Digital potentiometer for AC signals with low THD / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Digital potentiometer for AC signals with low THD

1 2 next

Move Thread

LAN_403

Linas

Mon Feb 01 2016, 10:52AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

I have such a strange problem:
In my design, i need to control volume from microcontroller. All sounds OK, but audio sygnals are completely AC (+-1V).
I need at least 100db dynamic range, because this is what i get from my dac, but i was unable to find any digital potentiometer that does that, closest is AD5262 , and it can only give 0.014% THD 77db

Can any one give me idea, how to control sound, with more dynamic range, with complete AC input ?

Sulaiman

Mon Feb 01 2016, 11:18AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

Just to check, what is the maximum ac signal pk-pk voltage and bandwidth ?

...

Mon Feb 01 2016, 01:00PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

It seems unlikely that you will find a single chip solution that will give you that kind of performance.

There is a reason that people use relays and resistor banks for attenuators, they really are hard to beat in terms of performance.

You might try looking for 'programmable gain amplifiers', there are some pretty high end devices like the PGA281 which claims 0.0001% nonlinearity over a +-18v swing (!), although it doesn't have many gain steps and you will have to mess around with the levels to make use of them all (it goes from gain = 0.125 to 128 in 11 steps).

Linas

Mon Feb 01 2016, 02:23PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

... wrote ...

It seems unlikely that you will find a single chip solution that will give you that kind of performance.

There is a reason that people use relays and resistor banks for attenuators, they really are hard to beat in terms of performance.

You might try looking for 'programmable gain amplifiers', there are some pretty high end devices like the PGA281 which claims 0.0001% nonlinearity over a +-18v swing (!), although it doesn't have many gain steps and you will have to mess around with the levels to make use of them all (it goes from gain = 0.125 to 128 in 11 steps).

0-20kHz ans signal should be around 1V or more. ( what is better for low noise)

Yes, VGA would be nice, but i need to reduce gain, not apply it !

johnf

Mon Feb 01 2016, 06:25PM

Registered Member #230 Joined: Tue Feb 21 2006, 08:01PM
Location: Gracefield lower Hutt
Posts: 284

Try a PWM or DAC to a LED's that shines on a LDR's set up like a pot

Sulaiman

Mon Feb 01 2016, 07:17PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

100 dB SNR implies En = (1v/2.sqrt(2)) x 10^-5 = 3.5 uV rms noise voltage

and En = sqrt(4.k.T.R.BW) = sqrt(4 x 1.38 x 10^-23 x 300 x R x 20,000) = sqrt(R) x 18 nV rms

R <= (3500/18)^2 = 37 kOhm

and, good op-amp En = 1 nV x sqrt(BW) = 141 nV rms = negligible

OK, sanity check pass

NEXT: can you measure if you achieve your goal ?

So a resistive divider is feasible provided that the resistance is less than 37 kOhm effective.
I would use signal relays with gold plated contacts (because mercury-wetted are uncommon and a nuisance)
but mosfet switches may be adequate, not sure if low-noise switching mosfets are available
and a resistive divider.

That would be bulky, costly and unreliable if considered commercially.
Try contacting Analog Devices, (or Maxim, or even Texas Instruments, Vishay or maybe Infineon) etc.
for technical help, they probably have something.

100 dB isolation is quite challenging, even for an attenuator, due to parasitic coupling,
intelligent routing of conductors, digital and analogue 'earth' and lots of screening probably required.

Please publish your methods and results, even failures, so I (we) can learn.

Sigurthr

Tue Feb 02 2016, 12:46AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

Is it bad the first "solution" that came to mind was a magnetic amplifier? Haha.

In all seriousness though it may be worth doing a stepper motor controlled potentiometer.

Hazmatt_(The Underdog)

Tue Feb 02 2016, 04:07AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

I would implement a Pi type attenuator with two N channel fets, and a P channel fet as the signal pass element. This makes a voltage controlled pad. You may have to do some clever things with the voltage control to get linear attenuation though, and its fairly simple idea, but the implementation will be difficult.

It will probably require some op amps to swing voltage from negative to positive to give you full control of the fets.

On a positive side, if one Pi stack only gives you 50dB attenuation, you can stack two together with the same voltage control lines, so it can be cascaded.

On the downside, the fet selection may be difficult. Rds On will need to be low, like 30 ohms.

The fets may also add noise to your signal, but I don't know, I've never tried it. I've only done a SPDT fet switch for audio, but it seemed to work okay.

Linas

Tue Feb 02 2016, 06:49AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

At work i have scope with -130db noise floor

So i think i must use DC blocking capacitor, to make it work. Best idea would be ( i think ) is to take 1k digital pot that can operate at 0-10V, use another one good 1k resistor to Vref, lets say ultra stable 5V supply ( i have nice LDO for that, LT3042 :) and all chips will be using this LDO to make it perfect ) , and put like 10-100uF capacitor ( film, not X7R or anything stupid, maybe good tantalum) into middle point, and that should work a treat. I am right ?

1454395741 1143 FT175214 Digital Pot Amplifier

op amps will be all OPA1612AID and similar grade

hen918

Tue Feb 02 2016, 08:23PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

0.014% THD is practically inaudiable. There are very few humans on earth (or possibly none) who can tell the difference between 0.14% and 0.014% THD in a blind test. (Except if it was made up of entirely one harmonic, possibly)

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob