Time Varying Amplifier

[dknguyen]

https://www.electro-tech-online.com/custompdfs/2008/01/vca810.pdf

I am trying to make a time-varying gain amplifier for my ultrasonic receiver. I have chosen the VCA810 so far because I don't want to deal with feedback resistors and figuring out the gain-"control voltage" curves that seem to be there in other VGAs and I don't want to manually make voltage-controlled resistors with JFETs, or MOSFETs, etc.

Now, I know I could use a DAC and an MCU to incrementally incrase the gain. Though I would prefer the gain to be contiously variable (otherwise I could have just used an op-amp with a serial-controlled digital pot rather than one of these VGA ICs). The gain control voltage has to go from 0V to -2V within a fixed time interval, except I'm not sure how to do this.

I was thinking a about using 555 timer circuit (or discrete comparators since the lower threshold voltage is different from that of a 555 timer and is inaccessible via the CONTROL pin), but the accurate capacitors are too small and result in resistor values that cause currents comparable to the input bias currents of the control voltage.

 

[Hero999]

Use an analogue multipllier.

 

[Roff]

https://focus.ti.com/lit/ds/symlink/vca810.pdf

I am trying to make a time-varying gain amplifier for my ultrasonic receiver. I have chosen the VCA810 so far because I don't want to deal with feedback resistors and figuring out the gain-"control voltage" curves that seem to be there in other VGAs and I don't want to manually make voltage-controlled resistors with JFETs, or MOSFETs, etc.

Now, I know I could use a DAC and an MCU to incrementally incrase the gain. Though I would prefer the gain to be contiously variable (otherwise I could have just used an op-amp with a serial-controlled digital pot rather than one of these VGA ICs). The gain control voltage has to go from 0V to -2V within a fixed time interval, except I'm not sure how to do this.

I was thinking a about using 555 timer circuit (or discrete comparators since the lower threshold voltage is different from that of a 555 timer and is inaccessible via the CONTROL pin), but the accurate capacitors are too small and result in resistor values that cause currents comparable to the input bias currents of the control voltage.

What control voltage parameters do you need to control -
Is it a linear ramp?
Is the time (slope) adjustable?
How do you control when the "ramp" starts?
When it gets to -2V, what should happen (hold, reset to zero, etc.)?

 

[dknguyen]

-I need a linear voltage ramp
-constant, fixed slope (the time and magnitude of the ramp is constant, though it may be nice to cut it off or reset it prematurely)
-the ramp starts needs to be triggered by a digital line
-The ramp needs to go to zero after it completes the ramp and stay there until the next triggering

 

[Roff]

-I need a linear voltage ramp
-constant, fixed slope (the time and magnitude of the ramp is constant, though it may be nice to cut it off or reset it prematurely)
-the ramp starts needs to be triggered by a digital line
-The ramp needs to go to zero after it completes the ramp and stay there until the next triggering

There may be a simpler way of doing this, but this should work. You didn't mention the time duration, so I used 10ms, but you can change C1 as noted in the schematic. I simulated this.
EDIT: Added plot. C1 was changed to 6.2uF to get the time to 100ms. A better solution would be to reduce the current and make the cap 1uF or less.

 

[on1aag]

Hi Dknguyen,

Something like this ?
It was only a simulation, I never built this ****.

on1aag.

 

[dknguyen]

Yeah, like that. Except when I look at those circuits I have no idea how they work! I don't have a very good analog intuition. It's supposed to quite a minor part of the circuit and it seems Roff's is quite a major part! I'm starting to appreciate something like a a PWM DAC right now, but I will sit down with these circuits and stare at them for a while. THe ramp time is 100ms BTW.

 

[Roff]

Yeah, like that. Except when I look at those circuits I have no idea how they work! I don't have a very good analog intuition. It's supposed to quite a minor part of the circuit and it seems Roff's is quite a major part! I'm starting to appreciate something like a a PWM DAC right now, but I will sit down with these circuits and stare at them for a while. THe ramp time is 100ms BTW.

I added a plot to my previous post. I agree that a microcontroller and a DAC of some sort would be better. I also considered an integrator (as on1aag did), but concluded that a current source with a discharge switch (the MOSFET) would give less zero offset, and faster reset time.

 

[dknguyen]

I added a plot to my previous post. I agree that a microcontroller and a DAC of some sort would be better. I also considered an integrator (as on1aag did), but concluded that a current source with a discharge switch (the MOSFET) would give less zero offset, and faster reset time.

Oh, is that what your circuit is? A current source charging a capacitor? I got lost in all the transistors and op amps everywhere. I was thinking about a current source (using a negative regulator or something). At first I was trying to charge a capcitor through a resistor but it wouldn't be very linear and the charging voltage would have to be about double or so of the maximum ramp voltage to keep it linear-ish. I dropped both ideas though because I ran into the problem of the bias currents on the amp's gain control pin being larger than the capacitor charging currents due to the small value of precise capacitors.

 

[Roff]

Oh, is that what your circuit is? A current source charging a capacitor? I got lost in all the transistors and op amps everywhere. I was thinking about a current source (using a negative regulator or something). At first I was trying to charge a capcitor through a resistor but it wouldn't be very linear and the charging voltage would have to be about double or so of the maximum ramp voltage to keep it linear-ish. I dropped both ideas though because I ran into the problem of the bias currents on the amp's gain control pin being larger than the capacitor charging currents due to the small value of precise capacitors.

On1aag's integrator is actually a current source charging a capacitor also. Just another way of doing it. His circuit either has to be controlled open loop, as he has done it, which may be OK, or it also needs the comparator and the flip-flop. I think you said you wanted it to be edge-triggered, That's why I put in the flip-flop, and the comparator to detect when the ramp reached 2V so it could reset the flip-flop and the ramp.

 

[dknguyen]

I just said edge triggered because it could be triggered by the ultrasonic square wave signal driving amplifier or a separate digital output from the uC that goes HI simultaneous to the start of the square wave transmission, and I didn't want to make it all wordy as I have done in this post. I guess saying edge triggered means the same as "rising edge triggered", which is what I would have said if I wanted it to be actually edge triggered...yeah. Lazy.

 

[Hero999]

Why can't you use a microcontroller?

You connect the output to a DAC which goes to an analogue multiplier, that will allow you to program the gain very easilly in software.

 

[dknguyen]

It's just that I wanted to having an actual continuously varying gain, though now that I've seen the complexity of it, I am considering just using a PWM DAC.

 

[dknguyen]

THis is my PWM ADC circuit. It has to be isolated anyways since the electronics on the DAC side of things are not isolatable from the 200V signals running around. And it helps me use a 0-5V signal to switch transistors with 0 and -5V power rails.

I figured the resistor placement would be a nice way to use the filter resistors to control shoot-through too so I wouldn't need fancy dead-time control (although I suppose I could do it through software anyways since I could just control the FETs separately through the gate driver by splitting up the gates and connecting them to different drive pins on the IC.

Does anyone see any problems?

 

[cjb]

Hi All,

This is probably coming in slightly late, but would a circuit like the attached not be a lot simpler?

The Run/Stop signal could be generated from a PWM on a Microprocessor, or a 555 running as a monostable. With Run/Stop held high, the output sits at 0V, when it goes low, the cap charges until it hits the current sink's compliance, or until Run/Stop goes high again.

 

[dknguyen]

Well, I need an isolator anyways and both circuits have two resistors. I was actually going to add an inductor in to make it an RLC filter at 30kHz, and switch it at 500kHz since it seems the MCU and FETs can handle the power levels, frequency, and transition times just fine.

Technically I need a 1-10% duty cycle sawtooth signal at 10-100Hz. So from Wiki it seems I need at least up to the 30th harmonic to make it a near perfect wave, so the filter and switching frequencies are way beyond what I need, but they make the inductor the same on as I am using in other parts of my circuit. In actuality, I doubt I need a sawtooth signal since I don't care what the second half of the waveform does after it ramps up, as long as it reaches the start value again by the time I need it. So if it's actually a triangle wave that just reaches the starting value by the next time I need it then the bandwidth requirements are reduced by 6x.

Thanks.