That is a really horrid design and for audio work it really isn't going to do at all. You may be able to understand a voice but it will be very hard to listen to.
No way this schematics is going to make 100 to 300W.
The first improvement is use the half bridge output stage instead of the single ended.
If you need a simple class D amplifier which really works, there is a bunch of single IC solutions from TI, NXP, ST, Freescale. If this is about the minimalist contest, a class D amplifier can be built using just one FET and few resistors and caps.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
That design is only suitable for a small demonstration amplifier to show the principle to students. In addition to all the other things which are wrong with it, it will radiate large amounts of radio frequency interference and must not be used for P.A.
--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
A basic oscillator circuit with hipass RC-RC-RC in the feedback from drain to gate. The speaker is the load in the drain. The hint is that at high frequencies the speaker inductance makes the resonance with the FET stray capacitance. The typical parameters make the resonance exactly where it is needed. So, the whole thing oscillates in the nonlinear mode, and the FET is switching. If the audio frequency is applied to the gate, that pulse width modulates the drain current. There is also the bias circuit to keep the operating point.
As simple as that: FET, speaker, 4 resistors and 3 caps.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
Here's a fairly minimal circuit, a lot more parts than under discussion but a lot fewer actives than anything using an IC. Class DA, the circuit itself has good efficiency but the output network ruins things. Excellent linearity though -- less than 0.1% THD!
There are plenty of examples of awful designs in both valves and transistors, but that doesn't mean one is worse than another.
If the final smoothing capacitor of the PSU were made from a pair of capacitors in series and output transformer primary were returned to their centre point instead of earth, that problem would be solved. (The
10uf 350v capacitor between the choke and the O/P primary would then be unnecessary, so there would be no increase in component count.)
If negative feedback were necessary, it could be added from the loudspeaker terminals or from a tertiary winding on the O/P transformer, tightly coupled to the output winding. There is nothing in this design to prevent NFB being added, but perhaps the linearity of the ramp generator was good enough for that particular application without it.
--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
Come to think of it, I don't know why I didn't do it like that. Oh well, supply ripple aside, it's three point five watts of some of the cleanest single-ended tube audio you can get, and with unmatched bass response. The plate output impedance is around 100 ohms, so it's quite easy to saturate the transformer at LF. Not to mention the entire amplifier is DC coupled up to the transformer, an unusual feature for a tube amp.
The ramp generator is just an RC hysteresis oscillator -- nothing particularly clean about it, just using a relatively small voltage range (~10% of the supply). Still a recognizable amount of exponential curl to it. Maybe it's just enough to cancel delay and slew in the later stages?
Full page on the project is here,
formatting link
I'm planning a larger PP design, but its linearity is probably going to suck a whole lot more because it will run class D-AB. These tube comparators don't like to turn on and off gracefully, as can be seen on the ends of the transfer curves above. Global NFB may be required to keep distortion within useful limits. Even if distortion is worse, power output will be unmatched: over 150W from a pair of sweep tubes, without even breaking a sweat! Do that in class AB and you'll be pushing red plates.
Much better to completely ignore the 555 altogether. The 555 isn't bad in and of itself, it's a fine switching device and oscillator building block. However, using pin 5 to control it is essentially stupid. You can linearize it, principally by controlling charge and discharge currents, but that's extra circuit. You can also use two, one as a clock generator and the other as a variable delay, but that's still more circuitry. You might as well start over with another general-purpose 8-pin chip, like so:
formatting link
Scrunched together on a bit of perfboard, this circuit hardly occupies a square inch, it's quite easy to build. Couple the logic-level signal into whatever output stage you want.
Or for that matter, if you're adventerous, you can use the LM311's beefier output stage directly for maybe 1W total output. ;-)
There are plenty of examples of awful designs in both valves and transistors, but that doesn't mean one is worse than another.
If negative feedback were necessary, it could be added from the loudspeaker terminals or from a tertiary winding on the O/P transformer, tightly coupled to the output winding. There is nothing in this design to prevent NFB being added, but perhaps the linearity of the ramp generator was good enough for that particular application without it.
--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
I cancelled that post because I realised that, even with a split capacitor, there would still be modulation hum unless a feedback loop were used. You must have read it before I cancelled it.
A stabilised supply would be the real answer, but it spoils the neatness of the original idea.
--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.