I was looking at a simulation of the circuit with a 10kHz square wave input. I am unable to give the difference figure this showed - I am away from home this week.
I will check this when I get back.
I still say the process of error correction involves the changing emitter currents. Both transistors are common emitter amplifiers when looking at signal inputs at the base. However these are supplied by current from the tail on a shared basis - the shared current amounts vary depending on the predominant voltage on the base of one of the two transistors.
Looking at collector currents says nothing about the path of the nfb signal. Re variations in the tail current caused by the nfb transistor, and taking this as a signal input to the input transistor what sort of amplifier structure do you see from that stand point.
I will look at the value of the load resistor in the LTP constant current source and try some alternative transistor models. This imbalance accusation has been raised on another thread some time ago. In that context I made a point that contrary to expectations of measuring the collector load resistor to deduce the current flow, the Miller capacitance in the nfb transistor would be draw some current exiting the collector. Since the voltage drop at the collector is greater than with the input transistor Miller capacitances will be disproportionate
Member
Joined 2009
Paid Member
All active devices are voltage controlled, transconductance amplifiers. So it's the voltage at the base relative to the emitter we should consider. However, the transconductance, gm is a function of emitter current and this is the basis for non-linearity.
Gm is roughly 40 per m.A . Of collector current and Av is the product of this and the collector load in kilo ohms by a rule of thumb method. If the Av or gain is considerable (loop gain) the input stage will be a small portion of an operating voltage versus current slope. Increasingly smaller sections of that will look increasingly more flat.
Last edited:
Some transistors e.g. BC 546 that are at relatively flat in their optimum range - a good starting point.
Member
Joined 2009
Paid Member
Flat as in resembling more closely to a straight line. You will have to look at a datasheet for typical output characteristics - showing Ic on the vertical axis and Vce on horizontal with IB shown for various plots. There is a strongly curved region for each plot - as far as fixing the operating point is concerned the need is to avoid this region.
Last edited:
Member
Joined 2009
Paid Member
Early Effect is modulation of the base width in a common emitter amplifier. The short explanation is the width of the collector and base layers change when an increased current is passed through a collector load. The full length explanations can be accessed from web sources.
There should also be sources dealing with biasing transistors for correct operating points. These should show the consequences of selecting an operating point too close to the curved area of a plot.
Member
Joined 2009
Paid Member
There are as many ways to introduce 2nd harmonics into the signal from an LTP as there are spices available for a tasty curry. Imbalance is one way, and there are many ways to unbalance an LTP, some of them are unavoidable in a simple design like this. You can also search up Gilbert Cell, which is another method to generate 2nd harmonics more appropriate for my TGM1 amplifier that does not use a CCS load for the LTP.
“Why should I want to add distortion? The distortion’s bad enough as it is without wanting to invent any more of it.” –Marvin the paranoid amplifier designer
Member
Joined 2009
Paid Member
Marvin is right, there will be distortion. Spices were used to cover up smell of old meat. So we add some distortion of the right kind to make everything nice 😀
It's clear enough that JV deliberately increased distortion in NAP models to his delectation, so it could only lead to a frankenstein if we then attempt to "improve" it along the lines of conventional audio engineering precepts. Some, like Greg Ball, have done this though I haven't seen much interest.
However, that suggests we've either misunderstood the products, the design intent, what sound qualities the market pays more money for or we've simply lost the cloning plot. Build a NAP clone or not but there's little point to building the generic 1970's design on which it is based.
However, that suggests we've either misunderstood the products, the design intent, what sound qualities the market pays more money for or we've simply lost the cloning plot. Build a NAP clone or not but there's little point to building the generic 1970's design on which it is based.
Member
Joined 2009
Paid Member
THD is one thing . However the measurent adds all the harrmonics to give a net result. High odd order harmonics 7th,9th,11th are unpleasant . There is a saying that all men are equal only some are more equal than others -you can apply that to high odd ordet harmonics..
There is a question here whether an LTP comparator treats the harmonics equally or compresses some more than others and if so what means can be used to preserve the harmonic structure of the input signal.
The nfb transistor Miller capacitance reduces the magnitude of the feedback signals at hf. There is a question if there are high order harmonics at hf and how these are affected.
That's not clear to me at all. Unless you mean the lower cost models. Would you explain what leads you to think this?
Honestly I have never understood the idea behind all of these series of TGM amplifiers.
For me, different topology will allow for different maneuver in using different transistors. It opens up possibility to get special amplifiers. Also, the quasi topology will allow me to make use of those NPN transistors I have on hand. For small power I have the JLH, but for high power (say 50-100W) I don't have any. So it is okay (for me) just to adapt the topology (tho I prefer singleton input ala NAD3202 or QUAD303), then build it according to common engineering wisdoms (low THD or whatever). If it is good, then it should be better than NAIM or NBIP.
I don't believe that properly engineered amps will be judged worse (by majority) than an effect box. If it is so, then I believe the amps are not really well engineered.
Good spectrum of THD for example, can be applied to amplifiers with PPM distortion.
May be he knows what to look for, or he just try to use the topology.
My interest is to find out if this topology is really suitable for slow output stage (I have many good vintage slow NPNs). I don't have interest with using fast transistor like C5200 (coz I have them in pairs) unless this is the right topology for such transistors.
I recall Bigun mentioning that from building a JLH69 he didn't understand the praise for the amp. But Bigun used slow device!! Bigun should have used the C5200 in the JLH. JLH69 is old, but it was a design that is way modern above its era. Many JLH builders were building worse amplifier than what JohnLH built. We should make a better one because we have more options and tools than JohnLH, not a worse one.
And so with the Naim clone! 😀
DD,
The JLH has high THD, and up to about 10W most is H2 and beyond it is H3.
If you want to 'engineer' an amplifier, maybe you should amp for less than 0.005%. The fact is by THD references the JLH quasi is an 'effects box', to restate your colourful (sic) phrase!
The real question should be:
'Why is an effects box sound so good, and why do so many people like it? Are they cloth-eared gits? Is the JLH an effects box, or something else? Why?'
HD
The JLH has high THD, and up to about 10W most is H2 and beyond it is H3.
If you want to 'engineer' an amplifier, maybe you should amp for less than 0.005%. The fact is by THD references the JLH quasi is an 'effects box', to restate your colourful (sic) phrase!
The real question should be:
'Why is an effects box sound so good, and why do so many people like it? Are they cloth-eared gits? Is the JLH an effects box, or something else? Why?'
HD
- Status
- Not open for further replies.