scratch built am radio...LC frequency not making sense

[KTP]

Hi guys,

The lab kit in my circuits I course (we just finished the topic of diodes) included some 1N914 diodes so I decided to try and build a simple am radio with one of them (not part of the class or lab, just to explore tank circuits somewhat).

Anyway, I wound some 26 gauge magnet wire around a 1 inch pvc pipe, about 4 inches of windings with multiple taps.  With my "Almost All Digital Electronics" L/C Meter IIB, I measured the total inductance of my windings to be 256uH.  The middle tap is about 124uH. (I sort of lost count while winding  ).  Anyway, I hooked up the full length of the inductor across a 7-100pF variable cap with one end of the parallel tank circuit grounded and the other end connected to a 20 foot antenna and the cathode of the 1N914 diode.  I then fed the demodulated output into a inverting op-amp with a gain of 100 and driving a small 32 ohm speaker.

At first I did not get any station at all, but after trying several taps I picked up a decent reception using the middle tap (about 124uH) and around 23pF on the variable cap.  When the station indentified itself, I looked it up and found out it was AM 1380 broadcasting about 30 miles away (but I am in a deep valley).  Anyway, this didn't really make sense to me.  My calculations show that with a 124uH in parallel with 23pF my LC tank should resonate at close to 3Mhz???

Since I now knew that there was a strong station at 1380khz I tried the full inductor (256uH) across 52pF but got nothing but silence.

Ideas on what is going on here?  Is there possibly hundreds of picofarads of distributed parallel capacitance across my wound inductor?  Am I picking up some harmonic of the 1380khz broadcast, and if so, why can't I get the primary signal?

Thanks for any help

[Simon]

what else are you using to demodulate the AM ? you should have a capacitor after the diode as basically the AM demodulation process is to low pass the signal so that you get the varying amplitude of the audio smoothed out without the HF component. how have you physically built the circuit, I ask because my attempts at making a RIAA preamp (with more input and output leads all over the place than recommended) and ended up picking up some foreign radio station.

[KTP]

I have the diode coupled to the inverting op-amp resistor with a 0.1uF capacitor

Anyway, as a slight update, I found some 3/4 inch diameter by 1 inch long ferrite rods from a scrap flyback transformer.  I wound another inductor with MUCH fewer turns (dc resistance less than 0.5 ohms vs 2 ohms) and tried to space the windings out this time.  The air core inductance of this new inductor was only 24uH, but when I insert two of the ferrite slugs, it goes up to 176uH.  I hooked this into my circuit which still had a fixed value capacitance of two 33pf in series (so about 17pf) and I was able to pull up maybe the same radio station with a stronger signal with the slugs about half way in.  I haven't heard the call letters yet so I am not 100% positive it is the same station.

I can only guess that my tightly wound air core inductor with many windings had a lot of interwinding capacitance which resulted in a total distributed capacitance of over 100pF.  I guess.

Update, the station is the same, AM 1380.  I measured the inductance of my new, more ideal inductor and it was 117uH with the slug out to a position with the strongest reception of the station.  The fixed capacitance in my tank circuit is 16.5pF.  These numbers are not what I was expecting...they seem to indicate resonance would be at about 3mhz. :-(

[KTP]

Yay.  Ok, I figured it out I think.

With an inductance of 117uH I *knew* that my capacitance should be around 120pF to resonate at about 1.38Mhz, so I ignored the working 17pF capacitance and replaced the two 33 pf series caps with two 220pF series caps (so 110pF).  Moving the ferrite slug slightly and I pulled in the same AM 1380 station.  This time the inductance and capacitance make sense for that frequency.

So this is interesting.  I was picking up the third harmonic with the old values, and it was coming in almost as strong as the base frequency!  I would have thought the third harmonic would be significantly down in power.

Ah well, now I at least have a start on a circuit that makes sense to me.

[Simon]

apart from the coupling cap to the op amp you should have a cap from the diode output to ground, but then as a an op amps input is considered virtually ground sometimes I'm not sure if you have achieved the same result, hopefully someone else knows more on that

[KTP]

Yes, I realize I *should* have a low pass filter at the op-amp, but when I tried a 1nF cap to ground there it destroyed my signal.  I haven't fully researched why this should be so.  I have very little signal coming from the demodulator diode because my antenna length is not super long and I do not have any rf amplification before the demodulation.

I am reading up on the superhet circuit methodolgy now.  I wanted to start with a circuit I could totally understand and model before jumping into something of that complexity (although I guess most would consider the superheterodyne am receiver to still be simple stuff...

[Simon]

maybe you need a resistor in series with the diode, make the low pass frequency to be above audio and it will filter the audio out of the HF signal

[marianoapp]

it could be that the high output impedance of the tuner plus the 1nF capacitor are working as a low pass filter, but with a much lower cutoff frequency, and that is what is destroying your signal. You could do the math or just try add a 1pF capacitor and see what happens.

and of course you can have something else doing the low pass filtering, maybe the op-amp or even the speaker itself

[KTP]

No, it was my error...I had a misplaced wire.  I now have the signal from the diode going through a low pass filter comprised of a 1nF cap to ground and the 10Kohm input impedance of my inverting op-amp configuration.

I recently tried adding two stages of broadband RF filtering using MPF102 jfets.  While it did sort of work (I was able to faintly pick up the AM1380 radio signal on a 3 foot antenna that I had zero chance with using no amplification before the demodulator), it added a lot of buzz to the sound that I have not been able to get rid of.  Better would have been having each stage be a tuned amp with it's own adjustable LC, but that is a lot of winding and tuning...

[Strube09]

I am by no means a professional in RF but here are a few things.

1. Increase the antenna length and get it up as high as you can. The more energy you antenna can absorb the better the signal and easier the filtering will be.

2. Are you using earth ground on the other end of the self wound inductor? I would bring you complete circuit back to earth via a water pipe or copper rod into the ground.

3. A self wound inductor has quite a bit of stray/parasitic capacitance. With the old "Crystal Diode" setups most of the time the stray capacitance of these self wound inductors were enough to create your tank circuit. This is why it was so easy for soldiers to do during the war. Not capacitors needed, just wire, pencil lead and rusty razor blade. I have even seen where people were trying too hard to calculate the tank circuit all to be ruined as soon as they put the shunt capacitor in place.

Hope you get it figured out.

[KTP]

I had approached this wanting a bit more than just hit or miss.  I wound my 2nd inductor with pretty heavy gauge wire keeping space between each winding, and I used a (probably) ferrite core with a high permeability so that I would not need as many windings (thus cutting down on the resistance and interwinding capacitance.

These factors did help and allowed my calculated capacitance for a certain resonant frequency to be within a few pf of the actual value.

I recently tried a higher inductor and larger capacitor to try and resonant the circuit at 60khz.  I wanted to see if I could detect the atomic clock time signal.  I did get a 60khz wave but it was in bursts of about 10mS spacing, which makes me think it is some industrial arc welder somewhere in town or something.  I think the atomic clock signal is supposed to be constant amplitude except for a small time period at the start of each second where it is reduced.

[M84AB1]

I too have recently started playing around with ferrite rod antennas and am pretty much experiencing the same problem as KTP. I have successfully used tank circuits before although the inductors were made of iron powder toroids and the frequencies were 20MHz - 30MHz.

I have wound about 80 turns of 4mm magnetic wire on a MW band ferrite rod (9mm x 100mm) from jaycar resulting in 430uF as measured by an LCR meter. I too have this in parallel with the same variable capacitor ( 8.5 - 100pF) that I have successfully used before. I am using a 1N60 germanium diode followed by a 1nF decoupling capacitor (for demodulation). The signal is then fed via a 0.1uF coupling capacitor into a inverting op-amp setup. Keep in mind that I have used this very same setup for ASK transmissions without any problems. The only difference now is the ferrite rod inductor instead of the iron powder toroid.

Pretty much like KTP, regardless of how I rotate the capacitor I always only pick up one station. I know for a fact that my detector, demodulator and amplifier setup works as it has proven itself and so it comes down pretty much to the ferrite rod antenna.

These guys here seem to be using a second coil right next to the centre coil. I have seen people do this if one wishes to attach an external antenna, but that I am not interested in.

http://www.saunalahti.fi/hohtola/ham/active-ferrite-rod-antenna/active-ferrite-rod-antenna.html

[Zero999]

It sounds like the parasitic capacitance is reducing the resonant frequency.

I suggest you measure the resonant frequency of the coil by exciting it with a squarewave and looking at the ringing on an oscilloscope.

[M84AB1]

Ok i just excited the ferrite rod and in fact it is ringing. The frequency of the square wave is 90KHz. The ringing obviously occurs during the off period of the wave even though (for convenience) I have positioned it in the middle.

Pic 1 = This is what I get on the oscilloscope with a 25Kohm resistor in series with the ferrite rod inductor.

Pic 2 = This is without the resistor.

Pic 3 = That is my ferrite rod inductor with the two circuits I used to test the ringing.

Hmm, how do I now eliminate the parasitic capacitance.

[M84AB1]

errrrr works great for me. Does anyone else have problems seeing the pictures?

[DJPhil]

errrrr works great for me. Does anyone else have problems seeing the pictures?

I can't see anything at all, no links or pictures. I posted a screen capture where I highlighted the post.

I'm not sure what could be wrong though.

[M84AB1]

Ok, I have now attached the files instead of using the - - HTML code.

Hope you guys can see it now.

[DJPhil]

Ok, I have now attached the files instead of using the - - HTML code.

Hope you guys can see it now.

Yep, works for me. Thanks for taking the trouble, I'm following along.

I wish I could help with your problem, but RF is over my head for now. 

[Zero999]

That's exactly what you shoudl expect to see.

You now need to zoom in on the ringing and work out the frequency of the ringing. I'll try to estimate it from your photograph but it probably won't be accurate.

From zooming in on the computer each cycle takes up just over ¹/₅ of a division, say 0.22 divisions.

If the frequency is 90kHz then one cycle of the squarewave is 11.11µs and it takes 2.8 divisions so your 'scope must be set to 3.97µs/div which is an odd setting so either the 'scope is way off or the signal generator is.

Each cycle of the ringing takes 3.97µs×0.22 = 873ns so the coil's self resonant frequency must be ¹/_873ns = 1.15MHz which is obviously wrong as you managed to pick up 1.38MHz (the maximum frequency you can tune to will always be less than the self resonant frequency) but at least you now know how to do it.

One more question?

Did you use a compensated 'scope probe or just a piece of coaxial cable?

The 'scope probe needs to be compensated (you know the divide by 10 probes with an adjustment trimmer?) otherwise the capacitance of the cable will add to the LC circuit and reduce the resonant frequency.

Zoom in on the ringing, make sure the 'scope is reasonably accurate and the probe is properly adjusted and you should get a fairly accurate reading.

Another option is to build an oscillator with the coil and measure at the frequency generated.

Once you have the self resonant frequency, you can calculate the parasitic capacitance and use it in the tuning capacitance calculation.

[Mechatrommer]

good info! hope i can get back to this thread when i play with this LC RF thing.

[rf-loop]

for simple crystal detector radio there is lot of good information example here:
http://www.crystal-radio.eu/index.html


go to technical information after pictures on side and you find huge amount of practical and theoretical information. Specially for LC circuits and diodes.

Read carefully specially thing for diodes as detector and LC circuits and also note all what is written about Q.

-------------

For scope probe. Do NOT load parallel LC resonant circuit with scope probe.  It destroy Q value and it add also its capacitance to circuit and change it resonant freq. (also if you adjust your probe low freq risetime, it do not delete (eliminate) this capacitance! )

-------------

Also antenna is capacitor. Diode is capacitor. Between coil windings are capacitances. (old crystal radios coils are not  "designed for some kind of beauty".... there are real physics reasons, specially for get high Q value for LC circuit and also keep parasitic capacitances low as possible. (also for air coils with quite high inductance come problem with wire resistance. There is reasons why they use "litz" wire in history...and in history they make very superior radios (if think how simple they still be... coil, capacitor, diode, earphones or special louspeaker. What is secret of this quality. High quality high Q coil, good capacitor (also high Q), special diode for detector... most good was hand adjustable "crystal" where electrical parameters may be most good for this purpose.  (specialy some mild radioactive materials...nearly zero bias )

Today we have "forget" nearly all these basics for radio LC circuits becouse electronic have go different road with amplifiers and filters and today specially soft radios (SDR) where can forget nearly all RF design basics. (but still my one analog receiver is better... year of light forward of these "digital" stuffs... but this analog old radio is heavy... 20kg... and SDR is maybe 20g.)

Simple crystal radio. It looks like simple on the paper... one coil, one capacitor, diode and earphones. Yes it is simple to draw.
But if make working and specially if make good... it is all other but not simple. Making high quality coil... it is challenging.

If someone tell me that LC frequency is not as L and C value give, there is only one posibility: wrong calculations or wrong value for L or C or both or all these three. If measurement show there is also something wrong. Simple: there is (or maybe) wrong measurements (RF measurements are not simple or you measure you probe and not DUT (device under test)). Every millimeter of wire is resistange, inductance and capacitance. Also high frequencies do not travel in conductive wire as DC. More high and more current travel only near surface. (and there come just this mystery: where high freq travel... what to calculate. Why they are so clever in history that they use "litz" wire for used frequency area.
Why they use different diodes. Why some diode receiver without amplifier use battery. Some magic? NO, later some find that new diode types can use better with small forward biasing.

[M84AB1]

Thanks alot for the input guys. I sincerely appreciate it.

I have been playing around with the ferrite rod inductor and the 10-100pF ceramic capacitor tank circuit. I have excited it again with a square wave and according to my oscilloscope readings the LC tank circuit is ringing at about 500Khz. I am picking up two stations, one at 744Khz and the other around 1300Khz, but that is IT. I have two amplifications stages of 100 and 200 gain equaling to a total gain of 20,000. That should be enough to amplify at least a few of the stations in between the 744 and 1300Khz range. The 744Khz station appears to be the strongest and whilst I am listening to the second station at 1300Khz, I can still faintly hear the other one in the background.

Maybe the ceramic trimmer is not suitable for AM applications. I am using this trimmer

http://www.jaycar.com.au/productView.asp?ID=RV5722&keywords=trimmer&form=KEYWORD

Maybe I should use one of these? Ill buy one tomorrow and Ill give it a try, see if the capacitor might be the problem.

http://www.jaycar.com.au/productView.asp?ID=RV5728&keywords=trimmer&form=KEYWORD

I have reattached new oscilloscope screens of the ringing. The first image is at 5uS and the second is at 2uS on the time/dev on my oscilloscope. Like I said before, according to my reading the ringing is at 500Khz, yet I am picking up 700 and 1300Khz signals and nothing else. Any clues?

[rf-loop]

One "problem" with your receiver is not how much amplification you have! There need be something what amplify. Problem is that there are not (enough) signal what amplify after diode. Amplifying only noise is not so good idea.  What is reason. Many, but one clear problem over all others is diode you use and how you use it. This is normal silicon diode with very bad characteristic for this purpose. With very extremely low signals it do not work well. (Recifying in the square law region and your available signal level is really very low...after diode you have nothing for amplifier (but only some extremely high signal level station) I do not know your situation but with good simple crystal receiver you can listen on MW many stations specially at night time (MW propagation in night time is better). After this diod ("attenuator") most of available stations signals are far away under noise level imho).

One, but not very good, posibility is small (DC) forward biasing diode if use this kind of diode you have.
In LC circuit you have now several radio sations signal but they do not go to amplifier becouse this diode totally wrong type.

If you can then better is change diode type.

Quote from http://www.crystal-radio.eu/endiodes.htm

"Silicon diodes have the highest voltage drop (about 0.5 Volt) and are for this reason not very useable for crystal receivers.
Unless we use a small DC bias current, which brings the diode already a little bit in conduction.

Germanium diodes have a low voltage drop (about 0.1 - 0.2 Volt) and are often used in crystal receivers.
The properties like voltage drop and reverse current can vary a lot between two germanium diodes of the same type.
In practice we can best test several germanium diodes in our receiver and then choose the best.
The diode resistance RD of germanium diodes is most times rather low, and only useable in crystal receivers with a low Q (low sensitivity and low selectivity).
For high performance receivers, we can better use a suitable schottky diode.

Schottky diodes have a voltage drop of about 0.25 Volt.
The differences in properties between two diodes of the same type are often small.
Schottky diodes with the correct resistance RD are very useable in high quality crystal receivers."

Maybe somebody think... not so big difference 0.2V or 0.6V. But there are many other things also... lot of.
agen: carefully read this: http://www.crystal-radio.eu/endiodes.htm



After you solve diode problem you can make better LC circuit IF still need. (need look how to connect antenna to circuit and how take signal out from LC circuit so that not destroy Q (it affects to available signal level and tuning selectivity.) Simple crystal receiver (independent of how much amplify you have becouse signal is under noise level and signal detection is with level what you can get with antenna) need good LC circuit and good antenna.

[M84AB1]

rf-loop,

With the amplification part I meant, what if the signal is coming through but the station is much much further away than the two dominant ones and as such needed more amplification to be heard. I have however in the meantime worked out that, that is not the problem and it is mainly my tuned circuit.

The diode in my case is not the issue either as I am using a germanium diode (1N60) which are primarily made to be used as detectors. I appreciate your detailed reply about the different diode types and their associated forward voltages. I have successfully used this diode(s) in my RF detector project which detects HF, VHF and UHF.



The problem is definitely to do with the ferrite rod inductor, as I have today purchased a 15-160pf Tuning Capacitor and I am still getting the same problem. I have noticed that I often hear two station superimposed on each other. That indicates that my Q factor is suffering.  I am using the exact same detector circuit as the file I have attached except that my diode is a 1N60. With the Tuning Capacitor I should be able to cover the entire MW band yet I am still only picking up two - three stations where there is 20 commercial AM stations present here where I live. I guess the only way to isolate the ferrite rod as the problem is to buy a commercially prewound aerial ferrite rod like this one and see if that solves the problem.

http://www.jaycar.com.au/productView.asp?ID=LF1020

[Zad]

Don't forget though, even germanium diodes require 0.5v or so to operate. The oscillogram of the resonant circuit does show it dropping off pretty quickly though, so yes, you need higher Q if possible. Ideally you need a higher L and lower C, the circuit you described doesn't seem to have many windings in comparison to the ferrites I remember winding as a kid.
 

[rf-loop]

Sorry, I have not read carefully... start of topic there was:
"1N914 diodes so I decided to try and build a simple am radio with one of them"

so I have this in mind when I write...

Your diode type is not problem number one but also it is not best.. (if it is shottky 1N60... FV 0,3V @ 1mA  (sure there do not go 1mA... in this case need think how diode work with extremely low current. (before I give link where can find really huge amount of theoretical AND practical information about crystal radio..  also about high Q resonant circuits and also what is effect of loading this resonant circuit. (oscilloscope probe btw destroy nearly all... if want measure HF LC resdonant circuits it really need also tools... not common scope probe. (yes it can use but it need understant what is LC+Probe circuit together.

Maybe your system loaded Q go very low. (detectorr load resonant circuit, antenna also maybe make heavy load. For HF (MW) frequencies need guite lot of wire for coil.. so it is better use litz (RF resistance is more low and loss is more low so Q go more high.) Also capacitor need be high Q. Many small "pocket radio style" tuning capacitors are not.  One old solution is also (specially if can not use good capacitor or if need do "small" dimensions. (High Q 100...400pF capacitors are big... tens of years ago I use this permeability tuning. Fixed high Q capacitor and.... tuning with inductance. (you know... ferrite rod can move inside coil... yes it works also. (permeability tuning is not bad as we know also it is used in some professional receivers)

From old times I remember one 20cm ferrite rod and 80 turns coil... it was fun to listen some utility stations. (yes many of these use this time AM modulation)


It do not cost much of if you try forward biasing this diode...  (this is classic "trick" from real life in time when there was no other radios)
example some small DCV through 10Mohm resistor to diode anode and also cut DC road to L with small capacitor. (of course best is do it with inductor for cut RF "road" but experimental test can do without...

[Zero999]

For scope probe. Do NOT load parallel LC resonant circuit with scope probe.  It destroy Q value and it add also its capacitance to circuit and change it resonant freq. (also if you adjust your probe low freq risetime, it do not delete (eliminate) this capacitance! )

Doesn't using a compensated 'scope probe minimise the probe's capacitance?

[jahonen]

For scope probe. Do NOT load parallel LC resonant circuit with scope probe.  It destroy Q value and it add also its capacitance to circuit and change it resonant freq. (also if you adjust your probe low freq risetime, it do not delete (eliminate) this capacitance! )

Doesn't using a compensated 'scope probe minimise the probe's capacitance?

No, it just equalizes the frequency response seen by the scope. It can't get rid of ~10-15 pF load capacitance which typical 1:10 probe has.

Regards,
Janne

[Zero999]

All right I didn't know about that.

Another question: does the capacitance still depend on the cable or is it solely dependant on the 10:1 probe?

If the capacitance of the probe is fixed, is it given on the datasheet (I've not seen one before)? If so then it will be predictable so can be included in the calculations.

Another option is to measure the capacitance of the probe so it can be subtracted from the calculated value.

[eternal_noob]

Hmmm... How can I tell this without sounding cruel.
First of all.. You should really take a second look at what rf-loop is trying to tell you. Forget that ringing thing on the scope.
Second. The most common mistake first-time builders of crystal receivers is to load down the tuning circuit. Crystal receivers requires taking great care in impedance matching in both front and rear of the tuning/detector circuit. You can't expect ANY good results at all if you do not. At best, you will have, as you already have achieved; one station in the middle of the dial.

Second. Crystal receiver building has nearly become a science in itself. Take a look at the impressive work that the experts in this field do. Here we are talking about coils with a Q well over 1000. Multiple front-end tuning circuits ahead of the detector. Variable capacitors with silvered vanes and modified insulators (even ceramics isn't good enough here) fort best Q possible. Math, talk and tinkering that I for one do not understand a ting of (you should take a look at their discussions).
What the result is, you ask? It is 200+ stations spread over the whole tuning range on  the Medium Wave band, with a reasonable good selectivity.

Third. I will guess you are still loading that tuning circuit to much. This time with the detector. Try to tap it waaaay down on the coil.

Fourth. If you really want to achieve some results with a simple receiver, try a regenerative one instead and get rid of some of the Q problem. Use any JFET from the junk box. Audio and switching JFET's works good.