I believe the NAP 160 used a quasi-complementary NPN bipolar pair.
Technically MOSFETs are transconductance devices rather than voltage amplifiers.
Yes, originally BDY56s, then from 1982, improved with custom made NA002s.
Hi jan,
I read this differently than all the follow on replies… rather than seeming to assert that mosfet amps could not be made to sound good, I thought your story indicated that the sales guy was saying to make your choice by listening to music rather than deciding based on technology.
Maybe he truly did not know or care whether he was selling mosfet or bipolar amps, maybe he made his selection of brands based on what he heard, maybe he felt more confident that his customers would be happier in the long run if they chose something that sounded good, and if they thought about factors - like the (noisy) wireless home phone - that might compromise the sound rather than internal design choices from any given manufacturer.
I’ve been down both roads, of course, as have many in this thread I imagine. Cool new tech is cool…
But, I think it’s important to link your anecdote to the parallel discussion of the importance of the overall circuit design and engineering when building a DAC, as compared to the temptation to focus solely on the specific IC that does the conversion. This comes up routinely on the forum given Naim’s choice to buy up discontinued parts that lie at the heart of their overall design (even if they are beavering away to develop something entirely new based around a different part, or a completely different technological approach, in their lab).
I’ve read with interest the myriad pre-treatment tweaks needed for improving (especially) USB data input signal quality to get better sound output. I am definitely with the RME boys here, who note that it is not only possible, but a reasonable expectation, to design the DAC input circuit in such a way as to be immune to these effects, or indeed to sound the same irrespective of whether the input comes via USB or SPDIF, optical or coax. That they have achieved this in practice demonstrates the validity of their approach, and calls into question the implementations of other equipment that does not.
Your old dealer, with his “listen and decide” perspective had it right imho. We are buying things to listen to, not (just) to stick on a test bench.
Regards alan
2 Likes
Both bipolar and FET transistors can describe gain using transconductance (gm) , although the transconductance amplification curves have different properties between the two types of transistor.
The FET is essentially a voltage amplifier as the drain to source current, and therefore voltage across the drain-load resistor is controlled by the gate source voltage, there is very little gate source current, essentially this is just leakage current. Where as with bipolar the the gain is controlled by the base emitter current amplifying the collector emitter current. The collector emitter current through a bipolar transistor is proportional to the base emitter voltage, hence both are transconductance devices in an amplifier
But yes your post did prompt me to blow off the dust of my copy of Horowitz and Hill and University notes to make sure I still had it correct in my head and reminded me of the transconductance differences and similarities between bipolar and FET devices 
Sort of yes and sort of no.
In the bipolar case the device itself is a current amplifier (collector - emitter current is dependent on the base - emitter current); there is a relationship between the base - emitter voltage and the base emitter current, but the current in the collector - emitter flow is dependent on the current not the voltage (there may be correlation, but the relationship doesn’t cause the change in the emitter - collector current), so the device itself has a current to current relationship - i.e. a dimensionless gain.
In the case of the FET, the source - drain current is dependent on the field generated by the gate (which is a function of the gate voltage and, in essence, not related to the gate leakage current - although they may be correlated the relationship doesn’t cause the change in the source - drain current), so the device itself has a voltage to current relationship - i.e. transconductance.
In the case of a bipolar transistor you need to include an external impedance (typically the base resistor) to get a transconductance relationship across the resulting component network.
1 Like
That’s not actually correct, the base - emitter current is that of a forward biased diode - a logarithmic relationship, rather than proportional.
Exactly which is why as amplifiers they are both transconductance devices -though there are some pit falls with FET transconductance where as bipolar transistors tend to be better behaved.
I used to love my electronic engineering labs at Uni - and I still have and frequently refer to very well thumbed Horowitz and Hill - the electronic and digital design engineer’s bible… albeit not on bipolar transistor transconductance until your post
Good technical description and explanation of differences between transconductance between bipolar and FET devices in amplifiers in section 6.07 of Horowitz and Hill (published 1983)
Correct the relationship is logarithmically proportional - which is why they are effective amplifiers… as you see on any bipolar transistor gain chart. [Ic vs Vbe]
In respect of component networks, yes both are transconductance (but not when considering the devices in isolation).
The main reason for the use of power MOSFETs was their much greater usable frequency range. In the 80s and 90s this allowed for simpler and more stable designs compared to the much slower bipolar devices. Their resistance to thermal runaway is a secondary advantage. However as bipolar power transistors are now becoming a lot better, the case for power MOSFETs in audio use is rapidly disappearing (but still good for high current HF switching).
The NA009 is an excellent example of this - a high power bipolar device with good extended high frequency performance.
All of this is way above my head, but if it adds anything to the conversation, my Nagra amp (MOSFET) runs a lot hotter than my 250DR ever did. It has a built in fan, that blows hot air, which is the only downfall in my eyes.
But Xanthe - is that not being a little pedantic? - the devices don’t operate in isolation - they need to be setup and biased one way or the other even really basically or they likely would be damaged…? hence my original point of voltage amplifier vs current amplifier which is how ones starts describing the differences to students. Both have transconductance properties as devices.
Anyway I think we agree… I think…
Is it class A?
For Class AB, there is another property of power MOSFETs that running them at a much higher quiescent current than would be used (or useful) for bipolar devices has significant advantages in open loop gain linearity. This approach results in a much more ‘musical’ amplifier when used in a relatively low feedback design.
This is one of the principles I used for my later power amp design.
You’re the one who usually insists on not over simplifying!
There are limits 
I hadn’t noticed them previously! 
1 Like
It is class AB, this quote is from the manual
“OUTPUT STAGE
The MPA is equipped with a class AB push-pull output up to 250W maximum power per channel. The output stage uses low level of feedback including static and dynamic limitations. The advantage of such an output stage allows the MPA to accommodate a wide range of loads (capacitive, inductive, or varying resistance) without the risk of oscillation or intermodulation.”
Simon you need to quit while you’re behind 
Xanthe is absolutely correct.
You need to move out of the 80’s and experience the current (pun intended) MOSFET designs from Pass. There’s a reason why Nelson doesn’t use bi-polars anymore.
That explains the heat then.
To get it to operate well with low feedback you run MOSFET output stages at a very high quiescent current (high in comparison to bipolars that is). It’s possible to do this due to the negative temperature coefficient of the source drain current - if you tried the same trick with bipolar transistors it would work for a some milliseconds, and then thermal runaway would destroy the transistors!
(On the other hand the trick is both less effective and less necessary with many types of bipolar devices.)
1 Like
I believe we were both right although I state both devices are transconductance devices, and the Ebers-Moll equation and model for IC in a bipolar transistors is determined by VBE, hence transconductance, and in FETs ID is determined by VGS, again transconductance (all small signal values)
Horowitz and Hill is still in print in its third edition , and Paul Horowitz is Professor of Physics and Electrical Engineering, emeritus, Harvard University and it is his texts and ideas on transconductance in bipolar and FET devices I was referring to… so he might be interested to hear from you why his principles are now out of date. Technically both bipolar transistors (VBE/IC) and FETs exhibit transcoductance… and my simplification of bipolar transistors are current amplifiers and fETs are voltage amplifiers is a generalisation and yes doesn’t apply in all cases … It applies to many applications such as many amplifiers but would not apply to logarithmic converters, differential amplifiers and temperature compensators for example.
I was a fan of MOSFETs and JFETs when younger and designed some FET amplifiers using low noise opmamps I was rather proud of… although one did have stability issues, but thanks to in part the wonderful electronic engineer’s bible The Art of Electronics, of which Horowitz co wrote.
It was also my introduction into digital electronics design and build… invaluable and set a whole career.