Hi Alan

I did finally build one of these receivers and got it working on new year's eve, which I took as a positive sign.

Overall, when I can tune it into a station, the sound quality is very good.

However, compared to the very simple super-regenerative reciever that I had previously built, it is very difficult to tune and does seem to need a well positioned arial.

I am trying to figure out how to improve it.

When I tried to look at what was going on with my oscilloscope, it seemed that the IF was working at about 100 KHz, which seems odd. It also does not look like it is really being driven fully into limiting.

All of this is so new to me and there are huge gaps in my knowledge, so I am not quite sure how to proceed.

I will try rebuilding it more carefully and maybe look at using a varicap for the oscillator. I did use a J310 for the mixer and wondered about including an RF stage or trying a design based on the SA602. So many choices and decisions to make.

Cheers

Peter

Hello Alan,

Great project by the way!

You mentioned about debugging using oscilloscope. I'm working on same and what I don't get is the scope bandwidth required for these kind of experiments. I can see a 10MHz scope in your video, but ya, the IF in this project is in terms of few hundred KHz only. But my question is what will be the minimum analog scope bandwidth required for a 10.7MHz IF project?

Planning to buy a one :-)

Peter,

Glad you like my website mate.

Pulse-counting FM detection basically works by converting a limited FM signal into a PWM signal which is then low-pass filtered to recover the modulation.

First the IF signal is squared off (limited) to a constant-amplitude, basically 50% duty cycle pulse train. Then one of the edges of this signal is used to trigger a monostable which produces a pulse of constant length (shorter than one cycle of IF). When the input frequency is constant the duty cycle of this signal is also constant and once low-pass filtered you get a DC-voltage. When the IF increases the triggering happens a bit more often on average which means the low-pass filtered voltage rises (the PWM duty cycle increases). Similarly when the IF drops in frequency the rate of triggering drops and the voltage after the LPF follows. Effectively it averages (counts) the number of IF cycles over some time (the LPF cut-off frequency).

I'd recommend you build the whole thing point-to-point over a ground plane. I only did the VHF bits that way, but the IF has lots of gain, so it can't hurt to be a bit careful with it too. The layout can be a bit critical, but it shouldn't be too difficult to get going, especially if you have an oscilloscope to help debug it.

I've been long interested in building vacuum devices, but I haven't invested in the vacuum system to really experiment with it yet.

Regards,

Alan

Hi Alan,

I think your website is great. I teach science and have spent hours looking at your different projects.

I want to try building this pulse counting fm receiver. I still don't understand how the pulse counting works though. Is there anywhere I could read more about it?

You said that when you built the circuit on overboard a lot of time was spent on rf hygiene. I can see thar you built the mixer and oscillator on copper clad board. Was there anything else you did

or would recomend doing in order to achieve good stability. Cheers and keep up the good work!

PS. have you tried any home made CRT projects, or near space weather balloons. They seem like the sort of thing that you might also be interested in. Peter H

Hi Alan,

thank you for your design.

It seems that your separate mixer is better than an autodyne because it would then seem difficult to set the transistor polarisation to get both stable oscillation AND efficient mixing.

However, your IF amplifier does not do much filtering (besides the natural transistor frequency limitation that seems too high)... This seems important, because this is what rejects signals that are not in the chosen frequency channel.

Of course, in such a design, the VHF frequency channel is shifted down almost to base-band; therefore the traditional IF band-pass filtering is replaced by a much simpler IF low-pass filtering.

It is not possible to catch up on this after FM demodulation which is non-linear.

A simple way to get this LPF is to add a capacitor from collector to ground (or to base, using the Miller effect) on each IF stage, hence adding a pole for each one.

This would also improve the overall IF amplifier stability.

Regards,

Alexandre.

Petr,

Yeah the gate capacitance of the JFET becomes part of the tuning capacitance of the antenna circuit (limiting the max frequency with a specific inductance). The LO signal injected into the source leaks into the gate circuit via the device capacitances and is well within its resonance so grows quite strong. The Miller effect isn't a huge deal for this mixer as the output is only a few hundred kHz and we really don't want RF on the drain at all. I'd have to simulate how much effect is has on the sensitivity, but as long as there is some mixing going on there will be a difference signal at the drain.

A cascode arrangement of two JFETs would reduce the reverse coupling and capacitive feedback effects improving the stage performance over the current single JFET. I use it in some of my other projects and it works well. It is still unbalanced, the LO passes straight though into the IF port with quite a bit of gain (lower LO drive is required - a feature too), but for a device like this where the IF frequency is so vastly removed from the LO and RF frequencies the lack of LO suppression isn't a big deal.

Regards,

Alan

hello Alan!

about the mixer input C: shouldn't that be mitigated since the gate is a part of the input tuned circuit? but what about reverse capacitance then, I guess that would still increase capacitive loading due to Miller effect in CS stage, or..? I'm asking this bc I don't think it's ever covered in textbooks and I figured you might know. thanks.

Soumya,

Yeah I probably should have used a cascode front-end with better JFETs. The 2N5484 is a fine VHF mixer but its input capacitance and noise figure are not ideal for this application... If I built it again I'd use the headphone lead as the antenna and use hotter FETs with a cascode amp+mixer front end.

One crazy idea for the minimalists is to build an autodyne converter-style mixer + oscillator front-end. It would be deaf as a post, but simple.

Sensitivity wise it is a bit deaf, but I live in a strong signal area so that isn't a huge issue for me. I should measure it formally, but I'd need to shield it and couple to the loop antenna in a meaningful way which is a bit non-trivial. Maybe a helmholtz pair with the inductance tuned out - tough at 3 metres VHF for a device large enough to bathe the antenna is a uniform field and not interact with it enormously.

Regards,

Alan

Great work mate, but I don't get why you haven't used J310 itself for mixer part and how was the sensitivity of this receiver?

Claudz,

You know where I live. You are welcome any time.

Bring a hack, or something to hack.

Regards,

Alan

OMG, you are AWESOME.

I totez need to see your lab. You are more than welcome to say stuff and explain how it doesn't work. I will pretend I get it but I wanna play with all this stuff.

SO COOL.

oh, I didn't notice that, RC tau looked fine at first glance. :) now it'd seem that the circuit was designed with ±50kHz deviated signal in mind. or at least that's my theory, though I'm pretty sure non-standard FM radio broadcast deviation levels were used in some parts of the world...

given that the receiver uses 120kHz IF (IIRC) one would rather aim for roll off at about 210kHz. stereo BW is of course much higher than 180kHz, says Carson, but 180's still a reasonable compromise for local reception of standard ±75kHz deviated stereophonic broadcasts.

Petr,

Thanks for that schematic, interesting circuit... Interesting that I arrived at something vaguely similar all by myself with just spice and a hint about the pulse integrator topology...

That IF limiter simulates as 90 dB gain peaking at about 80-90 kHz... It is fairly beta specific to the BC547Cs specified, replacement with 2N3904s results in cut-off devices and no gain at all... At least in simulation, of course you can fix the bias... I'll build it and give it a try.

The proceeding RC filter simulates as rolling off before 100 kHz and having about 12 dB loss when you include the mixer output impedance and an estimate of the IF limiter input impedance. Not really sure I understand why it was designed like that?

I think I'll sit down with spice and design a better IF system. I'll still give this one a try though. With 90 dB of gain I can't imagine it is all that easy to tame either. It is simpler though, being direct coupled. They also manage to avoid the need for a final squaring stage before the common-base pulse integrator. Mine needed that for AM rejection when I first breadboarded it.

The audio filter seems like a bit of a hack, no worse than mine though.

The cascode front end and varicap tracked-tuning is something I was looking at doing myself, but I wanted this first prototype to be as simple as possible.

Regards,

Alan

About the FM-IF limiter part... wouldn't you be better off with something like this?

http://img529.imageshack.us/img529/6588/schc.jpg

GNFB should rid the amp stages of any instability.