Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
I think it's a 'Red Herring' to look at the part numbers in this case.
An LTP without degeneration resistors running at <1mA with high value collector load resistor has high gain allowing more loop feedback and this has been a popular approach. It also has a slightly different sound signature as a result when compared with designs using emitter degeneration.
Over time it was discovered that certain distortion mechanisms could be addressed by introducing emitter degeneration to the LTP with higher current. I think it also helps reduce LTP device matching requirements slightly. The degeneration improves linearity of the LTP devices but the higher current means a smaller collector load resistor and this lowers the gain and hence the loop feedback factor. In general, it's advisable to include emitter degeneration but if you're making a clone you must leave them out or you are not going to have the same sound - regardless of transistor type. A compromise might be to introduce small value degeneration, such as 10R. If I remember correctly, this was my approach on some early projects such as my TGM and TGM1 amplifiers.
An LTP without degeneration resistors running at <1mA with high value collector load resistor has high gain allowing more loop feedback and this has been a popular approach. It also has a slightly different sound signature as a result when compared with designs using emitter degeneration.
Over time it was discovered that certain distortion mechanisms could be addressed by introducing emitter degeneration to the LTP with higher current. I think it also helps reduce LTP device matching requirements slightly. The degeneration improves linearity of the LTP devices but the higher current means a smaller collector load resistor and this lowers the gain and hence the loop feedback factor. In general, it's advisable to include emitter degeneration but if you're making a clone you must leave them out or you are not going to have the same sound - regardless of transistor type. A compromise might be to introduce small value degeneration, such as 10R. If I remember correctly, this was my approach on some early projects such as my TGM and TGM1 amplifiers.
Last edited:
There is debate. If you add emitter resistors it widens the linearity window of the LTP but also reduces loop gain and loop feedback.
https://www.diyaudio.com/community/...tead-of-emitter-resistors.274117/post-4321046
There are other considerations like dc offset and system stability. I would recommend omitting the resistors and matching the transistors.
https://www.diyaudio.com/community/...tead-of-emitter-resistors.274117/post-4321046
There are other considerations like dc offset and system stability. I would recommend omitting the resistors and matching the transistors.
Member
Joined 2009
Paid Member
I am not sure, but a diode in the rail is likely not the same as a lower voltage transformer without a diode when it comes to noise and clipping behaviour. I think it’s it‘s relatively easy to improve on the performance of the amp but then it’s not so clear whether you are preserving the sound. I have noticed that one of my first amplifiers, an AKSA clone, sounded quite different with a lower voltage power transformer. In my NAIM clone (you can search up the “TGM10” thread) the main thing I was willing to ’improve’ was the Vbe thermal control, which resulted in a very stable amplifier that didn’t depend on the physical enclosure etc. for proper behaviour. I have not yet boxed up that amp but will be looking carefully at proper choice of power supply.
"...comes to noise and clipping behaviour." , clipping is happen when close&open the diode, like in ac/dc convertion. Here, voltage are likely the same, not oscillating. I think there is no clipping. I beware the noise. But, real world there is no noise. I've still been listening. With/without zener sound looks like the same. and ZTX's are still very hot 
Member
Joined 2009
Paid Member
https://www.diyaudio.com/community/threads/tgm10-based-on-naim-by-julian-vereker.302454/
I'll ask the Mods to re-open my thread once I'm ready to finish up and box it.
If your ztx's will burn out, then you can try zeners. Only, front end is working with lower voltage. ZTX's cooler than 40v. ztx's driving (collector) voltages are nearly the same (1-2mv difference, 855&858mv). Also, idle current and idle output voltage changes little bit. Sound is the same up to now.
Sorry, I might have confused my rememberings... Here another record of a higher output stage: what about the claim that it makes the output act as a voltage regulator...?
Member
Joined 2009
Paid Member
AFAIK the idea is that the higher voltage is used for the front-end. This promotes linearity and means headroom is no issue. Because the front-end is low power the higher voltage doesn't bring much penalty in dissipation. The output stage is where all the heat is generated so you tend to limit the voltage to the design target. Of course, circuits can be designed without a higher voltage for the front end and the use of a capacitor based bootstrap gives the additional headroom instead. Most common is a single bootstrap on the collector of a single-ended VAS but others (e.g. Hugh Dean of AKSA fame) has used a capacitor bootstrap on both rails to ensure the front end can drive the MOSFET output stage to the rails despite several volts of gate threshold voltage.