Fixed attenuator for balance control?

[phatfish]

I have a bit weaker hearing in my left ear and was happy to add a balance control to my crack.
https://forum.bottlehead.com/index.php?topic=15470.0

Well, I've got the upgrade bug and I'm planning on a mainline.  I'm looking for the best way to attenuate the right channel a fixed 1 - 1.5db between my EROS2 and future mainline.  I'd prefer not to add a balance pot to the mix with the assumption that'll defeat the benefits of the stepped attenuators.  My thought was that I could add a bit of fixed attenuation via an RCA adaptor.  Something like this:
https://www.parts-express.com/RCA-Line-Level-Attenuator-1-dB-266-230?quantity=1

I haven't been able to find something like this in stock anywhere (only much larger values).  I assume it's just a resistor in there and is something I could create myself, but I have no idea what value I would need.  Any advice on the best way to accomplish this?  I'm also curious the ideal place in the audio stream for something like this: Turntable > Phono Preamp > Amp > Headphones?

Thanks for reading.

[hmbscott]

It contains a very simple circuit called an LPad, that consists of two resistors. One connects signal to ground (R2) and the other connects inline in the signal path (R1).

The trick is to figure out what two resistor values to use, which is complicated by the fact that the LPad R values depend to a degree on the impedances of both the source and driven components (driven impedance typically is the most important).

It's possible to buy the RCA shell and add your own R1 & R2 resistors wired as described to make your own LPad attenuator with the attenuation you want.

Assuming the impedance your driving is 50kOhm. A pair of resistors that will preserve that 50KOhm impedance while introducing approximately 1.5dB of attenuation is R1 = 23.2kOhm and R2 = 124kOhm. I believe both values should be available as 1% tolerance resistors. If anyone would care to double check my numbers, I'd appreciate it. However, I think if you build one using close to those values, it should be about the right attenuation.

[Paul Joppa]

The problem with that kind of attenuator is that the impedances are important but never provided by the maker. (I think that's because the explanation it thought to be too complicated for the average customer.)

I have some 12dB Harrison Labs units with 10K input impedance, which is pretty low for Eros. One person, somewhere on the Web, has measured a Rothwell - I think it was a 10dB unit - at 100K which is good for Eros, but it's output impedance would be around 30K, which means it would likely change the attenuation of the switched controls.

Technically, you are better off rolling your own.

[phatfish]

I'm good with rolling my own.  It sounds like an LPad would be more appropriate than a single resistor?

I did find a calculator, but it seems geared towards speakers:
https://sengpielaudio.com/calculator-Lpad.htm

Using 50,000 Ohm and 1.5 db came up with pretty different numbers than hmbscott.
R1 = 7930.24
R2 = 265248.86

How would I go about calculating this for the EROS2 > Mainline?  Sounds like the mainline impedance is variable due to the attenuators I assume "25Kohms minimum" from the product page.  Is 50K more typical or was that just an example?

[2wo]

How about adding a pot in place of L-pad to the desired channel and dial it in.

[phatfish]

I figured throwing a pot in the mix would largely negate the benefits of a stepped attenuator in the Mainline. 

That said, maybe a mono stepped attenuator would be a good option.  I'm not sure how a mono (pot or attenuator) would interact with the EROS2/Mainline loading.  Perhaps as long as it's large enough of an impedance for the EROS2 it wouldn't matter since I'd be balancing out the levels (to my ears) anyways. 

[Paul Joppa]

I'm good with rolling my own.  It sounds like an LPad would be more appropriate than a single resistor?

I did find a calculator, but it seems geared towards speakers:
https://sengpielaudio.com/calculator-Lpad.htm

Using 50,000 Ohm and 1.5 db came up with pretty different numbers than hmbscott.
R1 = 7930.24
R2 = 265248.86

How would I go about calculating this for the EROS2 > Mainline?  Sounds like the mainline impedance is variable due to the attenuators I assume "25Kohms minimum" from the product page.  Is 50K more typical or was that just an example?

Yes, you have identified the critical problem - the Mainline level control's input impedance varies with different gain settings. I don't have the numbers but hopefully PB can supply them. Eros output impedance is about 4K-5K ohms. Once we have those numbers, we can optimize the L-pad resistances for minimum deviation from the target 1.5dB and decide if that's close enough.

[hmbscott]

phatfish,

The calculation for attenuation for an LPad is A = 20 x Log10(R2/(R1+R2)), where A is attenuation in dB. The R1 and R2 values (7.9KOhm and 265kOhm) from the sengpleaudio calculator do not give 1.5dB attenuation but rather only -.26 dB. I'm not sure exactly why they are not right. I tried to derive the equations he shows on the site, but my algebra just cannot come up with them. Although his values seem to give the wrong attenuation they do preserve the load's 50kOhm impedance.

That said, I also made a mistake with my initial R1, R2 values. I misused my own spreadsheet. My corrected R1 and R2 values are 17.5kOhm and 92.8kOhm. They do give -1.5dB attenuation using the above equation. They also preserve the 50kOhm impedance of the load plus attenuator, I'm calling that combined Za. The equation I used to calculate that is  Za = R1 + 1/(1/R2 + 1/Z).

[Paul Joppa]

phatfish,

The calculation for attenuation for an LPad is A = 20 x Log10(R2/(R1+R2)), where A is attenuation in dB. The R1 and R2 values (7.9KOhm and 265kOhm) from the sengpleaudio calculator do not give 1.5dB attenuation but rather only -.26 dB. I'm not sure exactly why they are not right. I tried to derive the equations he shows on the site, but my algebra just cannot come up with them. Although his values seem to give the wrong attenuation they do preserve the load's 50kOhm impedance.

That said, I also made a mistake with my initial R1, R2 values. I misused my own spreadsheet. My corrected R1 and R2 values are 17.5kOhm and 92.8kOhm. They do give -1.5dB attenuation using the above equation. They also preserve the 50kOhm impedance of the load plus attenuator, I'm calling that combined Za. The equation I used to calculate that is  Za = R1 + 1/(1/R2 + 1/Z).

Your A is only correct when Z is infinite and R1 = 0. The calculation that phatfish made would be correct if the source (Eros) output impedance were zero, which is not the case here.

For the full calculation, R1 is in series with the source (Eros) output impedance and R2 is in parallel with Z. This gives the attenuation relative to the unloaded voltage output of the source. That means the other channel is also attenuated and you want to get 1.5dB more attenuation in the target channel. It gets complicated quickly! You can see why manufacturers of attenuators don't want to talk about impedances ...  :^)

[hmbscott]

Thanks Paul. I appreciate the lesson.

[phatfish]

Good grief.  Sounds like I need to take my right ear to the front row of a concert to even things out. 

Given the variable impedance of the mainline is this even possible to get right?  Would I be better off putting a preamp with balance control between the Eros2 and mainline?  Should I just avoid stepped attenuators altogether?  I did ask about getting a Crackatwoa (one of the attractions being the integrated balance control).  Doc made it sound like it would be possible to get together a small batch, but the thread hasn't been updated in a while, so I shifted my attention to the mainline. 
 

[Paul Joppa]

I expect that we can find an L-pad that is close enough to 1.5dB with all settings. With your Crack, do you adjust the balance control frequently?

[phatfish]

Ok cool.  No, i leave the balance control alone hence asking about something fixed. 

My dac has a digital balance and that's where I'm getting the approximate value from.  I might want to craft a 1 or 2db one as well depending how the 1.5 one sounds but intend to pick one and leave it in place.

[Tom-s]

A preamp can "flavor" the sound to your liking. My suggestion would be to try one.
In my experience with Eros + SII and Eros + Mainline the MorePlay was a nice addition to the chain in both cases.
And since I put in the MorePlay I never had the urge to take it out. Even when some DIY DHT preamps did better, most notibly with my speaker setup.
Personally I use the balance control every session. But have to note that I the adjustment is pretty much set to a "standard for me" position once in use.
It only changes when switching from speakers to headphones. With other DIY preamps the DAC's balance control is used.

[hmbscott]

Just to close the loop on my calculations: Treating the load impedance Z as a resistor in parallel with R2 of an Lpad, I calculate a combined impedance of Z2 = 1/(1/R2 + 1/Z), and using Z2 in place of R2 of the attenuation equation I get this equation, A = 20 x Log10(Z2/(R1+Z2)). I can confirm that the R1 and R2 values given by the sengpleaudio calculator do give the correct 1.5dB attenuation using Z of 50kOhm with this equation.