Compound Power Amplifiers

[Antoinel]

The F6 Amplifier thread is an intellectual stimulation and true inspiration. I understood the operation of its output stage, and thus generated a model Compound Power Amplifier. This is not a simulation; but a working power amplifier which can be assembled in a short time. Take a look at post #38 in the F6 Amplifier thread. In it I reported the schematic OTLAmp1.pdf of an amplifier [attached below]. Basically, I replaced the NPN output BJT [Q3] with one channel of a THRESHOLD S/150 stereo power amplifier. I characterize [S/150] as a phase non-inverting voltage source amplifier or [VSA]. Conceptually, [S/150] is a building block which models the upper JFET in the simplified F6 schematic. I also replaced the NPN output BJT [Q4, in OTLAmp1 schematic] with a phase inverting transconductance power amplifier [TCA]. This TCA happens to be a diy. Other famous TCA building blocks in the hands of DIYers will work equally well and better. Conceptually, this phase inverting TCA models the bottom JFET in the simplified F6 schematic.

The schematic of this Compound Power Amplifier is shown below as OTLAmp6A.pdf . I'll pause my discussion at this point for DIYers to mull over the strength and weakness of this Compound Power Amplifier. I assembled it a couple of days ago, collected the indicated voltage measurements, and listened to it [one channel]. Here is the pleasant surprise: It sounds great. Hint: Imagine using your choice of amplifier building blocks [VSAs and TCAs] like those found in www.firstwatt.com

 

[Zen Mod]

from quick look at schm - you are listening difference between these two blocks ?

 

[Antoinel]

from quick look at schm - you are listening difference between these two blocks ?

A closer analysis of the schematic [ATLAmp6A] shows that the measured output voltage to be the sum contribution of both amplifiers; just like in the output stage of simplified F6 which are are in-phase.

Trust me ZM, and I do not use this term loosely, I am listening [and you, other DIYers can easily too] to the sum and not the difference.

 

[kasey197]

Antoinel, sorry I don't get it.... The upper fet and lower fet in the f6 simplified schematic are both operating as transconductance amps and operate in the same mode if there is no nfb right .... In the classic Lin circuit they have different transfer functions but here I don't find that to be the case....

Notwithstanding that, this is an interesting experiment

 

[Antoinel]

Antoinel, sorry I don't get it.... The upper fet and lower fet in the f6 simplified schematic are both operating as transconductance amps and operate in the same mode if there is no nfb right .... In the classic Lin circuit they have different transfer functions but here I don't find that to be the case....

Notwithstanding that, this is an interesting experiment

I like these words by Mr. Pass from his recent post in the F6 Amplifier thread :

I just hook these things up and see what I get.

• The VSA [S/150] has a published voltage gain of [Vo/Vi] of ~21-22. From the schematic [Vo] = 0.61 V and Vi = 0.027V. Measured Av = 23; close in value within experimental error.
• The VSA [S/150] has transconductance too. First; its output current = 0.03V/0.5 Ohms = 0.06 A. Its transconductance or gm = 0.06A/0.027 V = 2 Ao/Vi
• Its output signal is in phase with its input signal or the signal phase at its secondary winding [solid triangles].
• The TCA inverts the phase of the signal at the secondary winding. But; the phase of the signal at its secondary winding is already out of phase with that at the upper winding driving S/150. So the phase of output signal from TCA is overall in phase with that emanating from S/150. A consequence of 2 successive phase inversions.
• The transconductance of TCA is calculated like that for S/150. First, output current = 0.033/0.5 = 0.07 A. The value of its tranconductance is 0.07 A /0.07 V = 1Ao/Vi
• The voltage output of TCA = Output current of 0.07 A times 4.5 Ohms [load resistor] = 0.3 V. Its voltage gain is Vo/Vi = 0.61Vo / 0.066Vi = 9; or roughly 1/2 the output voltage of S/150.
• Since the load resistance [4.5 Ohms] is lower in value than that of the TCA's output impedance, the current from S/150 [0.06A] sums with the contribution of TCA = 0.07 A to equal 0.13 A. This is sensed across the load as the output voltage = 0.13 A times 4.5 Ohms = 0.6 V as measured.

Clearly, I chose equal output currents to emanate from S/150 and TCA. I can choose other ratios [more or less from S/150] by tweaking the 25 Ohm potentiometer at the input of S/150. Or manipulate load resitstance [future post].

 

[Antoinel]

ACA [Amp Camp Amplifier] is a candidate to plug in the circuit of OTLAmp6A.pdf in post #1. ACA with loop feedback is a phase-inverting VSA. By contrast, ACA without loop feedback is a phase inverting TCA. The phase of the input signal to VSA [upper winding] now needs to be aligned with the phase of the input signal to TCA [lower winding]. The overall phase of the output signal is out of phase relative to that at the primary winding of the transformer. This ACA Compound Power Amplifier inverts the phase of its input signal.

 

[Antoinel]

The schematic which goes with post #6 is attached and is named OTLAmp6B.pdf. The only circuit parameters which are needed to write [and describe] it follow. They are published and/or known for ACA. Use them to understand the circuit.

• The closed loop voltage gain [Av], and transconductance [gm] of ACA operating as VSA.
• The transconductance [gm'] of ACA operating as TCA.
• The value of the load impedance. Use 4, 8, 16, and 32 Ohms. Why this broad range? Possible variation in the impedance of a practical loudspeaker load. Plug in the circuits parameters for each value of load impedance. Notice the relative contribution of the output current from VSA and TCA to the load. At each of the above load values, Vo will [and must] remain constant; because it cannot violate the constant voltage gain equation of VSA.

 

[Zen Mod]

check phase symbols on each secondary (not triangles , but sines)

 

[Antoinel]

So, the voltage gain [Av] of VSA is a constant [fact]. The transconductance [gm'] of TCA is simultaneously a constant too [fact]. By contrast, the impedance of the loudspeaker varies as a function of the excitation frequency in a predictable [measured] or not manner. What gives? Does this compound power amplifier named OTLAmp6B have an operational problem? Absolutely not. The answer lies in the transconductance [gm] of VSA. It will decrease as the impedance of the loudspeaker increases. This also means that the contribution of the current from VSA decreases as the loudspeaker impedance increases, and clearly vice versa. The output impedance of VSA is inherently low by design. Thus, VSA operates as a continuous donor/absorber of output current; a dynamic ballast of sorts.

So at a certain high impedance of the loudspeaker, the contribution of the output current by VSA will reach zero. At a still higher value of a loudspeaker impedance, the VSA will shunt the excess current [from TCA] or divert it from the load. Otherwise the resultant voltage drop across the load will violate the constant gain equation of VSA. Throughout this exercise, TCA did not give a damn. Its output current is only dependent on the value of its input signal signal.

 

[Antoinel]

check phase symbols on each secondary (not triangles , but sines)

Thank you ZM for catching my mistake. The corrected schematic is attached as OTLAmp6B1. The signals at the primary and the secondaries of the transformer are in-phase [triangles and sines]. The output signal of the compound amp is out of phase with the primary and the secondary signals [sine]



Noteworthy is that the general circuit format of OTLAmp6B1 has 4 possible variations. Thus, it is versatile. But first, the circuit constants are: a transformer with 2 independent secondaries, a VSA and a TCA. The 4 variations follow. In each variation, the circuit is made operable by manipulating the relative phase of the input signals [to VSA and TCA] at the secondary windings.

1. A non inverting VSA and a non inverting TCA.
2. A non inverting VSA and an inverting TCA [e.g. OTLAmp6A]
3. An inverting VSA and a noninverting TCA
4. An inverting VSA and an inverting TCA [OTLAmp6B1]

Interestingly, the four variations operate in a similar fashion.

 

[Antoinel]

Why compound amplifiers? Please track this discussion using the attached OTLAmp1.pdf, and OTLAmp6A.pdf.

• First reason. A single transistor be it a BJT or a FET is not an ideal gain device. By contrast, a stand-alone power amplifier be it voltage source [VSA] and current source [TCA] are [can be] almost perfect gain devices. Question: Which gain device would you use in an actual power amp build of your design?. A perfect or an imperfect device?
• The Second Reason is the simplicity and effectiveness of compounding. You may say: a single transistor is a simple device compared with a stand-alone power amp be it VSA or TCA. True when you look at these two gain devices [e.g. a transistor and a VSA] side by side. But the simplicity of a single transistor disappears when one builds a power amplifier around it. Take a look at NPN transistor Q3 of the output stage in the schematic of the OTLAmp1.pdf. It is appended to bias circuitry, other active and passive devices on a PCB. One then slaps Q3 on a heat sink, and then attaches a power supply to it and then encloses the whole circuitry in an enclosure. The makers [including you] of stand-alone power amplifiers [VSA, and TCA] have already taken care of the complexity like that associated with integrating Q3 in its own power amp. The stand-alone VSA and TCA are off the shelf, in your trophy collection and thus available for immediate experimentation and use.
• The Third Reason is versatility It means many ways to integrate and operate a Compound Power Amplifier comprised of a stand-alone VSA and TCA ideal gain devices. By contrast, the compound amplifier comprised of the discrete Q3 and Q4 NPN transistors in the schematic of OTLAmp1.pdf are "locked" in one operation only. In the Compound Power Amplifier of the attached OTLAmp6A.pdf, the VSA can be a Class D, Class A Class AB stand-alones, and at any voltage gain [Vo/Vi], output impedance and %THD. Similarly, the stand-alone TCA can be Class A, Class AB, and with several choices of its transconductance Gm = Io/Vi, output impedance and %THD. As far as I can speculate, one may be able to integrate a stand-alone VSA [perfect device] and a single or multiple transistors as the companion TCA. Open boundaries for experimention, DIY and use.

I'll continue to nurture this thread with working examples, data and listening assessment.

 

[Mark Allen]

I just hook these things up and see what I get.

I tried that too... almost burned the shop to the ground!

 

[Antoinel]

I tried that too... almost burned the shop to the ground!

I use in-line fuses to protect the gear and loudspeakers. A bit of know-how helps too. Please recount the event above with minimum pain; I hope. There's a learning in it.

 

[Antoinel]

I am renaming the attached files beginning this post to CompoundAmp [A, B, ..] instead of OTLAmp. I am still on the main subject of mating a non-inverting Voltage Source Amp or VSA [in my case a Threshold S/150] with a phase inverting DIY current source amp; which is also known as a Transconductance Amp or TCA for short.


Fig. 1 in the attached file CompoundAmp A.pdf shows a highly simplified schematic of the TCA that I am using. Its output stage is comprised of a complementary pair of BJTs which are simultaneously operating in a common emitter configuration. The emitters are grounded. The junction of the opposed collectors of the BJTs is the output node Vo. The properties of interest are:

• The output signal is inverted relative to that at the input.
• Output Impedance Zo = Zl.
• Has voltage gain Vo/Vi [to be measured]
• Has a transconductance Gm = Io/Vi [to be measured]

My scope is referenced to ground, and does not measure differential voltage. Fig.2 in the attached file is the tool that I assembled and use to interface my scope to the floating differential voltage [shown across a 0.5 Ohm resistor]. It is a step-up [10X] torroid power transformer. Clearly, it provides voltage gain and ground isolation. Its frequency response is flat between 20Hz and 1KHz. But; its output voltage droops a bit at ~10KHz. Nonetheless, it is an acceptable differential voltage amplifier.



I'll continue this post in a follow-up and discuss these two topics:

• Determine the Intrinsic Output Impedance of the subject TCA; meaning in the absence of the load Zl.
• Demonstrate the versatility and value [performance wise] of the Compound Power Amplifier [CPA] under study.

 

[Antoinel]

Compound Power Amplifiers. Determine Intrinsic Zo of TCA

I determined the Intrinsic Output Impedance of my TCA as follows. But first, I upgraded the sensitivity of my voltage differential amplifier in Fig. 2 of the above post. I substituted a small iron core power transformer [Radio Shack 0.3 A/12.6 V] for the toroid power transformer [2A at 12.6 V]. The resistor across the [high voltage] coil is now 22K//100K. It gave the highest voltage step up ratio [amplification] and a flat frequency response between 100 Hz and 20KHz. Fully adequate for this experiment.

The attached CompoundAmp A1.pdf shows the schematic of the circuit I used to determine Zo[Intrinsic] of TCA as a function of frequency; with and without a Zobel at its output which is normally used for its stability against oscillation. The Intrinsic Zo of TCA is primarily the load resistance of S/150 [VSA]. The Table below the schematic shows the data I collected. The results are calculated by dividing Vo' by Delta V as in schematic. Delta V across 1 Ohm is also Delta Io which is the output current of the VSA driven into the Zobel and TCA. All of the voltage measurements were done with the iron core step up transformer. They are amplified or inflated by a factor ~10. So; a Delta V of 180 mV at 20 KHz [with Zobel] is actually ~ 18 milliVolts or ~18 milliAmps Io current output of VSA.

With Zobel: It ranged between 2,000 Ohms at 100 Hz and 60 Ohms at 20 KHz.
Without Zobel. It ranged between 2000 Ohms at 100 Hz and 550 Ohms at 20 KHz. Much much higher than the impedance of a typical loudspeaker! The source of this TCA impedance is ~ a capacitor of a rough value ~ 0.01uF.

So what is the value [and principal learning] of this experiment? In actual operation, the VSA will pump its output current exclusively into the loudspeaker and not the TCA.

To be continued...

 

[Antoinel]

Listening to CompoundAmp A2

This is it. Please take a look at the attached schematic CompoundAmp A2.pdf. Here's a brief summary:

• VSA is a voltage source amp. I happen to have and use a Threshold S/150. It has a voltage gain [Av = Vo/Vi] equal to ~21.
• TCA is a current source or transconductance amp. Mine is a DIY as I briefly described earlier. It has a Gm = 1 Amp out /Vi.
• Zl is my loudspeaker. It is an older A/D/S L-730 which has been with me since 1982. I am very familiar with its voice.
• The shown transformer is a toroid power transformer named earlier.
• The signal drive to the primary winding of the transformer is from a signal generator and the headphone output of a SONY CD player. Both sources are volume- controlled.
• The voltage gain, impedance and distortion at the output node Vo are exclusively those of VSA and not the TCA.

Let's go.

• Turn off Switch A. Listen to TCA exclusively. It expresses a characteristic voice through the loudspeaker. One can also generate the impedance plot of a loudspeaker as a function of frequency. After this is done, replace the loudspeaker with reference resistors; 4, 8, 16 Ohms etc. They will give flat and parallel lines versus frequency for one to "quantify" or gauge speaker impedance at any frequency. Note the maximum and minimum impedance of the loudspeaker. L-730 has a 4 Ohm minimum [at 100-150 Hz] and a 20 Ohm maximum in the 1-2 KHz range.
• Turn off Switch B and turn on Switch A. Listen to VSA exclusively. It also expresses a characteristic voice which is totally different from that of the stand-alone TCA.
• Finally, turn on both switches A and B. Listen to the Compound Power Amplifier. It expresses a voice which is a melange of VSA and TCA. It is different from the stand-alone VSA and TCA

I'll continue this discussion in a new post later on today.

 

[Antoinel]

Listening to Compound Power Amplifier

I failed to mention in the above post that Vo for the stand-alone TCA exactly mirrors the impedance of the loudspeaker as a function of frequency; ditto for the reference resistors.



I believe that each one us audiophiles believes that every power amp which has been created todate, has [or must have] a characteristic and discernible voice in a reference loudspeaker. Where is this unique sonic print of any power amp found? It is physically in its output current, and the special interaction between the amp and the loudspeaker. My loudspeaker receives a current contribution from VSA and another independent current dose from TCA. What is the relative current contribution from TCA [or VSA] in the sum flowing through the loudspeaker? Easy to calculate once you know Av of VSA [21], Gm of TCA [1A/Vi] and the min [4 Ohm] and max [20 Ohm] impedance of my loudspeaker. One rule prevails: Vo must always satify the law of VSA which is Vo = Av of VSA multiplied by Vi. Here are examples: Assume Vi = 1 V. Clearly both amplifiers are operating simultaneously in parallel.

• Let Z speaker = 4 Ohm. The transient Vo due to TCA only is 4 V. It is calculated by multiplying 4 Ohms by Io = 1 A. VSA instantaneously says; violation or no way, this transient Vo due to TCA only does not cut it. It must be 21 V instead [ and it is]. So the VSA backfills the defiencient current which must flow through the loudspeaker. The net current through the loudspeaker is 21V divided by 4 Ohms = 5.25 Amperes. So the % contribution of current from TCA to the loudspeaker is 1A divided by 5.25A = 19%. This is a 19% sonic print of TCA to the voice of the loudspeaker.
• Let Z speaker = 21 Ohms. The transient Vo due to TCA only is now 21 V. No violation as experienced by VSA. The % contribution of current from TCA through the loudspeaker is 100%. [sonic print too]
• Let Z speaker = 30 Ohms. The transient Vo due to TCA only is 30V. Violation says VSA. It must be 21 V. So the actual current flowing through the loudspeaker is 21 V divided by 30 Ohms = 0.7 A. But the TCA put out 1 Ampere. Where did the balance of 0.3 A go?. The VSA absorbed it; afterall it has a low output impedance. Is this shunting good or not?

To be continued. In the interim experiment with this Compound Power Amplifier and determine its value.

 

[Antoinel]

Compound Power Amplifier

The VSA in the Compound Power Amp of the upper post dumps a small amount of its output current [at best] into the TCA and its Zobel. Because the impedance of the TCA is high [e.g. 130 Ohms at 10 KHz with Zobel] by comparison with that of the loudspeaker [4 -21 Ohms].

What if I have a Phase Inverting TCA of Gm = 1A/Vi; but with an intrinsic output impedance which is low e.g. 10 Ohms, and no output Zobel. A practical example is the DIY SIT amplifier "L'Amp; A Simple SIT Amplifier" by Michael Rothacher . The load of the SIT device is two or more light bulbs of net impedance equal to ~10 Ohms. The light bulbs [~10 Ohms] are connected between its output port Vo and V+ which is AC ground. Let's take this 10 Ohms as its Intrinsic Output Impedance to work through a paper analysis of a Compound Power Amplifier using it.

Will this SIT amp [a la Rothacher] operate in a satisfactory manner in the above Compound Power Amplifier Type A? The paper analysis begins by asking: Suppose the impedance of the loudspeaker is 4 Ohms...

To be continued..

 

[Antoinel]

Compound Power Amplifiers. Listening tests

In Post #17, I asked whether it was "good or bad " for the TCA to dump excess current in the VSA ? Recall in the example of post that it was 30% [0.3 A] of the TCA's must output current of 1 A. I let my ears be the judge. Please consult the attached redundant schematic [CompoundAmp A3.pdf] on testing the hypothesis and thereafter listening to music. Note first that the declared or published minimum impedance of most loudspeakers [mutiway and full range] happens in the 100 - 500 Hz range. So :

• Open switch A to isolate VSA from the output
• Generate a low level output signal [Vo'] at the loudspeaker using the frequency you have already determined to give minimum Z in the 100-500 Hz range [impedance plot vs frequency]. Or find at this point the minumum output voltage as you slowly sweep the frequency between 30 and 500 Hz.
• Measure Vo'. I used my AC multimeter. I only needed a number.
• Adjust the volume of the 25 Ohm pot so that the output voltage Vo of VSA is numerically equal to Vo'
• Close Swich A. Vo' or Vo [joined power output] must read exactly the same value measured for each stand alone amp.

Thus the output voltage of TCA bucks the output voltage of VSA at that specific and minimum speaker impedance. We already know [or measured] that higher values of loudspeaker impedance exist. Case in point, as you swept the generator frequency from 30 Hz up to 100 Hz, you hit the high impedance of the woofer at its resonant frequency; for example at 60 Hz. We also know [above posts] that the VSA will absorb excess current from TCA; otherwise the new output voltage [Io from TCA times higher impedance] is higher in value than Vo [same Io of TCA multiplied by a lower impedance]. This situation will violate the constant gain equation of VSA. So with switch A closed, sweep the generator frequency again from 30 Hz up to 100 Hz. The joined output voltage at the loudspeaker is constant throughout.

So, the tones you now hear [20 Hz and upwards] are a 100% contributed by TCA. You are now hearing TCA [of a certain sonic print in A/D/S/ L-730 loudspeaker] operating as a VSA [a different voice from the same loudspeaker]. The output impedance [<0.1 Ohm], damping factor [>50], and distortion [v. low] of this Compound Power Amplifer [at the joined output port] are those of the stand alone VSA. Because it has intrinsic error correction capabilty absent or unattainable in TCA. The pristine output current in the loudspeaker came from TCA only; because TCA must put it out in a load [and VSA] whenever it has an input signal Vi.

I connected the headphone output of my SONY CD player and listened to the voice of the Compound Power Amplifier via the A/D/S loudspeaker in mono. I heard a clean, articulate and well detailed new voice of my DIY TCA.

To turn off the CPA, open switch A and power down the amplifiers in any sequence.

To be contd..

 

[Antoinel]

Please go to post # 1399 of the Thread L'Amp: A simple SIT Amp in the Pass Labs Forum. Since I do not have a SIT amp, I suggested to Mr. Pass, Mr. Rothacher and DIYers who are intimately familiar with the SIT amp to consider exploring this new application of it. The application's wiring diagram [or schematic] and operating procedure of the compound power amplifier [CPA] are also shown. They are essentially like the ones you have already seen in the above posts of this thread.

What is the output power of my CPA?

It is the same as, and must be that of TCA. This TCA has +/-25 V power rails. A comfortable maximum output voltage swing into an 8 Ohm load is 30 V{p-p}. It is ~11V rms to cause a current flow of 1.3 A rms in the 8 Ohm load for a total power of ~14 W rms per channel. Threshold S/150 [VSA] has +/-49 V power rails. It is capable of 75 W rms per channel in 8 Ohms. Overkill for this application. Best to have the output power of VSA close in value to that of TCA. This will minimize one amp inadvertently overdriving the other. In my CPA, I use a fast blow 2A fuse between the output of the S/150 and the Joint Ouput as an added precaution.


The transformer In the CPA.

I got the idea of using the transformer from the F6 Amplifiier thread as I already said in an earlier post. It readily can be the Jensen transformer suggested for use in diy F6, or a suitable other of your choice. I use a step-down transformer in my application as a safety measure to lower the signal drive so as to limit the output power of the prototype CPA. Here is the additional value of the transformer besides step up/down and 1:1 conversions.

• Electrically isolates the signal source at the primary from the amps at the secondary windings. This is the case when I drive the primary winding with the volume-controlled headphone output of my CD player. The ground of the CD player and that of CPA are not connected, and do not need to be anyway. The signal source may be equally powered by batteries instead of the power line.
• The secondary windings do not introduce the dreaded ground loops in their input signals. My CPA is quieter than a mouse.
• Has a phase inverting function by one secondary winding relative to the other; where needed as in this application. Why use an electronic phase inverter? This passive approach will do nicely.

I'll build a second channel of my CPA. To be continued..