JG´s Nobrainer and Nobrainer Discrete

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Here is the current schematic. Explanation follows after the listening test.
 

Attachments

  • SBP Schematic Voltage feedback.jpg
    SBP Schematic Voltage feedback.jpg
    48.5 KB · Views: 558
I made now the first listening session. Speakers are Suesskind Tangram ( 90dB @ 4 Ohm ) that i line filter with -3dB @ 80Hz. I use my active subwoofers so this little amp has not to work too hard in the bass.
The result is very nice. Very neutral and low distortion, good stage. The sound is not harsh
at all, there is even some inviting warmth. I am very happy that the amp does not sound sterile and anemic, quite the opposite.
Volume is really decent. Quite surprising. When i first listened i could see the peaks in the power supply go to 1A. Then my wife came in and complained about the too high volume ! I then sat my wife in the sweetspot and adjusted the volume to her taste. That was maybe half ( -10dB ) the volume i had listened before so for many listeners the power of this amp is enough, even on normal sensitivity speakers.
By the way i set the PSU to plus-minus 14V and the idle to 250mA, which is quite high.
 
How does the amp work ? Quite simple, there is nothing new here. Only common sense, well proven topology and good choices of components plus some fine tuning.
I did not want to build the next Blameless and Chip amps ( Gainclones ) can be good but there is already a lot of work done about this. I decided on a Power Opamp topology, first because PSU voltage is limited to what the solar battery can supply and second because a modern power amp is mostly a high power Opamp anyway.
The Opamp is the LME49710. It has very low noise and distortion plus good driving abilities. It can be run on plus-minus 17V, more then enough. I already mentioned that there is a version of this opamp that can be run on plus-minus 22V for people that want to try this amp with more power.
The Opamp drives a CFP ( Szikley ) stage. This has the advantage of high input impedance so the Opamp does not have to drive a low input impedance. Another advantage is only one Ube drop and that is in the drivers. If the drivers are NOT mounted on the heatsink we only have to deal with the temperature coefficient of the drivers. Because the CFP ( Current Feedback Pair ) has internal feedback this stage has also less distortion as a comparable darlington ( about 1/2 ).
There are 4 extra resistors per side in my design. The 300 Ohm resistors are base stoppers. A CFP is fast so it can oscillate. The base stoppers avoid that. The 100 Ohm resistors set the idle current in the drivers and speed up the sitchning time in the output. The 115 Ohm resistors supply some of the driver current into the output and can be used for fine tuning the idle. I may go back to 100 Ohm. The 0.2Ohm resistors do a measure of current feedback. An ideally biased Class AB amp should produce 24.5mV over this resistors. In my case it is around 60mV so the amp is a bit over biased. I will experiment with the bias and look what sounds and measures best.
Bias is done by Leds that are supplied by a J-Fet CCS. I found a combination of an Infrared Led and a red 2mA Led gave the disired result. TemCo is still a bit positive but in praxis the amp drifts low enough so that the output never saturates.
This Leds can also be substituted with conventional diodes or transdiodes.
Any desirable idle can be set that way. For example the case we have here:
The Leds drop around 3V that is 1.5V before each driver. The driver drops around 0,7V and the 115 Ohm resistor around 0.74V so we have 1.5V - 0.7 - 0.74 = 60mV over the 0.2 Ohm. Is is an estimate and can be fine tune by the current over the Leds and the 115 Ohm resistors. Expect 20% error.
Feedback is taken from the output to the Opamp, in this case 30kOhm/1.8Kohm giving a gain of 30/1.8 + 1 = 17,66 x.
The plus - minus 12V version saturates at 0.9V input so a pot meter can be used with sources that pull out at least 1V.
The 50pF cap is the Miller cap for stability. On the schematic you can see a more sophiticated output inclusive compensation that i have not died so far. The amp can also be set up as current out and open loop. When i have got it working i tell you more.
 
Sofar i used my lab supply as PSU for the SBP. I need that of cause for other projects too so i build a power supply. I only have 2 x 15V transformers of decent size so an unregulated supply would do too much voltage. that would be around 2 x 21V, too much for the Opamp. I had made good experiences with cap multipliers and use them in my version of the TSSA. I use a bipolar darlington design for low drop. Here i have some voltage to burn so i decided on a Mosfet multiplier with Zeners. I drop some 7V in this multiplier so i end up with slightly under plus-minus 14V. Quite optimal. Under this conditions the SBP can do 2 x 15W into 4 Ohm.
Schematic and pictures tomorrow
 
I listened more to the SBP, this time with CD. I use a Mc Cormack transport and EUVL´s
ESS board with my buffer-filter at the output. I designed a PSU for the filter-buffer not to waist too much 9V batteries. I also think it sounds better.
What i now find typical for the SBP is a very high amount of transpancy. A lot of details are audible. Tonal balance is spot on with CD and a little warm with vinyl, how it should be.
I still listen with the 90dB 4 Ohm Tangram, so volume is a bit restricted.
Tomorrow Martina comes to listen to the Tangram and later i put on my 95dB efficient dipoles.
Temperature coefficient with the Led bias is still a bit positive so the cooling fins get rather hot when i listen slightly under clipping. I plan to substitute the Led string with a conventional bias spreader for some measure of temperature control.
Holger asked me if i can make a bridge circuit for more power output. I think i can with the help of DR134 converters. Then 30W into 4 Ohm is possible. That should be enough for most purposes.
 
Somewhere in the thread I did read that it is possible to use the power pins to drive the output stage (as you wisely did not do).

I would like to add a little note on this :)

It is not advisable to use the power supply pins of the opamp to drive an output stage (in my opinion), all nasty’s are pushed up to these pins (and should be destroyed by the de-coupling capacitors and/or the power supply), and there are better ways to drive an output stage.
 
Frans, i could also use the Opamp inverted. That will lower common mode distortion.
There is a limit then of how high the input impedance can be before it gets noisy.
Also a wanted to include a rather high impedance volume control at the input to avoid a preamp and that is more difficult with an inverted design.
I also came up with a much simpler way of bridging.
 
Frans, i could also use the Opamp inverted. That will lower common mode distortion.
There is a limit then of how high the input impedance can be before it gets noisy.
Also a wanted to include a rather high impedance volume control at the input to avoid a preamp and that is more difficult with an inverted design.
I also came up with a much simpler way of bridging.

But you can always spend one more opamp to drive it from a low impedance.
 
Here are the options. Somehow i like the extreme simplicity of the inverted opamp. Typical D.Self.
Of cause i could build one amp inverted and the other non inverted. AS Audio idea.

Spend the extra opamp, invert the signal, and feed it into the second (identical to the first) amp. Using one non- and one inverting amp (for me) is the wrong idea. For me it sounds like using two different amps where two exactly the same amps are needed.
 
When i understand you correct i whould then modify the amps for inverted. See Old - New.
Then i whould add a buffer for both channels and an inverter to one channel to make a bridge.

Yes, that is what I would do :)

The other thing is, I would try to run the inverter with no compensation at all.

If bandwidth compensation (or limiting) is needed I would do this on the power amps (both the same). Keeping the design as balanced as possible.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.