How to make an Headphone amp on breadboard...?

[sgrossklass]

Originally Posted by ddietz

Are you talking about a discrete Diamond buffer?

Yes, he is. (@Antoinel, this is what the circuit in BUF634 etc. is called.) TBH I'd prefer a conventional A/AB buffer stage with medium power transistors though (idle current of anything between 30 and 100 mA depending on loads and cooling). It still has sufficient current gain for any modern opamp and better stability and distortion properties AFAICT.

(For an amplifier to be built on breadboard, the next step after a Cmoy would be an Apheared-47. That's a rather nice little amp already.)

Originally Posted by ddietz

Would there be a benefit to biasing the OPAx (in my case 2132) to Class A by using a resistor from the negative rail to the output of the OPAx?

One would have to try and measure. The effect of class A biasing is not the same for all types of opamps. It seems to work well for types that make up for low output stage quiescent current by being fast (e.g. LM6171), but may actually be detrimental when used on more oldschool designs (RocketScientist once mentioned that NJM4556 performance suffers with output stage imbalance).

Some measured data for an OPA 690 can be found in this article. Results look quite good for that one, though - not unexpectedly - maximum voltage swing suffers a bit.

[Antoinel]

Thank you sgrossklass for the information. I have always liked this buffer. In my opinion it is a diamond and not a Zirconia performer. Please shed more light on the Apheared-47 amp. Thank you. Best regards

[Antoinel]

The attached schematic shows one channel of an OP Amp driving a buffer pair of complementary output transistors (TIP 120, and TIP 125 Darlingtons). This output stage is operating in the common emitter rather than the customary common collector configuration. Thus it is an inverting amplifier with voltage gain if desired. This transconductance or current source amp requires 2 separate power supplies for proper operation. The schematic shows one +/- 6V supply to the OP Amp and another higher voltage +/-15 V for the output stage. The output of this amplifier is the common port or center point of the +/- 15V supply. Thus the headphones are connected between this port and ground. By ground I mean the center point of the +/-6V supply. The prototype was assembled on an experimenter's proto board. The attached images are not pretty; but they confirmed its feasibility. One image is a top view and the second is taken from the side of the volume control. The components are discernible. The above power supplies I used happen to be (ginormous) surplus from spent computers.

This amp is capable of a high outupt current into headphones. Note the resistor of 220 K which attenuates the input signal so as protect one's hearing. The quiescent conditions of the amp are:

• 10 mV dc offset at output port
• 5 mVpp noise/hum at output port
• Output transistors idle at ~23 mA each
• Op Amp is RC 4560 which sounded slightly better than OPA2134
• Must have output Zobel, a 10 pF ceramic cap (between inverting input and output of Op Amp) and the 100 K feedback resistors to prevent oscillation
• Headphone was Grado SR 80 connected in parallel (mono). It must be connected before turn on so as to complete the circuit of the output stage.

Two independent power supplies of +/-15 V and one common lower voltage +/-6-9 V for the OP Amps are needed to assemble a headphone stereo transconductance amp. Clearly, these power supplies can be compact in size.

I believe DIYers (and I) will run with this "square one" amplifier and tweak it so as assemble in the longer range the best there will be.

[jackinnj]

Originally Posted by Antoinel

. By the way the circuit of BUF634 was originally invented/used by National Semiconductor.

Perhaps we can convince TI to give us a SPICE model for the LME49600 now that they own Nat Semi.

I am using a BUF634 as the output stage of my Borbely All FET buffer for headphones -- it sounds great! I chose OPA2604 for the servo. I chose the BUF634 over the LME49600 as the former is available in TO-220.

[Antoinel]

I am glad to find you excited about BUF634 as I still am. My interest is in its application as a Tansconductance amp. I have been prototyping transconductance or Current Source Amps (CSA) by using discrete buffers including this versatile Diamond. Please see the articles on CSA by Mr. Nelson Pass at www.firstwatt.com to grip their general attributes and limitations; if any. The discrete version of BUF634 is easy to breadboard. The schematic I used is attached as Headphone 002. This buffer can be used as a stand alone like BUF634. It is DC coupled and has a relatively low input impedance. Or the Diamond buffer can be appended to an Op Amp like in Figure 5 of the Burr-Brown application. The said application gives a voltage source amp (VSA) of a low output impedance; shown in Figure 5 driving an 8-16 Ohm speaker. Everywhere I look, I see VSAs driving headphones. Why not a CSA instead and/or expand the application scope of the Diamond buffer? By contrast to Figure 5, the same discrete Diamond buffer circuit can readily operate as a CSA as in the schematic of Headphone 002. It has a high output impedance and could have voltage gain; which are desirable. This high output impedance emanates from using "the opposed collectors" of the output stage. Note the output port is the center point of the +/-25 V supply powering the output stage. This supply is wholly dedicated to the output stage. The other +/-15 V supply is dedicated only to the Op Amp and the Diamond circuit which is biasing the output stage.

So, take the discrete BUF634 prototype and append it to an Op Amp so as to get a VSA like in Figure 5 of the Burr-Brown App. Or use the same circuit as a CSA instead in the circuit I showed in my previous reply (Headphone 001). The schematic of Headphone 001 (CSA) is attached here for convenience.

A VSA and a CSA of the same same circuit sound different as they drive a loudspeaker and a headphone. This discrete circuit of BUF634 (with Op Amp) is capable of high output current into headphones which do not need a lot to begin with. For example, a 100 mA peak to peak of a sine wave across a normal 30 Ohm impedance of a headphone is an output "peak to peak" power equal to 0.1 X 0.1 x 30 X 1000= 300 mW. This is a lot of power into any headphone. This requires particular attention to protecting one's hearing by limiting input volatge!

My apologies!. I have an error in the schematic pdf Headphone 002. Please swap the names of the transistors. 2N3440 is NPN and 2N5415 is PNP instead of 2N5410.

[ddietz]

This is all really cool and I love seeing it here. I hope this discussion will continue!!! As a novice, I am only just starting to delve into the theory and "mechanics" behind how circuits work and seeing descriptions like this are really helpful.

In fact, I have been meaning to post a thread asking someone to walk through the description of some circuits in very great detail -- imagine you where teaching a class about a single circuit. I know its a lot to ask but if you, Antoinel, wanted to walk through one of your circuits in that fashion, I would love it!!! By that, I mean imagine you went to a lecture at a conference where the speaker talked only about this circuit.


Anyway, hope you all keep this going, I am definitely following and learning.

[Antoinel]

Gladly, I will explain (within the next 24 hours) the circuit of the Diamond buffer like the one in BUF634. In the interim, the site www.firstwatt.com has useful articles which discuss simple circuits. Actually, I advocate the design philosophy of Mr. Nelson Pass which is to keep the circuit simple.

[ddietz]

I just found that site yesterday, by coincidence, and have been reading there as I have time.

It may sound like hand-holding to some, but I'd love to read someone's circuit description that explained what every component does. I learn that way, and I teach the things I know that way (woodworking handtool skills previously and biology today). Anyway, looking forward to it

[Antoinel]

Originally Posted by Antoinel

Gladly, I will explain (within the next 24 hours) the circuit of the Diamond buffer like the one in BUF634. In the interim, the site www.firstwatt.com has useful articles which discuss simple circuits. Actually, I advocate the design philosophy of Mr. Nelson Pass which is to keep the circuit simple.

Ddietz, here is an explanation of the Diamond Buffer. The schematic is attached. The input consists of complemetary transistors (2N3440 NPN, and 2N5415 PNP) having their base ports tied together. In an ideal world, these 2 BJTs have identical characteristics. So the current leaving the base of PNP and that entering NPN cancel out completely. So the input can be DC coupled, and there is no offset voltage appearing across Rin (no current passing through it). Technically, the joined base ports are at 0 Volts ground. Let's focus on the operation of NPN 2N3440. The discussion applies fully to its complement 2N5415. The NPN transistor is operating in the common collector configuration. Its collector port is connected to +15 V which is at AC ground. It is at AC ground because the + 15 power supply has ~zero internal impedance. The emitter port is fed from a constant current source wich is the N-channel JFET. A constant current source is inherently high impedance. It resists a change in any additional current that is forced through it exactly like a very high value resistor. In operation. the +15 V supply pushes current through the collector port of the transistor. It flows through the physical body of the transistor and exits at the emitter port. Emitter current is roughly equal to collector current because one can neglect the miniscule contribution of base current to it. This emitter current passes through the silicon diode 1N914 and is the same value as the constant current sunk by the JFET. The flow of current through the NPN transistor causes a voltage drop between its base and emitter Vbe = 0.6 V, and another 0.6 V from the silicon diode for a total of 1.20 V. This is the bias supply for the PNP output transistor TIPab where ab is a 2 digit number (Texas Instrument BJT). The PNP transistor utilizes 0.6 V from the bias circuit to turn on. The remaining 0.6 V appears across the the 30 Ohm resistor. Ohm's Law says that 0.6V = I X 30 Ohm and thus gives a current I of 20 mA passing through the 30 Ohm resistor and through the transistor. In an ideal world, the complementary NPN TIPcd transistor will also be passing 20 mA from the analogous operation of 2N5415. Thus the output is the junction of the two resistors (30 Ohm). It centers at zero volts; meaning no offset. The output also has a low impedance of 15-30 Ohms.

Dynamically speaking; when the input signal voltage goes positive (above zero ground), PNP transistor 2N5415 begins to turn off. This means less standby or quiescent emitter current begins to flow through it. But the constant current source resists this, and pushes this excess current (disposed off by 2N5415) through the base of the TIPcd thus turning it on for a positive signal the output. The exact opposite happens when the input signal goes negative.

Regards

[ddietz]

Wow, awesome!!! I'll need to read this angew times to let it all sink in.