Forums :: Resources :: Features :: Photo Gallery :: Vintage Radio Shows :: Archives :: Books
Support This Site: Contributors :: Advertise


It is currently Mar Mon 18, 2019 11:19 pm


All times are UTC [ DST ]





Post New Topic Post Reply  [ 7 posts ] 
Author Message
 Post subject: richfair's audio processor for part 15 transmitter
PostPosted: Feb Thu 14, 2019 9:13 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 617
Location: Equinunk PA 18417
Attachment:
processor prototype sml.jpg
processor prototype sml.jpg [ 60.66 KiB | Viewed 319 times ]

This is the audio processing front end for my recent am transmitter, http://antiqueradios.com/forums/viewtopic.php?f=12&t=354147. It has enough gain range to accommodate a weakling Samsung smartphone, a more powerful laptop output, and typical line level audio sources. (Connecting to a headphone jack is discussed later in this thread.) A microphone would still need a preamp. I am not the neighborhood DJ and do not need a full-featured broadcast processor. I just want something simple and flexible and more importantly something that will mount within my little transmitter. Last year I hoped to build a super-simple limiter but have since traded that plan away for this more flexible concept. The next step up would be a multi-band compressor, which would add even more cost and complexity. This unit performs at a very high level. It is designed for a single-voltage supply 9-12 volts 27ma (including LEDs).

Attachment:
processor signal flow diagram sml.jpg
processor signal flow diagram sml.jpg [ 72.87 KiB | Viewed 319 times ]

This is a feed forward compressor design. “Feed forward” considers only the incoming signal. Results of processing do not affect the operation of the processing itself. It is a different approach than compressor/limiters with a “feedback” design, where their detection circuit(s) react to the signal they are processing, so a sort of feedback loop is created. Both approaches have advantages and both have been around forever. I originally conceived a feedback design and tried both with this circuit. The hands down winner was feed forward.

Considerable effort was put into frequency response tailoring with an NRSC mid-frequency boost and high frequency attenuation. In a part 15 transmitter it is okay to exceed commercial bandwidth restrictions and I take advantage of that. There is a standard pre-emphasis boost followed by a low pass filter network that is modeled after NRSC guidelines except it allows additional audio energy above 10kHz, with a more gentle (better!) sounding high frequency response roll-off than the NRSC guidelines allow. I am very pleased with the results but to be honest I don’t hear much above 11kHz anymore so I’ll be very interested in comments from anyone who wants to build this thing!

Attachment:
Xmtr processor freq response with highlight.jpg
Xmtr processor freq response with highlight.jpg [ 35.64 KiB | Viewed 319 times ]

As good as it is, this processor is not designed to make me the loudest on the planet. Without a multi-band compressor, no phase rotator, no digital look-ahead peak limiter, no weird non-symmetrical processing (except for the clippers!), this is just a full-bandwidth compressor with a fixed equalizer that works as well as any in its class. THAT Corp has done a lot of heavy lifting in analog volume controls by advancing/improving iconic DBX vca technology onto a new range of chips. The most daunting aspect of working at a hobby level with their parts is that DIP packages are nearly impossible to buy. I bought a couple THAT 4320 chips in surface mounted QSOP-28 packages and soldered them to adapter boards for prototyping. I chose 4320 because it is optimized for single voltage battery supplies and has everything onboard for high level compressor/limiter functions. The VCA is actually the only part of the 4320 chip that touches the audio signal, and as VCAs go, it is impressively clean and quiet. The bulk of the chip just works with control voltages. I’ve “borrowed” much of the control voltage workings from THAT’s many technical documents.

Power supply should be 9-12vdc although all of the processing works well at only 5 volts, with reduced audio headroom of course. Current draw at 9 volts is 27ma including LEDs. The two LED display circuits draw as much power as the remainder of the processor so if power consumption is a concern, you may want to switch the 0 volt supply rail of IC3, to turn off the displays when not needed.

All comments welcomed. You may wish to download construction/setup notes below. Enjoy!

Attachment:
File comment: download schematic below
audio proc schematic sml.jpg
audio proc schematic sml.jpg [ 89.85 KiB | Viewed 319 times ]

Attachment:
audio proc schematic 2-14-2019.pdf [146.81 KiB]
Downloaded 22 times

notes in 2 parts due to ARF size limits:
Attachment:
Richfair's companion audio processor notes pt1.pdf [252.84 KiB]
Downloaded 14 times

Attachment:
Richfair's companion audio processor notes pt2.pdf [252.73 KiB]
Downloaded 12 times

_________________
-Richard


Last edited by richfair on Feb Tue 19, 2019 4:37 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject: Re: richfair's audio processor for part 15 transmitter
PostPosted: Feb Mon 18, 2019 5:54 am 
Member
User avatar

Joined: Aug Wed 26, 2009 1:50 am
Posts: 1448
Location: Mesquite NV 89027 (from Coventry, UK)
Nice work Richard. One question: if my audio source is stereo, as most are nowadays, what changes would need to be made to the input to sum both channels? The same question applies to the transmitter circuit that you posted, if it were to be used without the audio processor.
Thanks,
Colin


Top
 Profile  
 
 Post subject: Re: richfair's audio processor for part 15 transmitter
PostPosted: Feb Mon 18, 2019 4:01 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 617
Location: Equinunk PA 18417
Hi Colin, that's a good question and you're pointed out something that I meant to mention but somehow just never did. Most people use headphone outputs, and the two most popular methods each work perfectly, both mentioned here: http://www.antiqueradios.com/forums/vie ... 2&t=298678 I'll update the transmitter thread later today.

Personally I prefer the two resistor method. The processor works best with a low impedance source. A pair of 1k resistors becomes a 500 ohm source, which is plenty low for the processor. Depending on what oddball transformers might get used, I'd say resistors are the better generic choice.

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: richfair's audio processor for part 15 transmitter
PostPosted: Feb Mon 18, 2019 4:34 pm 
Member
User avatar

Joined: Aug Wed 26, 2009 1:50 am
Posts: 1448
Location: Mesquite NV 89027 (from Coventry, UK)
Thanks Richard.


Top
 Profile  
 
 Post subject: Re: richfair's audio processor for part 15 transmitter
PostPosted: Feb Tue 19, 2019 4:35 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 617
Location: Equinunk PA 18417
A more complete answer to hooking a stereo output to a mono input:

Most people are satisfied to use headphone outputs from their phone or computer. I do and I use the most simple way, a pair of resistors as shown below. A variation of this using a 1:1 or 2:1 transformer to provide DC isolation between the output and input should work well also. Headphone driver outputs are extremely low impedance and drive ‘most anything. Source outputs that are not super-low impedance, such as those from some hi-fi preamps, may require a different approach because they may not be happy pushing audio into 1k or even 2k resistors. I would try the resistors first. Depending on how, exactly, they were designed they may be more happy with 10k resistors, but then my processor’s input would not be seeing a low impedance source and the pre-emphasis curve may be skewed. So, while it is safe to assume a headphone driver’s output is very low impedance, for others without knowing what their impedances are, a simple summing network may not be appropriate. One “guaranteed to work” way is to build a unity gain summing amp. The easiest way to do that is with another opamp. Google will yield dozens of ideas. I considered adding one to my processor’s input but, besides using even more parts that for many users not be necessary, I convinced myself that this is a homebrewer's forum and it would be okay to leave that solution for the builder.

Speaking of transformers, there is enough gain within the processor that no voltage step up by a transformer should be needed. So, although a transformer ratio like 10:1 could step up the voltage a lot, in so doing it may reflect a too-high impedance into this processor’s input, which could manifest as a somewhat-skewed pre-emphasis curve (among other things). Unless DC isolation is required, I wouldn’t bother with a transformer.

I hope this helps answer the question more thoroughly!


Attachments:
Headphone to mono..JPG
Headphone to mono..JPG [ 50.84 KiB | Viewed 177 times ]

_________________
-Richard
Top
 Profile  
 
 Post subject: Re: richfair's audio processor for part 15 transmitter
PostPosted: Feb Sun 24, 2019 12:14 am 
Member

Joined: Jun Mon 24, 2013 3:00 pm
Posts: 1238
Location: Champaign IL 61822
What's the designed use of this thing? Voice? Music?

What are the time constants?


Top
 Profile  
 
 Post subject: Re: richfair's audio processor for part 15 transmitter
PostPosted: Feb Sun 24, 2019 4:50 am 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 617
Location: Equinunk PA 18417
Using a 10uF cap at C24, I measure attack time (to 60% of final reduction) as about 4 ms (which is pretty close to what the data sheet says I should expect). Accurate measurement of release time is much harder on my limited test bench. I can "eyeball" a release time to be around 250-300ms, but this is in no way a proper scientific measurement. Attack and release times are both determined by the value of C24. A smaller value allows faster attack and faster release, a larger value gives an opposite effect.

A limitation of this design is that attack and release times have a fixed relationship; one component sets them both. I clipped different capacitors into this circuit while listening, until settling on 10uF as giving characteristics I prefer for the range of material I tend to listen to, which is dialog podcast and classical music (although I did test with some punchy guit-synth material as well). You might have different preferences and can audition C24 values for yourself. I built this to heavily control dynamic range which makes everything audible at a very low receiver volume. I considered adding a front panel switch to place a 22uF cap in parallel with the 10uF, which slows the release time very noticeably, which is good for some things, but I eventually decided that, for me, it wasn't a setting I would actually bother to change. Also, I feel that the action of the compressor section changes perception of attack and release times in a way that is hard to describe. Life is compromise.

For extra credit one could add some automatic adjustment of attack and release times, or define alternate processing modes, by further manipulation of the control voltage. I tried a couple of things, from simple resistor-diode additions (to alter C24 charge/discharge behaviors) to a "non-linear capacitor" circuit. The latter looked great on paper but sounded awful. I would like to digitize the CV in a processor such as a PIC (I have experience with AVR devices) in order to make real time changes to the CV. It would be very easy to add hold times and selectively alter the attack and release times. But, that is beyond what I wanted to put forward in this thread and to be honest, I don't know that I'll ever get to it. It sure would be cool though.

_________________
-Richard


Top
 Profile  
 
Post New Topic Post Reply  [ 7 posts ] 

All times are UTC [ DST ]


Who is online

Users browsing this forum: No registered users and 2 guests



Search for:
Jump to:  


























Privacy Policy :: Powered by phpBB