Active Lowpass Filter and AGC Noodling – QSD Receiver

Click Here to see the project from the beginning

Now that I have the signal chain functioning from antenna input to speaker output, I’m starting to work on the final missing function block – the steep-skirted audio lowpass filter. One of the PCBs I developed is an 8-pole active lowpass filter.

The 8-poie lowpass filter PCB

This will go just before the audio power amp which is presently an EBay LM386 module. I have a PCB for a TDA7052A which is an audio power amp with 80dB of voltage controlled gain reduction.

The TDA7052A Audio Amp PCB

One of the things I want to try is sampling the audio amplitude with a Nano A-D input and a software peak detector and using that number to index through a lookup table which gives a control voltage value to keep the final audio output constant. This “feedforward” method, if I can make it work. has several advantages which I’ve written about.

I hope to have the lowpass filter built up and in the signal chain this week.

Stay tuned

Joe

Advertisement

Gain Re-Distribution – QSD Receiver

Click Here to see the project from the beginning

I went back to the drawing board to avoid what could have been a train wreck. My original design premise was to keep the amount of gain before the steep active lowpass filters as low as possible in order to prevent strong off-channel signals from overloading the subsequent high-gain stages. I neglected to model the noise performance of the unity-gain phase-shift network and associated active filter stages, thinking the 30+ dB of gain in the QSD post-amp would get the signal strong enough that subsequent noise wouldn’t be an issue. When I did model it late in the game, it was a revelation! So I went back and re-purposed one of the pre-filter stages immediately following the phase-shift/lowpass network to turn the last stage into an amplifier with a voltage gain of about 10x. (“20 dB” if you misuse “dB” as is done in common parlance). Later, I want to explore lowering the QSD pre-amp gain and raising the gain later in the chain. It’s a trade off between best sensitivity and best overall dynamic range.

I haven’t measured the audio output S/N ratio, but my ear test says it’s about 10dB at 1.0 uV or better. I’m going to go with that for now.

I posted the modified schematic on the PSN board. You can see it here.

Next up is the final steep lowpass network that will set the overall bandwidth to about 2.5 kHz with pretty sharp roll off.

Noise ! – QSD Receiver

Click Here to see the project from the beginning

This morning while driving to the office, I had a frightening realization. The post amplifiers at the I & Q outputs of the QSD are a nice low noise design with an intended 30 dB gain. It’s closer to 40dB at the moment, but I really don’t want it that high in the long run. Without analyzing carefully, I’ve been assuming that much low noise gain would get weak signals (~ 0.5uV) up to an amplitude where I didn’t have to be concerned about the noise contribution of succeeding stages. Wrong!

The audio phase shift network (PSN), and especially the subsequent proposed active audio filters all have relatively large value resistors in the signal path. Each of those resistors is a noise generator. When I did an LTSpice noise analysis today, it became frighteningly obvious that I was headed for disaster. I now need to go back and re-scale the R and C values, and possibly adjust the gain distribution, to fix that. My major concern is the proposed audio lowpass filter design, since preliminary testing with my oscilloscope indicated that the sensitivity up to and through the phase-shift network is acceptable. However, that measurement was quick and dirty and was limited by how low my scope could read, so I’m not so sure now.

To verify that, I want to build a low noise test amplifier to boost the PSN output up to the point where I can more accurately measure the S/N ratio on the scope. With a test amp that can drive 50 ohms, I can also measure true audio S/N with my Textronix 495P spectrum analyzer, as that instrument can measure down to below 100 hZ with 10 hZ resolution. That will tell me how much sensitivity I’m actually losing in the PSN network, and what I need to do going forward in order to not degrade it much more.

Two steps forward and one step back. Typical.

QSD Receiver – Audio Phase Shift Network Added

Click Here to see the project from the beginning

Today I added the audio phase-shift / lowpass filter board and did a quick sideband suppression test. It looks like 45-50 dB across the audio band, which is close to what the LTSpice analysis predicts. The receiver is now a single signal receiver; ie, a direct conversion receiver with a suppressed image response..

After the present lash-up is tested, I’ll add more audio lowpass filtering and an audio inter-stage amplifier. Gain distribution will be an important issue as I proceed from here. Too much gain at the wrong places and strong signal handling will be affected. Not enough gain  and the sensitivity will suffer. I’ll write about this as I proceed.

I’ll report more details as I get time to work on it.

Click Here for design details on the phase-shift / audio filter board

Click Here for details on all the boards

 

 

QSD Receiver – Antenna to Speaker Test

I got my replacement LT6231 op-amps and installed them on the QSD board, but only after mangling the fragile pads and traces. I had to install about 5 ‘fly wires’ using thin wire-wrap wire. I put a glob of hot melt glue over the part and its fragile connections to make it less susceptible to damage.

I had a little EBay LM386 amplifier module laying around, so I decided to put everything together on a piece of copper-clad board and do a little antenna-to-speaker test. No phasing network, no audio filter, no audio post detection filter yet. Still, it receives a clear audio tone with a 3uv signal from my HP8640B signal generator.

Next step is the audio phase-shift module, which will make this a single signal receiver. Then an active lowpass filter and about 20db to 30db more gain before the audio power amplifier.

Click Here to see the project from the beginning.

QSD Radio Update

 

The display running thr system GUI

I’m somewhat dead in the water right now, as I managed to damage my last LT6231 op-amp that’s used as the QSD post amplifier. This is a really nice low noise, high bandwidth rail-to-rail part, but it costs upwards of $5, so I only keep a few in ‘stock’. Over the months, I burned through all of them (sometimes almost literally), so now waiting for a delivery from DigiKey.

I added some details about the boards. Click here to see the revised pages.

Click Here to see the project from the beginning.

Corona Virus Quarantine Project! – The QSD Receiver

Click here for more details…

Click here to follow the blog…

It’s been a while since my last post. Over the past year or so, I’been quietly running simulations of circuit ideas I come up with and designing PCBs to test out those ideas. Consequently, I now have a bunch of boards for a lot of radio sub-systems and I’m finally getting around to putting some together to make functional radios. During this Corona Virus quarantine, I decided to put some of them together to form a high performance QSD phasing receiver. It’s now somewhat functional, but I want to see it through and eventually turn it into a multi-band transceiver.

Check back here regularly as I’ll be updating all the pages as I progress.

The QSD Receiver Project

The display running the system GUI

I’m especially smitten with the quadrature sampling down-conversion (QSD) method of receiver design, with hardware as well as DSP backends. In the coming days, I’m going to roll out a receiver (upgradable to a transceiver) using the ‘phasing’ method of demodulation/modulation. I have a cute mini-Altoids size demod board that implements a QSD down-converter and an  on-board Si5351 LO module directly outputting quadrature LO drive for the QSD.  I also have a board for a hardware phase-shift network as well as boards for the audio stages. I wrote my own ‘slim’ Si5351 handler functions for that (no pesky libraries). Another board has a color TFT display with 2 rotary encoders as well as touch-screen and hardware push-buttons.

It’s all controlled by an Arduino Nano board that piggy-backs on the display board. Code is written for the system and today I put all the pieces together and demodulated a 40M signal from my signal generator, outputting I-Q audio signals to my scope. These will go to a hardware audio phase shift network, an audio lowpass filter an an audio power amp. Alternatively, the I-Q signals can be inputted to an Arduino Teensy to implement a full DSP receiver. All of the PCB blocks have been built and tested, so now it’s a task of system integration, testing, tweaking and implementing changes.

I’ll start off in a few days with photos and descriptions of some of the boards, then regular progress reports as the system comes together. I’ll post schematics and PCB gerber files as soon as each block matures to the point where I think others can reproduce it. Along the way, I intend to write some ‘tech notes’ about how things work in a non-mathematical and intuitive style. I can’t promise how quickly I’ll progress, especially with Summer and my outdoor activities, but I’ll be spending more time on this as I ease into retirement from my company.

Click here for details…

Update: April 13, 2020

As of today, the system is built up to the audio output, minus the final 8-pole lowpass filter and AGC loop. All the printed circuit boards are mounted on a temporary breadboard. Sensitivity (by ear) is around 0.3 uV in a 2.5 kHz SSB bandwidth. There is some digital noise down around the microvolt level which I’ll be working on. Next major step is to build up the 8-pole audio lowpass filter and insert it in the signal chain. This will set overall system selectivity and provide very sharp skirts. I also want to start making accurate instrument measurements of performance, which I’ll post here.

Stay tuned!

Joe
W3JDR

 

Shut down with test equipment problems

My silence recently hasn’t been due to lack of interest. For the last 3 weeks or so, my bench has been filled by the innards of my Tektronix 495p spectrum analyzer. A short in one of the PCB cards took down the power supply. In the process of troubleshooting it, I damaged one of the other cards and had to buy a new one. The unit is now alive again,  but I’m working out a few kinks that might have been there since I purchased it about 5 years ago. I really love this instrument. I considered buying one of the new Rigol units, but the 30 year old 495P outperforms it in every respect except cool display functions.

I also got involved with a high priority project at work which has consumed much of my time. I hope to be reporting on a tested and working Raduino Pill in the next few weeks. I’ll clean up the PCB design, buy/test some more prototypes, and then release the design files for public use.

Joe

Raduino Pill – All Systems Now Running

Today I got the Si5351 running, so all the sub systems (LCD, encoder, Si5351) are now operating. Next step is to see if I can build stock uBitX code to run on the new platform.

Raduino Pill

The board is now working with the stock LCD and encoder. Next step is to test out the Si5351 circuits, then convert the stock uBitX20 firmware and see if I can make the whole radio play like it did with the original Raduino board.