Analog Sample-and-Hold for a Noise Source; faulty circuit?

[funtentrains]

Hello everyone,

I'm new here, and have not much practical experience with electronics, so please be kind .

So, I have a project for which I need low-frequency (i.e. less than 10Hz) noise to drive a PWM circuit connected to an LED for a 'flickering candle' effect.

Since most noise source circuits (using Zeners and opamps) are guaranteed only for at least audio frequencies (20Hz and up), I decided to use the circuit presented in another thread here (https://www.electro-tech-online.com/threads/analog-digital-noise-downsampling.117339/#post962741) to try and sample noise so as to bring it down to a usable frequency.

The original circuit used a 555 timer chip and a 4066 analog switch to perform clocking and sampling functions. Unfortunately, due to the constraints of my project (it's for school), I am not allowed to use any IC more complex than an opamp to perform the functions.

(Also, no digital ICs are allowed, which means that I'm not allowed to simply use an LFSR to do the flickering candle effect, hence this roundabout way of generating noise)

So, I replaced the 555 with an astable multivibrator feeding into a monostable, generating a 100us pulse every sample period, and I replaced the 4066 with a single NMOS.

The modified circuit is pictured here:
View attachment Analog Sample-and-Hold.png

(As an aside, I'm also using a strange hybrid of single- and double-sided supplies for the various opamps, which is a result of the circuits for the astable and monostable multivibrators working only when they're supplied with a double-sided supply, whereas the original circuit was designed using a single-sided supply.)

Unfortunately, the performance of the circuit is terrible; the sampled waveform is not the stair-step expected, but a strange ramp/sawtooth shape. (voltage across the sampling capacitor C1 rises and falls even when the current through it is zero! )

Simulation results, original circuit:


Simulation results, modified circuit:
View attachment Analog Sample-and-Hold Simulation Results.png

Voltage across capacitor C1 vs. current through C1. Note that voltage changes even as current is zero:
View attachment Voltage vs Current, Capacitor.png

What am I doing wrong here? Is there any way I could improve the circuit's performance? Should I pursue some other approach to the 'flickering candle' problem instead?

I would be grateful for any advice.

The circuit's schematics and relevant component models are attached.

 

[crutschow]

I think you are making it more complicated then you need to. Random noise sources such as from reverse biased zener diodes (12V zeners give higher noise than lower voltage units) have no particular lower noise frequency limit. It just limited by the coupling capacitor RC (R is output resistance) time constant (it should be a least 1 second). You should be able to amplify this low frequency noise with an amplifier and then filter to get your desired noise spectrum. Add a pot in series with the zener to find the optimum bias current for most noise (add a fixed resistor in series to make sure you don't exceed the zener power rating).

 

[Roff]

I agree with Carl.
The ramps you are seeing are due to the input bias current of U5. Do the math. You need a FET-input op amp for U5.

Further comments on your circuit, for future reference:
1. The Veb breakdown voltage of most BJTs (Q1, in this case) is around 5 to 6V (see the datasheet). You should have a base current limiting resistor, and a diode from base to emitter to protect it from reverse voltage.
2. You could probably turn off Q1 with a differentiator to the base (I call this combination a half-shot), instead of the U2 one-shot. The falling edge of the collector waveform won't be really fast, but you don't need really fast turn-off time for M1.

All in all, a pretty nice design effort for a novice, even if it is overkill.

 

[funtentrains]

Thanks crutschow, Roff for your feedback. I'll shelve this (admittedly overkill) circuit and focus on a simple(r) Zener noise source + filter circuit. I'll probably be back for help, though .

Again, thanks.