Speech Amp Audio Questions

[W1UJR]

Gentleman (and ladies), now that we have the 30K-4 exciter working well, the next challenge is getting some acceptable audio from the unit.
On-air testing with NE1S seems to indicate that something is lacking in the audio from this unit.

Appreciate your input, take a look at the speech amp circuit below, from a Collins 30K-4, and tell me what you think of the audio passband.
Audio reports are "communications quality", punchy, but not hi-fi.

I swept the unit with my AF generator and a dual channel scope on the grids of the 75ths, and found little above 3700 cycles.

If I check pin 5 on the 6H6 tube, the audio seems to pass right up to 10 KC.
However, on the grid input of the 6B4G driver tube shows very little above 3700.
I'm assuming that L301 and the capacitor network attached to it are designed to shunt anything above a certain freq. to ground.

Is this correct, what are my options?
I'm not looking to do major mods, as I like to keep the unit close to stock, but would like to improve the fidelity.

Thanks!

[WBear2GCR]

Bruce, (right?)

this is essentially the same circuit as used in the Heath Apache. If you search on that speech amp, you'll get back useful stuff. I posted a cute mod that uses the clipper components differently, search for me and that here...

Bypass the clipper part, remove the tubes, and jump from the input cap R309? to C310?? It's hard to see ur jpeg, and there is a glitch on the secondary of the driver transformer, there's a cap there, not sure where it goes, might be a "splatter" filter... might need attention too.

Edit: ok you can't actually remove the 6SN7... you have to make appropriate mods there... you can use the second half of the tube if you need the gain, if not then an internal bypass to the cap .without the plate connected will do.

Alternately, model the network in the clipper in PSPICE and adjust the values to be more like 8khz.? (see my mod)

That should fix the problem, unless the iron won't pass anything...

If it won't do bass, then you might want to parafeed the SE driver tranny that comes off the 6B4 to get the DC off it...

 

Post some images of the full transmitter??

            _-_-bear

Oh, that filter you asked about? Look at it again, it's a pi-filter but with a cap across the inductor. Without the cap across, it is a 3rd order, 18db/oct Low Pass filter. The cap pushes some extra HF through, and actually is resonating the inductor, making it a more complex function, looking more like a bandpass, but still can be intially considered as just a 3rd order audio filter.

[Steve - WB3HUZ]

Just bypass the entire clipper. Jumper from the plate of V303/6SN7 section 1 to the grid of V303/6SN7 section 2 with an appropriately sized coupling cap. The clipper does little to increase the average level of modulation when feeding the lowpass filter formed around L301. Most of these clipper filters I've modeled  in the past have terrible impulse response, thus negating any modulation level improvement provided by the clipper.

In other words, clippers blow. Bypass it. I did this with my Eico 730 and it worked FB.

http://www.amwindow.org/tech/htm/eico730.htm

[W1UJR]

Ok, what value do you suggest as a coupling cap if bypassing the clipper?


What about the grid input resistor R302, currently 1 Meg, should I bump this to 5-7 Meg?

[WBear2GCR]

If ur gonna use a D-104 sure... if not, not required.

That pot that goes to the grid of the 6B4?
I'd drop a 1 meg or similar to ground from the grid, that way if the pot opens up, you don't have a grid floating up to the potential on the other side of the coupling cap... and I am wondering what is doing the biasing for the 6B4?

Whatever C310(?) the cap after V302 ought to be ok. You can safely increase it by one order of magnitude, if you really want to pass bass an octave lower than now... I'd sweep it and see where the LF rolloff is now. IF it is ~40Hz or so, then leave it alone. Imho there is absolutely no benefit to permitting much below that frequency into the modulator - as it can cause you to waste significant power on inaudible stuff.

Just my 2 cents.

              _-_-WBear2GCR

[Steve - WB3HUZ]

Not knowing the value of the grid resistor, I couldn't say for sure. But what BEAR said is good advice. No need to dump too much low-end into the iron that follows (probably can't handle it anyway).

[W1UJR]

Ok guys, thanks.
Will play around with it until I am satisifed with the audio.
Right now I can sweep down to 300 cycles or so without any loss, that's low enoiughf or me.

Bear, the bias for the 6B4G is off the center tap of the filament xfomer T303.

Here's your "money shot" Bear.


And the wavefrom at the grids with passed audio.





Distorted waveform.

[WBear2GCR]

Bruce, nice pix!

If your LF point is 300Hz. I'd do the mods and see where you are then. And IF it is still at 300Hz., I'd fatten up the signal path caps. How much? Well if you increase them by 10x you'll drop an octave - that's potentially 30Hz. (if you assume each stage is tuned the same, not a safe assumption - check) So you want to go a bit less than that, but not much, so that your LF point isn't too low. One way to do it is to test each stage's output and just parallel a cap or caps and see where the - 3dB point is. It is usually a good idea to not have the 3db points dead bang on top of each other exactly... I'd aim for a -3dB point of 40-70Hz. That'd probably sound very good.

Most of your passband is probably being determined by the clipper filter now...

          _-_-bear

[w3jn]

REALLY low end autio isn't good.  I can defnitely see Harleys riding by the house on my scope    THe audio on my GPT750 sweeps pretty good down to 10 Hz.  Way too low, IMHO.

[W1UJR]

Harley's are just too loud, no matter what!

What is acceptable "low end" sound John?

What do you consider to be a good low and high freq points for decent sounding audio?
My concern is that I like to have simple audio chain, and just run the D-104 into the rig, no fancy EQ or processing in front.
I don't want to make this thing so wide as to be counterproductive.

Tnx - Bruce

[w3jn]

I agree with Bear.  -3 dB at 70 hz is probably fine.

[k4kyv]

If you use a D-104, don't worry about too much lows. Its response is down quite a bit below 100~, but still has enough low frequency response for the audio to sound balanced and not too tinny. But as others have already suggested, by-pass the clipper and filter altogether with a simple coupling capacitor.  If you want the modification to be reversible, simply leave the clipper components in place but don't use them.

It is always best to maintain as wide a frequency response as possible, regardless of the audio range you wish to transmit.  The best place to shape frequency  response to your wishes is right at the microphone, or outboard equaliser feeding the mic input.  The wide frequency response of everything that follows assures that the audio waveform leaving the mic or equaliser will be preserved all the way to the mod transformer.  This is particularly important to observe if you want to preserve the asymmetry of your voice waveform.

The UTC Transformer Company recommends maintaining a flat frequency response to at least one octave above and one octave below your intended audio range.

Personally I would think cutting off at 300~ would make the audio sound too tin-can-telephonish or too slopbucketish.  Aiming for flat response down to about 150~ would be better.  But that means the audio chain should be flat down to at least 75~, or one octave below the 150~ target frequency.

To get good articulation, you need to boost the upper midrange, starting at about 1 kHz and steadily rising up to about 2 kHz, and then levelling off past that up to the highest frequency you wish to transmit. 8-10 dB of boost would be a good figure. But to avoid phase shifts and other waveform distortion, the transformers and coupling circuits should be flat to at least one octave above highest frequency you wish to transmit.  So, if you wish to go up to a maximum of 5,000 Hz, try to get the audio chain to sweep flat up to at least 10K.

I ran tests on mine, which uses transformers of the UTC Linear Standard series or better, and I found some phase shift problems above 11 kHz.  I connected a scope to the pushpull grid winding of the LS-49 driver transformer, with the horizontal input connected to one end of the winding, and the vertical input to the other.  I got a perfect diagonal line up past 10 kHz, indicating good phase response, but past 11K the diagonal line began to open up into a slightly oval  or elliptical shaped pattern, indicating that the output to one grid was slightly out of phase with that of the other grid.  In the worst case, if the relative phase shift became 90 degrees, it would display a perfect circle.  It never got anywhere close to that, but it indicated that 11k was about the limit of the flat frequency response free of phase shift distortion with that transformer, even though the manufacturer's specs say it is good to 20,000 Hz.  But since I use a low-pass filter in the speech amp to gently roll off the audio starting at 5 kHz, I didn't worry, because it showed perfect response up to a little beyond one octave above the intended high frequency limit.

I have seen very few audio transformers, even the best broadcast quality ones that are supposed to be flat from 20-20,000 Hz, that have actually tested out to be that good.  Many of the UTC LS series have peaks and dips, especially between 10K and 20K.  I have some very old 1930's era ones that roll of substantially above 5K.