The AM Forum
March 19, 2024, 12:55:58 AM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: 1 ... 5 [6] 7   Go Down
  Print  
Author Topic: Building my first oscillator  (Read 151733 times)
0 Members and 2 Guests are viewing this topic.
W4AMV
Member

Offline Offline

Posts: 672


« Reply #125 on: September 16, 2011, 03:51:12 AM »

YEAH Cheesy The AL on your toroid is 350, then the calculated L is 42 uH. So no issue here. SO-239 are ok, a bit huge. If you are an audio guy, you can use the ol'  RCA-phono plug and jack. They are easy to solder to coax, make a nice secure connection between male-female, and guess what, they are nearly 50 ohm characteristic impedance.

I measured my homebrew mixer last night, it works very nice with loss at about 6 dB. The IF is easy to see and extract, its balance is AWFUL. That is to say, if you turn either LO or RF off and look at the IF port, the LO or RF is blowing through. That is not an issue, your LPF will take care of that. Of course without any IF filter, you would be in trouble, since looking for the small IF signal in competition with the LO blow by, will saturate your IF amplifier. Thus the need for the IF filter.
Logged
ssbothwell SWL
Guest
« Reply #126 on: September 16, 2011, 04:28:46 AM »

yeah maybe so-239 is a bit overkill for benchtop gear. i rca would be easy but i want to stick to a connector that is more standard for rf gear so that i dont need to use quite so many adapters. maybe bnc would be a good idea?

i was looking around for info on IF amps in the handbook and was disappointed to find very little information. i searched on the web and it sounds like they are normal transistor amp stages with an LC circuit tuned to the IF frequency on the collector. is that correct?

edit: DMOD, is this the bias circuit you were describing the other day? http://i.imgur.com/lDlx1.png

edit2: just for fun i put together the mixer in ltspice and attached it to the spice mode of my oscillator. in spice the mixer is well balanced. i am going to try rebuilding it with shorter leads. http://i.imgur.com/AbGjf.png http://i.imgur.com/DeekF.png
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #127 on: September 16, 2011, 05:03:54 PM »

Very good. To your Spice entry, add another port and attach your LPF here, and port out the IF signal in a clean manner.

On the IF strip, most folks buy a commercial IC. There are a number of old great ones, like the MC1590, but there are very excellent new ones from several companies. In fact, todays operational amplfiers have significant gain-bandwidth and could be used in an IF. You certainly could build your own with bipolar, MOSFET, or other semi-conductor choices. The major design challenge there is achieving stable gain, a gain value of at least 90 dB, a reasonable noise figure, AGC, and a good grasp of impedance matching. The filters used in the IF strip will have to properly matched if you are to preserve their frequency response and selectivity.

Now you can get by with alot less, say 20-40 dB of gain, no AGC, and some simple ceramic filters, centered at 455 kHz, with bandwidths of 6 or so kHz for AM. The remaining gain can be achieved at audio after demodulation. I would poke around the library, the ARRL handbook and other texts. There is plenty of reference material available.
Logged
ssbothwell SWL
Guest
« Reply #128 on: September 16, 2011, 06:50:35 PM »

the filtered IF output in spice looks really nice. in fact it is almost incomparably nicer than my real life results. my mixer has too much loss.

http://i.imgur.com/i9qds.png
http://i.imgur.com/s5oDJ.png

i bought some ferric chloride and am in the process of etching a very simple pcb for the mixer. i used a sharpie to draw the circuit layout. hopefully this will improve my real life results.


Do you recommend the MC1590 or do you think i should try something more modern?
Logged
ssbothwell SWL
Guest
« Reply #129 on: September 16, 2011, 08:27:05 PM »

here is my new mixer: http://i.imgur.com/IDN4W.jpg
there is a missing jumper in that photo. i added it after i took the picture but before i tested the circuit.

i still do not get a balanced output but there is slightly less loss: http://i.imgur.com/xmmcn.jpg

is this the best i can hope for? LTSpice still gives way better results. Sad
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #130 on: September 16, 2011, 09:17:30 PM »

I agree, you should be able to do much better. Again, I looked at the HB mixer that I posted, the IF signal and conversion loss is in line with a quality commercial mixer. Yes, the balance is not great, but that is not due to layout. It is due to the nature of that particular circuit design. There are modern IC's, the older Motorola IC's work fine, but devices from Analog devices and TI are extraordinary leap forward. Switch to coaxial interconnect and minimize the use of wire jumpers. Here are some pix of the technique I use. The amplifier and filter card are all built on FR-4, dremel tool and took about 2 hours to complete.

Here is a hint to construction. Draw your schematic out in a manner this is electrically proper to minimize stray coupling, parasitic C and L, and just flows properly. Use the schematic as a guide to laying out the card. With your parts in hand, transfer the parts from the schematic to the card, pencil outline their position on the card and trace the part outline. Then with the dremel tool, begin cutting your circuit pads. You will see the filter card and the power FETS are sitting on dremel cut outs, all point to point wire connection and you can build your circuits in 3-dimensions! Yes, you have length, width and height available. I'll post a converter module built in that manner.


* Mixer 077.jpg (141 KB, 640x480 - viewed 1328 times.)

* Mixer 078.jpg (148.74 KB, 640x480 - viewed 1209 times.)

* Mixer 079.jpg (155.76 KB, 640x480 - viewed 1287 times.)
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #131 on: September 16, 2011, 09:24:05 PM »

Here is a converter module, and a Tx exciter, ALL built with the same technique. There is not one PCB in the lot. All cut with either knife or dremel tool. Once you get the hang of it, you can think a schematic in your head and transfer to the PC card in one motion.


* Mixer 075.jpg (151.82 KB, 640x480 - viewed 1238 times.)
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #132 on: September 16, 2011, 09:27:59 PM »

oops here is the 3-D converter card. A clever idea for a VFO is also shown, this one from Rick Campbell who writes for QST and QEX. A clever application for a SOUP CAN!



* Mixer 010.jpg (145.48 KB, 640x480 - viewed 1195 times.)

* Mixer 010.jpg (145.48 KB, 640x480 - viewed 1233 times.)

* Mixer 008.jpg (139.28 KB, 640x480 - viewed 1385 times.)
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #133 on: September 16, 2011, 09:35:03 PM »

boy, sorry, here is the converter which shows the stacking of modules. The whole assembly slides into a JPL extruded case that cost Uncle SAM a fortune from the Apollo days!!


* Mixer 014.jpg (145.56 KB, 640x480 - viewed 1384 times.)
Logged
ssbothwell SWL
Guest
« Reply #134 on: September 16, 2011, 10:34:14 PM »

wow really nice circuits. so the advantage to your dremel method is that you can put all the wiring in 3d space and have a giant groundplane? i noticed you have baluns on all your jumper cables. is that to prevent them from radiating and causing intereference? will coaxial with grounded shielding achieve result?

drawing the circuit with a sharpie and then etching was really easy for me and i would like to keep working making my circuits in that way. i'm worried that i will be sloppy with a dremel.


i looked at my ltspice model again and i realized that i added the mixer to the wrong file. the file i used didn't have an amp stage on the LO and i forgot to include a load.

here is a corrected spice model: http://i.imgur.com/uuhUZ.png

i think i actually get comparable if voltage. the main difference seems to be in the filtering. my butterworth filter doesn't clean up the signal nearly as much as in LTSpice. how do i go from a composite wave to a clean sinewave like in ltspice?

edit: i put the IF filter together on copper with very short leads. the board is fused onto the mixer's board. it looks goofy but it enabled me to get rid of one coaxial connector. i get a clearer difference frequency and am able to detect it with the RF signal as low as 130mVRMS. http://i.imgur.com/bYf9H.jpg

now i am curious to see what kind of improvements i will see if i get my oscillator off of the breadboard and onto copper. i am hesitant to do this until i know exactly how i want to setup my bias the 2n3866 and how i want to setup the tuning on the colpitts oscillator.
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #135 on: September 17, 2011, 05:23:09 AM »

Hi. Early AM! Good questions. First, I suspect that if you drove your mixer with a set of perfect sig gens it would look quite nice. I still suspect cross talk and some level of injection locking or pulling of your osc is present. Also, NOT terminating the mixer IF port into 50 ohms, put terminating in a larger impedance, will provide a larger output IF voltage. Now their are other issues that occur if you do not terminate into 50 ohms, however for your purposes this is not a severe issue. So, that said, you might at least look at your mixer IF output by connecting it directly to the O'scope through a short (3 foot) coax cable. No need to use the 10:1 scope probe divider and/or terminating the IF into 50 ohms.

The filter performance will be significantly hampered with poor layout and construction, you are only using a 3 section, 18 dB/octave rate of roll off beyond the cutoff point. So you can't afford fouling up the construction. The Spice simulator has no stray coupling. Its perfect! You can add in electrostatic and magnetic coupling, a lot of work and not needed.  If you can sweep your filter on its own, do that, use your scope and sig gen and measure the cut off freq, -3 dB down point and roll off, make sure the filter is functioning.

The single 1/2 turn wire through the toroid or bead is a simple RF choke. It provides a broadband reactive impedance that helps isolate two nodes. In many cases, an RF node from a DC node. To a limited value it provides some shielding but not nearly as well as coax. However, coax would not provide the required impedance disconnect between nodes, EXCEPT under certain conditions.

The main thing about these ferrite beads is their lack of low frequency resonance. They are lossy RF paths (providing several hundred ohms) of series reactance with a small DC component of resistance. You could use "RF chokes" so called lossy inductor, or just a good quality inductor to accomplish the same job. Just be careful about their self resonant point! In any case, the bead is easy to use. Don't put them in unless you find a need, they are easy to place,  cut a trace and run a pig tail wire through. They can introduce their own set of problems, so experiment first.

Practice with the dremel moto drill. It is faster, but your technique is fine. Grab some scrap PC card material and try cutting some simple rectangles, then fancy islands. The small grinder tipped dremel bits are great and they are inexpensive.
Logged
ssbothwell SWL
Guest
« Reply #136 on: September 17, 2011, 07:03:55 PM »

dropping the 50ohm load increased the voltage (as expected). at higher voltage the composite wave dissipears and the if frequency looks much smoother it is still not perfect but is a lot better. this is a good sign.

i was looking again at the handbook. would replacing the butterworth filter with a diplexer on the IF output improve my IF signal?

would one of these work as part of a diplexer? http://www.kitsandparts.com/455KHz.htm

maybe before i start working on the IF stage i should try to beef up my oscillator. is injection locking a symptom of an inadequate buffer stage? what could i do to improve that?


Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #137 on: September 17, 2011, 10:43:51 PM »

The diplexer is a network that provides a constant resistive termination, typically 50 ohms. A bandpass diplexer is an appropriate approach, although a low pass is fine as well. The purpose of this network is to minimize the INCREASE in the IF termination impedance for the mixer, which will occur if a FILTER is used, for example a xtal filter. The problem that occurs, is the voltage gain (as you see) of the mixer rises off channel (i.e. from the center frequency of the filter) and the distortion products of the mixer will increase as well. That said, don't worry about it. It is not your current issue. Yes, adding more buffer, another stage would help. A pad for example, a 6 dB pad, if you can afford some loss, would help. If you scope your oscillator when not connected to the mixer and then connected to the mixer, say at the oscillator source terminal of the FET, do you see the signal change in any manner? I'll take a pix of my homebrew mixer which has no IF filter and the IF goes directly into a O'scope via 3 feet of coax for reference. As I recall, the IF frequency is quite easy to detect, as well I will RC lowpass the IF and scope that as well.
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #138 on: September 17, 2011, 10:46:55 PM »

meant to say measure the osc NOT connected to the mixer, then connected, and note any difference in the oscillator waveform as measured at the source terminal of the FET osc.
Logged
ssbothwell SWL
Guest
« Reply #139 on: September 19, 2011, 04:13:42 PM »

when measuring directly off the oscillator FET i get 5VRMS with it disconnected from the mixer. connected to the mixer with the RF osc lturned on, i get a 4.2VRMS. in both cases the waveform was a nice sincewave.

a 6db attenuator after the LO lowpassfilter fixed the voltage drop on the osc. i used this calculator to make the attenuator: http://chemandy.com/calculatorsmatching-pi-attenuator-calculator.htm

when using the attenuator i cant find the beat frequency when running the mixer into a 50ohm dummy load. do i need more amplification of the LO to makeup for the attenuator?

i also noticed that at the LO filter output the signal is 70-80khz higher than when measuring directly off the LO FET. this is when disconnected from the mixer. why does this happen and is it a problem?
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #140 on: September 19, 2011, 08:51:17 PM »

You have more than enough LO even with the attenuator to permit mixing action. I assume you are looking at the IF (beat) after a LPF. The shift in LO frequency is minor and is due to the difference in load impedance presented to the oscillator, in one case your LO filter vs. no filter. Again, be sure your IF is below your LPF cutoff, check your LO and RF frequencies to be sure. I ran my mixer posted earlier with just 0 to -10 dBm LO, while the RF is at -20 dBm. The IF out and IF filtered out is very clearly seen. The conditions were LO at 2 MHz, RF at 1.5 MHz, the IF at 0.5 MHz. You can clearly see the RF and LO in pix below, and if you look carefully, the IF. However, through the IF filter, it is very clear. The IF filter used ~6500 pF and ~32 uH L. I'll post next.   
Logged
ssbothwell SWL
Guest
« Reply #141 on: September 19, 2011, 09:23:56 PM »

my filter is essentially the same as yours: 6300pF caps and 32.3uH (53 turns on a T68-1) toroid. i set my function generator to -20dBm and my LO to 0dBm and hooked them to the mixer. i used a potentiometer in series to reduce the LO down to 0dBm, could that cause problems? i also measured the dBm using a 50ohm load, is that correct?

i get out of the mixer's LPF a 5mV signal that the oscillator has a really hard time reading the freqencyon. it appears to be in the 100s of MHz.  Huh


i am really tempted to order one of these $8 diode mixer kits: http://kitsandparts.com/drm.php
i want to use a homemade one in my receiver but i would like to have a working mixer that i can compare my homemade stuff too.
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #142 on: September 19, 2011, 09:46:38 PM »

First, here is the pix's. I am beginning to suspect there is something amiss in your mixer. From the screen shots of your oscillator, it looks fine. The mixer is behaving now as though you have a considerable amount of conversion loss. Make sure your interconnect is clean. In may case, I just bolt the connectors (BNC) from the mixer card to the filter card. I BNC TEE off the IF direct from the mixer and then via the LPF and these go to the 2-channel inputs of the scope. No probes, just coax to the vertical channel inputs of the scope.


* Mixer 001.jpg (157.4 KB, 640x480 - viewed 2457 times.)

* Mixer 002.jpg (135.56 KB, 640x480 - viewed 1220 times.)
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #143 on: September 19, 2011, 09:52:48 PM »

The scope horizontal is 0.5 usec / division, so the IF is 500 kHz, 4 divisions. If you look carefully the RF and LO are seen. I amplitude modulated the RF signal with low level AM and it helps (sort of) mark the RF signal (a bit of a blur from the modulation), while the LO is not modulated. Again, the sources are sig gens, so they are decent sources. However,  so are yours,  unless something is significantly upsetting your LO, and we I don't see that issue.
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #144 on: September 19, 2011, 09:57:28 PM »

Yes, get a commercial mixer. Or, if you want, mail me yours, I would be glad to check it out. The problem could be as simple as your diode selection. You said that are hot carrier. It would be useful to measure one. See if the forward drop on your VOM is indeed 0.3 and not 0.7 V. 
Logged
ssbothwell SWL
Guest
« Reply #145 on: September 19, 2011, 10:14:40 PM »

oh wow that would be really amazing if you could checkout my mixer.

i dont have any bnc connectors so i cant set everything up as you have it but tomorrow i will go to the shop and some up. if the connectors dont help then it would be great to send you the device. can you private message me your address for shipping?

i'll also order a ring modulator kit online for comparison. the one i am looking at is really cheap and it has an included diplexer. to make comparisons to my mixer could i just bypass the diplexer? http://kitsandparts.com/drm.php


edit:

i think i blew out an mpf102 messing around with a clapp oscillator. http://i.imgur.com/V0YQ7.png
at first it was working fine and then it would only work if i turned the power supply off and on and then it stopped working. i tried putting the transistor back into the oscillator design you helped me with and it didnt work.

while it was working i was tried various values for C2 and i noticed that i needed at least 150pF to get it to oscillate. in ltspice increasing C2 that causes an increased spike in gate current when the oscillator is first turned on. do you think that is what burnt out the transistor?

now that i need a new transistor, would you recommend i go back to radioshack for another mpf102 or is there a better option i can order online? it would be nice to have a "go to" stable transistor i can use in a variety of HF situations.
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #146 on: September 20, 2011, 05:05:22 AM »

Yes, you can bypass the diplexer. Just need the mixer, that is 4 diodes and the set of xmfrs. On the FET, I have taken out a MPF102 many a time, it is a small geometry device, not a very big FET. Excessive gate current in forward bias mode will take it out. A good GP FET is the J309 or the J310. Either in plastic or metal case. If you want to mail, just look up my ham call on QRZ.COM. I have BNC etc... no need to buy them.
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #147 on: September 20, 2011, 05:10:50 AM »

The W8DIZ kit is fine. A bit more than you need. You might look into mixers from Mini Circuits Labs, not sure what their one up price is, probably less than the postage to mail!
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #148 on: September 20, 2011, 05:27:07 AM »

http://www.minicircuits.com/products/fm_pic_level_7.shtml

The SBL-1+ is 8 bucks. Same price, but just 1 part. The folks at MCL are very open to assisting experimenters, hams, etc... You might give them a call, tell that what you are up to and see if they would sample you one unit. You pay postage. It does not cost to ask.
Logged
ssbothwell SWL
Guest
« Reply #149 on: September 20, 2011, 11:25:07 AM »

thanks for the tip on getting samples from mini-circuits. i'll give them a call and try to get a test unit but i like the idea of the w8diz kit because it is discrete components. i am imagining it will be a better comparison to show what is possible with a home made kit. i also ordered the IF Amp kit from w8diz at the same time which is convenient.

the 9310 and 9309 FETs are super cheap. i'm gonna buy a big bag to keep lying around. i'll be ordering them from mouser. any other transistors or little discrete components you would recommend i order from mouser at the same time? i have a bunch of 2n3866 which i guess are good for amplifiers under 1W but it would be cool to have some sort of transistor i can use for QRP amps up to 5W.

what are these style connectors called? http://kitsandparts.com/iconnect.php
i want to buy a bunch that have 2pins but mouser has so many different connectors that i am having a hard time narrowing my search down.
Logged
Pages: 1 ... 5 [6] 7   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.079 seconds with 19 queries.