TI Active volume control board...tipd136

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2007
Paid Member
Around 4kHz is where the ear is at its most sensitive so you don't really want to be seeing noticeable distortion at all. Fortunately that's not a problem for most small signal stuff.

The specific questions aimed at this active volume control are a special case... it was whether the distortion rose significantly as the class ab opamp output stage fed an ever lower impedance as seen at its output, the effect of high attenuation levels.

Noise in front end circuits is worthy of consideration because its only going to get amplified all down the line. Matching guitar pickups to the most suitable amp stages isn't something I've any real knowledge on I'm afraid.
 
Administrator
Joined 2007
Paid Member
I think it needs all those Jan such that it is universally applicable and non inverting of overall phase. It needs to be driven from a low source impedance and also present a known and constant impedance to the outside world at its input. You could eliminate one of them and have the first stage as inverting. So three opamps in total would be do-able.
 
Hi Bob, just to clarify, this isn't a soundcard based PC measurement system. It's an Audio Precision APx555, which is capable of FFTs up to 1MHz. And you're right that the distortion harmonics are most likely increasing at 20dB/dec as the loop gain of the OPA1604 rolls off (35MHz GBW).

But, again, this behavior appears when using a low impedance potentiometer in the circuit (1kOhm) . Isn't this just another example of the design trade-off that often arises between noise performance and distortion performance? For example, what if we changed all the resistors in the circuit to 10 ohms? We would certainly reduce noise, but drastically increase distortion! It's up to the engineer to decide whether noise or distortion performance should be prioritized in their application :)

ChristianThomas:
I should point out that the idea of using a pot in parallel with the gain resistor of an inverting amplifier will have the same distortion behavior as the Baxandall circuit. You are attenuating the signal while simultaneously increasing the output loading on the op amp.

You're right; there is always a tradeoff in engineering, and audiophile circuit design is not exempt from this. An obsessive focus on low noise at the expense of other performance characteristics is unwise. Reducing resistances to get lower noise at the expense of making op amp output stages work harder is an example. Indeed, thermal noise goes down only as the square root of resistance, so going all the way from 10k down to 100 ohms gives only a factor of 10 improvement in noise while increasing circuit currents by a factor of 100.

Engineering is all about striking the best balance among competing priorities.

Cheers,
Bob
 
Indeed, thermal noise goes down only as the square root of resistance, so going all the way from 10k down to 100 ohms gives only a factor of 10 improvement in noise while increasing circuit currents by a factor of 100.
While illustrating the point, that's a bit apples to oranges.

I'd look at it like this:
You just increased signal power levels in the circuit by a factor of 100.
Lo and behold, the ratio of signal power to noise power goes up by a factor of 100.

Makes sense, doesn't it?

If you keep voltages constant but lower resistor values, power levels will increase directly proportional to the reduction in resistance. 1/100 the resistance --> 100 times the signal power.
A factor of 10 in noise voltage is a factor of 100 in noise power.

So you actually do get what you pay for.
 
While illustrating the point, that's a bit apples to oranges.

I'd look at it like this:
You just increased signal power levels in the circuit by a factor of 100.
Lo and behold, the ratio of signal power to noise power goes up by a factor of 100.

Makes sense, doesn't it?

If you keep voltages constant but lower resistor values, power levels will increase directly proportional to the reduction in resistance. 1/100 the resistance --> 100 times the signal power.
A factor of 10 in noise voltage is a factor of 100 in noise power.

So you actually do get what you pay for.

True. Look at it whatever way you want.

Keeping the nature of the units the same, dropping the resistance by a factor of 100 drops the thermal noise voltage by a factor of 10 but increases the required op amp output current by a factor of 100. Current-induced 2HD percentage could go up by a factor of 100 and 3HD percentage could go up by 10,000 if I'm not mistaken.

Not a good tradeoff in my opinion, especially because current-induced distortion from a class B or AB op amp output stage includes crossover distortion, which is particularly objectionable.

It is an example of where an obsessive focus on low noise is not a wise choice. Similarly, an obsessive focus on low distortion that compromises other performance metrics is also not a wise choice. An example here would be sacrificing gain margin, phase margin and stability in order to achieve a higher ULGF and more loop gain to reduce distortion.

Cheers,
Bob
 
Hi all, build 1 prototype, no hum when heavy attenuation is applied from 10k potentiometer(linear).

But there is a problem, when LM337AT/317AT PSU is turned off, i still hear the sound from amp(gain 27) or headphones(40 oHM) :D
Sound is full of DC, its somehow gets multiplied while Texas is powered off.(DC in 2.4mV, DC out 20mV jumping with heavy bass.....As soon as i turn power back on to the prototype, everything normalizes in no time and it attenuates music as well and there is only 0.8mV DC steady state!

Laboratory +/- split supply has no effect in curing power off sound leakage...
I have tried NE5532, OPA2134, with all these there is still power off sound leakage....
Changed input caps, from tantalum to MKS.

Someone, ideas ?
 
Thanks for answer (Y)
Yes, i will think of some relay muting/power off circuit.
Finished second one, will add third one, side by side on the same PCB.
6VA 2x18V AC, each winding 170mA transformer with LM AT-series regulators will do just fine :D

First there was a plan to implement them into AMP to get rid of HUM when using potentiometer without active devices.

I think they are better off in a tiny box as a preamp driving longer unbalanced distances.

Maybe someone knows easier way to implement a headphone stage to these Texas devices so one could be used as a preamp or headphone amp ?

Renu
 

Attachments

  • 150520151506.jpg
    150520151506.jpg
    813.3 KB · Views: 361
  • 150520151507.jpg
    150520151507.jpg
    833.5 KB · Views: 321
  • 150520151513.jpg
    150520151513.jpg
    878.5 KB · Views: 314
Finished the board

GND`s are separated (PSU and audio) as "stars" which are connected together @ 3.5mm jack output GND.
Decoupling paralleled capacitors are grounded @ PSU 0V star.

No hum or any kind of noise.
Total current draw when both channel are active is 33mA(opamps: opa2134)
Output DC offset is 000,0mV, i don`t even know why it dropped to zero from 0.8mV. But my guess will be in the schematic with changing some resistor values(gain of the first stage opamp).

Max signal input is 0.8V, output signal 1.2V.

Can i use here 1uF instead of 22uF @ both locations of Texas Active volume control ?
Texas recommends non-polarized caps, all my 2uF+ caps are polarized.....

Is it worth constructing better filtering then LM3xAT series can offer ?

Thanks in advance!
 

Attachments

  • 160520151525.jpg
    160520151525.jpg
    724.3 KB · Views: 67
  • 160520151519.jpg
    160520151519.jpg
    601.1 KB · Views: 73
  • 160520151518.jpg
    160520151518.jpg
    693.4 KB · Views: 70
  • 160520151517.jpg
    160520151517.jpg
    764.7 KB · Views: 293
  • wireing.jpg
    wireing.jpg
    51.6 KB · Views: 292
Last edited:
Played with the gain in the input stage, ended up with 24K to ground and 10k from the output to negative input.

Replaced 100k from Texas output with 56k to GND <- output impedance is now closer to the amp input impedance.

Why there is no significant DC on the output of Texas? I used to have 0.08-0.1mV with only 1 channel active.

Both channels active using 1 stereo potentiometer:
(Right channel: -000.2mV)
(Left channel: +000.0mV)

Polarity of DC on the output are somehow different ?

Maybe the reason lays behind the ground-plane that both channels share ?
 
Administrator
Joined 2007
Paid Member
All those values are insignificant. Any DC offsets are determined by the inherent characteristics (their internal offsets and so on) of the chosen opamp and any external imbalance caused by the circuit values allowing differing bias currents to flow into each input.

At those levels each opamp will be different and each channel different one to the other.
 
Still using OPA2134 chips.
Firstly, tried different PSU schematics on that Texas. LM317AT/LM337AT with the right ESR caps on bypass and output, results = OK.

Then build another LM317/337 psu with the shunt on the output. Results= Better, especially in highs and top end was more like "bounce", very melodic, maybe even better dynamics.

15-18V variations were used, ended up using Texas with +/-15.6V.

Tried 22uF 35V Tantalum capacitors as signal coupling on Texas. Result: Better then 1uF MKT for sure. After a while changed Tantalums to 22uF SMD ceramics. Results: Not sure, but it seems ceramics sound softer then Tantalums.

Still using 317/337 with shunt on the output:
Before mounting everything into amp chassis, decided to simplify Texas Layout.
Rebuild: less wires, more GND routing on the PCB, less electrolytic caps(cuz of the higher uF value per cap), bypass on electrolytics with 3.3nF MKT, 100nF bypass capacitors(X7R ceramics) from every supply pin and as close as possible. Output resistor is now 100R instead of 43R.

Signal GND and decoupling capacitors GND routes/traces are connected together only at 1 point: PSU 0V GND.

Power up time: to our suprise the DC offset is back with the new LAYOUT, both channels are equal -0.2mV.

Secondly, whole Texas draws less current lol. 34.6mA before, ~33.2mA now.
100R has nothing to do with schematic current draw.

Then, added one 10nF X7R capacitor on each opamp package as +/- supply coupling. Turned Texas on, again, the current draw dropped, now 32.44mA.

Question: duo to layout limitations in the case i cant mount potentiometers on the PCB, so i took them out from PCB via "shielded wires", the ones used in RCA audio cables. Only at one end Shields are connected to PSU star 0V gnd, where do you suggest to connect the other end of shields to avoid potential humming noise? See the last picture please.

Thanks
 

Attachments

  • 070620151541.jpg
    070620151541.jpg
    634.7 KB · Views: 64
  • 070620151542.jpg
    070620151542.jpg
    643.5 KB · Views: 51
  • 070620151543.jpg
    070620151543.jpg
    695.7 KB · Views: 52
  • 070620151546.jpg
    070620151546.jpg
    872.9 KB · Views: 67
  • texas123123123.png
    texas123123123.png
    112.9 KB · Views: 110
Administrator
Joined 2007
Paid Member
You can leave the shield floating at the pot end because its doing its job by being grounded at the other.

(It is also possible to drive each shield from the appropriate opamp which retains a low impedance for the shield, and because the shied is driven as well, the cable capacitance is cancelled out because there is no potential difference between shield and inner)
 
You can leave the shield floating at the pot end because its doing its job by being grounded at the other.

(It is also possible to drive each shield from the appropriate opamp which retains a low impedance for the shield, and because the shied is driven as well, the cable capacitance is cancelled out because there is no potential difference between shield and inner)

I was going to say that, too. But it might be worth having the body of the pot grounded too - assuming it's made of metal. That might just help a little bit. But keep away from making a circuit because, as Mooly says, it's doing its job already.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.