How to calc choke uH value for analog filtering?

[riotubes]

Hi,

I'd like to add a toroidal inductor after my dac's analog stage (discrete active IV and gain block) to help eliminate the digital images / HF hash as a result of the filterless design. It was suggested I try a range of values (20uH to 500uH) to find the best balance of reduced hash without flattening the presentation too much. Is there a way to model a narrower range of potential uH values without too much trial and error? I'm thinking of trying the J.W. Miller toroidal inductors available at mouser.

Dac:
10 ohm impedance seen by dac chip
1K ohm output impedance
twin tda1541a = 8mA
next stage: S&B TVC

[Rick Miller]

I am using 100uh before the I/V stage with good results.

[riotubes]

Thanks Rick for your feedback on 100uH inductor. Are you using passive or active IV? My prior tube stage used resistor IV and the simple RC filter was just after the IV but before the gain stage. On this current stage (active IV) the designer suggested I place the inductor between the output caps just before the RCA out jacks which is fairly unconventional I presume.

IV Output Stage.pdf

[infinia]

On this current stage (active IV) the designer suggested I place the inductor between the output caps just before the RCA out jacks which is fairly unconventional I presume.

Why didn't you ask the designer about this?
Your given criteria for "the inductor" is nonsensical maybe not yet unconventional. I would recommend use of a differential RC filter (slew rate limit) and a differential amplifier to isolate the digital noise from the quieter analog stuff (regular grounding stuff).

[jcx]

in audio signal processing we usually try to avoid small signal inductors as they are the worst class of passive component in terms of distortion and parasitics

lossy ferrite beads are sometimes handy to attenuate MHz digital hash but some (Curl, Hansen) claim they still cause audio frequency problems

air core inductors are used when nothing else does the job at high powers

[riotubes]

Originally Posted by infinia

Why didn't you ask the designer about this?

I did. He suggested I try various uH value toroid inductors (instead of RC network) after the analog stage and select the best value based on listening. Sorry my post wasn't clear.

The question in my original post was whether there is a way I could calc/simulate appropriate uH value to narrow down the range of uH values I experiment with. I recognize some folks prefer active filters while others prefer passive RC or CLR filters or no filters. The single inductor is the simplest solution so I thought I would try that first and proceed from there.

[riotubes]

Originally Posted by jcx

in audio signal processing we usually try to avoid small signal inductors as they are the worst class of passive component in terms of distortion and parasitics

Thanks JCX for the helpful feedback. My prior analog stage was a SRPP tube stage that followed passive resistor I/V. There was a CLRC filter after I/V before the tube. I removed the first cap and replaced the inductor with a 20ohm mills resistor at a more experienced DIY'r suggestion. So it was 20ohm in series with 150ohm R with a 1nF cap to ground. This sounded very good. But I understand these filters to be tuned circuits that rely on the specs of the specific circuit and the new analog stage is 10 ohm source impedance whereas the prior one used 44 ohm resistor. I downloaded filter pro but don't know how to use it to optimize for a 1st order passive filter.

Originally Posted by jcx

lossy ferrite beads are sometimes handy to attenuate MHz digital hash but some (Curl, Hansen) claim they still cause audio frequency problems

I see that I have a ferrite bead in each rail between the raw digital supply and the regs it feeds.

Originally Posted by jcx

air core inductors are used when nothing else does the job at high powers

My understanding is that the alpha core air core inductors are very good but large and expensive.

[infinia]

My main point I'm trying to communicate is the grounding and reduction of digital noise is most important at the D-A interface not filter topology per se.
Maybe you could investigate using common mode chokes here instead of single ended.

[Rick Miller]

Riotubes, I am using a AD811 as my I/V stage and have found out that even with its large GBW it needs some pre filtering to sound its best. The more filtering L I place in front of it the better it sounds. As of right now I have 100pf on the output of the 1541 DAC then a 16mh choke followed by a 100uh choke feeding the 811 and it sounds very good.

[riotubes]

Rick: thanks for the continued feedback on filtering, particularly given your experience with the tda1541

Infinia: I'm at ground zero regarding my knowledge of analog filtering. I googled differential RC network and differential amplifier and will read up on these aspects further. My current output stage is single ended though. Building this dac has exposed me to various rudimentary elements of design including psu, discrete regulation, i2s interface, analog stage, etc. but I have not addressed filtering up to this point. So I need to spend more time learning about filtering. Thanks for the suggestions.

thanks for participating...I'll follow up with some observations next week. Here's the output stage.