EMRFD Message Archive 13585

Message Date From Subject
13585 2017-02-11 17:08:07 w0pwe Trying to Improve Receiver IP3
Post for EMRFD
I recently built a Bitx like transceiver from scratch using the Termination Insensitive Amplifiers that Wes Hayward and Bob Kopski presented several years ago. I designed a board for the amplifiers and built them with surface mount components. The circuit basically follows Farhan’s latest schematic on the HFSigs.com page except for the bidirectional amplifiers and my filter is designed for 6 crystals rather than 4. I’m using a Si5351 for the VFO and BFO and I feed those directly into the DBM and product detector. The IF is 12MHz and the VFO tunes down from 5MHz.
 
The rig works great except for some annoying receiver IMD at night on 40M. I first noticed this when I heard W1AW code practice (7.0475Mhz) at 7.168. Other CW stations and digital signals from about 120KHz lower can also be heard in the phone band. I’m pretty sure I am also hearing some AM SW broadcast stations in band that are actually out of band. I was going to just put up with this behavior but I can’t resist the opportunity to venture into some areas of receiver performance where I have never been before and hopefully learn a little along the way.
 
So I studied a little on intermodulation distortion and discovered the two tone IP3 test. Conducting this test as described in EMRFD, I concluded that the IP3 is about -22dBm. That doesn’t sound like a good number so I started trying a few things. Below is a summary:
  1. Tried Si5351 drive levels of 2,4,6,8 ma. - no improvement 
  2. Add 5MHz lowpass filter between VFO and DBM – no improvement
  3. Replace homebrew DBM with Minicircuits TFM-2 mixer – no improvement
  4. Reduce gain of first amplifier (pre-DBM) from 24 to 15dB – IP3 improved to -15dBm but MDS is reduced from -120 to -115dBm.
  5. Change gain of first amplifier to 19dB – IP3 is -5dBm and MDS is now acceptable at -120dBm.
 
The slight improvement in IP3 hasn't made a significant difference in on air performance so I need to keep trying. I have samples of some nice Phillips RF transistors and I wonder if using those in the first amplifier instead of the cheap 2N2222 devices would yield any improvement? Any other suggestions?
Thanks and 73
Jerry – W0PWE
 

 


13586 2017-02-11 19:07:44 Lasse Moell Re: Trying to Improve Receiver IP3
Just a few random thoughts:

Have you looked at the spectrum out from your Si5351? I guess you can use any receiver to listen in on and around the carrier to verify no spurs causing problems with mixing in unwanted signals.

Verify that the LO power is enough. Too low will affect the performance!

24 dB gain before a DBM on 7 MHz? Even 15 dB is too much in my book. A proper design, looking at the IF port would have a noise figure less than 10 dB. Adding the DBM would set the NF to 15 dB or less. Having an antenna that is worth using should give a noise increase so no amplification needed.

As you found, gain before the mixer will lower the IP3. A simple DBM will have +13 dBm at the input with 7 dBm drive at the LO port. Lower LO power will affect the IP3.

73
Lasse SM5GLC

11 februari 2017 04:02:32 +01:00, skrev j.b.hall@ieee.org [emrfd] :
 


Post for EMRFD
I recently built a Bitx like transceiver from scratch using the Termination Insensitive Amplifiers that Wes Hayward and Bob Kopski presented several years ago. I designed a board for the amplifiers and built them with surface mount components. The circuit basically follows Farhan’s latest schematic on the HFSigs.com page except for the bidirectional amplifiers and my filter is designed for 6 crystals rather than 4. I’m using a Si5351 for the VFO and BFO and I feed those directly into the DBM and product detector. The IF is 12MHz and the VFO tunes down from 5MHz.
 
The rig works great except for some annoying receiver IMD at night on 40M. I first noticed this when I heard W1AW code practice (7.0475Mhz) at 7.168. Other CW stations and digital signals from about 120KHz lower can also be heard in the phone band. I’m pretty sure I am also hearing some AM SW broadcast stations in band that are actually out of band. I was going to just put up with this behavior but I can’t resist the opportunity to venture into some areas of receiver performance where I have never been before and hopefully learn a little along the way.
 
So I studied a little on intermodulation distortion and discovered the two tone IP3 test. Conducting this test as described in EMRFD, I concluded that the IP3 is about -22dBm. That doesn’t sound like a good number so I started trying a few things. Below is a summary:
  1. Tried Si5351 drive levels of 2,4,6,8 ma. - no improvement 
  2. Add 5MHz lowpass filter between VFO and DBM – no improvement
  3. Replace homebrew DBM with Minicircuits TFM-2 mixer – no improvement
  4. Reduce gain of first amplifier (pre-DBM) from 24 to 15dB – IP3 improved to -15dBm but MDS is reduced from -120 to -115dBm.
  5. Change gain of first amplifier to 19dB – IP3 is -5dBm and MDS is now acceptable at -120dBm.
 
The slight improvement in IP3 hasn't made a significant difference in on air performance so I need to keep trying. I have samples of some nice Phillips RF transistors and I wonder if using those in the first amplifier instead of the cheap 2N2222 devices would yield any improvement? Any other suggestions?
Thanks and 73
Jerry – W0PWE
 


 





13587 2017-02-11 19:17:24 Ashhar Farhan Re: Trying to Improve Receiver IP3
try throwing out the premixer rf amp. it is useless at 7 mhz. it retrospect, there is too much gain in the bitx.
- f

13588 2017-02-11 20:02:30 bob_ledoux Re: Trying to Improve Receiver IP3
I saw your Bitx Digi-TIA on Soldersmoke for January 22, 2017.  Do you have further detail on a blog or elsewhere on the web?

bob-N7SUR
13591 2017-02-14 11:07:47 w0pwe Re: Trying to Improve Receiver IP3

Thanks to everyone for the ideas and comments. I have made a little improvement but the SW broadcast stations are still hitting me pretty hard. There is a SW station above 7300KHz  (in Indiana I think) that is very strong when they start up in the evening. Last night I connected my antenna to the spectrum analyzer and saw them at about -25dBm.


Concerning the Si5351 output - when I first set this up I measured output with my AD8307 power meter and it looked like the 6ma output setting produced about 8dBm. But of course the power meter expects a sine wave. I read somewhere that second harmonic energy can cause the mixer diodes to remain in conduction when they should be off but the 5MHz low pass filter didn't help and neither did the adjusting the SI5351 output settings.


When I get back to this I will try it without the pre-mixer amp.


Bob - I do not have a blog. There really isn't much about my rig that is unique but I would be happy to provide some more details if you have questions.

73 - W0PWE

13600 2017-02-17 11:19:56 kb1gmx Re: Trying to Improve Receiver IP3
Some thoughts on improving the bitx...

First so there is no bad feelings Frhan did a bang up job with the design and its persistence
speaks loads of the reproduceability and utility of the design. 

The yabut..

The BITX was not intended to be the high performance transceiver only low cost and
its avoidance of hard to get parts.  So there are compromises, a hard fact of engineering life.

Things that may help or improve it based on the 40M version I built years ago.

The mixers using 1N914/4148 diodes are not level 7 mixers (nominal 7dbm drive) like commercial DBMs.
the reason for this is Schottky diodes have a lower turn on threshold than silicon junction diodes with the difference being .2V compared to .65V.  This means you need more LO drive for the terminal impedance 
of all ports to be 50 ohms.  It also means they are higher level mixers by about 3db.  So enough drive is important to intermod and overload.

Second the terminal impedance  of the crystal filter and the various RF/IF amplifiers were optimized for 200 ohms the mixers are better at 50 ohms and there is no impedance matching to correct for that.  This means that the DBMs are compromised in performance again for overload and IMD.  The fix here is to insure all ports
especially the IF port is matched to 50 ohms.

It many radios I've built the DBM to IF amp has a Diplexer.  Th reason for this is to insure the IF port sees a wide band (dc to 100mhz) match and only the desired IF pass though it.  This keeps reflected signals from reentering the mixer and adding to the possible products.

Choice of LO frequency.  Unless there is a defining reason I use a LO that is Higher than the IF as in 40M
with 12mhz if that would be 19mhz.  For  VFO that would be drifty but for a NCO (Si5351) its no big deal.
Why?  When mixing signals there are two known players (IF and LO) and many live signals in the band
and they all mix so you get sums and differences and then harmonics of all those making sums and differences.  A program like spurtune can list them all out and show what the result and likely strength 
of each will be.  When the LO is lower than the IF the likely possibilities and their harmonic mixes are 
more numerous at or neat the RX pass band and IF passband.    Again a low pass filter between the 
LO and the mixer can help sometimes.

RF amplifier...  For bands below 10mhz its likely not needed or needs to be very low gain as there are an abundance of strong signals.  RF selectivity before the preamp is an aid in this and even adding switchable
attenuators (I use 6 and 12DB so I can get 6/12/18db of attenuation) and in a strong signal cases that can 
help.   Lowering the gain of the RF amp (for RX) can help as well.  Another item is if the RF amp is not robust enough it can easily overload before the mixer, at that point all is lost to IMD.  What high end radios do is use lots of current in the RF amp so its not easily overloaded then use a mixer that can tolerate that as well as its no sense having the RF amp be clean and overload the mixer.

Its important to point out that for a given RF amp design and DC bias level there is a maximum signal level
that will exceed it distortion capabilities.  Differently said that amp has a maximum undistorted power out
that must be spread over all the accepted signals often that means the amp must be very robust.  An 
example is a RF amp I used in a radio that had to withstand 10dbm or more at the input in band 
and not overload.  The amp ran at 160ma and could deliver .4W of two tone signal undistorted (better than 35db down).  It actually used RF power devices (2xMRF584) to get that level of capability.  Of course the next stage had to deal with that.  The end result was a crunch proof radio but would be unforgiving about power 
used (RX needed 1.5A with .4A in the low level RF sections).  It is sometimes easier to attenuate the 
offending signals (as well as desired) as a strategy.  Why? because even if the RF amp is good enough 
and the mixer then the first IF (and maybe even second) need to be able to handle all that signal.  In the  end it tend to be a very "system" level problem rather than point solution.

Other tricks are front end preselection using narrow tuneable band pass filters (loss is tolerable)
to reject the offending players.  Notch filters as well though at higher frequencies they may not 
be effective enough.

Consider the case:

Offending signal of -25dbm which is very strong.  Add 17db of RF gain to that and its now -8dbm and any 
mixer below level 17 (50mw LO drive) will be overloaded.  To make matters worse if the RF amp is running 
less than a maybe 10ma it will be overloaded itself as I've seen this.  So we omit the RF amp and try again 
and a -25dbm signal is right below the limits for distortion for a level 7 (5mw lo drive) mixer.  At this 
point 6DB of attenuation of RX is an aid as then the signal is down to -31dbm.  Even without the RF amp 
the RX is sensitive enough to hear most likely signals your going to talk to and if need be you can even up 
the audio gain to compensate to a point.

They key is managing the levels of all the signal passing through the RF and mixer or overload will be 
a very negative result.   Excess gain often deemed desirable are not always helpful.

My first pass with mine was the described switchable two step attenuator and then the ability to switch 
out the RX RF amp completely.  Note that the Elecraft K2 (which has a very good receiver) took this 
path.  FYI: small sugar cube relays are handy for this as they can be placed close to here needed and 
powered from a front panel switch.

The BITX is a great experimenters radio and this is one area where experimenting can be useful of not 
required.  


Allison


13604 2017-02-18 09:36:29 iam74@rocketmail.... Re: Trying to Improve Receiver IP3
Allison --

It seems that I boasted about this info and yourself to the BITX20@io group (which is where the Yahoo
group moved). Many of them are not members of this group and have queried about it. Almost all of the
questions, etc. nowadays are about the BITX40 and the Raduino. It has made quite a splash.

It would be appreciated if you would post this or something similar to that group...or I could
(with your explicit permission) copy this message for them. They are very interested in
all things BITX (There have been over 3200 messages since AF posted about the new BITX40
in November 2016). They include a great many new hams and RF experimenters.

TNX 73

john
AD5YE
13605 2017-02-18 09:37:44 winston376 Re: Trying to Improve Receiver IP3
Allison,

Excellent tutorial.
For the readers, let me emphasize what Allison has stated in her email: A systems architecture approach, which includes the various signals in the passband environment is critical to a flexible and robust receiver design. 
 
An additional "Next Level" design element is the inclusion of interstage electric field shielding which goes to the heart of system design self compatibility and internal noise reduction.  You won't find this in most "Ham" receiver designs due to cost considerations.  How many amateurs have $ 10,000 +  for a ham transceiver?

For the foregoing "Next Level" performance levels, you need to look at professional systems that are used for 'instrumentation' / 'verification' in certain engineering areas.  Known system designs by the old Singer,  Singer/Stoddard, Watkins Johnson, Harris,  etc., would be highly instructive.  

Alex
13606 2017-02-18 09:40:13 w0pwe Re: Trying to Improve Receiver IP3

Thanks for all the info/comments Allison. And you are right, this discussion should in no way tarnish the reputation of the Bitx or the wonderful work that Farhan has done. In the 4 weeks I have had the rig on the air I have had many great QSOs and I have received excellent signal reports even under the close scrutiny of the many SDR radios out there. I could be very happy with it just the way it is, but this minor issue I am having is an excellent opportunity for me to learn more about receiver performance so I am taking advantage of that. If I improve the receiver a little in the process that is great. If not I will continue to enjoy operating the Bitx and being amazed by it’s performance.

 

I am using the Termination Insensitive Amplifiers that Wes and Bob developed in my build so I think the DBM sees a reasonably good match on the RF and IF ports. My IF filter with 6 crystals looked like it was going to be about 200 ohms so I am using  4:1 transformers on each side of it. The diplexer approach you mentioned is very interesting and one that I will try to remember for future projects.

 

I reduced the gain of the RF amplifier to 10dB and receiver IP3 improved from -22dBm to -14.5dBm. I forgot to check the IP3 of the amplifier itself after the gain change. The 19dB gain configuration had an IP3 of about -2dBm for the amplifier itself. I tried bypassing the RF amp but the receiver was pretty deaf. Turning up the audio gain picks up some noise from the Arduino, but that might be better once I clean up my wiring and put it in a box. I may try that again when I get everything situated.

 

I will try moving the LO up to 19MHz as you have suggested. Some discussion on the Bitx group also recommends that move as the 4th harmonic of the 5Mhz LO is especially troublesome with the SW stations above band. Running the numbers, this would explain why the station on 7385 shows up on about 7156. I have a low pass filter on the Si5351 output that could be made more effective at 19MHz but it is pretty easy to change the LO freq. Thanks for the spurtune suggestion. Sounds like a great tool to improve my understanding of the situation. I will give it a try.

 

As I write this on Friday evening I am listening to 40M and the ARRL CW DX contest is in full swing.  I have my step attenuator in line and it seems that many of the offending signals can be eliminated with 6dB of attenuation. 12dB takes care of all but the strongest of them. So the sugar cube relays with 6 and 12dB attenuators may be the way to go. I guess I could also just put a pot on the receiver input. I think the termination insensitive amp will tolerate the odd impedance input and look like 50 ohms to the mixer. Most everyone I talk to on 40M is running 1500 watts so I can spare a few dB on the receive end.

73 – Jerry – W0PWE

13608 2017-02-19 12:18:57 kb1gmx Re: Trying to Improve Receiver IP3
By all means copy and paste, permission granted.

FYI I am there but... time and all.

Allison
13609 2017-02-19 18:23:45 iam74@rocketmail.... Re: Trying to Improve Receiver IP3
Thank you, Allison.

Consider it done.

john
AD5YE