Anyone know the Siemens Sitral W295b EQ inductor values?

Hi there!

With regards to the Siemens Sitral W295b schematic from http://audio.kubarth.com/rundfunk/index.cgi, does anyone have measurements of the Inductors for Dr1 and Dr2?
Maybe from a real unit? Including inductance and DCR per tap?

Dr1 is a single fixed value inductor (which seems, from my circuit simulations, to control Q width) and Dr2 is what appears to be a multi-tapped inductor with 7 taps (one goes to ground and then there are 6 that are switched to choose the parametric presence band boost/cut) that helps, along with a resistor and capacitor, to set the presence centre frequencies.

If no-one has measured an actual unit, then can anyone tell me what configuration the presence band is (i.e. the technical name for the presence configuration so I have the right search words), since I'd love to look it up on Google and find a way to calculate the centre frequency manually and just reverse engineer it from the frequency markings on the schematic.

Just guessing with a values in the simulator via trial and error to get things to line up hasn't been going that well so far....

Thanks in advance!

the variable resistor adjusts the bandwith of each step, not center frequency.
The multi-tapped inductor can be calculated via series resonant LCR circuit from the frequency and capacitor value.

-max

OK cool, I should have recognized it's actually a passive LCR in a feedback loop to handle cut and boost.

So, some math leads me to the following conclusions:

L = 1.0 / (4.0 * PI * PI * freq * freq * C)

Where C = capacitance in Farads and freq is the frequency you are calculating the L for.

The resulting L is in Henries.

The resistance doesn't partake until you talk about the Q factor, which I'm not interested in just yet.

I'll get calculating and write up what I find.

thanks for the prompt it's just what I needed.

Here is some python code to calculate the required data:

# EQ helper functions

import math

def kHz( x ):
"""
convert kilohertz to hertz
"""
return 1000.0 * x

def micro( x ):
"""
convert microfarads to farads
"""
return 1e-6 * x

def calculateL( freq, C ):
"""
Given frequency and capacitance, calculate the corresponding inductor value that satisfies the constraints
"""
return 1.0 / ( 4.0 * math.pi * math.pi * freq * freq * C )

Here is a session showing the calculated values for the W295b:

>>> import eqCalc
>>> from eqCalc import *
>>> calculateL( kHz(5.6), micro(0.027) )
0.02991578787625714
>>> calculateL( kHz(3.5), micro(0.039) )
0.05301998097453574
>>> calculateL( kHz(2.3), micro(0.027+0.033) )
0.07980559518142548
>>> calculateL( kHz(1.5), micro(0.1) )
0.11257909293593088
>>> calculateL( kHz(1.0), micro(0.15) )
0.1688686394038963
>>> calculateL( kHz(0.7), micro(0.22) )
0.2349749156826015

The resulting values of inductance are in Henries.
Hope that helps someone.
I'll test these values later today.

Also it seems I was very wrong with relation to the purpose of inductor Dr1.

A quick Google search http://en.wikipedia.org/wiki/RLC_circuit would indicate that R101, C35 and Dr1 form a low pass filter.

The capacitor before it, C34, is therefore likely for AC coupling.

So, for the general purpose calculations of the presence filter, I believe that R101, C35 and Dr1 can be ignored, which does simplify my calculations somewhat.

Again, if I'm wrong, let me know ;-) - I'm still learning about filter topologies, especially where inductors are involved.

The bandaxall LPF and HPF at the front end is much easier to analyse.

Also, the Q is determined by the resistance of the inductor, and not the value of the resistors 77-84 - these purely control the gain and cut amounts.

DR1 just, IMHO, cuts supersonics.

It's feeding 620 ohms plus some other stuff.

1 Henry at 20KHz is 125K impedance, far too large, ~~200 times too large. Try 0.005H (5mH): 6.28*0.005H*20KHz is 628 ohms. -3dB essentially 20KHz. You might go 2mH for -1dB at 20KHz.

> Just guessing with a values in the simulator via trial and error

Don't use a jackhammer to touch-type.

The best tool here is a Reactance Chart. http://www.dmitrynizh.com/reactance-chart.gif

You know the rt to gert section of the coil resonates at 5.6KHz with 0.027uFd. Find 5KHz across the bottom and go right halfway to 6KHz. Find the \ slant line for 0.01uFd and go down/left a couple lines for 0.027uFd. Jab a pen there.

Follow a / slant line up/right to some H numbers. It's about 0.03H (30mH).

{EDIT} I see you compute 0.0299157.... which is the same real-world part.

Check: find 0.03 * 0.027^-6, take the square-root, invert, divide by 6.28. Looks like 5,595Hz on my abacus.

Follow a horizontal line over to "1,000 ohms", that's the nominal impedance. For a Q of 10, you want either 100 or 10K ohms, depending which way it's connected. In this case, it is a series-tank to ground. L-C impedance goes low at resonance. We don't want it to go to zero (infinite Q). In fact the peak boost/cut is about 2.5. The minimum resistance is something like 1K/2.5 or 400 ohms. The actual Q changes with +/-dB, right? Broad for 2dB, narrow for 8dB? R77-R84 set both the Q and the height. If you had a "perfect" coil of zero resistance, R77 or R84 would be set to 400 ohms. If you had a real coil of 400 ohms, R77 R84 would need to be zero. Anything between can b trimmed-in by looking for 8dB boost/cut.

Again: for 1KHz we use 0.15uFd and the rt-sw wires. From the chart: 0.15H and 1K ohms.

(BTW-- you might note that from 5.6KHz to 1KHz, both the cap and the coil change by factor of about 5 or 6, the resistance was apparently designed consistant.... well, of course, because you don't wanna change R77-R84 for every frequency.)

Bugger, since I keep learning throughout this post, I keep stating things as if they are fact, only to then discover they are bull moments later ;-)
Such is the joy of learning.

OK, so, I did notice the Q changing with cut/boost, totally.

I also noticed the effect of the DCR of the inductor in the Q value also, whose resistance component of the LCR circuit is related to the DCR of the inductor + the resistor along any given choice of presence frequency through the switch.

However, I'm assuming that due to the "typically low" (I seem to mostly see "hundreds of ohms" stated as DCR for most audio inductors, and now I think I get why they try and keep DCR low) resistances of the inductors, that their DCR will contribute very little change in the Q values right?

cheers, this is a great point for me, since it's the first time I'm really trying to not rely on the simulation and randomly plugging in values, but actually trying to understand the REAL math behind it all.

Thanks for your illuminating post. I'll be reading back through that a couple of times no doubt.

PRR said:

DR1 just, IMHO, cuts supersonics.

It's feeding 620 ohms plus some other stuff.

Click to expand...

It seems (and maybe the lack of parasitics is making the simulator lie) that without that supersonic rolloff, there is a definite kink high up in the frequency range. It seems that the LCR LPF network removes this kink. But again, that's speculation as the sim will clearly be "too good" for it's own good.

PRR said:

The best tool here is a Reactance Chart. http://www.dmitrynizh.com/reactance-chart.gif

Click to expand...

Great tool indeed!

This [reactance chart b/w] and this [reactance chart colored] are nice to print since vectorial.

1954U1 said:

PRR said:

The best tool here is a Reactance Chart. http://www.dmitrynizh.com/reactance-chart.gif

Click to expand...

This [reactance chart b/w] and this [reactance chart colored] are nice to print since vectorial.

Click to expand...

Wow, that coloured chart is very useful, printing that one!

> their DCR will contribute very little change in the Q values right?

I'm pretty sure that's what the trimmers R77-R84 are for.

An iron-core coil can be designed for specific inductance or specific resistance-- if you try to tightly specify _both_ you end up needing 35.67 gauge wire. Also the variance of iron may need a few more/less turns to get a specific inductance on a given handful of iron.

Since L is fussy but added R is cheap, you spec the L and a _maximum_ R, then trim-up the R as needed. For everyday work, you measure a few coils from each box then throw in (say) 220 250 or 270 ohm fixed resistor to hit your Q target. Apparently Siemens charged so much for this unit that they could test and trim trim-pots for every tap.

1954U1, thanks for the links to better charts.