Designing an AM Demodulator Circuit with No ICs

[Trancid]

I'm currently doing Electronics at the University of Eindhoven (Netherlands) and we need to make an AM demodulator (input signal lays between 875 and 915 kHz). We have made some sort of circuit (two to be exactly), but we are questioning the accuracy of this demodulator.
The main problem is we cannot use most of the normal ICs, however, we may use opamps and a Voltagesource (+15V...-15V).
One of the circuits we had designed was the input signal going through a diode and the outcoming signal is split by two condensators (one going to ground, the other delivers the outputsignal. This is somewhat inaccurate when there are big jumps between different freqs (the output signal is an audiosignal between 60Hz - 20 kHz).
Schematic:

in ----- (diode) ------- (condensator) ---- out
                    |
                    |
              (condensator)
                    |
                 (ground)

The othere circuit can be found at http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/amfmdet.html (the first circuit).

Like I said before, we doubt the accuracy of these demodulators. We don't need the values of the components. So if you can point out certain disadvantages or advantages of even new circuits, please do so. But keep in mind, no ICs...

Thanks for any help, Patrick

 

[Chen]

I can't sovle your problem, but you can use the [ code ] tag (without spaces):

in ----- (diode) xxxxxx (condensator) ---- out
                    x
                    x
               (condensator)
                    |
                 (ground)

 

[Trancid]

Well, that is one problem solved
Thanks

 

[turin]

Is the diode a silicon diode? I would use germanium. Also, you want to make sure that the diode is fast, the ac coupling cap is reciprocal (not electrolytic), and probably that it is much larger capacitance than the rectifying cap. I would put a large value resistor in parallel with the rectifying cap for good measure.

OK, I just looked at the link you gave. One other thing that came to mind: Are you using a buffer? You mentioned that you are allowed to use an opamp (which is techanically an IC, but, if that is against the rules, you can build a crude one out of bjt's or even scrap that idea entirely and build from the bjt level). The signal should go into a buffer for good measure. You may have heard of it called a "voltage follower." This just improves the robustness against load perturbations, which may be one source of the problem that you're encountering in the large frequency shifts. I think a 741 should be fine for this application, if you are allowed to use op-amps.

 

[Trancid]

Thanks turin for the info.
I think we could only use silicon diodes, because all of our lectures have subject of semiconductors made of silicon. And you're right about the fact that we can use opamps but IC's are prohibitted, while an opamp is an IC (I have some weird tutors;).
That buffer you mentioned, we have a build-in buffer at the recieverantenna (we also have to make the senderside). Did you mean that, or an additional buffer, because I don't know what you exactly mean by voltage follower. I could know the term and the concept, but I guess I only know it in Dutch. But if i understand correctly, the voltage follower solves the problem of let's say a musicnote at 80 Hz followed by one at 15 kHz, which a single cap can't follow correclty (am I making any sense here...).

Anyway, thanks for your help. I'll surely let you know how our presentation goes, which is in about 10 weeks. We'll be jugded at signal strength, clearness of music and another factor (read: three first prices).

Patrick

Btw.
Why is my post rated as a v or something (what does that mean?).

 

[turin]

We'll be jugded at signal strength, clearness of music and another factor (read: three first prices).

This sounds like some kind of competition, which implies that you want to push your project up against the imposed restrictions (i.e. no IC's except op-amps, so definitely use op-amps). I am making the following responses under that assumption, and with the explicit restrictions that you have mentioned. It would be most helpful if you had a hard/fast list of restrictions. Then, we could really crank your design right up into them. The most obvious restriction that I can think of would be cost. After all, if we're doing engineering, cost is an extremely important consideration. BTW, I think that whoever has imposed the restrictions meant "ASIC" instead of just "IC." That's just a guess. You may want to inquire for more clarification, since there could be a fine line in their mind between a certain level of sophistication in an op-amp and an IC.

I think we could only use silicon diodes, because all of our lectures have subject of semiconductors made of silicon.

If I were you, I would assume that you are allowed to use germanium until otherwise specified. I would get more clarification.

That buffer you mentioned, we have a build-in buffer at the recieverantenna ... Did you mean that, or an additional buffer, ...

You will have a load, presumably. Consider the diode part of your circuit as a stage (the stage that extracts the signal off of the carrier). Let's call it the "extraction stage" (until you suggest a different title that you would prefer). So, ideally, that is the last step that you need, but, in practice, your components are not ideal. Specifically, your load will have some capacitance, inductance, and, worst of all (in this case) non linear behavior.

You have two options for robust design: 1) design the extraction stage to better accommodate the load characteristics, 2) electrically separate these stages. While option 1) is possible, it restricts your load compatability and requires far more advanced/sophisticated modelling of the components. Option 2) is the generally preferred choice. That basically amounts to putting an op-amp between the extraction and load stages. If you do some internet searching, you can find op-amps that are custom made for this purpose (i.e. appropriate hysteresis, compression points, slew rate, frequency response, etc. etc.). A good design will still consider these non-ideal characteristics, even in option 2). But, the advantage is that option 2) affords load flexibility and enhances the modular design of the circuit (makes it easier to consider the circuit in modules/stages as opposed to having to consider how each stage affects the others requiring consideration of the circuit as a whole).

Most semiconductor manufacturers will have application notes and explicitly and specifically indicate the intended application for their products. If I remember correctly (it's been a while), national semiconductor has very thourough application notes. You will probably have to get your parts from a generic parts store, because the manufacturers will require orders of x1000. Just pay attention to the generic part number, and the parts store clerk will most likely be able to hook you up. I strongly advise AGAINST Radio Shack! They sell second rate parts at convenience store prices, and their staff has, in my exprerience, poor ability to accommodate component level project assistance.

All of the above goes for selecting the appropriate diode as well.

... because I don't know what you exactly mean by voltage follower. I could know the term and the concept, but I guess I only know it in Dutch.

It is just another name for "buffer" that implies the use of a generic op-amp. The function can also be accomplished by a transistor, but if I were allowed to use op-amps without restriction, then I would definitely go with a customized op-amp (as detailed in the previous response).

... if i understand correctly, the voltage follower solves the problem of let's say a musicnote at 80 Hz followed by one at 15 kHz, which a single cap can't follow correclty ...

No, not exactly. A strict buffer cannot solve this problem. This is where you have to use some judgement calls in your design, for which you can use the range secifications. The response will not be flat from the capacitor. You get approach flatness in at least one of two ways: 1) make the extraction stage a more complicated filter network, 2) use an op-amp that is designed to flatten the response. There is probably customized op-amp for every application you can imagine. Again, if I were allowed to do so, I would go for the op-amp solution (it would be more reliable by the way I design circuits).

Why is my post rated as a v or something (what does that mean?).

Beats me.

 

[Trancid]

I'll try to find a website where the presentation is put on, so that you can get a good impression on my assignment. I will also type over a part of a document we got, in which the restrictions etc. are written. But this is something for tomorrow, because I need to get up in les than five hours...
I don't know if I already said this, but this is (thank god) a group assignment in which I got to figure out the demodulation part for now.
The parts you were talking about. We are limited in a second way. We may only use parts that Chef (a guy in my faculty building) can give us. We cannot use resources that aren't available at our university.

Anyway, you'll see a better reply from me sometime on Tuesday.

Patrick

 

[turin]

... you'll see a better reply from me sometime on Tuesday.

I'll wait until then to comment.

 

[Trancid]

Description and Specifications

Sorry for the late reply, but I haven't had much time.

First a description of our task (I'm in a group of eight people):
Every group needs to realize a wireless transmission system, i.e. a radio transmitter and reciever combined with a heat alarm. It is used as a broadcast and alarmsystem in a nuclear power plant to forsee the personel in some musical entertainment during their work. A cd-player is hooked up to the transmitter, which the personel can receive at their recievers to enjoy good quality music, in the technical sense of the word.
Next to this, the transmitter has to have a temperature sensor to detect the heat of the coolant. If this gets too high, it needs to broadcast an emergency signal (a slow whoop signal). The trigger temperature needs to be adjustable. We have the option to cut the music and only broadcast the slow whoop signal, but we don't have to. The slow whoop signal needs to be at 5W and the music at max. 2W when it comes out of the speakers.
We need to design and build everything ourselves, except for the antennas, the speakers and the cd-player.

Specifications and restrictions:
-Transmission band: 875-915kHz.
-Max. 1% of the transmissionpower may be out of our band.
-Required stability of the carrierwave is less than 1 promille.
-Transmission and reciever antennas: H-field loopantennas, ready to use, only need to be adjusted to our band.
-Audio information: source is a cd-player with an analoge exit. Ready to use. A mono signal is transmitted within a band of at least between 60Hz to 10kHz (wider band is a higher grade)
-Temperature sensor: has to work in the range 0 - 100 degrees Celsius. The treshold value has to be adjustable in this entire range with a accuricy of 5K and is only manual resettable at the recieverside (i.e. if transmitted signal stops, the alarm still sounds until someone disables it at the reciever side)
-Temperature alarm information: will be send as soon the temperature sensor has reached the treshold value. It is free for the designer how to send this signal. (We chose for a signal at 20Hz which we filter out at the reciever side where the slow whoop signal is created.)
-Reciever: It has to deliver max. 2W to the speaker, adjustable with a volumecontrol. The quality of the signal had to be "good" (the designers decide on the quality themselves). The distance between the transmitter and reciever must have as little effect on the reception as possible. The reciever must only recreate music send by our own reciever.
-Alarm signal: slow-whoop signal with T=1 sec, ascending in frequency from 100Hz-4kHz. The power of this signal must be send to the speaker at a strength of 5W, independant of the volumecontrol. The slow-whoop signal me only start if the alarm signal is send out. It may not start because of a certain tone in the music.
-Voltage sources: these are standarized at +5V, +15V and -15V. The +5V may be derived from the +15V source with help of a voltagecontrol (type 7805).
-Available components: only discrete components may be used (resistances, condensators, diodes, transistors, etc.), operational amplifiers, simple logical components (gates, flipflops, counters, shiftregister, etc.), crystals, the received antenna's and speakers.
Available low power bipolar transistors are BC550, BC560; bipolar medium power transistors are BD139 and BD140; FETs are 2N5458 (or possibly 2N3819) and 2N3820; MOS transistors: BS170, BS250 and power MOS transistorsL IRF540 and IRF9530.
Available crystalfreqencies are: 8.000000MHz, 9.830400MHz, 10.00000MHz, 11.05920MHz, 12.00000MHz, 14.31818MHz, 14.74560MHz, 16.00000MHz, 17.73447MHz, 18.43200MHz, 20.00000MHz and 24.00000MHz (We chose the 14.31818MHz, because if you divide it by 16, you get our carrier wave.) We can use a voltage controller 7805.
Only in agreement with the project tutors we can use other components than above are discribed.
We chose AM-modulation, because it was the simplest to do.

I'll try to upload a schematic of our transmitter and reciever a.s.a.p.

Patrick

 

[turin]

FM would have better performance (I'm thinking specifically in terms of the distance and band cutoff), but if you want to do AM, far be it from me to disuade you.

You still need to be more specific about the parts list. There are literally thousands of different op-amps out there (some designed for your exact purpose I would venture to say). You have said that you may only use what Chef can give you. Does this mean that he has an inventory of parts, or that he has correspondence with a particular parts company, or what? Can you get a comprehensive list of op-amps from him?

I don't understand why you're worried about the crystals (for selective reception) or transistors (for output). Maybe I'm a little confused about your responsibility. As I understand it, you are a member of the receiver crew. Someone else will take care of the reception, and someone else will take care of the output, so your job is demodulation. Are you going to use an IF in your demodulation?