Thanks Guy and Don. Don, I apologize for my muddled email. My fault, not yours.
I will look closer at the system schematics so as to discover the flow of the process when I have some time. Thanks for pointing out some of the nuances from your perspectives. What I call "punching holes in the audio" is exactly that... If you have a short loud sound followed by a soft one, part of the soft one goes away due to the attack/decay ratio of whatever magic waveform modification process is at work in this rig. Analog monolithic compressors do this when the attack is set faster than the decay on short duration spikes. I have adapted to the way this process "thinks". I was just curious to the way it was programmed to "think" and looking for a way, other than hang a AGC at the mic input, to manage the process better. However it works, it works quite well. But the smoother the waveform going in, the smoother the waveform is going out. May we live in interesting digital times indeed! -lu-w4lt- -----Original Message----- From: guyk...@gmail.com [mailto:guyk...@gmail.com] On Behalf Of Guy Olinger K2AV Sent: Thursday, April 29, 2010 9:30 PM To: lrom...@ij.net Cc: d...@w3fpr.com; elecraft@mailman.qth.net Subject: Re: [Elecraft] W2IHY 8 band equalizer and EQ Plus Hi Lu. Schematics for the K3 are downloadable PDF on the Elecraft website. Many questions can be answered right off just digging into the drawings. I find the ability in Adobe Reader to search in the PDF text very useful for finding stuff. Device specifics are in the schematic, so I guess you could go looking for a manufacturers technical writeup and figure out the granularity from that. I think that some of what you are calling "punching" really is just the K3 dealing with sudden power spikes. Professional microphone technique would not include very soft followed by very loud. The K3 is interpreting the loudest audio as being the intentional "top" of your speaking pattern and is setting power management to NOT punch out the amp and cause ALC spikes coming back from the amp. This effect would be further exacerbated if all of the various TX power calibrations had not been done correctly, as it will be further exacerbated if the users manual MIC/CMP/PWR setting procedure is not used. The terms you are using to describe the K3's internal functioning will remain speculative unless Wayne publishes stuff, and our thought patterns still have have that analog, sequential function sound to them : >) K3 has a digital transmit envelope management function that should be preceded by proper TX gain calibrations, and the user manual MIC/CMP/PWR setting. 73, Guy. On Wed, Apr 28, 2010 at 11:03 PM, Luis V. Romero <lrom...@ij.net> wrote: > Some quick questions (yeah, I know I said I would stop, but I cant > help it, I want to understand!) > >>"...a Pre-DSP two band audio leveler/AGC and handles for Attack, Decay >>and > Ratio on the RF Clipper." >> >>That is, prior to Digital Signal Processing, or before analog is >>converted > to digital. > > Where exactly does the input to the TX DSP section (OK, IF section) > hit an ADC? Where do we move from the Analog domain to the Digital > domain? Where is the "quantizer" for the mic input? Would make sense > if it was before the Clipper, right? > >>In your headroom explanation, is there a common instance of microphone >>ADC > saturation that needs to be reported? >That's a pretty seventh-grade > mistake on Wayne's part if it's true. (Yeah, I know, the first Hubble > lenses, and >the unit snafu on the Mars landers, also very > seventh-grade.) > > I don't believe that's it at all. I can punch holes into the audio if > I speak softly then speak loudly. Sounds to me like there is some > kind of compression happening (that is why I first incorrectly assumed > that this thing used a compressor not a kind of IF/RF Clipper). What > happens is that the radio grabs the loud syllable and holds off from > releasing the gain reduction for a bit, then recovers. If you hit it > with something loud, then something soft, you hear the hole. The > decay of whatever is grabbing the peak is slower than the attack. > There is no overshoot that I can discern with my ears, as the firmware > must be dropping audio into a buffer and setting the response in kind. > I once heard people complain about delay in the monitor, I do hear a > very slight one. So there must be a look ahead buffer that computes > the response to the peak. All I'm saying is that it would be nice to > have a handle on at least the decay, so it can more match the attack. > Probably cant do that as it would create more delay in the monitor. It's a fine line. > >> > "What I was suggesting is an AGC funtion before the RF Clipping > section touches the audio waveform. Something to smooth out the > dynamic range of the audio input so that the DSP processing engine > would not have to work so hard dealing with peaks and valleys and can be "let loose" some." > > Elecraft is already accomplishing envelope leveling and shaping with > digital functions that don't appear to >resemble the sledge and wedge > of RF clipping and AF variable band amplifiers. SOME folks get > excellent results using the K3's leveling and shaping processes for TX > audio. I would hate to bring up RTFM on setting up K3 mic gain and > compression, but the manual procedure does seem to work. >> > > I would like to get a better understanding of the process that is used here. > I do use the manual settings, It does work well, and does sound good, > but I would like to understand what is happening under the hood better > so I can better adapt to what this rig likes to hear, which, to me, is > consistent levels. My old admittedly analog rig was less picky and > had much more room to play with (when I let it). Hopefully that is not Elecraft Secret Sauce. > >> > AND, since there is NO analog audio band circuitry in there anywhere, > BUT there ARE banded TX and RX equalizer functions being done in the > number soup, whose gains are being set by NUMBERS we enter in the > menu, what makes us think he hasn't already done something proprietary about "banded gain" > which he is developing further and is not about to reveal so the > competition can't copy it for free? >> > > This is probably the Secret Sauce I'm talking about. Are you implying > 8-band digital split band processing? > >> > You should have seen the look I got from the oldest one when I asked > him if the audio actually went through all the slide pots on one of > those big digital mixers he was running, the look that says "Please > don't talk like that when my friends are around." >> > > That's pretty funny. On my side, the video side, I was explaining to > a pretty hot non linear editor how we did non-b-roll match frame > editing with timecode on helical composite VTR's and the importance of > the color frame sequence across 4 fields (360 degrees of subcarrier > and matching subcarrier to horizontal sync) so that the picture > wouldn't jump at the match frame in NTSC. Couldn't understand the > process! Couldn't even begin to understand true A/B rolls either. So I know what you mean. > >> > there is no K3 RX AF analog circuit controlled by AF gain. There is > no K3 RX RF analog circuit controlled by RF gain. The AF and RF pot > settings are immediately turned into "advice" numbers and passed along to the MCU. >> > > Advice in the way of a VCA setting? Wonder what the granularity is. > It is an analog pot, so logically we read a voltage and digitize it, > then report it to the MCU. > > Thanks for pointing me in the right direction, Guy. Now I know the > probable true reason for the annoying "cant talk to the RX controls while in TX" > behavior. A hybrid Parallel/Serial signal bus. > > I'm afraid I pushed paper from the left side to the right side of a > desk for way too long. > > -lu- > > > > > No virus found in this outgoing message Checked by PC Tools AntiVirus > (6.1.0.25 - 6.14830). > http://www.pctools.com/free-antivirus/ > No virus found in this outgoing message Checked by PC Tools AntiVirus (6.1.0.25 - 6.14880). http://www.pctools.com/free-antivirus/ ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
