Hardware for COHIRADIA: Details and problems/solutions

ID: 567819
Hardware for COHIRADIA: Details and problems/solutions 
03.Apr.22 13:47
29

Hermann Scharfetter (A)
Moderator
Articles: 49
Count of Thanks: 1
Hermann Scharfetter

Dear friends of COHIRADIA,

In order to keep the main COHIRADIA thread compact, from now on all technical contributions about hardware will be concentrated in this new thread. So if you want to discuss technical problems and solutions, please reply here and not in the main thread. Examples could be:

  •     Use of baluns or other coupling methods to the radios,
  •     wideband amplifier for recording
  •     antenna coupling for own recordings
  •     ... and many more

The goal is to keep the main thread compact ad readable in the sense of general and historical aspects. Therefore in the future only issues like

  •     interesting comments on recorded material
  •     reports about the use of COHIRADIA on historical devices
  •     other activities involving COHIRADIA (e.g. in museums, at events or similar)
  •     general comments not involving technical details
  •     .....

should be posted there.

In any case, I am looking forward to exciting suggestions for improvement, problem solutions and discussions around the software.

With best regards

Hermann Scharfetter

To thank the Author because you find the post helpful or well done.

 2
Hardware for COHIRADIA: Details and problems/solutions 
16.Apr.22 09:39
29 from 934

Hermann Scharfetter (A)
Moderator
Articles: 49
Count of Thanks: 1
Hermann Scharfetter

Information about alternative hardware:

For own recordings to COHIRADIA the STEMLAB125-14 is not the only possible device. I have also tested the much cheaper STEMLAB125-10 (currently about 250 Euro), for playback of AM recordings with good dynamics it is definitely an alternative with comparable playback quality. I have not made any recordings with it yet.

Besides the STEMLABs, there are a number of inexpensive SDRs that can write between 12 and 18 bit IQ data in the range of the old AM frequency bands. However, to be able to output signals as well, you need TX-capable SDRs; there exist some products that are used in amateur radio. If you want to achieve appropriate digital resolution (i.e. 12 bit or more), only devices with a price comparable or more expensive than the STEMLAB125-14 appear suitable. A good overview can be found e.g. on the pages of rtl-sdr, especially there is a nice table with a quick comparison of the important key data and the current prices.

The advantage of these devices is a much more sophisticated software environment, which typically also allows to read COHIRADIA data files and play them back digitally on the PC. See my recent experience with SDR# in the software thread.

I intend to test at least one of these devices in the near future, e.g. the one from Airspy or the 2-channel device from SDRPlay. If anyone in the radio museum community has experience with either of these devices with regard to AM-band-recordings, I welcome comments and recommendations.


Best regards

Hermann Scharfetter

 

To thank the Author because you find the post helpful or well done.

 3
Hardware for COHIRADIA: Details and problems/solutions 
16.Apr.22 20:15
44 from 934

Stan Roberts (USA)
Articles: 35
Count of Thanks: 2

Hi Herman.

You requested comments from any of us who may have experience with some of the cheaper SDR receivers. I have an RSP1A from SDRPlay running SDRUno as the free software package that is readily downloadable from the vendor’s website. The latest version of the software has a function called Scheduler that allows you to record/capture the IQ data for the whole of the MW band in one swoop. In fact, the method is described in Section 2.6 of the latest user manual and I found that it worked the first time I tried it. The data are stored as a WAV file, but of course you need the SDRUno software running to replay it. I recorded about 15 minutes, which took up just over 4 GB of storage on two WAV files. (Why two files, I have no idea.) As far as I can tell, one can do this over any displayed section of the spectrum within the receiver’s limitations. The audible quality of the recording is, to me at least, as good as the original AM stations sound.

The one drawback, of course, is that there is no means to play back this record in a true TX mode in order to be able to tune an old radio receiver across its MW band as you described so well in your videos. After all, your solution is what we all want so that when all the AM broadcast transmitting stations have long been retired, then we can pseudo-resurrect them with hopefully, by then, cheap software designed low power broadband transmit solutions. Then again, with modern technology there are no dreamland solutions as the software code may be out of date before the AM stations’ eventual disappearance.

With due respect, I cannot expect the RSP1A to perform both RX and TX functions at the price it is offered, but perhaps before too long there may be some devices which can perform both RX and TX functions at a reasonable price point along with the software to match. To me, as an old-timer, the ease of software is key, and kudos here to the folks who keep updating and improving SDRUno.)

Much that I grew up watching MontyPython when it was originally broadcast on BBC2 television, and I love it still, I am not acquainted at all with the Python programming language. There’s my problem and I admit it! If I could write a script for the Red Pitaya device that you advocate, I’d no doubt call it Norwegian Blue.

Finally, thank you for kicking off this topic again. I find it of great interest and commend you for describing so well your work and progress to date!

Kind regards,

Stan

To thank the Author because you find the post helpful or well done.

 4
Hardware for COHIRADIA: Details and problems/solutions 
20.Apr.22 19:05
63 from 934

Hermann Scharfetter (A)
Moderator
Articles: 49
Count of Thanks: 1
Hermann Scharfetter

Dear Stan,

Thank you for your helpful comments and the recommendation of the RSP1A. I should mention for the other readers that meanwhile we had some email exchange and that you sent me a sample wav-File of an MW-spectrum recorded in Los Gatos (CA) with your RSP1A. I could convert it to the IQ raw data format which I am using and which can also be played back on the PC with the freely downloadable software sdr#. The spectrum has excllent quality and is very rich in broadcasting stations with clear sound. One remaining question is whether the RSP1A splits a longer recording in several subsequent wav-files or if it just stops the recording after some time - what is your experience with that issue ? If one can get a series of  contiguous files they could be easily converted and concatenated. Concerning resolution (14 bits) and frequency range (1 KHz - 2 GHz) the device would then be a very good and affordable solution for recording-only. If you don't mind I would like to upload the converted file to the COHIRADIA server so that other RM members can tune through a Californian spectrum - I found that experience really interesting.

By the way: I like your humor - Norwegian Blue sounds really good for a Python project - so when I need another name for one of my next apps I will ask you before ;-).

Best greetings

Hermann

To thank the Author because you find the post helpful or well done.

 5
Hardware for COHIRADIA: Details and problems/solutions 
22.Apr.22 02:04
78 from 934

Stan Roberts (USA)
Articles: 35
Count of Thanks: 1

Dear Hermann,

’ll try to answer your questions, and the first one really is two questions rolled into one, so here goes:

Q1a. Does the RSP1A split a longer recording into several subsequent wav-files?

A1a. Yes, the software package (SDRUno) that comes with the SDRPlay devices does split the recordings into what I assume to be contiguous files as soon as the file size exceeds 2 GB. However, you advised me that WAV files are limited to 4 GB by design, so I hunted through the SDRUno manual, which is very comprehensive. I found that the default maximum file size was indeed 2048 MB, but that it could be changed to 4096 MB max with a bit of menu fiddling. I was using the default settings in the graphical user workspace, but if you go to Main→Settings→Misc in the top left-hand box of the workspace, you will find a dialog box for WAV File Max Record Length, where you can increase the file size to 4096 MB maximum (just as you advised).

Q1b. Or, does it just stop the recording after some time - what is your experience with that issue?

A1b. To date, I’ve found that it doesn’t stop recording because one is obliged to enter start and stop recording times, or the recording duration as an alternative. As an example, I tried a 30 minute IQ recording of the whole AM (MW) broadcast band today. This resulted in three files, two of 3.99 GB and one of 386 MB. This translates roughly to 280 MB per minute of recording time. I think that suggests that it keeps recording, but expect big files because of it.

Q2. If you don't mind, could I upload the converted file to the COHIRADIA server so that other RM members can tune through a Californian spectrum?

A2. Yes, of course. And as you say, members will likely find this to be very interesting as they can literally tune through the waveband and listen to whatever station they stumble upon for a few minutes. I deliberately started the recording at one minute before the hour, at which time most stations here announce their call sign and often their location. So, initially I found it to be a little like turning back time because, as you also mentioned, the sounds and sound quality are very clear and realistic.

Kind regards,

Stan

To thank the Author because you find the post helpful or well done.

 6
Hardware for COHIRADIA: Details and problems/solutions 
22.Apr.22 11:36
88 from 934

Michael Watterson (IRL)
Editor
Articles: 1033
Count of Thanks: 1

The biggest problem with wideband is dynamic range. A wider band makes it worse for two reasons:
1. Peak level becomes higher
2. More probability of a mix of weak and strong stations.

14 bits might only be enough for part of the band. Regular recievers have AGC which can't be used on wide band recordings.

 

To thank the Author because you find the post helpful or well done.

 7
Hardware for COHIRADIA: Details and problems/solutions 
22.Apr.22 17:42
105 from 934

Stan Roberts (USA)
Articles: 35
Count of Thanks: 1

Hi Michael.

I always like to read your comments on this forum and I agree wholeheartedly with the two points you mention. In fact, I had not even thought of trying to ‘capture’ the whole MW band until Hermann started this column. But then I looked back to the RSP1A I bought a little over a year ago – I bought it as I wanted to learn more about SDRs at the time, not for serious use. Anyway, Hermann’s suggestion piqued my interest in re-discovering what the RSP1A might be able to do. In terms of dynamic range, there’s a 50 kW daytime transmitter at 1170 kHz about ten miles away from this location and it poses no problem with overload or splatter using the RSP1A. As an antenna, I use a ten turn square loop of about 0.5 metre per side on a tripod at 2 metres above the ground out on the patio for the MW band. I used a design that appeared in a 1922 Popular Wireless magazine for the antenna, so it’s nothing fancy, but it suits the few old radios from the late nineteen twenties that I have. However, the loop antenna performs far better than a long wire here because of all the E-field broadband digital noise from modern appliances in our home and also from the 12 kV electricity lines that run by the boundary fence. In terms of cheap SDR versus conventional vacuum tube receiver comparison, I have a Collins 51-J2, my pride and joy boat-anchor, not perhaps known for its wonderful AGC, but a serious receiver all the same even today. Yet the sophisticated filtering of the RSP1A software along with its conventional low and bandpass pre-selection filters make me think that it far surpasses the capabilities of even the best of our old-time radios. I can receive twenty AM stations comfortably on the RSP1A during the day and I can ‘see’ virtually the same number of more distant stations several dBs above the noise, but not very intelligible.

To your final point about 14 bits might only be enough for part of the band, perhaps Hermann can do a better job answering than me. He can give you an impartial opinion on the data file I submitted to him which covered about 1.2 MHz span from 550 kHz to 1.75 MHz. I’m not expert on anything SDR, but I believe the 14 bits refer to the analog to digital converter (ADC) and its performance.

I hope I have addressed the points you made.

Kind regards,

Stan

To thank the Author because you find the post helpful or well done.

 8
Hardware for COHIRADIA: Details and problems/solutions 
23.Apr.22 09:57
121 from 934

Hermann Scharfetter (A)
Moderator
Articles: 49
Count of Thanks: 1
Hermann Scharfetter

Dear Michael, dear Stan,

Resolution is certainly an important issue. Unfortunately I am not a communications engineer, but I try to cast some own ideas on the topic. Concerning the SNR of a broadcasting station I am usually happy with 50 dB (carrier peak/noise floor). In the spectrum provided by Stan the best carrier peak is a bit more than 60dB above noise floor. The quantization noise power sq2 of an ADC is known as A2/(12*2n) with A= dynamic range, n=number of bits. If I assume this as my optimum noise floor then a 14 bit ADC (considering the sign) provides an SNR of 2.8*104, i.e. ca. 89dB, unless I use oversampling.In the STEMLAB the basic sampling rate is 125MS/s, which allows for a 10-fold oversampling in the MW-band, thus, at least theoretically, providing ca. 10dB more SNR. At 14 bits the RSP1A sampling rate is not excessively high (max 6MS/s acc. to the dataheet), so that oversampling does not help a lot. Let's thus assume the 14 bits, i.e. 89dB, as given.

Of course I would leave some headroom for peak perturbations, fadings etc., so let’s generously assume a useful dynamic range of 75dB. Assuming a densely populated MW band with say 100 stations and equal SNR of 50dB each, the total band power would need 20 dB more dynamic range, i.e. 70dB – still the 14 bits would suffice by far.

I agree with Michael that we may face a problem with an extremely strong station. Assume e.g. one with SNR >80dB. Then I would have to reduce the overall gain to fit into the dynamic range and to avoid intermodulations. This of course goes at the cost of the SNR of weak stations. In this example I would lose 5dB unless I exploit the headroom. That is exactly the problem I faced when using he videorecorder with its poor dynamic range and linearity – in those days a 50dB SNR station already could mix into many weaker stations indeed - but that is another story.

AGC would be desirable, because this would reduce the required headroom. Of course AGC should react to the peak power in the spectrum  rather than to the average power – it then acts on the whole band and thus, in unfavorable cases, may adversely affect the weaker stations if there appears an extremely strong transmitter. However, when allowing for less headroom maybe I do not lose too much. Thus I would relax Michael's statement a bit: AGC can be suboptimal for the weaker stations, but it is not impossible.

In my STEMLAB-setup I am using a logarithmic amplifier for monitoring the peak power. I am adjusting the gain manually as I have not yet implemented an AGC, though this would be doable. I am not familiar with typical on-board SDR-AGC implementations, but I hope they can be configured appropriately.
My conclusion: The required resolution depends very much on the RX situation.
In Graz I cannot receive any nearby stations any more and I am happy if, during nighttime, I can get individual SNRs of 50dB. There are no strong peaks in my spectrum, thus I am happy with the 14 bits. However, I would very much appreciate if Stan could record an MW band during night when his Californian spectrum is fully populated – would be nice to see if there are some remarkably strong peaks.

I include a picture showing A Stan's spectrum and one from Graz for comparison. Please take into account that the scaling is not the same, I produced quick and dirty screenshots with SDR#.


If somebody has his/her antenna very close to a strong transmitter then I could in fact imagine that Michael's concerns bekome striking and in that case I would need more than 14 bits. Alternatively I would try to notch this station somewhat with a good filter in front of the LNA. Maybe I should also think about Stans Loop antenna for suppressing some local perturbers (I am curremtly using a long wire in my garden). However, then I would have some directivity, wouldn't I ? For historical recordings this may be unwanted, what do you think ?


I would also be happy with an SDR with more than 14bits. I found the Airspy HF+ discovery, which has 18 bits and promises more advanced signal processing options. The hardware architecture looks extremely advanced and I would prefer this device over the RSP1A, but as far as I understand the maximum recordable bandwidth is 768 kHz, which suffices for most SW bands and LW, but not for MW.

There is a 2-channel device on the market, maybe one can record two adjacent bands simultaneously with that device. If you know about other SDRs with both better resolution and bandwidth I would be extremely interested !

But as I said: I am not an expert and would be happy to learn more details on this issue. Please feel free to criticize my comments if I am wrong.

Best regards

Hermann

To thank the Author because you find the post helpful or well done.

 9
Hardware for COHIRADIA: Details and problems/solutions 
23.Apr.22 10:02
122 from 934

Hermann Scharfetter (A)
Moderator
Articles: 49
Count of Thanks: 2
Hermann Scharfetter

Just forgot to mention that Stan's sample can be downloaded from the COHIRADIA server now as 'raw' file with that link. A *.wav version is available from the link. Thank you, Stan, for the recording.

Best  regards

Hermann

To thank the Author because you find the post helpful or well done.