Problem with chained LM3915

[audioguru]

I am retired so I also don't have a job. But I am very busy and do not have enough time to do everything I want.

 

[chemelec]

I am retired so I also don't have a job. But I am very busy and do not have enough time to do everything I want.

Yes you spend Far too Much Time on these Various Discussion Forums in Helping Others.
So no time to do anything else.

But it Keeps the Brain Active!

 

[fezder]

I have many friends who are into electronics but NONE could shed light to any of these problems. so audioguru's help is/was priceless! and even google didn't help much

 

[audioguru]

Did you notice that this thread has over 100 posts?

 

[fezder]

hehe, yeah, but luckily not much off-topic

 

[fezder]

I was wondering, is it possible to change that your circuit so it shows average instead of peak? as i'm planning to place both peak/average, if that is easily made instead of peak/instant. Or then only peak, just thinking out loud here

 

[chemelec]

I was wondering, is it possible to change that your circuit so it shows average instead of peak? as i'm planning to place both peak/average, if that is easily made instead of peak/instant. Or then only peak, just thinking out loud here

Increasing the Value of the .33 cap on the Output should do it. (A 1 uF or 4.7 uF.)

 

[fezder]

Increasing the Value of the .33 cap on the Output should do it. (A 1 uF or 4.7 uF.)

i was thinking about similiar, as bigger capacity capacitor holds charge longer. If i'm not terribly mistaken it also charges slower but that's negligible in this situation, it's still avergae

 

[audioguru]

The datasheet says that feeding the audio directly to the input of the LM3915 without a rectifying peak detector and without a capacitor will allow some peaks to be seen dimly and average levels to be brighter when the LM3915 is in the DOT mode.

A true average detector charges its capacitor slowly and discharges it the same. Then the datasheet shows changing the 100 ohm resistor to 100k ohms, and the transistor is not needed. You will not see the peak levels that might be much higher than the average levels. If you increase the capacitor value then the discharge will be too slow causing many events not to show.

 

[chemelec]

That 100 Ohm Resistor will Not Slow down the Charge rate very much.

 

[audioguru]

That 100 Ohm Resistor will Not Slow down the Charge rate very much.

The low value of the 100 ohm resistor charges the peak detecting capacitor very quickly but limits the charging current. I said the datasheet says to change it to 100k ohms to see average levels then it charges slowly the same as it discharges.

 

[fezder]

Got mylars today and couple opamps, these won't help thought, they need dual supply but i noticed it after placing order (were cheap hehe....)....lf353 & lf348 (dual & quad) and offset too is too big, tomorrow i should get those proper opamps

 

[Mosaic]

This may be OT, but the newer enhanced 32Mhz 16F1xxx Pic series with multiple Internal Vrefs for $1.50 or so would be able to sample audio and drive an LED display with PWM and deliver any type of linear/log averaging or peak or RMS or simultaneous peak & RMS or whatever you could want. A 28 pin PIC could do all 20+ outputs under PWM with trailing FX etc, if a bright full bar mode is essential you could add 3 ULN2003 darlington buffers for another $1.65 total.
The LM391x series is a bit dated now, although fun to play with.

 

[fezder]

No worries, i'm open for ideas. i didn't feel like placing MCU for such simple project

 

[fezder]

Got today those better opamps, MCM model works better in this circuit, supply voltage is 5v
Hmm, still output from peak detector is 0.4 volts lower than input.
Resistors are 4.7k (r7,r8,r9,r11) and capacitors are mylars(c4 & c8=330n c5=100n), no more ceramics.
Thing of note, if i take output capacitor from peak detector output out, output increases by 0.4 volts
nvm that, i tested with too high frequency

with TLE in place, output at peak detector is at least 200mv all the time
bought these opamps from farnell

i noticed also that lm358's offset can be ''nulled'' in first lm3915, by inreasing lower level. it then drops its sensitivity though but at least bars won't lit without sound applied
but in thi case i need also to change other lm3915's lower lever too, but seems won't be needed

But. it works enough well for now, even though peak detectors output is bit lower than input signal. i'll solder this now to protoboard if that makes any issues better...

 

[Mosaic]

Not to belabor a point, but I am using a differential instrumentation mode jelly bean LM324 configured as classB to derive peak voltages so rail to rail is not required...that feeds into an ADC of a 16F886 via a 1k-.01uf-1K pi filter for noise free peak signal sampling (Nyquist oversampling as well) . I added a 1n4148 diode clamp across the cap. in case of any over range conditions to prevent ADC latchup. A trimmer pot is used on the diff. amp feedback to get approx 0.02V accuracy, tested with a new 6.5 digit benchtop DMM. No expensive opamp, or dual voltage supply required.

Are you configuring a peak detector diode as an ideal diode with the opamp? It should have enough drive (20mA) to 'fill' the capacitor in time. I= C dv/dt is the eq you can use to calc. that response capability.
https://en.wikipedia.org/wiki/Precision_rectifier

 

[audioguru]

The peak detector is a precision type with an output peak voltage within a few millivolts of the input peak voltage. Your last posted schematic 2 weeks ago and mine should work perfectly so please post the schematic that causes the 0.4V voltage drop.
Why are your resistors R7, R8 and R9 only 4.7k when mine are 100k? Then the cutoff frequency of C4 and R9 is 103Hz so it will not detect bass frequencies.

 

[fezder]

I opted that 103hz would be good for my purposes, at least it seemed to give best performance so far, but that can be false performance if this circuit is not working as intended, yet i mean
circuit for 103hz

scope shot, actually voltage drop was much lower....0.2 volts


and for comparison, original values of 100k resistors (4.8hz)

and there's some bumbs at output too


tested with 3 different NPN small signal transistors: bc547, 2n5551 and c1815, same results

 

[fezder]

here's couple shots from protoboard

output from opamp

inverting (-) input

non-inverting (+) input

 

[Mosaic]

The OPA FB pulling down your input sig it seems.
Use another OPA to buffer the input signal first or increase the input cap by about 10 times.
Also R11 may not be required as that OPA does not have crossover distortions.

Try the precision rectifier circuit from wiki...the input signal is isolated from the FB, none of these design complications that you have.