I found this circuit on the net. Accordingly i assembled it with 2N5019 fet. I changed the opamp to njr4558 also for better sound. But the tv sound is getting distorted after using this preamp. I am using as the input from satellite receiver & output to tv audio input.
Waiting for suggestions & solution.
Waiting for suggestions & solution.
The J-Fet used as a variable resistor, needs to be linearized, or distortion occurs above about 25mv input. Even then, it's best to keep the signal below 100mv. you can do this with an input attenuator.
I think it's Vishay that have an ap note about using J-Fets in this manner, but simply, you want to create feedback between the drain and the gate of the fet.
With this circuit, you would need to take a cap, say at least 0.1uf from the drain, with a 100k r in series with the cap, and connect the other end of the r to the gate. You will then have to put a 100k r from this junction, to the junction of the cap and resistor at the gate of the fet and the collector of the transistor, removing the direct link between them.
As it is, it is a very simple circuit, and not really meant for high quality repoduction. With these changes, you should be able to eliminate audible distortion.
jD
I think it's Vishay that have an ap note about using J-Fets in this manner, but simply, you want to create feedback between the drain and the gate of the fet.
With this circuit, you would need to take a cap, say at least 0.1uf from the drain, with a 100k r in series with the cap, and connect the other end of the r to the gate. You will then have to put a 100k r from this junction, to the junction of the cap and resistor at the gate of the fet and the collector of the transistor, removing the direct link between them.
As it is, it is a very simple circuit, and not really meant for high quality repoduction. With these changes, you should be able to eliminate audible distortion.
jD
Also, put a 1k resistor between the base of the transistor and the junction of c3/p1 and see if that lowers distortion as well.
jD
jD
A number of manufacturers now make digital potentiometers or level controls. While few, if any, qualify as the last word in audiophile signal purity, most will be better than simple FET circuits.
For example look at the LM1972 data sheet from National Semiconductor .003% distortion 78 dB of gain rage and a signal to noise of better than 100dB. This was just the first one I found. Other people make them such as Aanalog Devices, it might be worth a search.
For example look at the LM1972 data sheet from National Semiconductor .003% distortion 78 dB of gain rage and a signal to noise of better than 100dB. This was just the first one I found. Other people make them such as Aanalog Devices, it might be worth a search.
hermanv said:
I think the original poster is using the circuit as an agc preamplifier. The LM1972 is essentially just a volume control, and would not work (well) as an agc preamp.
Of course there are better ways of doing this than with a fet, but certainly not as cheap, and not within the scope of this thread.
For alternatives, see www.thatcorp.com
An AGC circuit has two basic requirements, a method of controlling gain and a method of detecting the output level. While ICs that do both exist, all the ones I know of are srictly Lo-Fi. By buying a quality solid state "volume" control and designing your own rectifier, a reasonable AGC circuit can be developed with a few Op-Amps or use discretes if the last word in quality is the goal.electrode10101 said:
Having designed a few commercially, there is a real issue of attack and decay speed to consider. AGC circuits for voice or music turn out to be tougher than they apear because our ears quickly pick up the "un-natural" resulting sound if the gain down and up speeds are not done very carefully. It is also necessary to limit the maximum gain or the output just become hiss and noise when no input is present.
I don't quite know what the scope of the thread is, the original poster asked for ideas or a working circuit.
mhtplsh said:
If you want to use the original circuit with the fet variable resistance, there is a trick that EVERYONE uses in this case. Put a resistive divider (2 x 100k should work) from the input to ground and connect the midpoint to the gate. That cuts the distortion by an order of magnitude. And, make sure your input level stays below 100mV (preferably below 10mV) and put some gain after it.
Jan Didden
mhtplsh said:
This circuit will work better, but is a bit more complex as well. It incorporates the fet linearization with feedback coming from the output of the opamp. This works fine if the components values are correct.
Not having built the circuit, I can't vouch for the operation, but it looks like it will work better than the first one you postd.
If you want to get into high quality compression/limiting/agc circuits and components, check out http://www.thatcorp.com
Browse around the websites, and look at the application notes; lots of information and detail about these circuits - not the last word, but enough to keep you busy for a while.
what you want for minimum distortion would be a compressor. instead of the jfet at the op amp input, it would be better to use a 2 quadrant multiplier to control the signal level. a multiplier is much more linear than the fet, and so has less distortion.
unclejed613 said:
An analog multiplier IC might work well. I used one, once, in a somewhat-different type of AGC application. It was not for audio, per se, but involved audio frequencies, in a test instrument, and required very good linearity and low distortion.
In that circuit, the AD633 was driving the input of an LM1875 chipamp. One of the multiplier's inputs was used for the signal and the other input was used for the AGC voltage that was derived from the chipamp's output, using an "adaptive envelope detector", and an opamp-based integrator, in a feedback loop.
I tried designing it (quite a few times 🙂 with various different controlled-resistance elements that included FETs as well as Vactrols, which are essentially just an LED encapsulated with a photocell (and are pretty darn handy, sometimes). But eventually I just used an AD633 analog multipler (analog.com; cheap and widely available, 8-pin DIP, and very easy to apply, at least if you remember that the result is divided by 10 🙂.
The purpose, in that case, was to try to force the LM1875 power amp to keep its output amplitude at a constant user-selected p-p voltage (from 1v p-p to 30v p-p), for any of several user-selected waveform types (sawtooth, triangle, or sine) at frequencies from 60 Hz to 22 kHz, with an input signal from an oscillator that could have an output amplitude that was anywhere from 2.5v to 5v p-p (but stayed approximately fixed, when selectors weren't being changed), for any load that the user cared to connect. (But at least there was also a current-limit selector, with 1.5A max. 🙂
There was a user-selected DC reference voltage that was used by a differential integrator in the feedback loop, which integrated the difference between the envelope detector's output and the reference voltage, to produce the AGC voltage for the analog multiplier IC's other input. It all worked pretty well.
The envelope detector was the biggest design challenge, in my opinion. It was a little difficult to get extremely low ripple but still also be able to slew quickly-enough when a different desired output voltage was selected, especially when going from a higher voltage to a lower voltage. For that, I ended up having to make the envelope detector "adaptive", by using a Vactrol (current-controlled "pure" resistance) in place of the envelope detector's "bleeder" resistor, so the circuit could automatically increase the detector's slew rate when the error was large and decrease it when the error was small, which gave both speed when needed AND very low output ripple just when it was needed the most.
The vactrol model that I used, the VTL5C2, can vary its resistance from a couple-hundred Ohms to a couple of megOhms, when its LED current is varied from 40 mA to 0 mA. They're only about $0.60 each, at bgmicro.com . Something very similar could also be DIY'd, using an LED and a photocell. Unfortunately, for some applications at least, their response is more logarithmic than linear. Also, their response speed is MUCH slower than that of an FET, or a multiplier IC. On the other hand, they present a pure resistance, so they don't introduce any "artifacts" or have some of the annoying problems that something like an FET circuit might have. (If anyone wants a VTL5C2 spice model, feel free to email me.)
The AD633 analog multiplier and LM1875 combo ended up giving very low distortion in the output signal, by the way.
I don't know if that long-winded description will help the original poster. But maybe it will give them some ideas. Sorry to have blathered-on for so long, about all of that.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
mhtplsh said:
Hi mhtplsh,
I don't think that my original circuit would be very good, for your application, since it was optimized for fixed input and output levels, and is probably more complex, for other reasons, than what is needed for a plain audio AGC.
But I just now cobbled-together an AGC type of circuit that uses the AD633 analog multiplier, and have used LT-Spice to simulate its operation, and tweak component values. (This is a fun type of circuit to simulate, too.)
I am not really sure how well this type of AGC will work, for your preamp application. And even if it WILL work, it also might need to have some of its component values adjusted, etc. (But you can easily do that, with the help of the LT-Spice model that I have provided, farther below.)
As it is now configured, the input can be up to 20 V P-P and the output will always try to be 1.2 V P-P. But that is easily adjustable.
Here is the schematic:
And here is an example of the transient response, for a 20V P-P input at 100 Hz:
Right-click and select "Save Target As" to download the LT-Spice files for the circuit above:
http://www.fullnet.com/~tomg/diyagc.zip
To download the LT-Spice simulation software package, itself, click on the link below. The direct download link is near the top of the page. (There is also a link to the users' group, and a link to over 20,000 downloadable spice models' links, on that page, as well as some other spice circuits that I have made available.)
http://www.fullnet.com/~tomg/gooteesp.htm
(Note: To see the THD results, after a Transient simulation run, select "View" then "Spice Error Log", from the schematic window.)
U1 is used as a full-wave rectifier. U2 is used as a differential amplifier, to reference the rectifier's output to ground. R2 and C1 filter and average the rectifier's output into a quasi-DC "output level" signal. U3 is an integrator, which integrates the difference between the rectifiier's averaged output and a reference voltage, creating the feedback voltage that drives one of the analog multiplier's inputs, and is multiplied by the input signal.
As configured, making the reference voltage more negative (by increasing R17, for example) makes the output level increase. However, if the reference voltage is too close to 0, or is not connected, the output level will be approximately equal to the input level. Note also that with this topology, at power-on the output will be at minimum attenuation until capacitor C1 is charged (etc).
This circuit is not yet well-optimized. The two main variables to be balanced are 1) the speed of the attenuation response and 2) the distortion, which is directly related to the amplitude of the AC ripple component of the integrator output. (With the circuit as shown, the output is attenuated 90% of the way from the maximum 20 V P-P to 1.2 V P-P in about 170 ms. I am not sure what the speed should be. So you might need to adjust it.)
If the response is made slower, then the AC ripple in the AGC signal will decrease in amplitude, which will lower the THD. Since the ripple amplitude is larger for lower frequencies, the THD is higher at lower frequencies. With the topology used, the only easy way to lower the THD at lower frequencies is to lower the response speed.
Decreasing R14 x C2 increases the speed of the response and increases the output distortion. Decreasing R2 x C1 increases the speed of the response and increases the output distortion. Increasing R4/R3 increases the speed of the response and increases the output distortion. Increasing the input level increases the speed of the response but also increases the output distortion. Changing the input and feedback divider ratios and values will have similar types of effects. Changing opamp types might also affect the circuit's operation, slightly.
(Note, also, that if R14 x C2 is decreased too far, the integrator opamp will saturate (i.e. try to hit the power supply rail, and maybe stick there for a while) and sometimes-hard-to-predict "bad things" will probably happen.)
So, anyway, you can try it, with LT-Spice, at least. I am sure that there are lots of improvements that could be made, in the circuit.
That's all for now. Have fun!
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
-
Oops. I didn't realize that we should have been trying to get attenuation times of less than 10 ms.
I just saw that the explanation for the limiter circuit at the westhost.com link that someone provided above claims that a 5 ms "attack" time is necessary. That circuit is reported to give about 0.3% distortion, above 500 Hz, with about 1.65 VRMS output level.
I will see if I can adapt the circuit I gave, for that type of speed.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
I just saw that the explanation for the limiter circuit at the westhost.com link that someone provided above claims that a 5 ms "attack" time is necessary. That circuit is reported to give about 0.3% distortion, above 500 Hz, with about 1.65 VRMS output level.
I will see if I can adapt the circuit I gave, for that type of speed.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
- Status
- Not open for further replies.