Morgan Jones step attenuator in Valve Amplifiers 4th ed

[SteveFMT]

On page 556-557 Jones shows a QBasic program for calculating a Type A step attenuator resistor stack. I copied this exactly as shown into an online QBasic compiler (qbjs.org) and I cannot derive anything meaningful. My results for a 10k 'pot', 2.5k output, 8 steps, 3 dB per step:

Note that instead of each line adding 3dB each time (0,3,6,9,12...), it instead concatenates the term into 0,03,033,0333.... and of course terminates after 333dB.

Anyone played with this before, or know more about QBasic that can spot an error in the book?

Here is a link to the qbs.org compiler:
https://qbjs.org/?code=Q0xTCkE9MApC...CPUIrQwpBPUErRApOPU4rMQpMT09Qzy8gICBSLULHNC4i

 

[rayma]

This is simple enough to just do the arithmetic yourself.
10k pot, 3dB steps, eight steps of attenuation

first step down from 0dB
-3dB = 20 log( R1/10k ) -> R1 = 7.079k

second step down from 0dB
-6dB = 20 log( R2/10k ) -> R2 = 5.012k

third step down from 0dB
-9dB = 20 log( R1/10k ) -> R3 = 3.548k

fourth step down from 0dB
-12dB = 20 log( R1/10k ) -> R4 = 2.512k

fifth step down from 0dB
-15dB = 20 log( R1/10k ) -> R5 = 1.778k

sixth step down from 0dB
-18dB = 20 log( R1/10k ) -> R6 = 1.259k

seventh step down from 0dB
-21dB = 20 log( R1/10k ) -> R7 = 0.8913k

eighth step down from 0dB
-24dB = 20 log( R1/10k ) -> R8 = 0.6310k

so bottom resistor = R8 = 631R
second from bottom = R7 - R8 = 891 - 631 = 260
third from bottom = R6 - R7 = 1259 - 891 = 368
fourth from bottom = R5 - R6 = 1778 - 1259 = 519
fifth from bottom = R4 - R5 = 2512 - 1778 = 734
fourth from bottom = R3 - R4 = 3548 - 2512 = 1036
third from bottom = R2 - R3 = 5012 - 3548 = 1464
second from bottom = R1 - R2 = 7079 - 5012 = 2067
top resistor = 10k - R1 = 10k - 7079 = 2921

These nine resistors in series add up to 10k.
A simple spreadsheet could also be used.

 

[SteveFMT]

Thanks rayma....yes, no doubt easy enough. I was just wondering why Jone's Basic example bombs out on me.

 

[ejp]

I transliterated it into Java and it worked for me. You must have a typo.

 

[SteveFMT]

No typo evident. It works fine if I just enter the step attenuation as a fixed value instead of a variable that gets incremented in each loop pass.
Thanks all.....onward!

 

[jwags81818]

Try an autoformer volume control. You'll never go back. Its like taking 2 bed sheets off of your speakers. Single biggest improvement I have made to my system.

 

[exlabs]

Try an autoformer volume control. You'll never go back. Its like taking 2 bed sheets off of your speakers. Single biggest improvement I have made to my system.

As this is a bit of a technical thread, I’m interested to know the differences between autotransformer approach and resistive voltage divider. Assuming you are in the pass band of the autotransformer what’s the major difference? Maybe input/output impedance?

 

[jwags81818]

Read this https://intactaudio.com/atten_FAQ.html
I like the fact that the unused voltage becomes available current. You are not wasting signal just repurposing it. The transparency improvement is significant. You WILL notice an improvement. The guy that runs that site will answer your questions and is very knowledgeable. His name is Dave

 

[Mark Tillotson]

resistive divider adds noise but is linear (using metal film), transformer will add significantly less noise but isn't very linear, and will have a limited bandwidth too.

 

[jwags81818]

Yes but how limited? And who cares if the volume steps are not linear? Turn it up or down another click and compensate. The sound is SO MUCH BETTER.
It really is like lifting 2 bed sheets off your speakers.

 

[exlabs]

Oh that’s interesting, I never really thought of transformers as low noise devices. I’m always worried about loop area and orientation when building with transformers. But probably because I’m a little scared of magnetic black magic.

Not too sure I buy the energy efficiency argument though. Why should the speakers care where energy they never sees goes?

 

[Mark Tillotson]

And who cares if the volume steps are not linear?

No, that's not what I mean, the signal gets distorted at any volume except 100% as inductors are far from perfect.

 

[jwags81818]

Oh that’s interesting, I never really thought of transformers as low noise devices. I’m always worried about loop area and orientation when building with transformers. But probably because I’m a little scared of magnetic black magic.

Not too sure I buy the energy efficiency argument though. Why should the speakers care where energy they never sees goes?

Not saying that it matters or makes an appreciable difference. I just like the idea of all the signal energy going to use as opposed to becoming a space heater. I would bet that it increases detail or maybe micro dynamics at lower volumes. Again not saying it does but it makes sense that if the signal has a larger buffer available that these things might be helped. Regardless of any of this. Listen to one. It is not subtle. Its one of those "I'll be damned !" moments. Pulling sheets off the speakers is the best analogy that I have.

 

[jwags81818]

In regards to it causing distortion, Resistors are noisy.. But regardless no one could say it doesn't sound better. It is not subtle. If its distortion I am hearing then so be it. I bet its a whole lot less than the tube amp its attached to is creating. Have you ever heard one? On a system where you could AB it? I don't think you have or we wouldn't be having this conversation. Never said it was perfect just that it sounds a whole lot better.

 

[Mark Tillotson]

I just like the idea of all the signal energy going to use as opposed to becoming a space heater.

You realize how ironic this sounds in the context of a valve amplifier where each device burns up watts of power in its heater... A typical audio signal of 2Vrms into 10k load is only 0.4mW "wasted"...

 

[jwags81818]

You realize how ironic this sounds in the context of a valve amplifier where each device burns up watts of power in its heater... A typical audio signal of 2Vrms into 10k load is only 0.4mW "wasted"...

Not at all.. Its not the SIGNAL being wasted. Its just B+ and heater power. Big difference if the signal is the information we are interested in amplifying.

Jeff

 

[davidjt]

It's strange how a new insight can look blindingly obvious afterwards, but this thread has made me re-consider my amplification set-up. Previously my 'ideal' set-up was going to be either a valve pre-amp + solid state power or vice-versa. I have several home-built preamps and incorporating an autoformer into the final choice seemed logical.
But thinking 'attenuator' as opposed to 'gain control' you hit the obvious anomaly: the need for an attenuator pre-supposes too much gain. Why are you wasting energy amplifying (and distorting, however slightly) a signal, only to throttle it (and distort it some more?) before it reaches the next stage?
Since I use JRemote to control my digital source I already have remote volume control. So this morning I removed the pre-amp altogether. And I think the sound is cleaner. (Volume matching without a meter is impossible, but that's my impression.) It's early days, the 'honeymoon' effect of any change is a potential trap, but removing a lot of circuitry (and a 'volume control') from the signal path makes perfect sense. One potential rabbit hole for further investigation is the JRiver volume control: what, if anything, is that doing to the signal? Would bypassing it and using an autoformer-based preamp be another step up?

(I sometimes draft a post simply to clarify my own thinking, and then delete it. I'll put this one out there in case anyone has any thoughts on the last point.)

 

[jwags81818]

Since you are down the rabbit hole this far....You should look into whether or not your digital volume controls are bit perfect. I have a digital volume on my PS Audio directstream DAC that is bit perfect. No matter where the volume is set the data is unchanged. Only the volume. But many are not like that. Talk to Dave at Intact audio and give an autoformer a test run. It is not a subtle difference. I am only pushing the idea because it was in the top 3 things that improved my system when changed. The Tent labs filament supplies, the autoformer volume controls and having a properly aligned stylus on a boron cantilever rebuilt onto my cartridge. My Dynavector DV-20X had a stylus that was tipped so far back I would have had to raise the *** end of my tonearm 2 inches to set the vta correctly. FYI it was rebuilt by Steven Leung at VAS in New Jersey. for $500ish. Those 3 things made very big improvements in my system. The autoformers was like lifting 2 bed sheets off my speakers. It wasn't subtle..

 

[Mark Tillotson]

That sounds like an analog volume control on the DAC reference? To attenuate a signal digitally the bits have to change. Or am I missing something?

 

[davidjt]

The only reference I have found so far states that bit perfect volume control is a contradiction in terms......but also suggests that in the case of JRiver any effect would be inaudible. (?) Maybe I'll get my head round dsp one day....... I can see a passive volume control/input selector being next on the list after I finish all the other projects here. I don't play vinyl any more and the valve amp doesn't use DHC valves so there are two fewer issues to deal with. 🙂

Meanwhile, Radio3 via the laptop is also sounding clearer without a pre-amp. Less is more. It's a shame I have so many pre-s to leave out.