About make an audio power amplifier.

[Nikolai Petrenko]

e.g. good quality 48V need two. modify to make common ground bipolar out to make 200W+ PSU for power Amp
Add plenum duct and 35mm fans or put into rack. terminal posts for AC in. chassis ground, normally output floating.

4 avail.

**broken link removed**

shop around. make vs buy ?.

For what, while my amp is only 100W?

 

[Nikolai Petrenko]

The trick I learned better than any PC box is small aperture duct over hotspots with high air speed >1m/s with turbulent flow is better than laminar flow over hotspots, rather than flow way over top of the tall parts to remove heat and resulted in only 15'C rise at max load loss, pushing air horizontally within 1cm above the hot parts and out exbaust grill. Usually fan never operate until it reached 45'C at high ambient and full power.

Of course turbulent flow is better, much better, they can create forced cooling instead of natural cooling by laminar flow.

 

[Tony Stewart]

For 100W RMS into 8Ω ,you need at least +/- 44V rails for an equivalent Vdc (rms) of 28.5V. with +\- 4V drop minimum for output stage saturation avoidance to deliver 40Vp sine.
With Irms of 28.5V/8Ω = 3.6A (rms) and a Ip= 40V/8= 5 Ap
The unregulated PSU with 44Vdc/5A= 4.4V ripple or 10% means it will saturate ,,so you now need 46.4Vdc min.

so why not get a pair of 48V cheap PSU, with <0.1% ripple?

Otherwise unregulated PSU needs RC=5T to get 10% ripple and 10T to get 5% ripple.
Thus for R=8 Ω and with 50Hz, T= 10ms then for 5T or 10% ripple you need, C= 5T/8Ω=>6.25 millifarad caps rated at >=70V for no load rise of 140%. But 10r ripple results in lots of distortion, so this is poor quality. Also the ESR must be <= 50 mΩ.

When I made a 150W amp at age 18 , I used 100mF (100k μF) pair of high grade caps with ESR ~ 10 mΩ. These were cheap at the time, since they were used in mainframes.

I doubt your Amp will deliver 100W rms unless you offer more details. THe Nakamichi design does not use full negative feedback from output to input, which results in low phase shift and low THD, low IMD in treble range,mwhich also gives good 2D imaging of source on stage.

So It depends on your requirements for quality.

 

[spec]

Here is the preamp: The star ground look like a spider web
View attachment 100040

Spiders web looks good

In general, nice schematic Nikolai (what software package did you draw that with?)

First thoughts- not checked in detail:

(1) Unless I have misunderstood your schematic, you can remove the first two opamps. The second two will do the job.

(2) What is with the four potentiometers?

(3) Can you list all the inputs and input voltages?

spec

 

[Nikolai Petrenko]

For 100W RMS into 8Ω ,you need at least +/- 44V rails for an equivalent Vdc (rms) of 28.5V. with +\- 4V drop minimum for output stage saturation avoidance to deliver 40Vp sine.
With Irms of 28.5V/8Ω = 3.6A (rms) and a Ip= 40V/8= 5 Ap
The unregulated PSU with 44Vdc/5A= 4.4V ripple or 10% means it will saturate ,,so you now need 46.4Vdc min.

so why not get a pair of 48V cheap PSU, with <0.1% ripple?

Otherwise unregulated PSU needs RC=5T to get 10% ripple and 10T to get 5% ripple.
Thus for R=8 Ω and with 50Hz, T= 10ms then for 5T or 10% ripple you need, C= 5T/8Ω=>6.25 millifarad caps rated at >=70V for no load rise of 140%. But 10r ripple results in lots of distortion, so this is poor quality. Also the ESR must be <= 50 mΩ.

When I made a 150W amp at age 18 , I used 100mF (100k μF) pair of high grade caps with ESR ~ 10 mΩ. These were cheap at the time, since they were used in mainframes.

I doubt your Amp will deliver 100W rms unless you offer more details. THe Nakamichi design does not use full negative feedback from output to input, which results in low phase shift and low THD, low IMD in treble range,mwhich also gives good 2D imaging of source on stage.

So It depends on your requirements for quality.

No, the amp is only 60W RMS into 8ohm, only 100W RMS into 4ohm so it only need +-35V. I even want to reduce supply voltage to reduce power dissipation.

 

[spec]

No, the amp is only 60W RMS into 8ohm, only 100W RMS into 4ohm so it only need +-35V. I even want to reduce supply voltage to reduce power dissipation.

You do not need to reduce the power lines to reduce dissipation. The higher the supply lines the better, within reason, especially with your latest transistor line up.

spec

 

[Nikolai Petrenko]

Spiders web looks good

In general, nice schematic Nikolai (what software package did you draw that with?)

First thoughts- not checked in detail:

(1) Unless I have misunderstood your schematic, you can remove the first two opamps. The second two will do the job.

(2) What is with the four potentiometers?

(3) Can you list all the inputs and input voltages?

spec

I use Eagle 7.6 Pro , export image in monochrome mode because I think monochrome give more convenience to see.
(1) As you know, the preamp was steal from Elliot P88 fully , except all capacitors have been kicked out . I think use only one gain stage will be fine, but should have compensate for -6dB on the crossover.
(2) The first two pot will act like balance control, I like to control like this, not like use one pot between two channel with middle terminal connected to the ground . The scond two pot is part of dual pot which is master volume control.
(3) - CD or DVD player (max 3V RMS)
- TV audio out; computer, mobile phone audio out............ (max 2V RMS)
- Mixed: 2X CD audio out, 2X TV or computer audio out, 2x microphones
- Phono preamp will be separately, phono preamp out will feed CD player input.
Any problem?

 

[spec]

I use Eagle 7.6 Pro , export image in monochrome mode because I think monochrome give more convenience to see.

Neat I had not realized that you could do that.

(1) As you know, the preamp was steal from Elliot P88 fully , except all capacitors have been kicked out . I think use only one gain stage will be fine, but should have compensate for -6dB on the crossover.

Not sure what you mean. The opamp frequency response will be flat from DC to well beyond 20 KHz.

(2) The first two pot will act like balance control, I like to control like this, not like use one pot between two channel with middle terminal connected to the ground . The scond two pot is part of dual pot which is master volume control.

Good about not using a single pot. Are the first two pots (balance) separate pots with their own knobs?

100K pots look a bit high. 10K would be more normal to keep the noise down.

(3) - CD or DVD player (max 3V RMS)
- TV audio out; computer, mobile phone audio out............ (max 2V RMS)
- Mixed: 2X CD audio out, 2X TV or computer audio out, 2x microphones
- Phono preamp will be separately, phono preamp out will feed CD player input.

But on the schematic there is only two inputs to the preamp: left in and right in.

I can't read the values of the gain changing resitors

How are you mixing the signals. I can't see that on the circuit.

I would advise normalizing all your inputs to one voltage level (300mV RMS?) rather than switching the amp feedback (always dangerous). Some of the input signals are too high anyway.

The input to the power amp is 1V RMS, if I remember correctly, so all you will need is a x3 opamp circuit.

spec

 

[Nikolai Petrenko]

Not sure what you mean. The opamp frequency response will be flat from DC to well beyond 20 KHz.

No, I mean signal strength will loss 6dB after pass through 2nd-order Linkwitz-Riley filter. So in preamp stage, gain should compensate this loss.

Good about not using a single pot. Are the first two pots (balance) separate pots with their own knobs?

Yes.

100K pots look a bit high. 10K would be more normal to keep the noise down.

OK.

But on the schematic there is only two inputs to the preamp: left in and right in.

I didn't put the dual 3-throw selector in the circuit.

I can't read the values of the gain changing resitors

The values is hided by other resistors: 4k7, 5k6, 6k8, 10k

How are you mixing the signals. I can't see that on the circuit.

I would advise normalizing all your inputs to one voltage level (300mV RMS?) rather than switching the amp feedback (always dangerous). Some of the input signals are too high anyway.

The input to the power amp is 1V RMS, if I remember correctly, so all you will need is a x3 opamp circuit.

I haven't drawn the mixer yet.
The input voltage is maxium 2V RMS, but in normal listening, I think is only few hundred mV.
Why need even 3 opamp circuit?

 

[spec]

No, I mean signal strength will loss 6dB after pass through 2nd-order Linkwitz-Riley filter. So in preamp stage, gain should compensate this loss.

Got it- changing the gain of the preamp is no big deal: just one resistor change.

I didn't put the dual 3-throw selector in the circuit.

ok

How about having a push button switch or any two pole 1 way switch for each input. That way you could mix and match any possible combination of inputs and have a simple circuit. Be a novel approach too and could be made to look pretty cool with blue LED indicators for each input.

The values is hided by other resistors: 4k7, 5k6, 6k8, 10k

It would be a help to move the values so I can see them . The schematic should always be Xtal clear and as complete as possible. It saves people spending a long time trying to figure what is going on.

I haven't drawn the mixer yet.

right

The input voltage is maxium 2V RMS, but in normal listening, I think is only few hundred mV.

You always go by maximum voltage to get maximum power from amplifier. 1V RMS is pretty standard as far as I can remember. It is also a good value for a power amp input voltage for maximum output. The gain can be changed by changing one resistor in the power amplifier

Why need even 3 opamp circuit?

You really need a low and constant impedance to drive the power amplifier.

spec

 

[Nikolai Petrenko]

How about having a push button switch or any two pole 1 way switch for each input. That way you could mix and match any possible combination of inputs and have a simple circuit. Be a novel approach too and could be made to look pretty cool with blue LED indicators for each input.

Coooool! Sometime I put my head into the bush.

It would be a help to move the values so I can see them . The schematic should always be Xtal clear and as complete as possible. It saves people spending a long time trying to figure what is going on.

I will do.

The gain can be changed by changing one resistor in the power amplifier

Change resistor in preamp will be easier.

 

[Nikolai Petrenko]

The preamp, mixer will be put in a separately case to protect preamp, esp. phono from hum cause by transformer while all transformers, active crossover and power amp will be put in main case.

 

[spec]

The preamp, mixer will be put in a separately case to protect preamp, esp. phono from hum cause by transformer while all transformers, active crossover and power amp will be put in main case.

Nice

 

[spec]

Coooool! Sometime I put my head into the bush.

I will do.

Change resistor in preamp will be easier.

It is better to make the input to the power amp 1V RMS. You need to make sure you maximize your headroom (before clipping) but either way it does not matter much

 

[spec]

Here is an outline schematic illustrating the topology I strongly recommend for your preamp. It only has rough order of magnitude values and decoupling etc have been left out (note use of monochrome as advised by NP ).

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

ERRATA
(1) Chang R10 and R1 to 47 Ohms

 

[Nikolai Petrenko]

So simple, I like it.
Resistor in each input I think will be 4k7 or 10k. How many gain at second stage? (I don't know how to calculate gain with NF path like that)
Problem: WIMA don't have 1% cap big enough to use in crossover . The biggest is 0.033uF, so resistor is 47k (cause current hungry).

 

[spec]

So simple, I like it.

Good. It is a pretty standard topology

Resistor in each input I think will be 4k7 or 10k.

I would nor worry about resistor values- this is just an outline schematic.

How many gain at second stage? (I don't know how to calculate gain with NF path like that)

Yes you do.Voltage Gain of second stage is 1+(R1/Infinity)= 1

Problem: WIMA don't have 1% cap big enough to use in crossover . The biggest is 0.033uF, so resistor is 47k (cause current hungry).

1% is not necessary 5% will be OK and possibly looser. Again don't worry about details for the moment.

Can you provide or point to this:
(1) Power amplifier circuit diagram
(2) Power amplifier supply line voltage
(3) Block diagram of system, including position of Xovers
(4) Itemized List of all inputs including RMS voltage.
(5) Schematic of low pass filter
(6) Schematic of mid/high pass filter
(7) Anything else that you think might be needed to complete the design

spec

 

[Tony Stewart]

Sorry, I am a poor man, my money still for speakers, not only can concern on output transistors.

Yes, I am going to make a thermal controlled fan speed for better thermal protection, it is also available to control speed by hand. If the fan can't cold down heatsink in some condictions, the mission is for thermal fuse..........

The pic I have shown only tell the mounting method, of course I never fool enough to turn my amp into a toast.

I have known about this way since I have learn about thermal and airflow. This way is much more benefit than using more output transistors in parallel.

Here is the preamp: The star ground look like a spider web
View attachment 100040

Nice spider but inductance of grounds and any current crosstalk or capacitive coupling to output stage voltage cannot be neglected with 100k input impedance and keep in mind all inputs and outputs shunt your spider with large ground loops.
So shield or good physical isolation, orientation layout is essential. Compute or measure crosstalk to be at least 60 dB below minimum signal,,pref.,> -85 dB. and same with pre-amp for rejecting AM stray RF noise.

If you wish to improve cooling, modify SMPS like the $15 EBAY units to desired voltage.

Cheaper and better than brute force unregulated 50 yr old design .Use 5A L and Plastic C instead of huge e-caps.

This all depends what expectations you have for woofer quality power signal distortion.
Since you have no specs,, I have no further comments.

 

[Nikolai Petrenko]

Yes you do.Voltage Gain of second stage is 1+(R1/Infinity)= 1

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

1% is not necessary 5% will be OK and possibly looser. Again don't worry about details for the moment.

OK, found FKP3 0.1uF 5%.

Can you provide or point to this:
(1) Power amplifier circuit diagram
(2) Power amplifier supply line voltage
(3) Block diagram of system, including position of Xovers
(4) Itemized List of all inputs including RMS voltage.
(5) Schematic of low pass filter
(6) Schematic of mid/high pass filter
(7) Anything else that you think might be needed to complete the design

(8) Preamp as you already posted.
(9) Passive crossover for 3-way system (or 2-way?)
(10) Speaker protector as you has mentioned in the conversation.
(11) Clipping indicator (neccessary?)
(12) Phono preamp
Anything else? Some miscellaneous like thermal fan speed controller or soft start are need to talk about?

 

[Nikolai Petrenko]

Nice spider but inductance of grounds and any current crosstalk or capacitive coupling to output stage voltage cannot be neglected with 100k input impedance and keep in mind all inputs and outputs shunt your spider with large ground loops.
So shield or good physical isolation, orientation layout is essential. Compute or measure crosstalk to be at least 60 dB below minimum signal,,pref.,> -85 dB. and same with pre-amp for rejecting AM stray RF noise.

Thank, I will do.

If you wish to improve cooling, modify SMPS like the $15 EBAY units to desired voltage.

Cheaper and better than brute force unregulated 50 yr old design .Use 5A L and Plastic C instead of huge e-caps.

This all depends what expectations you have for woofer quality power signal distortion.
Since you have no specs,, I have no further comments.

I certain use unregulated +-35V, inductor can be wind easily. CLC filter alway have advantage: simple, reliable, good ripple rejection.