envelope follower help?

[PRR]

> The 1K pot provides some attenuation to the ammount of current to the led as far as I can see.

In electronics we can often calculate part effects.

What is the base voltage? The output of the envelope follower, several volts. Let's pencil 5V.

What is the LED current in the transistor? Say zero to 20mA. What is the transistor's base current for 20mA? hFE of a Darlington may be 10,000. 20mA/10,000 is 0.002mA.

0.002mA in up-to 1K resistance is up-to 0.002V. This is insignificant.

There's other ways to trim LED current. The LED resistor R is an obvious point.

But get your envelope follower working, to the point that the cap voltage rises and falls as you plink.

I suggest plagiarism. The problem has been solved before. I know you want to grow your design skills, but studying and building-upon the work of those who have gone before us is an important technique.

In this thread we find several commercial designs which "DO" bias an old-type opamp to the negative rail. In volation of opamp specs. Apparently this does work. It probably only works right for that specific opamp, expoiting some undocumented side-effect.

[daverdave]

Well I've changed the schem with a few suggestions made but I haven't built it yet. I did try biasing the opamp inputs at 4.5V as well and there was no improvement to the operation of the circuit.



Thanks for the replies Paul, it's really appreciated. I am trying to improve my design skills but this circuit was whipped up pretty fast withough much thought. I don't own my own test gear asides from a DMM so it's hard to see what's going on in the circuit. I just used the rectifier from the neutron project which seemed like point to start from, then added pots to change the attack and decay near the reservoir cap. The 1K pot is just taken from the meatball envelope, thought it would be a decent control to include.

I'll see if can get the circuit above built up and see how it goes, does it look plausable at least?

[daverdave]

ok, so you're all probably getting sick of me posting about this. But I've made a bit of progress.
I hooked up the meatball envelope follower to see how that functioned, and then tried the LM1458 in the circuit I've drawn up, I hooked up an LED to use as a probe. So the rectifier section of the circuit worked with the non-inverting input going to ground through the 1M resistor, but the following inverting / non-inverting amplifier didn't. So I hooked up the 10k resistor / LED voltage divider to get roughly 1V, and used that to bias the extra input of the amp. That worked perfectly when UP was selected, but not when down was, the LED just stayed continuously lit.

What I did notice though, was when I touched the ground resistor, that would have been connected to 1V, with a wire connected to ground it started working, anyone have any idea why?

I'll keep at it and see if I can get it working better.

[daverdave]

Right, so I've changed the circuit quite a bit, still not working perfectly. The inverting 'down' amp is working, but the non-inverting 'up' amp isn't, I have to really whack the strings to get a response. I'm pretty sure my problems are there. I put in the 1M resistor on the non-inverting input becasue I thought the amp was being affected by the 100k decay pot. Can anyone look this over and give me some advice? Pleeeeaaassssee!!!

[Gurner]

Your pinouts on the ICs are wrong on your schematic vs the datasheet. (for both TL072/LM1458 variants) pin 2 & 3 are fine but pin 6 is not the output (it's actually the 2nd opamp inverting input - http://focus.ti.com/lit/ds/symlink/tl071.pdf) ...you may know this already, but when seeking help, you should really iron out the obvious errors on your schematic ...else you're just wasting people's time (ie like mine by typing this telling you about the error!!)

What DC voltage are you measuring across that 10uf cap under healthy incoming signal circumstances? (there'd need to be about 4.5V there quiescently - as that's what your second stage is biased at). Also, none of your opamps in that circuit are providing any gain, so your incoming signal is gonna need to be pretty chunky to get any meaningful DC level extraction onto that 10uf cap.) A tip here would be to use your PC as a sig gen as an input ( http://www.dxzone.com/cgi-bin/dir/jump2.cgi?ID=17728) to give you a nice steady input signal to start debugging DC levels.

When you talk of 'up amp' & 'down amp' - what do you mean? (in other words what are you hoping the end result is when switched to inverting vs non-inverting)

[daverdave]

Yeah sorry, I couldn't find the TL072 on the program so I used a 741 single, the connections are correct on the circuit. I'll try to ammend that when I get home.

I'll also measure that cap, I do know that the signal passes through when the 10uF cap is removed, I socketed it to try different values. As I said before, the inverting setup is working, but not the noninverting. Basically the switch toggles the opamps inputs between input and ground (well 4.5V). The inverting amp setup gives a DOWN sweep (ie, makes the LED go dimmer from bright).

I thought the rectifier was providing gain? with the 4.7K input resistor and the 4.7M feedback resistor, that would give a gain of 100 right?

[Gurner]

Yeah sorry, I couldn't find the TL072 on the program so I used a 741 single, the connections are correct on the circuit. I'll try to ammend that when I get home.

I'll also measure that cap, I do know that the signal passes through when the 10uF cap is removed, I socketed it to try different values. As I said before, the inverting setup is working, but not the noninverting. Basically the switch toggles the opamps inputs between input and ground (well 4.5V). The inverting amp setup gives a DOWN sweep (ie, makes the LED go dimmer from bright).

I thought the rectifier was providing gain? with the 4.7K input resistor and the 4.7M feedback resistor, that would give a gain of 100 right?

Yeah, I glanced past it too quickly & was focusing more on the 2nd stage, indeed that 1st stage is providing a gain of x1000  ie 4,700,000 / 4,700 (which frankly, is a stupid amount of gain)

Re your expectations for the circuit - well since the LED is hard wired one way, I don't think it's going to work quite as most would want it (or more to the point, when you flick between inverting & non inverting, you're gonna have to tweak that pot significantly each time)  With the switch in the inverting position the second stage TL072 output will decrease from 4.5V towards ground as the incoming signal increases. Quiescently the LED will be bright, and with the strongest signal, the LED will dim..

However, without touching the pot, when placing the switch in the non-inverting position, then quiescently, the second stage is still gonna have 4.5V, therefore with no incoming signal, the LED will be on and bright - and this brightness  will increase along with the signal towards 9V. So you'll need to adjust the pot with every switch flip.

If you're saying that the non inverting isn't working that well, I'd say that quiescently the 10uf cap has not got sufficient voltage across it, which points to biasing issues on your first opamp stage vs the 4.5V your second opamp stage is biased at.

[daverdave]

That would make sense, there not being enough voltage across the cap, but with the gain of 1000 I'd have thought it would have been enough, I know it's large but it was more an experiment. I took it from the max gain of the neutron. I could cut it down to 500 or so.

Which pot are you talking about adjusting? I can see what you mean about the non-inverting being biased at 4.5V, but the LED isn't lit halfway. It stays off, and as I said, takes a bit to ghet going.

One thing I did notice is that when I measured the voltage on the inverting input with a multimeter (red probe to the input, black to ground) the follower jumped into life. I didn't have enough time to mess around with it, but at a guess I'd have thought that the resistance through the multimeter to ground was pulling the inverting input lower, does that sound plausableor am I way off?

[Gurner]

Like I say, looking at how your first stage is biased, there ain't gonna be 4.5V across that 10uf cap .....so there's gonna be come contention at the second stage as the extracted dc (from the AC signal) across the cap 'fights' with the 4.5V biasing on the second stage opamp.

The gain of 1,000 (or 500) is a poor workaround for having inappropriate biasing on the 1st stage opamp & will also mean you've got no nice steps/gradients going on. What you really want is say 4.5V quiescently and ramping up towards 9V slowly depending on input signal magnitude. What you've got is almost a switch ....ie a quiescent DC level across the 10uf cap, which will essentially shoot towards 9V when a signal comes in ...ok so the CR constant of the 'attack' pot & 10uf cap will slow this somewhat 'binary' condition ....but it's a poor implementation. Far better to have the gain set proper so that only the maximum signal input sees the cap charging towards 9V

(btw:  I'd still say that even a gain of 500 is way too much – what size/type of signal are you putting into the input?)

The pot I was talking about you having to adjust when switching between inverting & non-inverting is the intensity pot.

Re your measurement affecting the circuit conditions – it shouldn't so you've either got a really bad DVM/Multimeter that's loading the input ....or a bad joint in that area that's being 'made' when you probe it.

So a couple of thing I'd check...

1.   DC across the cap with no signal
2.   DC across the cap with large signal.
3.   DC output from that second opamp output for both the above conditions.

[daverdave]

Just measured the voltages around the 10uF cap, without a signal I'm getting about 2.3V out of the rectifier, just before and just after the cap.
Out of the inverting amp I'm getting about 4.5V, dropping to about 3.5V when the guitar is strummed. And I'm getting 0.9V out of the non-inverting amp, rising to about 3V.
I'm using a kramer 450g for the input signal, apparently the pickups don't have a very high output.

I've been looking at a few other envelope schematics, the boss TW-1 uses a similar circuit for it's follower, here's the schematic.


http://www.experimentalistsanonymous.com/diy/Schematics/Filters%20Wahs%20and%20VCFs/BOSS%20TW-1.jpg

The non-inverting input of the rectifier is biased at about 3.8V if the Vref is 4.5V. The pinout for the next amplifier is confusing, as it says the inverting input is biased at Vref, but the labels for up and down on the switch would suggest that the non-inverting amp is biased. I can't understand how the UP switch works to switch the second stage into a non-inverting amp, it shorts the non-inverting input to the inverting input's input resistor!

[slacker]

When the switch is in the up position the second opamp works like this the signal goes through the 100k resistor into the inverting input, that combined with the 100k resistor in the feedback loop gives it a gain of -1. The signal also goes into the non inverting input and the two 100k resistors set the gain in exactly the same way as if the left hand end of R29 was connected to vref. This gives a non inverted gain of 2, so the overall gain is -1 + 2 = 1, so it acts like a non inverting buffer.
In the down position the non inverting input is connected to vref through R30 so it acts an inverting buffer.

[daverdave]

Ah right, that makes so much sense, I've seen that before with the mutron III as well. Thanks alot for that. So I'm gueesing the trimmer after the 2nd amp is to move the quiescent signal back to 0V? I'll have to try that, it seems alot simpler.