Author Topic: Questions about receiver IF strip  (Read 3532 times)

0 Members and 1 Guest are viewing this topic.

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Questions about receiver IF strip
« on: May 30, 2016, 12:00:37 pm »
Hello

Lately I've been trying to make superheterodine receiver and i'm having some problems with the IF chain stability.

the receiver's first frequency block is:
 - input cascode amplifier, gain controlled by split emitor resistor - working
 - 3db 50ohm attenuator pad
 - SBL-1X mixer - working
 - 10.7MHz band-pass tunable diplexer, 50 ohm terminated - working
 - BC547 untuned amplifier, gain controlled by split emitter resistor - working
 - 3dB 200ohm attenuator 
 - 10.69850 2KHz crystal filter
 - 3dB 200ohm attenuator

This entire block is rock stable and because of the good filter, very narrow band.
I've tested it until overload and will not oscillate / do anything strange

The IF chain was build using 3 tuned stages that provided some decent gain and an untuned stage connected to the mixer via 3db pad
Each stage followed this schematic:



L1 / L2 / L3 where arranged in such a way that it presents ~ 5000 ohms in the transistor's collector.
Those are not the actual values used on the board.
Also, the Q is quite high - 10-20kHz bandwidth.

The first problem: when connecting the output of stage 1 (L3 ) to the input of stage 2 ( C3 ) the entire thing begins to oscillate at 100-150MHz.
The oscillation is strong enough to be seen just by keeping the scope's probe next to C2.
The same behavior was apparent with only one stage connected to the signal generator: when I pushed the RF-OFF key the entire
thing begins oscillating at almost 200MHz.
The generator shorts it's output to ground when the RF output is disabled.

The only idea i had was that the thing would become something close to:

( the same type of oscillator is used by the frequency generator )

I could stop it oscillating by adding a resistor in series with C3 -> i could get 3 ~ stable IF amplifiers to work stand-alone.
The next step was too add a buffer amplifier( the same configuration as the ones above) that is feeding the second mixer via a 3db 50ohm pad.

The problems started when i begin to apply the second IF oscillator signal: the IF chain restarts oscillating almost on the 2'nd IF frequency and acts as a mixer by itself
Because of that the 2nd mixer's output is just junk - it's mixing 3-4 very close frequencies.

If i remove the 2nd IF oscillator then the IF chain either becomes stable or keeps oscillating at 10.7MHz.

My question: if I have that narrow band filter, is there any need to have a narrow band IF ?
Can I get away with 3-4 untuned amplifier stages that will also allow for AGC ( by changing the bias point )
Or should i keep 1 tuned stage just before the 2nd mixer to clean up added noise ?

The entire board is build dead-bug style on double sided copper clad, with the mixers and filter on the back, pins passing trough the PCB

Also, any hints on how to make a DIY diode ring mixer that will work on 10.7MHz ?
It seems that some of the mixers I have are junk and I can't properly test them.

I have some T50-2, FT37-43 and FT37-61 cores and some small signal schottky diodes.
I've seen W2AEW's clip

and notes
http://www.qsl.net/w/w2aew//youtube/Diode_ring_mixer_operation.pdf but I don't know how to choose the number of turns for the transformers. 20 turns on T50-2 cores should work ?
 

Offline German_EE

  • Super Contributor
  • ***
  • Posts: 2306
  • Country: de
Re: Questions about receiver IF strip
« Reply #1 on: May 30, 2016, 04:19:28 pm »
Firstly, your schematic doesn't show any decoupling on the supply rail so I hope that it's there!

Assuming that decoupling is there you have two quick cures for an oscillating amplifier, either a) a ferrite bead on the base lead or b) a small value resistor. The ferrite bead is easiest as you don't have to alter your layout. Also, depending on amplifier gain, some screening between stages may help.
Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.

Warren Buffett
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 14091
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Questions about receiver IF strip
« Reply #2 on: May 31, 2016, 12:50:43 am »
Probably trying to get too much gain with too little neutralization.

You can apply neutralization in a few ways (google for ideas), or use a cascode (which doesn't really neutralize, but at least reduces the offending Ccb so you can have higher stable gain).

If the oscillating frequency is very different from the tuned frequency, you've probably also got parasitics in play: i.e., your circuit is more than what you show here, whether you think so or not. ;) Examples are coupling capacitors having ESL (due to wiring length and whatnot), which resonates with Cbe or something like that.

Tuned common emitter stages generally have negative input impedance, so be careful. You may need extra damping/padding between stages anyway.

On the upside, after the first stage (which can be a nice low noise grounded-base or something), you don't have to worry about noise, so you can afford to waste gain, at the expense of perhaps one more stage required.

If you'll be applying AGC, note that the load resistance (base input) and gain vary with bias, so your bandwidth and stability will vary as well.  Padding helps further with this.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Re: Questions about receiver IF strip
« Reply #3 on: May 31, 2016, 05:11:11 am »
thanks for your replies.
each stage has 100 ohm and 100nF decoupling from the main 12V rail.

regarding the other question: can i skip all the tuned stages after the main filter ? or i should keep the tuned stages and make them work properly ?

i will add some resistors at the output of each stage -> make each stage see something like 100-200 ohm output load so i can stabilize the gain.

is there any "recommended" gain / stage that i should aim for ? something like 10x per stage ?



 

Z80

  • Guest
Re: Questions about receiver IF strip
« Reply #4 on: May 31, 2016, 11:56:54 am »
Hi,

Tuned IF strips are old school and not really needed if you have a good crystal filter.  Keep the gain post filter and make sure the filter is well terminated and shielded.  Here is a good article by Wez Haywood of 'EMRFD' fame about a hybrid cascode variable gain IF strip.  I built a variation of this recently and it works like a champ.

http://www.ka7exm.net/hycas/hycas_200712_qst.pdf
 

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Re: Questions about receiver IF strip
« Reply #5 on: May 31, 2016, 01:17:48 pm »
that looks interesting, thanks.
i will take a closer look once i get home.

question: can i replace  the 2 transistor with dual gate mosfets ? I've played with http://www.nxp.com/documents/data_sheet/BF998.pdf with a basic config and the AGC range ( manual multi-turn pot ) was quite large

i'll try to prototype a stage once i have some available time, right now i'm wishing for 48 hours days ...
 

Z80

  • Guest
Re: Questions about receiver IF strip
« Reply #6 on: May 31, 2016, 03:58:50 pm »
You could use a dual gate mosfet with suitable changes.  There are lots of examples of that type available on the web, but have a read of the article, Wes explains why the hybrid arrangement gives better voltage stability.  The claimed AGC range of the final circuit is 100dB.
 

Offline w2aew

  • Super Contributor
  • ***
  • Posts: 1746
  • Country: us
  • I usTa cuDnt speL enjinere, noW I aR wuN
    • My YouTube Channel
Re: Questions about receiver IF strip
« Reply #7 on: May 31, 2016, 04:04:48 pm »
One good rule of thumb with BJTs is... ...you want a "real" impedance in at least two of the three legs to help avoid the negative impedance issues that can lead to oscillation.  Try a small resistor in series with the base of Q1 in your schematic.  This cures a lot of ills in BJT circuits.
YouTube channel: https://www.youtube.com/w2aew
FAE for Tektronix
Technical Coordinator for the ARRL Northern NJ Section
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 14091
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Questions about receiver IF strip
« Reply #8 on: June 01, 2016, 03:09:01 am »
One good rule of thumb with BJTs is... ...you want a "real" impedance in at least two of the three legs to help avoid the negative impedance issues that can lead to oscillation.  Try a small resistor in series with the base of Q1 in your schematic.  This cures a lot of ills in BJT circuits.

Yup, and this goes hand-in-hand with the parasitic paths being not what you're expecting.  A series resonant path on one terminal, with bypass caps on the others, makes a lovely Colpitts or other oscillator.  (There's a name for an inductive linked oscillator that's not Hartley, but I can't remember it.)

A small ferrite bead does a good job, because it acts as a L||R network.  The impedance and frequency need to be selected appropriately, though, otherwise you'll dampen the intended mode as well!

Tim
« Last Edit: June 01, 2016, 03:15:51 am by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Re: Questions about receiver IF strip
« Reply #9 on: June 01, 2016, 07:46:45 am »
quick update: i've changed all IF stages to untuned amplifiers ( using the old transformers ) and it looks like it's stable now. I could see ~ 20-25mV at the input of the 2nd mixer (2nd LO disconnected) with 100uV input, ~48dB of gain

i'll come back with some circuit values / schematics - for some reason the first amplifier after the filter seems to distort the signal if it's input is over 10-15mV. I did not have time to adjust biasing, etc so there must be a reason for it, i need to put it on paper and figure out.

also, DS1052 is not that usable at small signal levels.
 

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Re: Questions about receiver IF strip
« Reply #10 on: June 01, 2016, 12:57:57 pm »
After running it in ltspice i found out why it's distorting:

There are the actual values on my board:
Rb1 and Rb2 = 10K - the divider used to set base voltage
Re = 900 ohms
L1 = L2 = 23 turns on T50-2 - just reused the existing transformers -> ~2.96uH, 174 ohms at 10.7MHz

There are the calculated parameters:
Vb = 6V
Ve = 5.3V ( assuming Vbe = 0.7 )
Ic = 5.88mA
gm = 0.232
rpi = 467 ohms
rin = Rb1 || Rb2 || rpi = 427 ohms = Rin. this is the load seen by the previous stage

the transistor in the current stage sees a load RL = Xl1 || Rpad || Rin = 80 ohms
=> Av = -gmRL = -18.9 / stage. ( in simulation it's about 24? )

Assuming a 10mV signal at the input of stage 1, stage 2 sees Vin = 189mV and tries to output 3.6V.
The base voltage goes close to Ve -> it starts distorting.

It's interesting to see that the 2nd stage input resistance is no longer liniar -> the 1st stage must work in a non-linear load -> it distorts even more.

I think the large input signal is breaking the "small signal" assumptions.
Is there any generic rule ? if your input signal is more than .... you are in trouble ?
I've looked at the small signal model and Dic ~= 20e-3*DVin , DVin - voltage change of V2.
This explains why, for large input values it tends to saturate - the change in collector current is comparable to the bias current.

This is how 2 consecutive stages look like, with real values. Also, attached
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 14091
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Questions about receiver IF strip
« Reply #11 on: June 01, 2016, 01:33:37 pm »
You don't need nearly so much bias voltage... for a 12V supply, 2V on the base (in practice, ~1.2V emitter) is fine. This gives plenty of dynamic range (collector can pull down to ~1.3V, giving >10V of peak swing).  This won't affect your distortion though, unless that collector voltage range is being violated (i.e. saturation is occurring).

Which is a bit of a concern at this frequency, as Ccb(V) is pretty strong in the low-volts range.  You want to avoid approaching saturation, because the transistor slows down considerably at low voltages.  This may not be modeled very well in SPICE, or be very accurate with the default model.

I think the large input signal is breaking the "small signal" assumptions.
Is there any generic rule ? if your input signal is more than .... you are in trouble ?

On the order of 20mV.

Everything centers around:

Ic = Is * (exp(Vbe/Vth) - 1)

Vth is 26meV at room temperature, give or take the N (ideality factor) usually 1-2.

So over 26mV, about e (~2.7 times) has changed.  Which is rather a lot.  It's not a mere 10%, and the derivatives are all off by the same factor, for the same reason (d(e^x)/dt = e^x...), so the whole thing looks pretty dodgy by such signal levels.

In practical circuits, limits range from ~10mV (about as far as you can expect reasonable linearity) to ~60mV (for OTAs with compensation diodes), to a bit more for lower gain devices (FETs).

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Re: Questions about receiver IF strip
« Reply #12 on: June 02, 2016, 07:19:01 am »
update:
got the biasing point as recommended by T3sl4co1l:
Vb is ~1.6V , Re = 220 ohm -> ~ 5mA per stage
now each stage has a gain of ~10-15 and everything is nice and stable with 3 stages connected.

now i have to rewire the 4th stage that drives the second mixer and check everything again.



 

Offline ealex

  • Frequent Contributor
  • **
  • Posts: 281
  • Country: ro
Re: Questions about receiver IF strip
« Reply #13 on: June 03, 2016, 07:23:36 am »
finally got something working.
i can see the filter's narrow BW - 2KHz.

the entire thing is noisy, etc but it's at least doing something :)
i can see that some stage is overloading very easy - a weak input signal generates a lot of harmonics on the output.

i have way too much gain in the front-end.
also, i used some cheap ceramic caps for inter-stage coupling -> it acts as a microphone.

now that i know it's working i will rebuild it from scratch.
i will try to follow this example : http://www.tuberadio.com/new.html
it's nicely explained and now i can understand better why things are done that way

everyone, thanks for your help.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf