Time to start talking more about the amp now that the protos seem to be remaining stable.
Remember that the design goals were to create a reasonably priced, but good stat amp. Part of the original desire was to not require hard-to-get components and/or expensive iron. Well, it's hard to know when components will become hard to get, so you do the best you can with what there is currently around and what seems like it will be around for a while.
We also wanted easy-to-build and easy-to-bring-up for a moderately experience builder.
The goal became simple-but-good.
When this particular design concept came to thought I scoured the web for anything that I could find that would be similar or the same. I was not able to find anything using the keywords that made sense to me. But, the web is a big place and there are plenty of good designers in the world so I haven't really got a clue if there is an amp like this somewhere else. And, of course, I am not privvy to the commercial amps, so who knows. I'm sure that you all will not hesitate ....
Many of the stat amps that I have seen use three stages. Since stat amps are balanced they use a differential front end, an intermediate gain stage, and then an output stage. Sometimes the output stage is coupled to follower to reduce the Zo of the amp so that it can push more current easily. This also increases the speed in most cases.
This amp only uses two stages to simplify the topology. Andreas Rauenbuehler, however, has a nice two stage hybrid amp
here. His transistor amp is a three stage amp as is the KGSS (although it could be argued that the diff amp and current tunnel are a single unit), for example.
All of these amps are good stat amps each with different qualities. And since I haven't heard any of them I can't say how they sound, but others can comment on this.
To make a two stage amp we only have two things to work with: the input stage and the output stage (followers are part of the output stage). So how do we do this?
Well, we know that the front end will be a diff amp and that the diff amp must translate the signal down to the bottom rail. The bottom rail because HV N-type devices are easier to find than P-types and so the output device will be at the bottom rail. Fortunately for us, we have p-type devices that can make the diff amp so that their collectors/drains point down and not up.
A classic way to handle high voltage with solid state devices is to use cascoding. Cascoding was first developed in tube topologies to generate high gain, but since tubes are usually HV devices they were not often used for HV protection. In sand circuits, however, cascodes are very handy for using an HV device to protect some LV devices that do the actual work.
Andreas uses a circuit with pfets and HV bjts. I did not know about this when I suggested the same idea in the previous thread, but really, there is hardly anything new in circuit design. Much of what we do today is assembling basic building blocks in ways that suit our purposes. Nonetheless, pfets with cascoded HV bjts is a no brainer for the front end of a simple-but-good stat amp.
Now that we have the signal at the bottom rail we have to reflect it back up to the top rail. One way is to provide a high impedance load for the diff amp and then to drive the grid/base of the output device. But this scheme has other drawbacks.
The most simple way that I can think of to reflect the diff amp off the bottom rail is to use a current mirror. And since we need a cascoding effect we can use a Wilson current mirror. Wilson mirrors have also been around for a long time, we're just coupling one to the front end.
And, that's about it for the amp topology. Schematic will be in the next post.
Current Sources
There was a great temptation to use current sources in this amp. The first, not-so-good design, used them. But after some thought, it became clear that for simple-but-good, we really don't need current sources.
In this amp, for example, the tail of the differential amplifer is connected to the positive rail which is sitting at 300V. If the front end runs at a total of 2mA the resistor at the tail can be 150k. And while this is not the 1G dynamic resistance of a current source, it is plenty good enough for the operation of this amplifier.
Same is true on the output followers. It is tempting to load them with CCSs, but it isn't really necessary to do that to get good performance.
Not having CCSs really simplifies the amp. Hence, this amp has no CCCs. In fact the amp has no caps either because it does not have huge open loop gain which, when tamed by NFB, leads to possible instability. It is stable without caps while still having a gain around 1000.
The tradeoff is that at the limits of its excursion it will evidence more distortion than an amp with high NFB and high OL gain. OTOH, high NFB has its own drawbacks to the sound of an amp so, for simple-but-good, this was an easy trade to make.
Next post, schematic.
