the F7 Power Amplifier
Now For Something Completely Different:
the F7 Power Amplifier
Introduction - Short Story Long:
Conceived in 2007, the F5 was a push-pull Class A amplifier employing eight
semiconductors and 23 resistors to achieve 25 watts output with good
specifications and good sound. You can read about it in detail in both the
owner's manual and DIY construction article, both posted at www.firstwatt.com
The F5 received good reviews, sold well and received plenty of attention from
the DIY audio community (as of this writing, the F5 thread on www.diyaudio.com
is cresting 3 million views). Like other First Watt amplifiers, it was a limited
release, and was discontinued after 100 pieces were built.
Subsequently there was a more powerful version released to the DIYers, the F5
Turbo, which offered more power and numerous options for experimentation.
The F5 was noted for detail and neutrality, the result of simple Class A operation,
wide bandwidth, and a generous amount of negative feedback. With some
material and equipment, the sound was exquisite, but the amplifier also tended
to highlight the faults of recordings and the rest of the system. A poorly
engineered record, a second-rate DAC or peaky loudspeakers and the amplifier
was merciless.
Since then I have spent time mostly exploring other kinds of design approaches
to amplifiers, but I always meant to come back to the F5 to see how it could be
improved. Mostly I was looking for two things – better sound and an even
simpler circuit. The desire for a simpler circuit is self explanatory – apart from
the aesthetic, I imagine that simpler circuits tend to sound better. (Because I
build them myself, I will add cheapness and laziness to my motivations.)
I envisioned a circuit with only four transistors and four resistors – the bare
minimum for an amplifier of this type, where all the Fets are operated in
Common-Source mode, giving both voltage and current gain. Two
complementary input Jfets drive two complementary power Mosfets, and the
output voltage is fed back to the Source pins of the Jfets in what is commonly
called “Current Feedback” (CFA).
The schematic looked like this:
In such a simple circuit, there are opportunities for improving performance by
careful choice of transistors, resistor values, voltage and current values and
precise matching of parts. I built up a number of such amplifiers evaluating both
the measurements and the sound, and after a while a very nice little amplifier
emerged. That it (and the F5) resembled Plantefeve's Profet was testament to
his fine work.
It had most of what was on my wish list:
Very wide bandwidth
Low distortion and noise
Large Class A operating region
Less feedback
No degeneration in the output stage
Very low thermal distortion and drift
No capacitors or transformers (apart from the power supply)
Looks like a great laundry list. I thought the amplifier sounded pretty good, but
after a time the consensus was that it was kind of polite, not as musically
involving as some other examples. One area where the amplifier fell
“objectively” short was the output impedance. Most of your “better” amplifiers
have higher damping factors like 20 or 100, and this amplifier was only about 5
or so, typical of a simple Common-Source topology and low negative feedback.
The damping factor, which is the inverse of the output impedance, determines
the flatness of the amplifier response when the load impedance varies and is
important to the transient response of reactive loudspeakers, which is just about
all of them.
You could argue that single-ended tube amplifiers sound good and don't have
much damping factor, and you would be right, but I'm not trying to duplicate
single-ended tube amplifiers here – the First Watt SIT amplifiers, doing a decent
solid-state emulation of Triodes, represent that genre nicely. I'm shooting more
for the sound of The Beast With a Thousand JFETs but with obtainable parts.
I tried other Fets, varied the the feedback network values and tweaked the gain
symmetry. These efforts delivered a little more warmth and dimensionality, and
about a year ago I decided to go ahead and build a small pilot run in anticipation
of a product release – after all, it was a perfectly nice amplifier.
Still, I was not really satisfied, so I created new printed circuit board artwork,
adding cascode operation to the input stage and doubling up the output devices.
This version had a little more control, but still fell short.
So...
I went back to the original simple design. On the wall in my lab is a
little box with a glass window labeled:
DANGER - POSITIVE FEEDBACK
I broke the glass. Inside was a single resistor.
And I put that resistor in the amplifier...
A little explanation is in order. The relationship between an amplifier and a
loudspeaker is a bit like a dance. Both sides have their own complexities, but
the point is for them to get along well. The amplifier designer, not generally
having control of what speaker is used, usually chooses the amplifier as the
dominant partner by making it a pure voltage source. In the typical voltage
source, negative feedback in the amplifier is used to define the voltage across
the loudspeaker regardless of the current through the speaker.
This represents the “have it my way” approach to amplification, and large
hardware with lots of feedback are good at this. Most loudspeakers are
designed around the assumption of a low impedance voltage source.
The F7, a nice little Class A amplifier with hardly any feedback does not have
the brute force advantage. It resorts to a stratagem that makes the dance a little
more like a Tango.
Modest amounts of negative feedback are balanced in counterpoint to small
amount of positive current feedback, creating an equilibrium where the output
impedance approaches zero, improving transient and frequency response.
Of course you can achieve a similar effect with tons of negative feedback, but I
think this is more elegant and sounds better. For brevity, I will call it “PCF”.
Also, I put more capacitance in the power supply and found a clever way to
further reduce the effect of high frequency DAC noise and environmental RF.
This is a different amplifier. The diversity of audio taste being what it is, not
everyone will prefer it. I presume that a portion of audiophiles will like it.
Nothing But the Facts:
Distortion curve vs output power (1 KHz):
Spectrum of the distortion at 1 watt, 8 ohms and 1 Khz, noting the 10 dB ratio
between 2nd and 3rd harmonic:
That same test, showing the waveforms of the fundamental 1 watt tone and the
distortion:
Output current through half the output stage, leaving Class A at 50 watts peak:
Smooth clipping at 40 watts into 4 ohms:
Response vs Frequency:
10 Khz Square Wave at 28 Watts:
Here's where we get to see some of the differences afforded by a little positive
current feedback.
I created a purely reactive loaded using a 1 mH coil in series with a 50 uF film
capacitor which resonates at about 700 Hz to an impedance dip well below 1
ohm. Here you see the amplifier driving that load with a 180 Hz square wave –
showing the ringing induced. The improvement in control is readily apparent.
I also created a difficult test load reflecting an amalgam of the sorts of things
seen in several popular loudspeakers. Here is the impedance of that load,
dipping down below 3 ohms at two points:
And here is the response of the amplifier into this load with and without pcf.
How's the distortion when pcf is applied? Here is a distortion vs frequency plot
of that amplifier into 8 ohms and also the tough load. With or without PCF, we
see that the distortion goes up a bit when confronted by 2.5 ohm loads, but this
is expected from any amplifier – more power is being delivered at those
impedances, particularly from the amp with PCF, whose output voltage is not
sagging.
It is understood that pcf allows for slightly more distortion in proportion to the
amount of positive feedback employed. Another way of viewing it is that the
amount of positive feedback is subtracted from the negative feedback, and so
the result is similar to simply having less negative feedback in the first place, in
this case the amount being about 2 dB, which is not much at all.
Conclusion
In case you haven't noticed, I enjoy amplifiers with some personality. They don't
have to measure perfect, they just have to sound good. This is a very simple
little Class A amplifier with lots of personality, and I hope you enjoy it as I do.
Nelson Pass
12/5/16
Nominal specifications:
Measured at 120 V AC and an 8 ohm load:
Distortion @ 1 watt
0.05%
Input Impedance
10 Kohm
Gain
14.5 dB
Input Sensitivity (1 watt)
0.53 V
Input Sensitivity (max output)
2.7 V
Damping Factor
>100
Output power 8 ohms
20 watts @ 2% THD, 1KHz
Output power 4 ohms
30 watts @ 3% THD, 1KHz
Class A envelope
40 watts peak @ 4 ohms
Frequency response
DC to -3 dB @ 100 KHz
Noise
100 uV unweighted, 20-20 KHz
Power consumption
170 watts
Fuse
3AG slow blow type 2.5 Amp for 120VAC
1.25 Amp for 240 VAC
Warranty: Parts and labor for 3 years, not covering shipping costs or
consequential damages.
Copyright 2016 General Amplifier
Contact:
www.Firstwatt.com
nelson@passlabs.com
Now the following is for your protection –
Do not defeat the AC line Earth ground connection on the
amplifier power cord. It provides an extra barrier to prevent
potential shock hazard.
Do not replace the fuse with a type other than specified.
Do not operate the amplifier outside in the weather, or in and
around water or anything resembling water. If you spill a drink
in the amplifier or if your dog/cat/child urinates on it, turn it off
immediately, unplug it, and do not operate it until cleaned by a
qualified technician.
If something gets loose or rattles around inside or smells
funny, or if you can’t touch the heat sinks for 5 seconds or so,
then turn it off, unplug it from the wall, and contact First Watt.
There are no user serviceable parts inside. Do not open the
amplifier, and if you do anyway, don’t operate it with the cover
off. There are hazardous voltages inside. If you need to
change the operating AC voltage, contact First Watt.
Once Again:
If you have a problem, contact First Watt. We are much
happier helping you solve problems so that we can be certain
that it’s done properly. If you are far away and don’t want to
ship the product for repair, we will assist your technician with
information and parts.
Contact: www.Firstwatt.com nelson@passlabs.com
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising