I am curious is there is a schematic for an amp where you can swap output devices to determine if they produce a good real world result. In my head I imagine a simple single ended amp where you can pull out one mosfet, bjt, etc. put in another one. I am sure that you would need a robust resistor decade box or two to make this work but would you really need anything else? I don't want to reinvent the wheel so if there is something out there that does this already, I would love to know about it.
Most audio circuits are tolerant of different devices. As long as they aren't obviously exceeding some parameter such as voltage or current you find that you can swap with the most unlikely of devices.
Take the generic blameless class B. It would work with BD/MJE or even 2N2955 devices for the input LTP. It would also work with those for the outputs too within their limits.
You could even replace the outputs with signal types such as BC547/557's as long as you didn't exceed their ratings. The majority of circuits would "work" just the same.
That said, for optimum performance circuits are designed with specific devices in mind.
Never be afraid to experiment though.
Take the generic blameless class B. It would work with BD/MJE or even 2N2955 devices for the input LTP. It would also work with those for the outputs too within their limits.
You could even replace the outputs with signal types such as BC547/557's as long as you didn't exceed their ratings. The majority of circuits would "work" just the same.
That said, for optimum performance circuits are designed with specific devices in mind.
Never be afraid to experiment though.
I would think the power requirements could be dealt with using an adjustable power supply. I wouldn't want to swap devices while the unit was on. I would just want to be able to pop one out, pop the next in, setup the circuit and power supply, check everything with o-scope, and lastly hook up speakers to listen to it. I would imagine that the front end would have to be rather robust to handle the large range of devices. I guess you could just make a single stage amp but somehow I think that might actually be harder to have the adjustability needed for this idea.
That would work for most cases as long as you observe the device ratings.
On a practical level most circuits are designed so that the actual device characteristics play a relatively small part (notice the word most... not all) and the circuit design itself determines the basic performance to a very large extent.
The Circlophone is such an amplifier: you can simply drop in any active device, and the bias servo will take care of everything.
You do not need to modify any component, although realistic values will certainly help if optimum performances are desired: a power pentode is not exactly like a germanium transistor.
You do not even need to readjust the bias setting: there isn't one.
Here is the darlington version:
http://www.diyaudio.com/forums/solid-state/189599-my-little-cheap-circlophone-2.html#post2585337
And here is the MOS version:
http://www.diyaudio.com/forums/solid-state/189599-my-little-cheap-circlophone-7.html#post2593162
You could even mix the technologies, if you felt so inclined: why not try a PNP Ge device with a NMOS for example.
Some Japanese amps in the seventies experimented the mix of Si with Ge, with great pains. You could do the same and even go further without any difficulty
The linear A version of SRJLH is also such an amplifier.
And for much the same reason, realtime bias servo...
Mismatched, SI, Ge, MOSFET, IGBT, it wouldn't care.
Current sensing networks becomes your virtual output
devices. Active devices still do the bitchwork, but they
are not making current running away or crossing poorly
decisions. Matching active devices becomes irrelevant.
Beware my square law Schottky AB variant remains unstable.
Some have built and say it worked, some reported oscillation.
You want AB? Spend the extra transistors and go with Elvee's
Circlophone. Proven stability is worth it.
If linear class A is your thing, SRJLH might satisfy with
a lot fewer parts...
If the device under test cuts the mustard and can keep up with
the sensing network, one device should sound pretty much like
any other. The sensing network as virtual output devices, might
add a bit of flavor, but thats not the intent.
Any amp that can handle mismatched active devices is likely
gonna kill any flavor the active devices might have of their own.
Rolling becomes a simple go or no-go, with not much inbetween.
And for much the same reason, realtime bias servo...
Mismatched, SI, Ge, MOSFET, IGBT, it wouldn't care.
Current sensing networks becomes your virtual output
devices. Active devices still do the bitchwork, but they
are not making current running away or crossing poorly
decisions. Matching active devices becomes irrelevant.
Beware my square law Schottky AB variant remains unstable.
Some have built and say it worked, some reported oscillation.
You want AB? Spend the extra transistors and go with Elvee's
Circlophone. Proven stability is worth it.
If linear class A is your thing, SRJLH might satisfy with
a lot fewer parts...
If the device under test cuts the mustard and can keep up with
the sensing network, one device should sound pretty much like
any other. The sensing network as virtual output devices, might
add a bit of flavor, but thats not the intent.
Any amp that can handle mismatched active devices is likely
gonna kill any flavor the active devices might have of their own.
Rolling becomes a simple go or no-go, with not much inbetween.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.