Regen II: High Performance Rig

Created @ 2014/6/13

Third round building season, it's time to get the regen receiver more amazing, target to high performance.

in the << Regen: Aamazing Regen Receiver >>, several pro and cons founded.

* it's simple, less component, but still high gain.

* decoding AM/FM/SSB/CW...

* band covering from VLF to VHF/UHF even microwave

simple Regen i ever built is N1TEV recommended <<ARRL Regen for beginner>>, www.arrl.org/files/file/Technology/tis/info/pdf/0009061.pdf . several cons i feel:

* audio stage use LM386,  i did not like, motor-boating some time, and little bit noise.

* Regen control is not smooth, when you cross the oscillating point, you might need tune back and forward several times.

* I use band set, band spreading and fine tune knob, still will loss some station, and very hard to find the 'right' tune point.

* seems very hard to find a CW station. 

* sensitive can not competing with the superheterodyne receiver.

Key for building Regen 

N1TEV published article on ARRL said the regen receiver can compete most of heterodyne receiver actually. several key point for this,

1. Fist of all, use capacitor as throttle regen control, this is the key factor for smooth regen control.

2. use  tickle coil  as regen feed back, at high frequency this increase the feed back then improve the smooth.

3. use pre RF stage to reduce aeronautical effects, prevent antenna absorb the tank energy, isolation the detector, provide a constant resistance. provide some gain at high frequency.

4.  use  power transformer as audio signal extractor, less noise and more audio gain. 

then let's exploring some ham's favorite modern regen receiver.  

common Regen receiver cover the VLF to SW band, VHF and UHF regen is seldom seen.  most of circuit on internet is almost same one, and some other's is really had something new.

(L1 is a small T50-2 (red) toroid with around 18-20 turns (40m) or 35-40t (80m) on the main winding connected to the FET with a single turn coupling loop for the antenna connection. The tap is about 20% up the main winding. C1 is adjusted to set the radio into the 40m band and C2 then acts as a fine tune within the band.)

from active component used in detector, there is BJT version and FET version 

 An FET version of the 1-V-1 Regenerative Receiver, designed and built by C.F. Rockey,  is directly mapping from value tube version.

most of BJT version will suffer cause BJT low input impedance and hard to control the regen, and most of them use resistor as re-gen control.

and lots of builder like make the regen receiver 'art end', and very very beautiful. some geek like Alan' Lab use a different oscillator as Regen:

WBR receiver is another project must mentioned, that's a something new type of  regen structure, should build one another day.

there is really novel Regen receiver, i.e the Parametric Regen receiver experience by  Michael J. Rainey:

N1TEV 's High  Performance Regen receiver

(published by the ARRL in QEX for Nov/Dec 1998)

this version of Regen receiver is much famous, and built by world, like:

http://www.w5jh.net/regen.htm

http://www.aa5tb.com/regen.html  by AA5TB , with pre RF stage use JFET and different power regulator.

http://sm0vpo.altervista.org/rx/regen2.htm   by SM0VPO, also similar.

every time, every time i make some mistake, in another way, i learn something new, that's good and painful. i made two Regen receiver, every time i think it can pick up station without long wire antenna, i mean i use very short wire, about 50cm,  and no any station at all, this hurt me every time, then i realized, the SW band, must have a antenna, a long wire, 2 meter for starting test. so decide write it down: Need  Antenna.

I does not use the N1TEV's audio amplifier, i use the SM0VPO audio amplifier,  and connect the speaker, not ear phone.  I had used ear phone in making my first Regen, experience is good.  in second Regen(rebuild first one) use speak instead of ear phone, i recall, that's bad,  this time i use speak again, and bad again.  so suggest use earphone, at least for debugging stage. cause speak is very un-sensitive for man's ear,  you might missing the station, and don't know does the detector is starting oscillation.

also tried to make it sensitive by adding a pre-amplifier, single BJT, you might know, it's severally motor-boating and i can not stop it,  pre-amplifier need carefully design and thinking, that's might not a good idea, because when tune on a station, the detector output will increase so much, then overloading the audio stage, AGC of audio might needed.  So i give up, and try use earphone.  very unfortunate, my earphone burned when i plugin the ear phone, cause my audio amplifier connect the speak from VCC.... what a shame.  finally i soldering high-z earphone to output of audio stage....

ear phone is very good, you can know when it just begin oscillating. Ok, then i can hearing something. but, there is other big problem bothering me. first one is throttle  control is not functional, another one is when tune to middle point of my capacitor, oscillation stopped, only high range, 10 Mhz to 18 Mhz can oscillating. let's me introduce the big problem in new section.

My building of N1TEV, The big problem

another problem is when C4 is reach it's middle to max value, oscillation never starting, even you adjust the R4, control the gain of JFET. after several times trying, i feel bad and trying to resolve this.

first of all, why C6 can adjust Regen and why C6 in JFET Drain is ok also?  

when you Connect to Drain, C6 will pass  some positive feed back signal to ground, then stop the oscillating. Connect to Drain eliminating my one problem, there is no bump now. think, why C6 connect to RFC1 does the regen control, that's because RFC1, RFC block Tickler's signal, let it can not reach the GND( in another word, all signal voltage taken by RFC),then when C6 reach some value, the capacitance provide a low impedance,  then C6 let signal pass to GND,  most of signal voltage shown on Tickler(L3), positive feedback enough, it start oscillating.

 So what will be the problem?   i suspect the RFC, 2.5 mH.  if RFC1 buggy, can not block the signal, C6 loss it's functionality, and if C6's SRF(

Self-resonant frequency)  located in the oscillating range, there might bump to start oscillating. I do remember ARRL talk this problem much, but i misunderstanding something.  should DIg into the RFC and SRF(jump there).   

finally, this inductor work as RF choke very good from 3 Mhz to 15 Mhz, after replace use this, regen control is working perfectly, and no bumping, now only one problem left, low frequency range, 3 Mhz to 8 Mhz is still not working,  but better, with buggy RFC only work down to 10 Mhz. which mean my original RFC  only work to around 8 Mhz, from 10 Mhz, it can not block the signal, loss it's choke function, for safe, suggest use that toroid only for below  3 Mhz - 5 Mhz. 

now its time resolve why 3Mhz not starting oscillation: 

  1) increase the tickler coil to 10T,  with L2 is 20T

  2) reduce R4 to 680R

  3) worth to mention, i use MPF102 instead of  J310 or 2N4416, which provide much gain than MPF102.

Well all problem solved, now.

My layout

I built this audio amplifier into above board,  no negative voltage need. 

Testing version

the long connection line of coil. and too loose couple between L3 to L2,  violate the shortest line rule of HF.  and when you try to change the coil, you encounter the big problem: too many screw there and the electronic contact is not perfect.

Perfect layout regen core use 8 pin sockets

@2014/6/24

 changing coil, this is the regen receiver's BIG fun.  strong willing motivate me to make a cute design, so in a long time period, i will play with and enjoy the design.

Final layout use 8 pin socket, upside-down, right small one is coil plugin forms layout:

then show time:

How this version performance?

* basically it's good,  regen control became a little bit smooth. 

* sensitive is also good. had received many SW station and some 40M band CW signal.  (In door, CW receive is hard.. even through use long wire, but very close to wall)

* quick coil change capability and very tight and small layout i ever build.

* full frequency range 5 MHz to 15 Mhz, get perfect sine wave output. [ serial 470pF capacitor with 360pf varactor ]

I  would get a cute case for this receiver.

fixing gradually stop oscillation problem

@2014/6/26

occasionally,  the oscillating stopped, checking the status show  Q2  Drain voltage dropped to around 5V and even lower and lower with time. this  behavior confuse me long time.  it work well after a long time run,  just stopped.   checking C6 RFC1 C7, D1, even remove D1. all seems can not bring back it to work

improve my bias of JFET

at the begin, i thought this is  JFET bias problem. and try to get it better, this build configuration:

     Q2  MP102

     Zender D1:  7.8V

JFET character vary from components to component, they might had very different Idss. 

so i changed my zender to 6.8V and change R4 to 3.3k, and fix this problem, at least i think i fixed it. 

here is the estimation of bias:

MPF102:  

                Vgs(off)= -8V

                Idss =  2mA - 10 mA

                VCC=11.9V

                without load  zender draw 2mA current, Iz need keep >1mA for normal 0.5W zender. ( actually  suggest Iz is 2-5mA for 0.5W series zender, you should check the datasheet), so i suspect when provide > 1mA for oscillator, the zender actually not working on correct bias.

if zender is 7.8V:

              assume Idss is 2mA,    Rs (R4)    2.7k will bias the MPF102 at Id=0.9mA, Vgs around 2.5V, this is my version of configuration, it works.

              if Idss is 5mA,           keep Rs as 2.7k, now, source voltage will need around 4V to pull 1.4mA current. this is near the zender current source                                                          capability,  i believe from there,  the zender can not provide bias current, voltage then loss it's stability .

if change to 6.8V zender, current maintained by zender is 2.5 mA,  given more margin, use 3.3k Rs reduce bias current to 1mA(Idss around 5mA).  

Finally Got the ROOT cause

after tear down and reassemble them several time to looking for the reason, i decide to believe my analysis.

1. the voltage is gradually reduced at gate D, and conjunction of r6 and RFC1

2. i notice when voltage begin low down, current increase, and Drain to ground  resistance reduce to few hundred ohm.

3. while voltage begin to drop, the Source voltage is keep stable, 2.5V.

from 1, 2, i know there must something to ground resistance reduced, i eventually disconnect the JFET, and other circuit, this phenomenon still there. from 3 i also should confirm the JFET is actually work well.   

  seems ALL problem solved now. 

Performance tune

before going further discuss, one thing to listen SW: i use this regen receiver, when daytime, only copy few local station and very weak signal, but after 6:00 afternoon, there is bunch of stations arise, well,  to debug and test SW receiver daylight is nightmare meanwhile my poor antenna.

this rig still need tune to get comfortable listening.

* the 20pf-360pf with a 60:1  vernier  still hard to tune, 4:1 band span is really wide band.(a vernier cover whole band need 30x360 for 180 degree air capacitor ) 

* regen is smooth, frequency drift is very little. but i want another type smooth: oscillation should start gradually. 

there is a ugly solution is add fine tune and fine regen control, i'm not fan of this, but it does work well. i try to series capacitor with 360pf air capacitor to balance between wide-band and the tune Experience. and also try to change jfet bias. 

source bias resistor  testing :1.8K, 2.2k, 3.3k

limiting the tune range, series with: 470pf/330pf/220pf 

i found limiting the tune range and change bias does help a little, but really tiny  little.  

for the second problem, daylight felling and night feeling is totally different, when the signal strong,  you feel like the regen is gradually get strong, but daylight, suffering..  i believe a AAC feed back will make it better, but i don't want to do this this time, my built is compact and not easy to use as  a test bed for AAC. 

exhaust to tune like this,  stop here, it's still FUN.

Ture High performance (PVC really not good for coil Form)

@2014/7/1 

How to wind a coil?  for every version your rig, the coil might different, because of each rig had it's owe character, different JFET, different  mechanical structure different layout,  so  there is reall need tune the coil for your rig.  

here is my Steps to Tune the Coil.

Step 1:  Tune band coverage

•  choose a coil form,  i run out of my PVC tube,  so choose a unknown plastic coil form, it's hard than PVC.

•  guess and calculation the turns you need.

• Initially wind more Tickler coil(L3), like 1/3 of the main coil(L2), and same for Input coil

• testing frequency cover, reduce or increase the main coil(L2)

Step 2: tune the performance 

• reduce the Tickler turns, until at the low band, the receiver just could start oscillation

• reduce the input turns, make balance between strong signal and selective. (too strong input signal pull all station come out together)

Coil form Diameter : 2.7cm

                        input :  5T

                        main : 14T

                      tickler : 4T

              cover band: 5Mhz-18Mhz  (cover all band i like.., one coil, make me suspect plugin is useless)

Then i use this new coil, the dramatically high performance.  sorry to use so large font to express feeling. 

• on whole band it can cover, it's easy start oscillation, NEVER need limited it's band coverage, nothing like the original coil which low frequency hard to oscillating, every tickler is almost 1/2  of tank's turn

• on whole band ,  very sensitive, pickup too many local noise, compare to original coil which so quietly, make people feel like this is another kind of high performance, but it's not.  how sensitive this could be?  from far to start oscillation, there is so many strong SW station it can pick up, and very LOUD than original one. and i begin to believe the input attenuation and volume control is useable. 

• tune experience is very good,  a station is gradually pulling in and gradually fade out.  (compare to original coil)

• Regn control get much better, because it's sensitive, make the regen tune like very gradually get strong and strong. (compare to original coil, this is amazing) 

I begin to suspect i Never make this rig to be high performance till now.

What's the reason?  i change the schematic: remove 1M(R3) resistor and the 100pf(C5) capacitor,   does this make it so differently?  No  because i change back to my original coil, the low band still can not start oscillation.

I now very sure the coil form, the PVC, is very bad (at least my pvc tube i believe there are many type of pvc actually).  i had read several time, how bad the PCV is, now get really sense of this. not just make low Q, also absorb the energy so much , even prevent  oscillation. 

there does another little pitfall to show you: 

on very low band( 5-5.5Mhz), when oscillation is weak, there is parasitic high frequency oscillation, but advance rege a little the high frequency despair.  

another problem is while advance regen too much, the oscillation wave will became totally chaos.  covering such wide band, this pitfall is really acceptable for me.

Trying High frequency regen

@2014/6/29

on WWW, most of regen work up to 18Mhz - 20Mhz,  more than 20Mhz,  difficult thing arise, like mechanical design, layout,  parasitic capacitance / inductor. My design is compact and short lead, but air variable capacitor is connecting via long wire. but anyway i try:

T68-6  12T， tickle: 3T,    ~0.7uH

Experience:

    * this coil cover 14Mhz to around 30Mhz.

    * from 14Mhz to 20Mhz the regen control basically work.

    * above 20MHz,  the feed back is so strong make regen control is very hard, and with strong regn, there is low frequency stray oscillation.

    * from 14Mhz to 20Mhz,  there is no such many  SW stations like 6Mhz to 12Mhz.  and seems single station span more bandwidth than 6Mhz-12Mhz, might because the selectivity is decreased in high frequency, or the signal is more strong use the same antenna.

    * above 20Mhz, here is almost no station... or my regen is does not work correctly. 

 i realize the tickler coil is too much turns, 1 turn might enough for such high frequncy. @2014/7/2

take a glance at hybrid feedback oscillator

i received comments from qrp-gaijin, and recommend me a good site http://theradioboard.com/, and good design. vladn offer a hybrid feedback oscillator configuration to achieve flat amplitude response in a wide range band.  compare to AGC amplitude control, this is useful when use as DIPer/Regen, cause hybrid feedback get the flat response via frequency flat response feedback path,  not by gain control which will make ‘dip’ unnoticeable.

I don't want to wait, so just change a little the wiring, get a hybrid feedback design. most of  change is done by changing coil plugin.

schematic change:

plugin Coil change:

Performance review

* this is not a regen for now, just a oscillator

* use this simple configuration, from 5Mhz to 12Mhz,  the oscillator amplitude keep almost flat: only about 3dB's change in the middle  of  it's band.

anyway this is really a good feedback method.