EMRFD Message Archive 1902
Message Date From Subject 1902 2008-08-07 18:49:21 Craig Johnson, AA... EMRFD Fig 1.17 Audio amp Hi,
I've been playing with an LTSPICE model of the audio amp in Figure 1.17
of EMRFD. It's not working as expected. Maybe I missed something in
the model but I can't see it.
Here are the problems I am having with it:
1) I'm only getting an overall gain of about 2.5 from the amp. Is this
right? THe Q1 bias is correct at about 2ma (with supply voltage of 12)
but the voltage gain from this stage alone is only 2 instead of the
expected 20 per the text. I assume this causes the relatively low
overall gain too.
2) I am not seeing the expected push-pull action that I expected to see
from the complementary action of the two output transistors. Probably
because the drive is too low.
I've posted the LTSPICE model in the files section, file named "EMRFD
1.17 audio amp LTSPICE".
1903 2008-08-08 04:50:21 Allison Parent Re: EMRFD Fig 1.17 Audio amp 1904 2008-08-08 06:49:22 Nick Kennedy Re: EMRFD Fig 1.17 Audio amp Hi Craig-
I had an LTSpice model of almost the identical amplifier, the audio amp from
the Tin Ear receiver, which I think came from that circuit in emrfd.
I see the voltage gain about the same as you do -- maybe 2 to 2.5. The
power gain comes mainly from the fact that the load resistance is much lower
than the input resistance.
Cranking up your input voltage from 1 mV to 0.5 V peak, I see a pretty nice
sine output. However, a 1mV it looks pretty good too. And plotting the two
individual currents of the final transistor pair, they look about equal and
180 degrees apart. Oh, I see what you mean, they're both sines at that low
level. You'd expect each to be a half sine. At 0.5V peak, they're closer to
>[Non-text portions of this message have been removed]
1905 2008-08-08 13:51:24 Craig Johnson, AA... Re: EMRFD Fig 1.17 Audio amp Hi Allison,
1906 2008-08-08 14:06:29 Craig Johnson, AA... Re: EMRFD Fig 1.17 Audio amp Hi Nick,
1907 2008-08-08 14:59:07 John Levreault Re: EMRFD Fig 1.17 Audio amp Sorry I've been away or I would have responded to this post earlier.
The circuit is not too difficult to understand. It's a simple discrete
Its open-loop gain is approximately equal to R8/R4, perhaps a bit less,
or ~20x (26dB). The circuit comprised of Q2, R7, R6, and C3 is called a
"Vbe multiplier". It multiplies the Vbe of the transistor (Q2 in this
case) by the quantity (1 + R7/R6). This Vbe multiplier provides bias for
the emitter-follower output stage (Q3/Q4), which has approximately unity
gain. For the purposes of circuit analysis, it's a very low resistance
and its effect on Q1's gain can effectively be ignored.
The feedback loop is comprised primarily of R3 and R2. This sets the
overall gain, since it's an inverting amplifier, to R3/R2 or about 2,
perhaps a bit more owing to the finite input impedance of Q1 (~R4*Beta
or about 20K). This feedback network also helps stabilize the DC
operating point of Q1.
However, whatever we drive this circuit with will have an impact on the
closed loop gain. If we connect it to the output of an opamp stage,
which has a very low output impedance, then, for the purposes of closed
loop gain calculation, R1 effectively gets connected in parallel with
R2. Therefore, the net closed loop gain will increase to about 5x
(14dB). As the source impedance increases, the closed loop gain will
This circuit is intended for as a low-gain output buffer for driving
headphones, as described in the text. Its function is to provide a big
increase in output current but only a modicum of gain. If you need more
gain, I'd recommend an opamp to drive this circuit.
So yes, what LTSpice is telling you is basically correct, but you might
want to observe the effect of a finite source impedance by sticking an
opamp buffer in front of this circuit.
Hope this helps.
de John NB1I
Craig Johnson, AA0ZZ wrote:
>I've been playing with an LTSPICE model of the audio amp in Figure 1.17
>of EMRFD. It's not working as expected. Maybe I missed something in
>the model but I can't see it.
>Here are the problems I am having with it:
>1) I'm only getting an overall gain of about 2.5 from the amp. Is this
>right? THe Q1 bias is correct at about 2ma (with supply voltage of 12)
>but the voltage gain from this stage alone is only 2 instead of the
>expected 20 per the text. I assume this causes the relatively low
>overall gain too.
>2) I am not seeing the expected push-pull action that I expected to see
>from the complementary action of the two output transistors. Probably
>because the drive is too low.
>I've posted the LTSPICE model in the files section, file named "EMRFD
>1.17 audio amp LTSPICE".
1908 2008-08-08 16:54:58 Allison Parent Re: EMRFD Fig 1.17 Audio amp 1909 2008-08-08 17:46:36 Nick Kennedy Re: EMRFD Fig 1.17 Audio amp I think you'll need up around a volt or so. A few mV at 60 ohms would
mean microwatts of audio. I think milliwatts are more what it takes
for these little phones. Seems like they're spec'd at 1 mW for a
certain output, but I'm not certain.
Plug a set into the line out jack of a music player and it'll be about
right for rock music, maybe a trifle on the loud side. Line voltage
is something like a couple volts. So I do think you need some voltage
gain ahead of this stage as it's a power amp based more on current
gain than voltage gain due to having high Z input and low Z output.
It could have a voltage gain of 0 and still have OK power gain.
BTW, on not seeing the complementary action at low levels -- I'd say
that's because you bias push-pull stages so that both are normally
"ON" a little with no signal. That's to prevent the dreaded
"crossover distortion" that takes place between the Vbe turn-on
voltages of the two transistors.
A fun amp to analyze. I think I finally figured out the Vbe
multiplier studying one of these things a few months ago.
1910 2008-08-08 19:51:51 victorkoren Re: EMRFD Fig 1.17 Audio amp Craig,
Could you post the LTSPICE schematic of the amplifier in a JPEG or
equivalent file? (i dont use LTSPICE so I hope it has a graphics
Victor - 4z4me
1911 2008-08-08 21:06:29 Craig Johnson, AA... Re: EMRFD Fig 1.17 Audio amp Victor,
1912 2008-08-08 22:30:04 bob bailey Re: EMRFD Fig 1.17 Audio amp thank you, I grabbed it too.
1913 2008-08-09 08:27:47 Craig Johnson, AA... Re: EMRFD Fig 1.17 Audio amp Allison,
--- In firstname.lastname@example.org, "Allis
1914 2008-08-09 08:29:39 Craig Johnson, AA... Re: EMRFD Fig 1.17 Audio amp John,
Well done analysis and explanation.
1915 2008-08-09 08:52:23 Craig Johnson, AA... Re: EMRFD Fig 1.17 Audio amp Nick
1916 2008-08-09 09:42:38 flyboy2high Re: EMRFD Fig 1.17 Audio amp Headphones require almost no audio power so a power amp such as the
Fig. 1.17 amp is a poor match for your requirement. Why not use a
two transistor voltage amp driving an emitter follower? The voltage
gain will be in excess of 100 and the current draw about 10-20 mA,
maybe less. Perhaps even better would be to use a small audio output
transformer in a single ended configuration.
I tested a cheap dollar store pair of low Z headphones and they were
about 32 ohms Z.
Years ago I determined that 1mV of audio across a pair of 2000 ohm
headphones would produce a usable signal level. That is something
1917 2008-08-09 14:07:42 Nick Kennedy Re: EMRFD Fig 1.17 Audio amp A while back on one of the lists some of us were fooling with "CW
crystal sets", which was just an antenna to a mixer to headphones. Of
course you had to have BFO input to the mixer or product detector to
make it switch, but the point was, the audio all came from what was in
the antenna. And we could copy CW signals of moderate strength OK.
That's a testament to the range of the human ear and also the
sensitivity of some headphones.
With that said though, I don't think microwatts of audio will cut it
for ordinary, every day ham use. In my "measurements" spreadsheet, I
have some notes on estimated loudness using Koss "the plug" phones
(which are quite sensitive) for various levels of power and a 600 Hz
Loudness Vp-p Power mW Ip-p (mA)
barely there 0.0125 0.001 0.368
low, but OK 0.02 0.012 1.67
loud enough 0.225 0.19 6.6
extra loud 0.5 0.92 14.7
For general usuage, I'd want the margin for the extra loud level, or
more if you want dynamic range. Maybe 6 dB above that highest level or
about 4 mW. I think the little push-pull circuit is ideal. There are
probably ICs that do the same thing more easily, but I'm sure the
EMRFD circuit and the Tin Ear were both done to show how it's done
with discrete parts.
I had a hard time coming up with a single transistor class A solution
for the Walkman (now iPod) phones driver, whether emitter follower or
with the phones in the collector lead. Maybe with a transformer added
it would be easier.
1918 2008-08-10 07:50:34 victorkoren Re: EMRFD Fig 1.17 Audio amp One remark.
John's analysis is almost totally correct. I have only one remark.
The AC feedback circuit is mainly R3 and R1. This is an inverting
amplifier, if its open loop gain was very high the AC voltage at the
input transistor base would be very small and this point would be a
"summing point" or "virtual ground" just as it is at the inverting
input of an operational amplifier circuit. In such a case the
1919 2008-08-11 17:28:28 Glen Leinweber Re: EMRFD Fig 1.17 Audio amp Nick's table of "loudness" measurements suggests that for his
32 ohm 'phones a class-A audio amplifier is a decent approach.
For "extra loud", he suggests that 0.5v peak-to-peak, with 14.7
mA peak-to-peak is about right, so let's see how a class-A amp
design proceeds from there...
Assume a 9v DC supply. Transistor biasing subtracts a bit from
that, just to establish stable DC operating point. Sacrifice about
1.5v for bias, using a single silicon transistor (2N3904 or equiv).
This design begs for a collector transformer to drive the
phones, since the 'phones only need 0.25 v peak swing, while
we have about 7.5v available. A step-down transformer having
turns ratio of 30:1 would be perfect. For class-A, we need a
steady collector current that can accommodate the full 14.7 mA
peak-to-peak swing. This gets scaled by the 30:1 transformer
to 0.5 mA. A little extra margin is advisable, say 0.7 mA DC
collector-to-emitter standing current.
BTW - this is a GREAT approach for a direct conversion
receiver where audio gain is very high. The higher AC currents
that cause so many ground-loop problems are isolated by the
transformer - the low-Z transformer secondary needn't be
connected to the receiver's ground system at all.
The transformer needs sufficient inductance so that its
secondary reactance is higher than 32 ohms at about 300 Hz.
This winding should have inductance greater than 17 mH. The
primary winding should have more than 15 H. With inductance
this high, its not likely you'd wind this yourself!
For DC bias, I've left 1.5v available between base to
ground. About 0.64v is taken by base-emitter junction, leaving
0.86v from emitter to ground. With 0.7mA emitter current, a
resistor of 0.86/0.007 ohms (that's 1.2K) is needed from emitter
to ground. If left unbypassed, stage voltage gain would be
almost 24 (from transistor base, to 32 ohm load)....
32 ohm load x 30^2 turns ratio divided by 1.2K
The downside is that you're not likely to find a 30:1 audio
transformer. A transformer having less turns ratio could be used,
requiring higher class-A DC current.
Could all 'phones be as sensitive as Nick's? If not, a
smaller turns ratio (with more DC bias) would be good. 32 ohm
'phones are quite common. A push-pull class AB amp is a lot
more complex, but can be made to have lower distortion than
this amp, with a careful design. A LM386 would be good, but
would probably take more current from the supply.
1925 2008-08-12 08:32:19 bobtbobbo Re: EMRFD Fig 1.17 Audio amp A sentence near the end of the referenced article might be
appropriate: "An IC is usually the preferred solution".
I have found that the LM4865 from National Se3miconductor to be an
excellent general purpose audio IC. It will output .75 watts into 8
ohms with THD = 1%. It is a 5V part. Switching between bridged
speaker mode and headphone (single-ended)mode is accomplished via
headphone sense pin. It can be set into low current consumption
shutdown mode of .7 microamps by lowering the DC Vol/SD# pin to below
0.3V.It was designed as a boom-box amp so battery current conservation
was an important consideration, as it is in portable battery operated
One of the features I like is the DC volume control. It also has click
and pop suppression and thermal shutdown. It is a SMT 8 pin part. I
mounted one in an 8 pin DIP header to make it easier to work with. I
think this is an excellent replacement for the LM386.
1926 2008-08-12 08:42:01 John Levreault Re: EMRFD Fig 1.17 Audio amp Also, the NJM2135 is a good headphone amp and runs off +12. They're
usually in stock at Mouser for 47 cents apiece. It's available in an
de John NB1I
>A sentence near the end of the referenced article might be
>appropriate: "An IC is usually the preferred solution".
>I have found that the LM4865 from National Se3miconductor to be an
>excellent general purpose audio IC. It will output .75 watts into 8
>ohms with THD = 1%. It is a 5V part. Switching between bridged
>speaker mode and headphone (single-ended)mode is accomplished via
>headphone sense pin. It can be set into low current consumption
>shutdown mode of .7 microamps by lowering the DC Vol/SD# pin to below
>0.3V.It was designed as a boom-box amp so battery current conservation
>was an important consideration, as it is in portable battery operated
>One of the features I like is the DC volume control. It also has click
>and pop suppression and thermal shutdown. It is a SMT 8 pin part. I
>mounted one in an 8 pin DIP header to make it easier to work with. I
>think this is an excellent replacement for the LM386.
>Yahoo! Groups Links
>Internal Virus Database is out of date.
>Checked by AVG.
>Version: 8.0.100 / Virus Database: 270.5.12/1597 - Release Date: 8/7/2008 5:54 AM