whither regenerative amplifiers?

Hi John,

But the first wto ar not tunable. At least not without using sandpaper ;-)

Direct DSP would be prohibitively expensive at VHF. Probably you'd have to assemble an array the size of a large pizza and call the power company before turning it on.

Regards, Joerg

I think its possible it might become popular technology again. The advantage of high gain per cost is attractive, and it has other benefits. Stability is the prime issue: it can be controlled by low frequency quenching, which can of course be adaptive. The question now is whether a simple scheme can be devised that ensures consistent stability and gain with small quench control circuit cost. If solid stability is achieved, the rf emission problem is eliminated at the same time. If its achieved at minimal cost, it may be a winner.

I know I've seen it used in some modern receiver chips, but damned if I can remember the details.

NT

Although superhets were conceived early on, they were absurdly expensive. For most of the 30s the reaction set was still the prime configuration, superhets stayed in the luxury bracket for some while. Anyone that built a superhet in 1925 had to be rich.

NT

In article , Joerg wrote: [...]

Thats the easy part. First, you have to run the signal through an ADC. Doing an ADC that has very many bits is tricky at high frequencies. You would end up with a combination ADC and spaceheater.

How about a large copper tube as a wave guide and a plunger driven by an electric motor? You can get sharp peaks out of such things. The bad part is you never get just one peak. Perhaps using two or more you can arrange it so that only one peak lines up in both (all three).

No I mean positive. Lets take the simplest example:

-------------------------------- ! ! --- ! --- C1 ! R1 ! R2 ! \ ! ---/\/\/\------+------/\/\/\/----+---! K >----- +---- ! ! / E1 ! --- --- C2 ! GND

Feed this circuit into spice with

R1 = R2 = 1 Ohm C1 = C2 = 1 Farad

Try the circuit with K= 0.5 1 2 and 2.7 2.8 and 2.9

See what happens to the 0.03Hz to 0.3Hz band as you raise the gain and hence the positive feedback.

An engineer I know attempted a design like that and found that (A) 14 bits isn't enough and (B) 50 MSPS (his case) isn't enough either. Unless you put an analog filter infront of the ADC, a strong station dominates and the performance just isn't there.

You need to oversample enough that the Nyqist of the sampling system is well above the highest frequency you intend to receive. This way, a simple fixed anti-alias filter can keep the UHF band out of the front end.

The ADC's nonlinearity tends to intermix the various signals. Unless there are enough bits or you can assume something lucky about the "noise" of the other stations, the intermix products end up in the pass band.

No, he ended up doing an analog superhet. down to an IF he then digitized. The circuit was working great on the day the project was cancelled. It was great electronics attempting to implement something the physics said wouldn't work. He knew the idea was daft but he needed the work and they wouldn't listen to him.

No, the boss say it would. It was a classic E=MC^3 sort of project.

Why ?

If we are talking about amateur band systems only, using suitable high-Q front end band pass filtering (helical or cavity resonators), you can limit the bandwidth to a few hundred kHz (e.g. the SSB or satellite band). Then you only need a sampling frequency of the order of 1 MHz and undersample (decimate) it and process it at that sampling frequency.

Paul OH3LWR

In article , Paul Keinanen wrote: [...]

This method tends to increase noise unless something clever is done in the area of the sample and hold. The device that does the sampling, must be fast so there are a lot of sqrt(Hz) included. The sampling will alias the high frequency noise from the device down.

You are better off to mix down to a lower frequency using something that acts multiplying sinewaves so that there are no high frequency components to cause the mixing of high frequency components. You can make sample and hold circuits that act this way but you need more than one to make it work right.