The New Radio Log

#2

--- One of the circuits I saw in SSD that seemed really strange at first is the Class C amplifier. I have done a bit reading in the past on amplifiers but I didn't see class C amps discussed much. Even The AoE barely mentions the m. I tried simulating this thing on LTSpice but I get a collector current o f a few Tera Amps. Still looking on Digikey for transistors with enough pow er rating to accommodate :P

Going from there, I am currently looking into the universal QRP transmitter (3rd printing, page 26). I tried simulating this thing too but I can't get oscillations going on LTSpice if I connect the class C amplifier output ci rcuit. If I disconnect the coupling transformer and just take output from t he oscillator tuned tank, oscillations start happening again. My guess is t hat class C amplifier loads the tank too much so I'm not very confident rea lity will be much different when I build the circuit. Of course it could be that my crystal model is not adequate. Well, we'll see!

Those things should work just fine in SPICE, you're probably just missing subtleties about simulation, or reality itself. Example: oscillators don't start from DC, you have to give them a push. In reality, there's noise to perturb it; in SPICE, it's perfectly happy sitting on the quiescent operating point, even if it's an unstable point. Easiest way is usually setting initial conditions to "set to zero", or setting an initial voltage in a capacitor or current in an inductor, or adding a voltage source (such as VPULSE or noise) of a few mV or uV to the loop.

Likewise, the models are just that, .MODELs: no parasitics whatsoever, just a model of the die itself (and even then, usually a pretty crude one unless you sign NDAs to get the big foundry BSIM3's and such). Die and terminal capacitances are often modeled poorly, and resistances and inductances are modeled as lumped elements (if they are modeled at all; power MOSFETs with .SUBCKTs usually include these at least to some level, but .MODELs have only the R and C elements in the internal model, no L).

As well, schematics vary in their representation of current flow and reactive components; the most obvious disconnect being, attempting to schematically represent something like a cavity amplifier. Even a voltage source is rarely so; the parasitic inductance (or more accurately, distributed transmission lines) and bypass caps which characterize a power distribution network guarantee nonzero supply impedance, whereas a SPICE VSRC is God and utterly absolute. (Likewise, an ISRC is perfectly happy delivering negative voltage, something which regularly catches newbies using them.)

Tim

Please post the .asc files, so we can see what's going on.

Thanks. I tried adding some noise or a pulse generator with no luck. Then I realized I was using the plain vanilla NPN model. I changed that to the 2N

I can only guess here but probably the parasitics in transistor model provi des the feedback necessary for oscillation (because I can't see any other f orm of feedback). Using the old model, I tried adding a small cap across th e base-emitter but that didn't help.

I'll be spend>

Here it is. It's sorta working now.

I realized I was using the plain vanilla NPN model. I changed that to the

vides the feedback necessary for oscillation (because I can't see any other form of feedback). Using the old model, I tried adding a small cap across the base-emitter but that didn't help.

t a bit more and I have some confidence it will work once I build it.

sc?dl=0

It also turns out that if I use the plain NPN model but add a 1 pF across t he Base-Collector I can get oscillations starting relatively quickly! I gue ss that's what's missing when I use the standard model (among other things)

In Class C, the collector current only flows at the positive top of a large input voltage/current. At zero or negative input, no collector current will flow. This also means that feeding it with a very small AC waveform, no output will be created.

Especially bipolar amplifiers can draw a lot of base current when conducting and practically nothing when not conducting. Feeding a Class C stage with a small waveform, the stage doesn't conduct and neither does it draw base current, so no loading on the signal source. With a large input voltage, the base current flows on part of the positive half cycle, changing the input impedance radically.

This may also complicate your simulations.

The change of amplifier input impedance depending on input voltage may explain something. Try inserting an attenuation pad (say 6 dB) between signal source and amplifier.

Note also that a Class C transmitter will need a tank circuit (tuned circuit) to distribute the energy over the whole RF-cycle and not just the positive peak of the waveform.

Due to the tank circuit requirement, it is hard to make broad band amplifiers running in Class C. In addition, Class C transmitter amplifiers are usable for constant amplitude modes, such as CW and FM, but not for AM and SSB.

Except by modulating the power supply.

John

Google for 'EER amplifier'.

Won't help. A Class C amp is in cutoff for small input signals.

Cheers

Phil Hobbs

#3

---

I haven't been able to get back to my radios for some time now. Lately I ha ve been doing a good amount of reading on various topics. I have been readi ng on tubes. I understand that I may never ever touch a vacuum tube in my l ife, but it always bothered me to look at tube circuits without understandi ng what I'm looking at.

It turned out that tubes are a lot simpler than I thought. At least from th e outside. Tubes in many ways behave like depletion mode FETs with some sub tle differences. Of course things get complicated when you start looking at tetrodes, pentodes, etc with all the biasing requirements and complication s. But in essence, if I look at a tube schematic in one of the nice older b ooks I can probably have some understanding of what a circuit does.

I have one such a book that I picked up at Goodwill. It's called "Electroni c Communication" by Robert Shrader, 4th edition. The book looks promising a nd I have been wanting to delve into it a bit deeper but the tube circuits have been putting me off. I think now I should give it another try.

Progress has been made!

Tubes are surprisingly good at stuff like radio. The performance specs aren't anywhere near close anymore, but they were fantastic at the time, when the technology, tools and needs matched.

I've been playing with this lately:

It's useful to think of tubes as a cascade of virtual diodes. Between any two electrodes, there's the potential of an outer electrode, and the current arising from the inner one.

In a diode, that's just the anode (plate) and cathode.

In a triode, the current from the cathode is controlled by the grid voltage. The current passing through the grid, however, is controlled by the plate voltage.

In a tetrode, the current from the cathode is controlled by the grid. The current passing through the grid is controlled by the screen. The current passing through the screen is controlled by the plate.

And so on.

It's important to note that, electrodes more negative than the cathode, will experience very little current flow (just leakage), while electrodes more positive can intercept current from the beam.

When an electrode pair acts to oppose current flow, the electron beam is absorbed by the nearest, most positive electrode.

A useful example of this is using the suppressor (grid 3) as a single-input differential: when positive, current is allowed to the plate; when negative, the plate is cut off, and all the current either enters the screen (beware screen dissipation!), or is turned back to the cathode (where it remains in equilibrium with the surface of the cathode itself, called space charge).

In a pentagrid (heptode) tube, there are two control grids (g1 and g3), two screen grids (one following each control grid -- g2 and g4), a final suppressor (g5, usually tied to the cathode) and the plate. Therefore, you can figure that g1 controls cathode current, period; g3 controls g4 and plate current (in a positive sense) and g2 current (in an opposing sense).

Tim

That is some cool stuff. Thanks for sharing. Yes Id like to get my hands on them one day.

I'm afraid to ask how long it took you to finish all the construction? It would've taken me a year. Also what is the maximum voltage you have to deal with?

Thanks. Only a few days total labor. The hard part is planning it out so it looks *and* works good, without being so cramped that you can't change things later or anything.

The DC-DC converter makes +100V and +6V, one of these,

Tim

Very nice, thanks for posting. Just curious why you chose to regulate the 100V rather than the filament supply?

piglet

More critical (at least as far as ripple rejection, if the 12V is dirty), but I don't think that was strictly the final circuit either... think I ended up with joint regulation. Otherwise, you need a 10k 1W preload on the high voltage to get it to even start up.. :^)

Efficiency is reasonable, over 70% I think. With various tweaks, shouldn't be a problem to get to 90% or more.

Tim

#4

--- A while back I wanted to construct the direct conversion receiver radio fro m the first chapter of "Experimental Methods". I had to think long and hard about mounting the huge variable capacitor required for tuning. I don't ha ve the small 30pF and 50pF that the design calls for so I'm using the 365pF caps that I got a few weeks ago from tubesandmore.com with some padding.

This led me to look for enclosures and such and I found aluminum relatively cheap at home depot, so I bought a 3ftx3ft sheet. When I got home, I reali zed I have no idea how to shape it into something. After some search it tur ns out Harbor Freight sells this work of wonder for bending metals.

Both 18" and 30" are available. I tried it on a little aluminum piece and t his was the result:

I was also motivated to figure out how to assemble my dremel press that I b ought 2 years ago and never used. The instructions were terrible and someti mes wrong. After a few agonizing hours and some youtube checking, I figured it out. I used my dremel to drill some holes in aluminum and copper clad b oards and results were good.

So back to the receiver. I spent a few hours this morning to assemble the d arn thing. It's based on a NE602 chip. This is the final result

Needless to say: it doesn't work