No mystery about cascodes. It's simply a two stage amp with a grounded cathode stage driving a grounded grid second stage. Like every other multi-stage amp, start from the end and work backwards.
Here's how I designed a cascode. (Attached) Start with the second stage. In this case, assume a Vpp= 275Vdc, and play around with loadlines until an Rp= 68K gave the best compromise between THD estimate and output swing. (In this case, THD= 5.4%, which would be
barely acceptable. 6BQ7s are hard to design for, having no audio pretenses whatsoever, and really require a higher Vpk to linear up. However, in this case, that THD estimate applies to just the one stage, not both. As with the solid state version, cascoding improves overall linearity. Also, this particular cascode was half of an LTP which nulls out even order harmonics, and if the THD is mostly h2, then you lose most of it anyway.) Next, use the plate characteristic to estimate u and g(m). From that, calculate r(p). You need to know that in order to figure out what your r(k) is likely to be. r(k) then becomes the plate load for the grounded cathode stage.
Next, run a loadline for the input stage. This is always a good idea to see if the first stage has enough swing to swing the output to its theoretical maximum. Here, we're OK in that regard, and good to go. No need to move the input loadline by adding external parallel resistance.
Finally, figure out what your point-to-DC ground voltages are. Then you're done. Build it to see if it really works as designed. In this case, it did, and measured results were within 5% of design nominal values.
Quote:
Originally Posted by Skitch72
I have been just taking some stabs in the dark on component values and its funtional but the sound is very rolled off and muddy. B+ is about 115 volts.
Thoughts comments suggestions?
|
The A Number One bugaboo with cascodes is their
eeeeeeeeenormous output impedance. Combined with the capacitances of a load, this will not play well at higher frequencies. After completing this cascode design, connecting the o'scope probe directly to the output gave results that looked positively hideous. A 1000Hz square wave had very tilted tops, the measured f(h)= ~20KHz. It looked like the
slowest thing I'd ever built, either solid state or hollow state. This was, of course, an illusion caused by the capacitances of the scope probe, the connecting cable, and the Ci of the scope's vertical deflection amp.
Once connected to a cathode follower, the true picture emerged. The measured f(h)= 117KHz (-3.0db) and square waves had nice, flat tops with excellent rise and fall times, when the measurements were taken at the output of the cathode follower.
22nd September 2009, 10:20 PM
6021 Cascode
Similar Threads
