EMRFD Message Archive 9629
Message Date From Subject 9629 2014-01-01 18:03:40 artus1947 Measurement Receiver I have been posting to the PHSNA Yahoo group and thought I might post something about my project over here. It started with the PHSNA project and the desire to look at transmitter spurious emissions. I followed the EMRFD receiver design on pg 7.33. Wes provide me some guidance on the filter programs and I was able to select 5 fairly well matched crystal as measured with the PMSNA and a crystal jig. I used an ADE-1+ and SA614AD. Also change the V+ to the LM358 to 6 volts for the Arduino ADC range.
I posted a few images in the Photos section. One is a plot and the others are of the setup. I used a DDS-60 mounted on an Arduino shield with a UNO. I am using the PMSNA latest sketch modified to shift the IF, and generate the second and third harmonic as well as the fundamental. I find the shift is frequencies of the harmonics shown on the plots curious. I don't know it this may be a error in the algorithm used to generate the 9851 frequency or some other non linearity in the system.
The plots start to flattop above an input signal greater that -60 dBm. I suppose I can work with that but perhaps increasing the pi pad on the receiver may be warranted, The 2n5109 stage shows a +20 dB gain. The pi-pad and crystal show a -6 dB. I still have to calibrate the baseline then I'll try a real world transmitter.
73 and Happy New Year
Jack (W0FNQ)
9630 2014-01-01 19:03:16 Ashhar Farhan Re: Measurement Receiver
Have you posted the circuit in emrfd or group's file sections as well?
Or do you have it on a website that doesn't need subscription like
yahoo groups do? You may want to consider hosting the schematic and
writeup on a website/blog with a link to the yahoo group.
I really want to see your work. It is very exciting to me as measuring
IIP3 through homebrew test equipment can help us raise our art. I will
venture to say that there are three grades of home labs :
first, those who build with a VOM and probably a frequency counter.
These labs can measure DC and frequencies of oscillators. This is good
enough to tune up kits, etc. Adding rudimentary crystal oscillators
and power meter can up their capabilities significantly. This is the
setup I had about 15 years ago.
Second are those who have an oscilloscope. Oscilloscope makes it
possible to see what's going on, measure gain, detect spurs, etc. But
it is a rough measure.
Third are those who work with two clean oscillators and spectrum
display. This makes it possible to accurately measure distortion,
phase noise etc., pushing the performance even higher.
I've spent most of my time last year trying to figure out a $100
solution to the third grade lab.
Measurement receiver under arduino control is an excellent approach.
The arduino's 10 bit adc can help eliminate a difficult to obtain log
amp chip. You can use a germanium diode with very slight bias, then
caliberate the adc output using quick n dirty attenuators made from
regular resistors. Another approach, a little painful on your
solderinf hand is to make a log amp out of twenty odd 2N3904s. The
details are here http://lea.hamradio.si/~s53mv/spectana/sa.html
- f
On 1/2/14, artus1947@yahoo.com <artus1947@yahoo.com> wrote:
> I have been posting to the PHSNA Yahoo group and thought I might post
> something about my project over here. It started with the PHSNA project and
> the desire to look at transmitter spurious emissions. I followed the EMRFD
> receiver design on pg 7.33. Wes provide me some guidance on the filter
> programs and I was able to select 5 fairly well matched crystal as measured
> with the PMSNA and a crystal jig. I used an ADE-1+ and SA614AD. Also
> change the V+ to the LM358 to 6 volts for the Arduino ADC range.
>
>
> I posted a few images in the Photos section. One is a plot and the others
> are of the setup. I used a DDS-60 mounted on an Arduino shield with a UNO.
> I am using the PMSNA latest sketch modified to shift the IF, and generate
> the second and third harmonic as well as the fundamental. I find the shift
> is frequencies of the harmonics shown on the plots curious. I don't know it
> this may be a error in the algorithm used to generate the 9851 frequency or
> some other non linearity in the system.
>
>
>
> The plots start to flattop above an input signal greater that -60 dBm. I
> suppose I can work with that but perhaps increasing the pi pad on the
> receiver may be warranted, The 2n5109 stage shows a +20 dB gain. The pi-pad
> and crystal show a -6 dB. I still have to calibrate the baseline then I'll
> try a real world transmitter.
>
>
> 73 and Happy New Year
> Jack (W0FNQ)
>
--
Sent from my mobile device9631 2014-01-01 19:30:52 artus1947 Re: Measurement Receiver I haven't posted the schematic because it is in Chapter 7 of the EMRFD book. Wes Hayward did all the main work and I just built on his concepts. I would not want to post copyrighted material. His writeup was a sufficient start for me to start the build. Many of the parts were no longer readily available so I used mostly SMD components and did a quicky etched board. This is a purely learning process for me at this point; I would be glad to share other information there are specific questions.
Jack (W0FNQ)
9632 2014-01-01 19:48:17 w7zoi Re: Measurement Receiver Hi Jack, (and Farhan and group)
The overall gain of the front end in Fig 7.60 is not high. The mixer has a loss of 6 dB while the post mixer amplifier has a gain of 20 dB. But there is a 6 dB pad following the amplifier. The crystal filter has a loss of 2.3 dB with the high Q crystals I had available, Qu=240K. That loss goes up to over 5 dB with a more modest Qu of 100K. The result is a net gain from 3 to 6 dB. There should be no "flat topping" or gain compression in the front end at the levels you describe.
The way to build a circuit like this is to begin by testing the NE-604 part. Note L1 and the related L network capacitors. They will transform a 50 Ohm signal up to 500 Ohms. This can be used to bypass the front end and the crystal filter. Just drive from a test generator into the L network and then directly into the 510 Ohm termination that sits at the input to the NE-604. This can be used to calibrate the system at 5 MHz. Once you have this in place and are happy with the results, the front end can be added and used with a VFO to turn it all into a measurement receiver. Be sure that you have the required +7 dBm or so of LO injection for the ring mixer.
This system was used with a wide range step attenuator. This moves the signals to be measured into the dynamic range of the receiver.
Also, if you are interested in transmitter evaluation, don't overlook the scheme of Fig. 7.64. An example of this scheme in use is shown in Fig 6.142.
Good luck with the experiments.
73, Wes
w7zoi
9637 2014-01-05 09:36:58 Ashhar Farhan Re: Measurement Receiver
to measure IIP3? I would seem to be an easy setup to me : two RF
signals combined and fed through an attenuator to a dc receiver. The
output is monitored on an audio spectrum display. Toggle the RF
attenuation to figure out how the peaks increase or decrease and
calculate the IIP3.
- f
On 1/2/14, wesw7zoi@gmail.com <wesw7zoi@gmail.com> wrote:
> Hi Jack, (and Farhan and group)
>
>
> The overall gain of the front end in Fig 7.60 is not high. The mixer has
> a loss of 6 dB while the post mixer amplifier has a gain of 20 dB. But
> there is a 6 dB pad following the amplifier. The crystal filter has a loss
> of 2.3 dB with the high Q crystals I had available, Qu=240K. That loss
> goes up to over 5 dB with a more modest Qu of 100K. The result is a net
> gain from 3 to 6 dB. There should be no "flat topping" or gain compression
> in the front end at the levels you describe.
>
>
> The way to build a circuit like this is to begin by testing the NE-604
> part. Note L1 and the related L network capacitors. They will transform a
> 50 Ohm signal up to 500 Ohms. This can be used to bypass the front end and
> the crystal filter. Just drive from a test generator into the L network
> and then directly into the 510 Ohm termination that sits at the input to the
> NE-604. This can be used to calibrate the system at 5 MHz. Once you have
> this in place and are happy with the results, the front end can be added and
> used with a VFO to turn it all into a measurement receiver. Be sure that
> you have the required +7 dBm or so of LO injection for the ring mixer.
>
>
> This system was used with a wide range step attenuator. This moves the
> signals to be measured into the dynamic range of the receiver.
>
>
> Also, if you are interested in transmitter evaluation, don't overlook the
> scheme of Fig. 7.64. An example of this scheme in use is shown in Fig
> 6.142.
>
>
> Good luck with the experiments.
>
>
> 73, Wes
> w7zoi
>
>
>
--
Sent from my mobile device9638 2014-01-05 11:25:08 w7zoi Re: Measurement Receiver Hello Farhan and group,
That is exactly what Figure 7.64 is all about. Exactly.
When I read your note, I wondered just what I had done wrong when I wrote the text for EMRFD Fig 7.64. So I pulled the book off the shelf and read what I had written. As the kids say these days, "My Bad." I goofed, evidently in a major way. I said so much about DFT and FFT and Baseband and other such technical terms that the basic goal of the box was lost. The reason that the circuit of Fig 7.64 exists is to do exactly what you describe. That is, it was built to examine the spectrum of a SSB transmitter with two-tone excitation.
The physical box is shown in the photograph on the left side of p 7.34. The two tone audio generator that I used for transmitter testing is the one in Fig 7.26 on page 7.14. For a result with a real transmitter, take a look at page 6.81, Fig 6.142.
The original application I envisioned used a two tone audio generator with the audio frequencies within the normal voice band. But it really makes no difference where the two signals originate. One could just as easily use a pair of signal generators. They would be combined with some sort of a hybrid circuit such as the 6 dB hybrid of Fig 7.42. This is, of course, nothing more than different application for a Return Loss Bridge. A 3 dB hybrid, a.k.a. "Magic Tee", could also be used to combine the two generators. See Fig 3.81 and related text.
Don't just hook two signal sourced together, either directly or with a resistor T or Y circuit. The reason you want the hybrid is to isolate the generators. If you just had the two generators hooked together through a few small valued resistors, nearly the full output of one generator would be available at the other generator. But if you use the combiner of Fig 7.42, the output of generator #1 would be attenuated by some 30 dB or more at the output of generator #2, yet would only be attenuated by 6 dB at the load.
The system of Fig 7.64 has limitations. The major one is that the RF signals must be close to each other in frequency. Moreover, they must be close to the frequency of the LO used for the block converter. These requirements result because the audio spectrum analyzer has a limited bandwidth. I did my measurements with an old laptop and FFT program that had only a 20 kHz bandwidth. Modern laptops with more up to date programs may be better. The FFT program that I used, which is no longer available, was capable of a resolution bandwidth of just a few Hz.
Throughout EMRFD, intermodulation distortion measurements are quoted. For my part, most of those measurements were done with crystal controlled sources like that of Fig 7.32 with amplifiers like that of Fig 7.35. More recently, I've used the generator of Fig 7.27 along with a HP generator for the measurements. This was not possible with early versions of this signal source, for there was not sufficient isolation between the generator output and the oscillator itself. Even small signal from a second generator applied to the output of the Fig 7.27 generator would cause FM to occur in the oscillator. This problem was fixed with the common base buffer stage following the diode band switch, D1 and D2. The common base stage has virtually no net gain, but has very low gain in the reverse direction. It's a wonderful buffer.
Whatever generator pair you use, it is important to adjust the signal levels that you apply to the circuit under test and the levels applied to the block converter that forms the spectrum analyzer front end. If you overdrive the analyzer, you will end up measuring the input intercept of the analyzer instead of the circuit under test. (Extra step attenuators are always useful.) This applies to whatever spectrum analyzer you might be using -- overdrive can always be a problem.
So Farhan, you are exactly correct. Build up the circuit of 7.64, hook it to a computer with a suitable FFT program, and give it a try. Thanks for tweaking us about this. Sorry I was not more explicit in my explanation.
73, Wes
w7zoi
9639 2014-01-05 17:40:50 Ashhar Farhan Re: Measurement Receiver
I thought the text was quite clear 'in the context'. Thanks for
pointing out that the word 'combined' should be used carefully. A
hybrid combiner is needed.
For audio two-tone test, I use an audio file played from the PC.
One has to be a little careful with the dynamic range of the PC too.
Just yestedray, on the bench I used this scheme to characterize a
little one transistor phase shift oscillator. On the 465, the waveform
looked fairly clean. But the spectrum display on the pc showed the
second harmonic to be down by just -16dbc. But lowering the mic input
brought further down to -27dbc.
The average sound card's dynamic range on the mic input is quite
limited. A 16 bit card should theoretically 'read' signals with the
precision of 32,786 values. That should give a power ratio of 90 db.
This never happens. The analog circuitry is often noisy and
non-linear. In fact, the non linearity may even be desirable. The
prime usage of the mic input is voice comm using skype. Voice clarity
works better with compressed audio. Some of the mic processors even
have an on-built agc! I have long struggles with bad mic inputs at my
mng qth where we were developing VOIP software. They are a modest
design.
- farhan
On 1/6/14, wesw7zoi@gmail.com <wesw7zoi@gmail.com> wrote:
> Hello Farhan and group,
>
>
> That is exactly what Figure 7.64 is all about. Exactly.
>
>
> When I read your note, I wondered just what I had done wrong when I wrote
> the text for EMRFD Fig 7.64. So I pulled the book off the shelf and read
> what I had written. As the kids say these days, "My Bad." I goofed,
> evidently in a major way. I said so much about DFT and FFT and Baseband
> and other such technical terms that the basic goal of the box was lost.
> The reason that the circuit of Fig 7.64 exists is to do exactly what you
> describe. That is, it was built to examine the spectrum of a SSB
> transmitter with two-tone excitation.
>
>
> The physical box is shown in the photograph on the left side of p 7.34.
> The two tone audio generator that I used for transmitter testing is the one
> in Fig 7.26 on page 7.14. For a result with a real transmitter, take a
> look at page 6.81, Fig 6.142.
>
>
> The original application I envisioned used a two tone audio generator with
> the audio frequencies within the normal voice band. But it really makes no
> difference where the two signals originate. One could just as easily use a
> pair of signal generators. They would be combined with some sort of a
> hybrid circuit such as the 6 dB hybrid of Fig 7.42. This is, of course,
> nothing more than different application for a Return Loss Bridge. A 3 dB
> hybrid, a.k.a. "Magic Tee", could also be used to combine the two
> generators. See Fig 3.81 and related text.
>
>
> Don't just hook two signal sourced together, either directly or with a
> resistor T or Y circuit. The reason you want the hybrid is to isolate the
> generators. If you just had the two generators hooked together through a
> few small valued resistors, nearly the full output of one generator would be
> available at the other generator. But if you use the combiner of Fig 7.42,
> the output of generator #1 would be attenuated by some 30 dB or more at the
> output of generator #2, yet would only be attenuated by 6 dB at the load.
>
>
>
> The system of Fig 7.64 has limitations. The major one is that the RF
> signals must be close to each other in frequency. Moreover, they must be
> close to the frequency of the LO used for the block converter. These
> requirements result because the audio spectrum analyzer has a limited
> bandwidth. I did my measurements with an old laptop and FFT program that
> had only a 20 kHz bandwidth. Modern laptops with more up to date programs
> may be better. The FFT program that I used, which is no longer available,
> was capable of a resolution bandwidth of just a few Hz.
>
>
> Throughout EMRFD, intermodulation distortion measurements are quoted.
> For my part, most of those measurements were done with crystal controlled
> sources like that of Fig 7.32 with amplifiers like that of Fig 7.35. More
> recently, I've used the generator of Fig 7.27 along with a HP generator for
> the measurements. This was not possible with early versions of this signal
> source, for there was not sufficient isolation between the generator output
> and the oscillator itself. Even small signal from a second generator
> applied to the output of the Fig 7.27 generator would cause FM to occur in
> the oscillator. This problem was fixed with the common base buffer stage
> following the diode band switch, D1 and D2. The common base stage has
> virtually no net gain, but has very low gain in the reverse direction.
> It's a wonderful buffer.
>
>
> Whatever generator pair you use, it is important to adjust the signal
> levels that you apply to the circuit under test and the levels applied to
> the block converter that forms the spectrum analyzer front end. If you
> overdrive the analyzer, you will end up measuring the input intercept of the
> analyzer instead of the circuit under test. (Extra step attenuators are
> always useful.) This applies to whatever spectrum analyzer you might be
> using -- overdrive can always be a problem.
>
>
> So Farhan, you are exactly correct. Build up the circuit of 7.64, hook it
> to a computer with a suitable FFT program, and give it a try. Thanks for
> tweaking us about this. Sorry I was not more explicit in my explanation.
>
>
> 73, Wes
> w7zoi
>
>
>
--
Sent from my mobile device9658 2014-01-17 05:02:28 artus1947 Re: Measurement Receiver Folks,
I posted a few more photos of my measurement receiver adventure. The dB scale is arbitrary because I used a loose link couple to sample my dummy load. The system is too sensitive at this point -- working on that issue. At least its starting to look like what I expected. Holidays and a Panama Canal cruise got in the way of progress.
73,
Jack (W0FNQ)