View Single Post
  #6   Report Post  
Arny Krueger
 
Posts: n/a
Default Arny doesn't understand that secondary breakdown=thermal runaway


"Bob Morein" wrote in message
...

"Arny Krueger" wrote in message
...
"Bob Morein" wrote in message


"Chad Williams" wrote in message
om...


In researching solid state integrated amps/receivers I've come across
several proponents of old receivers, circa '75-early 80's, who say
that these solid state systems are every bit as good as anything
being made now.


Certainly by the mid-late 1970s and early 1980s really good-sounding SS
power amplifiers had started to really proliferate.

To your ears, perhaps. I put it early eighties.
While the Phase Linear amplifiers had good sound, they couldn't manage
thermal runaway.


While I don't disbelive this statement, I'd like to
understand why this is the case. Was the build quality simply better
back then? Are the transformers higher quality???


[snip]
However, in terms of quality of the amplifier, it is definitely
false, for several reasons:


1. Designs in the 70's suffered from transient intermodulation
distortion. Around 1979, this was discovered and eliminated.


TIM is just excess nonlinear distortion at high frequencies.


No, it's not. The definition includes the word transient.


So what? If the problem shows up only with instantaneous, one-time
transients, then it isn't a problem in audio because there are no truely
instantaneous transients in audio signals (they are always band-limited) and
if they are one-time events they aren't heard because there is no mechanism
in the ear for perceiving extremely short-term one-time events. Make a pulse
narrow enough (i.e, instanteous) and it isn't audible.

It is
distinguished from steady-state TIM in that it is measured when the
amplifier is not in steady state. Read he
http://www.zero-distortion.com/techn...cations_04.htm


They get to be wrong. Just because its on a web page doesn't mean it has to
be right.

In the early
days of SS amps there was a tendency to give power ratings for power

amps
that were too close to the actual capabilities of the equipment. Most SS
amps have output decoupling networks that improve amplifier stability,

but
also cause losses that decrease the maximum undistorted power output of

the
amp at 20 KHz by 0.5 to 1 dB.


This is obfuscation.


No, it was a cause of misundertanding.

The cause of TIM is excess loop gain at high
frequencies, and there is a specific remedy.


Since we can't agree on what the problem is, it is very impossible to
discuss what caused it or what cures it.

However, the cause of TIM was stated by Otala in Circuit Design
Modifications for Minimizing Transient Intermodulation Distortion in Audio
Amplifiers JAES: Volume 20 Number 5 pp. 396-399; June 1972 as not being
excess over-all loop gain, but excess gain in certain parts of the loop,
and/or inadequte gain in other parts of the loop.

[snip]
The TIM myth was
effectively deconstructed at that time.


I disagree.


You get to be wrong, which I have documented with authoritative independent
references many time, such as the example just above.

2. Bipolar transistors suffer from "thermal runaway", which occurs
when a small area of the junction heats up locally and becomes more
active than the rest of the transistor. Once it starts, the
transistor is quickly destroyed.


Again this is false. Thermal runaway is a large-scale effect that

involves
the entire transistor junction. It has been known and managed about as

long
as there have been bipolar transistors, or more than 50 years.


Permanent damage of transistors due to small-scale localized heating is
instead known as "Secondary Breakdown".


IMPORTANT***IMPORTANT***IMPORTANT:

Arny, in a BJT, THERMAL RUNAWAY AND SECONDARY BREAKDOWN ARE IDENTICAL.


No they aren't. They have well-known and distinct definitions. I clairy
this below using one of your referneces. I suggest that you search google
and inform yourself properly, or just read your own references.

Secondary breakdown is merely the physical phenomena which causes loss of
current control.


It's a specific failure mode of output transistors that happens quite
rapidly due to excessive short term power dissipation. It is generally
independent of transistor biasing although class A bias can lead to
transistors becomeing prematurely overstressed.

In contrast, classic thermal runaway is basically a failure of biasing
circuits to maintian normal quiscent, long-term current levels.

In this case, you are confusing the same phenomena viewed differently.
"Secondary breakdown" is the term used by a device physicist.


They are two different things, which both equipment designers and device
designers call by the same names.

"Thermal runaway" is the term used from the systems & control point of

view.

Nope, its defined differently as I already explained.

The only solution available in the 1970's was brick-wall current

limiting.


Again this is false. In the 1960s and to this day, SOA protection

circuits
were widely used, but these circuits monitored both the voltage being
dropped across the output devices and the current flowing through them.

If
one finds audio transistor design manuals from the 1960s as well as

modern
manuals, SOA protection circuits that are controlled by both voltage and
current are described.


In my experience, I have not found an amplifier made in the 60's or 70's
which did anything other than brick-wall limiting.


This sentence is irrelevant to the discussion since it does not contain the
phrase "currrent limiting" which was present in the original claim and my
response to it.

I'd be interested in some examples.


First, its time for you to start talking about the same thing you started
out talking about, i.e, "current limiting" and that I responded to.

However, amplifiers which use
this kind of protection cannot handle the dynamic range of a CD at
greater than low volume.


This is also false. SOA limiting parameters are determined by the SOA

limits
of the amplifier's output devices.


Based upon my experience with Marantz and Heathkit, I disagree.


SOA is not just an issue relating to a couple of equipment manufacturers.
However Heathkit SS amplifier schematics are online and all but their
initial non-performing germanium-transistor instant flame-out junk used SOA
limiting circuits that monitor both current and voltage. I built and owned
AA-22 (germanium junk) as well as the properly-designed silicon-based AR-15,
AR-1500, and AA-1640 Heathkit amps. I also own three different Dynakits, one
of which has a reputation for extreme fragility and lacks standard SOA
protection. I own numerous other SS amps as well.

But as I've said, I'd be interested in some counterexamples. Crown,

perhaps?

Crown amps from the DC-300 on incorporated some kind of SOA protection. In
later Crown designs the output stages can be very complex and analyzing the
function can be very difficult, but in the DC-300 the SOA circuits are
pretty straight-forward.

SOA circuits tend to be activated by real-world speaker loads, but are

less
likely to be activated in resistive load testing. Therefore, some power

amps
with relatively high power ratings such as the original Crown DC-300

were
sold that arguably lacked sufficient SOA for handling loudspeakers that

had
the deadly combination of low impedance, high reactance, and low

efficiency
at frequencies where music tends to have a lot of energy.


SOA limiting was often observed while playing LPs, not just CDs as

stated
above.


For example the "Some Amplifiers Sound Different" article that I
co-authored and appeared in High Fidelity News and Record review is
basically a story about an expensive (Audio Research) solid state power

amp
that had just been highly reviewed by TAS, but in fact had SOA issues

with
certain (Acoustat) speakers. These listening tests were as I recall,

based
on playing a LP of the Eagles' "Hotel California" at a pretty high

level.
Back the volume off a dB to a more modest but still loud level, and the
problem went away.


3. The noise figure of bipolar transistors dropped about 10 dB around
1980. Prior to that, equipment had an S/N ratio of around 70 dB.
After 1980, S/N ratios of 90 dB and greater became the norm.


This is a meaningless statement because SNR is only meaningful when
referenced to an operating level. If one measures the SNR of amps and
preamps made in the 1960s, 1970s or even 1990s the line level circuits

tend
to have SNRs that are better than tubed equipment and are in the 90+ dB
range.


Apparently, you can't hear hiss.


Gratuitous and irrelevant personal attack noted. SNR is a measured
parameter. Whether or not you have audible hiss depends on many things, not
just SNR in the preamp.

If one measures their phono inputs the SNRs are more like 70 dB and
up which is logical because the operating voltage levels are lower.

There
have not been any difficulties with well-designed transistor amplifiers
being noisy excessively noisy since no later than the late 1950s. So not
only is it a meaningless statement, it's just plain wrong.


4. In 1981, David Hafler was the first to build an audio amplifier
with Hitachi's new "power MOSFET."


Again the date is all wrong. Trivial searching shows that the DH-200 was
introduced in 1979.


I'll give you the date.


I gave you the date, which is backed by independent references I didn't
bother to cite.

However the DH-200 was not the first Hitachi-MOSFET
power amp, just the first popular-priced kit.


This was a major watershed in amplifier design.


While the DH-200 and many successive MOSFET amplifiers are fine

amplifiers,
in fact they don't as a rule sound better than competitive,

well-designed
bipolar designs. MOSFETs have long been popular, but have never

dominated
the marketplace for high quality power amplifiers. They have their
advantages and their disadvantages...


Concurrently, new methods of protecting bipolar
transistors were implemented.


Previously debunked, and false.


No, true from personal experience.


You get to be wrong, again.

I have had several amplifiers of that era that clip hard as a result of

the
current limiting.


So what? I explain why more modern amps are less likely to be driven into
SOA limiting. BTW again, the claim that pure current limiting is the fault
is false. SOA limits are based on combinations of voltage and current not
just current alone. It is possible to approximate SOA limits with pure
current limiting, but this results in poor exploitation of device
capabilities. Nevertheless it is used in certain highly-overbuilt (from the
standpoint of SOA) power amps such as my two QSC units.

What has happened is that the SOA of SOTA
power transistors and ICs has undergone steady improvement.


True with respect to discrete devices.


True with respect for all devices, discrete or IC.

With respect to ICs, there has been
substantial work in more sophisticated safe area protection, which I
reference below.


This does not prove that IC SOA hasn't been improved. Certain ICs may have
more sophisticated SOA protection that better exploits device properties,
but the solid cure to SOA issues is simple and straight-forward: Output
devices with more SOA.

Therefore, it's
possible to build an effective high powered amplifier with fewer and
physically smaller output devices. For example I have an upgraded Dyna 4

00
that equals or exceeds the reactive-load handing capabilities of the

stock
Dyna 416 with half as many, but far more modern output devices.


That's a terrible amplifier. Your fundamental problem is a lack of hearing

acuity.

Gratuitous and irrelevant personal attack noted. Use an amp within its
limits and there is no audible distoriton.

Arny, you're SO full of ****.


Gratuitous and irrelevant personal attack noted.

Here's a reference to a National data sheet:
National Semiconductor's bipolar-output parts (Table 1) incorporate a
dynamic SOA-protection mechanism, called SPiKe, which stands for self-peak
instantaneous Kelvin temperature-protection circuitry. National claims the
circuitry makes the ICs nearly impervious to damage from instantaneous
temperature peaks and overvoltage and overcurrent conditions.


Largely irrelevant because protecting the device better does not increase
its basic capability. If you want to drive a certain speaker load with an
undistorted signal the only option is to deliver a certain set of currents
and voltages. Improved protection circuits don't increase basic device
capabilities, they only make incremental improvements in how the existing
capabilities are exploited. The major story in SS power amp performance over
the past 40 years has been basic improvements in output stage SOA
capabilities. Better SOA circuits provide detail-level improvements.


You can read the above at the EDN website:
http://www.e-insite.net/ednmag/archi...1795/17df1.htm


Nothing there that is consistent with what I said. Indeed it does a nice
job of distinguishing between thermal runaway and secondary breakdown. The
paragraph headed "Beware of Bipolar Burnout" covers thermal runaway and its
relationship to VBE variations with temperature.

Two paragraphs later a discusison of SOA and second breakdown. The
referenced document (AN-039) can be found at
http://www-k.ext.ti.com/SRVS/CGI-BIN...00000000346801
7,K=7800,Sxi=3,Case=obj(17402)


and the datasheet at http://www.national.com/an/AN/AN-898.pdf#page=9
Also of interest is http://www.national.com/an/AN/AN-261.pdf#page=2, which
dates an attack on this problem to 1981.


SOA protection circuits appeard in RCA transistor manuals during the
early-mid 1960s.

The Philips Power Division has a relevant document:
http://www.semiconductors.philips.co...es/APPCHP7.pdf


"When a power transistor is subjected to a pulsed load,
higher peak power dissipation is permitted. The materials
in a power transistor have a definite thermal capacity, and
thus the critical junction temperature will not be reached
instantaneously, even when excessive power is being
dissipated in the device. The power dissipation limit may
be extended for intermittent operation. The size of the
extension will depend on the duration of the operation
period (that is, pulse duration) and the frequency with which
operation occurs (that is, duty factor)."


Classic, completely supports my position.

and

"Conclusion
A method has been presented to allow the calculation of
average and peak junction temperatures for a variety of
pulse types. Several worked examples have shown
calculations for various common waveforms. The method
for non rectangular pulses can be applied to any wave
shape, allowing temperature calculations for waveforms
such as exponential and sinusoidal power pulses. For
pulses such as these, care must be taken to ensure that
the calculation gives the peak junction temperature, as it
may not occur at the end of the pulse. In this instance
several calculations must be performed with different"


Classic, completely supports my position.


IBM has a document dated 1977 that refines safe area calculation:

http://domino.watson.ibm.com/tchjr/j...8fcca51d85256b
fa0067f80c?OpenDocument


Per the reference cited above, this was old news at the time.

Both Hafler's and Strickland's MOSFET
designs had interesting qualities that raised the bar for bipolar
designers.


Credit needs to be given to Hitachi's engineers who laid out many of the
circuit designs and parameters for building MODFET power amps and

provided
them along with the devices. "Name" high end audio engineers generally

don't
innovate much of anything in the way of power amps, they just tune and
repackage circuits that are already widely used and/or suggested by

device
manufacturers.


This is bull****.


Sorry that the truth hurts so bad.

And what the hell is a MODFET ?


MOSFET + a typo.

DoooHHH!