EMRFD Message Archive 4728

Message Date From Subject
4728 2010-06-01 13:33:05 John Levreault AGC audio chain
Here's an interesting article here on a low-distortion AGC'd audio amp
with a claimed 85dB dynamic range using a photoresistor in the feedback
loop of an opamp:


I recall that the subject had come up here before, so just wanted to
pass this along.

John NB1I
4729 2010-06-01 15:31:11 ajparent1 Re: AGC audio chain
4730 2010-06-01 17:09:53 Lou Re: AGC audio chain
I couldn't find a schematic ...... does anyone have the schematic?


4731 2010-06-01 18:33:26 ajparent1 Re: AGC audio chain
Download the PDF versi
4732 2010-06-01 20:01:14 Dave - WB6DHW Re: AGC audio chain
Download the PDF versi
4733 2010-06-01 20:30:06 Don Hackler Re: AGC audio chain
Old broadcast audio mixer circuits sometimes used Vactrol devices with
an incandescent bulb in a similar circuit (sometimes with AGC, sometimes
not...). When the bulb would burn out, the gain would jump to max,
which would often cause havoc downstream. (one of your mixer inputs
would go "wide open" and there was no way to turn it off!)

This circuit seems to have the same configuration, but the LED is
probably not going to burn out any time soon... :)

4734 2010-06-01 21:20:16 Johan H. Bodin Re: AGC audio chain
A number of Vactrol devices made by EG&G Vactec (PerkinElmer?) are still
available, both incandescent and LED versions. One use for them is variable
termination in K9AY receiving loops although I made my own from a selected ORP12
CdS cell and 4 bright yellow LEDs. It went all the way down to 25 ohms!

Johan SM6LKM

Don Hackler wrote:
> Old broadcast audio mixer circuits sometimes used Vactrol devices with
> an incandescent bulb in a similar circuit (sometimes with AGC, sometimes
> not...). When the bulb would burn out, the gain would jump to max,
> which would often cause havoc downstream. (one of your mixer inputs
> would go "wide open" and there was no way to turn it off!)
> This circuit seems to have the same configuration, but the LED is
> probably not going to burn out any time soon... :)
4735 2010-06-02 06:48:19 Tim Re: AGC audio chain
I've done it using photocells from hardware-store $1.98 nightlights black-taped with LED's, and with Silonex CdS optocouplers. AGC works fine either way.

It's too bad that Cadmium gets such a bad rap environmentally. Optocouplers using cadmium are a perfect application.

4736 2010-06-02 08:07:03 drmail377 Re: AGC audio chain
Hi John,

Long post, broad but related topics - sorry I don't intend to hijack this thread...

I tried doing a home-brew "Vactrol" audio derived agc amp awhile ago with some LEDs and some junk box CdS cells. It worked, but I found the AGC time too slow for CW work due to the response time of the CdS cell over broad ranges necessary in a single stage.

There are plenty of Vactec (now owned by Perkin-Elmer) parts available on the Web, mostly used for music devices & analog synthesizers. The VTL5Cx parts are common - albeit a bit pricey IMHO. There is another company making similar parts today, but the name escapes me at the moment. Google is your friend here.

Sorry if I'm a bit off-topic for this thread with the rest of this post (non-Vactrol AGC approaches, but similar)...

Perhaps a better approach would be to use something like the circuit here:


Which is quite similar to the audio derived AGC in the Juma-RX1 receiver:


The Juma-RX1 claims around 90dB of AGC range. This approach uses an AGC (AVC) detector followed by a JFET in the op-amp AGC gain-loop. The first link above mentions you can use either P or N-channel JFETs in the loop if you flip the detector diodes around, even if your receiver design runs off a single-rail supply.

It should be noted that the equal-value relatively high-value resistors that form a divider between the JFET drain (or source) and gate act to "linearize" the voltage controlled resistor effect of the JFET somewhat. This is explained nicely in the article here:


You want to choose a small signal JFET part that is more rather than less drain-source symmetric for these AGC circuits, many - if not most, are.

I bounced off some LTSpice simulations of these JFET audio agc circuit awhile back. The AGC threshold seems to tight. That also raises concern about the AGC threshold changing with temperature as the JFET characteristics vary with temperature.

TI makes a nice Voltage Controlled Amplifier (VCA) the VCA810 with unity GBW of 35MHz and +/-40dB range. There are others in this series VCA8xx and VCA26xx series that go to higher frequencies. Most have about half the gain range. The VCA811 has an 80MHz unity GBW (but no PSpice model for the VCA811 it seems). The TI VCA8XX/28xx series VCA's require dual supplies, are a little difficult to find (but still in production), and perhaps a bit too pricey for some minimalist designs.

The VCA810 is fairly low noise, 2.4nV/sqrtHz & 1.4pA/sqrtHz, but if memory serves the data sheet doesn't say at what frequency and a plot isn't provided. So I'm a little concerned 1/f noise may be a bit high with this part at low (audio) frequencies, like you see with many instrumentation amps.

TI supplies a PSpice model for the vCA810. I modeled the part in LTSpice. However the model doesn't include noise characteristics. It seems to work OK otherwise. You can find my LTSpice version of the VCA810 model with a simple test jig in the LTSpice Yahoo Group in the Files > Lib area as VCA810_Test.zip.

Regardless of the AGC approach, do take a look at TI's AVC810 data sheet. There's a nice example AGC circuit that explains an approach for different AGC detector attack/decay times.

Some of these TI VCA's that go to higher frequencies, so they might be good candidates for IF agc amps in superhets, especially a general-coverage receiver with a 45MHz first IF. Although these parts are not rail-to-rail you might get good dynamic range, especially if using wide dual-rail supply voltages.

Analog Devices Inc. (ADI) makes some VCA's that may be suitable for AGC work. The venerable low noise, low distortion ADC602/ADC603 parts come to mind. If memory serves, they have around 42dB variable gain in two user selectable ranges. The ADC603 is 1.3nV/sqrtHz and goes to 90MHz. However, typically I find ADI parts difficult if not impossible to simulate, costly, and relatively hard to find.

With these VCA chips, use low-value resistors (noise) and pay attention to in/out impedance matching, even at audio as you will likely employ them earlier on in your audio stage. Test
4737 2010-06-02 08:31:54 John Levreault Re: AGC audio chain
Thanks for your comment, David. You're welcome to hijack the thread; I
was just passing along information, not endorsing the approach. I do
have a couple of comments, though.

1. I agree that Vactrols can be slow.

2. The JFET approach still has pretty high distortion, despite the use
of the "linearizing" resistors. Whether this is a problem in terms of
audible performance is up to the user. I've used them, and I feel that
the advantage (no more ringing eardrums?) far outweighs a few percent THD.

3. IMO, the TI and Analog parts you mention seem more appropriate for IF
amps, but there's of course no reason you couldn't use them for audio.

4. The best (in my experience) "pro-audio quality" limiter/compressors
IC's come from THAT Corp. I think Mouser carries them. They're not
cheap, but they're not too expensive either, when you factor in the
prospect of a twiddle-free design.

5. Most anything originally designed for bipolar supplies can be made to
run off a unipolar supply. However, you may experience limited signal
swing. I like some of the newer "rail-to-rail" opamps, which can
basically swing 12V (well, maybe 11.8?) off a 12V supply, whereas a
TL07x-type part might only be good for 7-8Vpp. Just a few dB in dynamic
range, but it certainly makes me feel better.

Anyway, all FWIW. Thanks for your comments! A lot of great information.

John NB1I

drmail377 wrote:
> Hi John,
> Long post, broad but related topics - sorry I don't intend to hijack this thread...
4738 2010-06-02 09:22:41 Tim Re: AGC audio chain
4743 2010-06-02 23:12:33 drmail377 Re: AGC audio chain
Hi Tim,

I wasn't happy with the decay time of the CdS VCR's. Decay times were running into the multiple seconds across the range I wanted. You are right, this is probably because I was using only one stage. The decay time is faster from completely off (megOhms) to around a 100kOhm or so, not totally on. The problem with this approach is that you have to use high value resistors in the stage, not optimal for low-noise designs.

Some commercial parts have impressive specifications though. Silonex (that was the vendor I couldn't remember in my first post) P/N NSL-32SR3 has a 25megOhm off resistance and decays to 100kOhm in 10ms. Today I found the NSL-32SR3 at around $4 USD each (qty.-2 min.) at only one vendor - Farnell-Newark, and it is stocked only in the U.K. and Singapore. I'm located closest to Singapore. I believe U.S. buyers get dinged something like $20 USD for deliveries from Farnell-Newark U.K.

Interestingly, for the home-brew vactrols I built I started out following the suggestions of online tutorials, primarily in the music community - use yellow LEDs. Then I tried other colors and found green LEDs were a bit more efficient. I studied the data-sheets of several CdS cells (e.g., Photonix PDV-P9203) and indeed the peak response is around 520nm; green is 495–570nm followed by yellow, YMMV.

Maybe I'll buy some of the Silonex parts and give it another go.

73's David WB4ONA

4744 2010-06-02 23:23:08 drmail377 Re: AGC audio chain
Hi again John,

I forgot about THAT Corp. I'll revisit them today and see what's up. Thanks for the tip.

To reiterate, I do have one concern about using the TI parts at audio frequencies; 1/f noise. No internals diagram and no noise plots in the data sheet adds to this concern.

Best 73's David WB4ONA

4745 2010-06-03 06:03:01 Tim Re: AGC audio chain
4746 2010-06-03 07:42:28 drmail377 Re: AGC audio chain
I think a (user-selectable) short AGC hang time for CW is preferable, especially with QSK, just my opinion though.

I agree, no AGC is quite good for CW (my preferred mode) in relatively quiet band and low QRM conditions. In this case a simple diode limiter earlier in the audio stage will save your ear-drums and/or speakers if something does "pop-up". But the limited high level outputs will be highly distorted. SO be careful where you put the hard limiter in the receive chain - proper gain distriubution to optimize dynamic range. Sorry to waste your time; this is comm
4747 2010-06-03 10:16:43 Tim Re: AGC audio chain
4748 2010-06-03 20:26:04 ajparent1 Re: AGC audio chain
4749 2010-06-04 06:39:23 Tim Re: AGC audio chain
4750 2010-06-04 08:25:17 KK7B Re: AGC audio chain
Hi Tim and Group,

The many benefits of completely independent transmitters and receivers have been known for 100 years--we keep forgetting and then rediscovering them, hi. One classic problem is making them all work together, including frequency tracking, from a single VFO.

If you choose the approach of using a DDS synthesizer as a transmit VFO, it is possible to simply count the receiver VFO frequency, do a few calculations in the digital hardware, and set the transmit frequency to the desired offset. That can provide seamless transceive operati
4751 2010-06-04 09:06:49 Tim Re: AGC audio chain
Actually Rick, I've thought about something similar. Except staying away from DDS's and using some sort of analog synthesizer to get to the transmit RF frequency while the receiver VFO stays fixed.

Inevitably these approaches involve I/Q mixers etc. They basically end up being phasing SSB transmitters except that the audio being transmitted is simply a sine wave at the XIT shift.

The receiver VFO already has I/Q outputs because I use it with my R2Pro :-).

I have tried different ways to do the "transmit shift" audio-range-frequency, with 90-degree phase shifts. For a single audio frequency an analog approach is the "Bubba oscillator" (e.g. TI app notes) which has quadrature outputs. But if I want to dial in the XIT frequency it gets more complicated... I've played around a bit with trying to phase lock two Wein bridge oscillators with a 90 degree phase shift but haven't been satisfied with the results. More digital and less analog, I've played with simple sine wave DDS-like audio synthesizers (circuit copied from Horowitz and Hill, the 8-stage shift register with "sine wave tuned" resistor chain) where the shift between the two shift registers is kept at a theoretical 90 degrees. But the audio DDS needs low-pass filtering after it, and the low-pass filtering for the two 90-degree channels has R's and C's in it, and those are R's and C's that have to be matched. It starts seeming that for typical XIT shifts (300 to 5kHz) maybe I should abandon the quadrature oscillator and just start with a good clean sine wave and run it through a phase shift network very similar to the R2PRO's receive networks.

No matter how I do it though, I end up with the image of the audio tone approx 30-50dB (depending on matching and phase shift accuracy) down on the other sideband, and I don't like the thought of that being transmitted. I've heard KWM-2's doing CW on the air and they did not sound good. I know a good SSB phasing transmitter can sound a lot better than a KWM-2a. (I know, I just pissed off a lot of Collins fans, but the KWM-2a is IMHO the world's CRAPPIEST CW rig and it uses the "transmit an audio tone" method that I'm thinking about about implementing myself.)

4752 2010-06-04 11:56:57 KK7B CW tone into SSB transmitter audio
Hi Tim,

I have built CW transmitters using the phasing method or filter method and a keyed audio tone into the mic input. I don't like that approach for HF CW on crowded bands, where someone may be listening to a signal 60 dB below mine a few hundred Hz away. But that is how all the modern sound card based digital modes like JT65 and PSK-31 work. I guess a computer DSP routine doesn't find the extra transmitted garbage as offensive as the human ear.

I ran a 75S3C and 32S1 as my HF station for a few years, and modified the 32S1 with a separate carrier oscillator for CW. I've pondered making a really fine CW transceiver out of a KWM-2. One reason I quit using Collins gear is that the Collins community reacts so negatively to modifications. Other than that, it's not bad stuff, and easily brought up to modern standards with relatively simple mods--maybe a few extra holes in the front panel, and be sure to scrape the paint off around the mic jack for better grounding. (Just kidding--I sold my Collins gear to collectors who would appreciate it for what it was.)

But running a clean keyed CW sine wave into the audio input of a clean SSB transmitter is an excellent way to instantly switch between SSB and CW on the VHF-UHF-Microwave bands. Several of my homebrew microwave transceivers have the mic and key both connected all the time, with no mode switch. Drift is usually a more serious problem than nearby interfering stations on microwaves, and keeping the other station tuned to exactly your sidetone frequency on CW guarantees that after you peak the antennas you can instantly switch to SSB and be perfectly tuned. Since most serious stations on VHF-UHF-Microwaves work SSB, CW, and Digital modes, it may make sense to design a very clean SSB exciter with several additional audio inputs. I've also experimented with a system that intentionally transmits a little carrier and keeps the receiver locked to it with a vary narrow PLL so that both SSB and CW have zero frequency error.

You can clean up the output of an IQ CW exciter with a PLL
4753 2010-06-04 12:08:02 ajparent1 Re: AGC audio chain
4754 2010-06-04 12:19:57 ajparent1 Re: AGC audio chain
4756 2010-06-07 08:31:56 kilocycles Re: AGC audio chain
Hi Rick,
Here's a possibly dumb question that I've been thinking about for a long time. What happens when you feed an on-frequency VFO (heterodyne VFO or DDS, for example) to a transmitter (DX-20, HT-40, Globe Scout, etc. using the normal VFO input) that uses multiplier stages, say for 20, 15 and 10 meters? It seems to mew that a rig that simply doubles, triples or quadruples (one stage multiplication) would simply pass the signal through the tuned circuit (it is tuned to 14mHz expecting to see 7 mHz, but sees 14mHz for example). For a rig that doubles and doubles again (7 to 14, and 14 to 28) wouldn't that be a problem? Then there are the rigs that use 80m crystals on 80, and either 80m or 40m
4757 2010-06-07 09:11:11 Tim Re: AGC audio chain
I'm not Rick but in my experience with the oscillator/multiplier/PA type transmitters, starting with my simplest: the Eico 720

The crystal oscillator is 6CL6. It is mostly (see below) untuned. It also serves as a VFO buffer when not using a crystal.

The next stage is a 6AQ5, followed by the 6146 PA. Traditionally the tuning between these two stages was thought of as being on the PA's grid, but I think that realistically it was actually on the 6AQ5's plate.

There is a control marked "grid tuning", and one adjustment criteria was 6146 grid current, so maybe that's not actually a wrong way to label it.

The 6146 has a pi network output: plate tuning capacitor, switch-selected inductance per band, and antenna load tuning capacitor.

Now as to harmonics etc:

The 6CL6 oscillator RF choke works as an RF choke with an 80M crystal. With a 40M crystal it works (due to winding capacitance) as a broadly tuned 40M resonant circuit. So if you have a 80M crystal, the output of the oscillator is on the 80M band; if you have a 40M crystal the output is on the 40M band.

The 6AQ5 on 80M with an 80M crystal, or on 40M with a 40M crystal works straight through, no multiplying. To get to 20M, 15M, 10M it works as a doubler/tripler/quadrupler. It also can double a 80M crystal up to 40M.

The 6146 works straight through. The instructions about grid tuning/plate tuning are written to discourage using the 6146 as a multiplier. i.e. there are explicit instructions "it might look like it peaks up there but that's the wrong harmonic!"

In the Eico 720 the grid tuning and plate tuning inductance switch wafers are ganged together, so by design it was to discourage you from multiplying there. In other transmitters the PA is used as a doubler on 10M.

Does any of that answer your question? I sometimes hook my homebrew UVFO-style VFO's up to my Eico 720 for 80M and 40M. I use a little FB37-43 step up transformer to boost the voltage up to what the 6CL6 likes (it likes many many volts of AC on the grid). This does not work well for 30M or 20M because these are above the 6CL6's choke self-resonant frequency, but it kinda works there.

4758 2010-06-07 10:36:54 ajparent1 Re: AGC audio chain
A jump into the frey..

The problem with multiplier rigs is that what may work for one may
be invalid for another.

All of those that I'm familiar with expect the input (Crystal, VFO,
DDS or whatnot) to be in some range usually in the 80/40m range
but that will vary depending on rig. The reason for this was the availability of crystals in the ranges noted at that time.

For the lower bands usually the first stage after the osc was
a buffer and for bands 40M and higher it was a multiplier
followed by a driver to a final.

Depending on where selectivity was placed and type using a higher
frequency crystal or source than the intended range could result
in either normal operation or none, The latter case being a result
of tuned circuits not passing the higher frequency.

If the source is high stability and in the basic crystal frequency range that is sufficient for all bands. However, on 10M that
source due to multiplcation better be stable (DDS is)
or drifty operati
4759 2010-06-07 12:31:03 Tim Re: AGC audio chain
I have used my Eico 720 on 30M and 20M "straight through" on the VFO input. I have also taken my f/2 output from before the UVFO style doubler and run it into the Eico 720's VFO input and doubled it in the Eico. The driver level setting knob on the Eico 720 accomodated either but I did need to use a step-up transformer on the VFO input especially when running straight-throu above 30M.

I think the "standard VFO output level" back then was circa 10V RMS, when I scope the output from my old Eico or Heath HG-10 VFO's they show circa 30V P-P. This must've been rather high impedance.