opamp so when I turn down the volume, I attenuate the noise from the preamp and DVD recorder. I don't have active circuit between the volume pot and the power amp.
 
Last edited:
Better performance if the vol pot is cranked to max and yet does not clip the transients.

Vol Pot at max gives the least source impedance presented to the Receiver circuit following the vol pot and this generally improves the performance of the Receiver.
Vol pot at max gives the highest signal to noise ratio for that Source.
Vol pot at max gives the strongest signal relative to interference from beyond the audio band.

Vol pot at max is a good way to set your listening level IF you can enjoy the output level and IF your system gain is elected such that no signal is ever clipped.

You want/need the lowest system gain that just allows the maximum signal to never be clipped and does not damage your ears and your speakers.

System gains selected such that the vol pot can never be taken above 20% of rotation are bad ! Sort the system gain.
 
Why have you proposed a pre-amp with multiple inputs and multiple adjustments when you only have ONE Source?

2Vpp and a gain of 21 in the power amplifier will give you a maximum output of 42Vpp at the speaker.
42Vpp is equivalent to 21Vpk and 14.85Vac at the speaker.
This is a maximum output of 27.5W (about the same as an F5).
2Vpp is the same as 707mVac. That is a low signal, but not the lowest if you ever acquire another source.
Many Sources will have outputs approaching 2.5Vac (+11dB rel to 707mVac)

You almost certainly will not need a +12dB gain stage, but an optional +6dB gain stage would allow Sources down to 350mVac to drive your power amp to 27W.
 
Why have you proposed a pre-amp with multiple inputs and multiple adjustments when you only have ONE Source?
Ha ha!!! I have 3 DVD recorder going and I can watch from any one of them. I don't watch live TV shows, I tape them all. I am still over a year behind in the program. I don't trust DVR or others that hold hundreds of hours because if it fail, I loss the whole year of programs. So, instead, I have 3 DVD recorders just in the living room recording and playing back.

Call me crazy, we have 4 Direct TV stations in different rooms, I still have to set up two stations with antennas to cover recordings for all the programs as they are airing at the same time and I have days that I run out of recording stations during prime time.



Vol pot at max is a good way to set your listening level IF you can enjoy the output level and IF your system gain is elected such that no signal is ever clipped.

You want/need the lowest system gain that just allows the maximum signal to never be clipped and does not damage your ears and your speakers.

System gains selected such that the vol pot can never be taken above 20% of rotation are bad ! Sort the system gain.
My listening level is so low that if I follow your suggestion, I would have to divide the signal even directly from the DVD recorder output. Then I might as well getting rid of the opamp. then take the signal from the recorder, drive through a resistor before even driving the volume control to divide it down.
 
Oh, and most of the time the preamp actually serves as an attenuator. Even if there is a volume control in the preceding signal chain, I would include something like switchable 0/-20/-40 dB at the very least. Otherwise chances are you'd be hearing plenty of source noise.

From Andrew's reasoning, attenuation is really what I need. I am setting the gain of the power amp very low at 21. I don't think it is safe to (stability issue) lower it. Plus it is not easy to keep the amp stable if you lower the close loop gain too much.

I am using +/-40V supply, I expect the amp to swing +/-35V. At gain of 21, the input should swing 1.75Vpeak(3.5V peak to peak) or 1.24Vrms maximum to get clean output all the way. This is a little higher than the normal output of DVD/CD player that is supposed to be 1Vrms max. So having a little gain like in my preamp schematic is not necessary a bad thing.

But I really never listen beyond say 5W. So an optional attenuator you suggested is really what I need.

Yes, the divider or volume control should be at as far back towards the power amp as possible to reduce the possibility of noise pickup. Seems like the best place to put the attenuator is AT THE POWER AMP!!!!

This mean I should put a selector switch at the BACK panel of the power amp right where the RCA jacks come in. Seems like a divided by 5 is plenty. I don't want to get too complicated, the more complex the circuit, the more potential of noise pick up. Using a 4K and 1K resistor to form a divide by 5 network, using a DPDT toggle switch to choose between full output( have 5K to ground as load) or junction between 4K and 1K resistor to drive into the power amp. In my case, I just flip the switch to divide by 5 so I have more range on the volume control of the preamp.
 
Status
Not open for further replies.