Simple Attenuators - Design And Testing

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
@JohnH
So, I'm doing a bit of sketching, but I'm having a little late night brain lock... You could just endlessly string instances of R7 and R8 for successive 3.5db drops, right? If this still makes sense when I look at it in the morning, it may actually be simpler in practice.

That would work jn theory, you can stack any combination of the stages 2 to 4 end to end to make a combination. But I don't see it as the way to go for working with rotaries because each stage would need to have switch, which if translated to a rotary controlling all the stages, would need to be a separate pole, one per stage.

My current sketching around this is to have Stage 1 as designed before (design M or M2), then a single stage after that where each of the two resistors is a chain, selected by a switch, so two poles, and as many steps as can be found based on switch positions. If an 8 position 2 pole switch can be found, then that would do it. If it's a 6 position switch, then Id augment it with another high/low toggle switch to extend the range while keeping rotary increments small.

I have a heap of maths built into spreadsheets to deal with this stuff, but the trickiest step currently is to find a recipe that works with the limited range of resistance values that are available from normal suppliers. Definitely solvable, just a bit of a PITA!
 

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
So FWIW folks, I use one of these units almost constantly, with various amps and must say that it is as close to perfect as I think we can get, for a generally "passive" unit! With that said, a tiny bit of highs (the "sparkle" if you will) seems lost at attenuation levels of more than -21db, but I belive that to be more of a physical/air movement phenomenom than an electronic one. I'm betting that any attempt to correct for it "electronically" within the attenuator, will end up being a "hit or miss" situation, depending on the amp used as well as the attenuation level. This design is truly one of the:

Greatest Things Since Sliced Bread! :fever:
Gene

Thanks Gene, seriously many thanks for your perceptions, encouragement and trust. You were the first to believe in this design!

Sparkle:
Many commercial units have some form of tone switch for treble. Usually this has to compensate for an inherent dullness that creeps in at more than a small amount of attenuation. I don't think this one has that fundamental issue. But there's lots of reasons why a treble switch might be helpful, just to give an option for a player who wants it. It might help to compensate for a particular amp, or adjust for FM effects.

Its not hard to do, particularly if the intent is to make that adjustment only at large attenuation more than -21db. In this range, the -14db stage will always be on, and below this it will always be off. So a bit of capacitor treble bleed around this stage would do it, and being downstream of at least Stage 1, the amp wont be messed up by it. Shall we work out a design? It could have a 3-position switch for 'off' and two options for treble lift.
 

Gene Ballzz

Well-Known Member
Joined
Feb 19, 2013
Messages
3,538
Reaction score
3,992
Location
Las Vegas, NV
@JohnH
As mentioned, on my 16 ohm, K4 "Gene" unit it does not show up at -21db, but does at -24.5db and more. It is extremely rare for me to use level this quiet and I certainly never gig at such, so not a big issue in my case! Understand that everywhere I go, I tout the beauties and usefulness of this design and provide links to many non-ttmembers, so we might be getting some new members, simply for being able to ask questions.

I was going to suggest that it may be helpful/beneficial to start a new thread not intended for discussion and/or comments but instead to simply list the different versions, along with possibly the diagramatic [sic] evolution of each. Then if someone wants the full history and testing details they can read this thread that is becoming rather monstrous!

Just The Thoughts Of A Crazy Man,
Gene
 

Marvelicious

New Member
Joined
Dec 1, 2019
Messages
13
Reaction score
1
@JohnH

Document 21_1.jpg

Excuse the chicken scratch, but this is based on the M design. This shows a 5 position rotary... The common of one deck wired to the output and the other to ground. The dots represent switch contacts... position one is bypass, two is the 7db reactive stage and each jump after adds 3.5db of drop. Add more stages as needed with more positions on the rotary.

Am I missing anything?

Edit: As it turns out, I am missing something... I'll leave this schematic up for the moment, since I think I'm on the right track, but the grounded side of the circuit doesn't work as is...
 
Last edited:

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
hi @Marvelicious

Thanks for the diagram, but I'm sorry that it has some hairs on it!

Its OK up to Stage 1, but when additional stages are engaged, the ground connections to each of the previous stages which are still active, need to be maintained. That cant be done with just one pole to deliver the ground connection.

Also, the design works with the series resistors (eg R8) on the speaker side of the grounded resistors (eg R7). Its the key trick by which the right impedances are maintained. As drawn its shown the other way around, similar to most l-pads, which suffer tone problems in amp attenuators.
 

Marvelicious

New Member
Joined
Dec 1, 2019
Messages
13
Reaction score
1
Yep, I noticed it had some issues right after I posted it... still, I think I'm close to it...

Edit: I'm into a rut on it at the moment. Pretty sure it can be done with a 3rd pole, but I'm not making any headway with two right now. Time to walk away for a bit.
 
Last edited:

matttornado

Well-Known Member
Joined
Mar 14, 2012
Messages
1,675
Reaction score
1,211
I kind of got the fan to work but with some weird problems.

1st: the fan circuit added a lot of extra attenuation when it worked. That's no good!
2nd: it would only work if grounded to the metal enclosure & I think the circuit's ground together?
3rd: when connected to just the circuit, not the enclosure, there was no sound. I guess it created a short?


These were quick tests so I need to go back and do some more experimenting. maybe the fan has to be connected to a different ground (enclosure) not sure what's up yet. I was kind of in a rush.
 

matttornado

Well-Known Member
Joined
Mar 14, 2012
Messages
1,675
Reaction score
1,211
I got the idea from this Marshall Powerbreak schematic. I'm not sure if this is entirely correct or not.
I connected the fan between attenuator R4 & R6 and got lucky with the correct DC voltage between 5 & 6 volts DC. I was using a 5 volt DC fan. The fan seemed to work regardless of what attenuation setting I was using. I didn't put a diode in yet, just a cap.
I'll post a few pics of my set-up as soon as I can.

Thanks, JohnH!

marshall_powerbrake.pdf_1.png
 

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
did you have the bridge rectifier BR1 and the 56 Ohm resistor? Need to see it all drawn on our diagram.
 

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
Here's how I reckon the fan would be wired:

fan191220.gif

the value of R would be depend on what works. the powerbrake shows a 56 Ohm there
 

matttornado

Well-Known Member
Joined
Mar 14, 2012
Messages
1,675
Reaction score
1,211
Yep that's what I did but instead of adding the resistor, & connecting right at the input of the attenuator, I tapped power between R4 & R6 at the attenuator circuit. I wonder if that has anything to do with why it added more attenuation?

I can't wait to play around with it some more. Hopefully this weekend.
 

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
Could be! It's better as I drew it, and then there's more juice avaikable and you can use a larger R to control it, so less attenuation added.
 

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
Rotary Switches

I've been thinking about how to do this design using a rotary, with the best balance of fine control, range and minimum switch complexity. If I built one, and if a suitable switch can be found, I reckon it might be like this:


It has a fixed Stage 1, based on design M2. Then a high/low range toggle in Stage 2, to change from say rehearsal or gig level, to home or studio level. I've made this a -10db stage. Finally, the rotary stage, to go from 0 to -15db in -3db steps. So the whole range is -7 db to -32db. The rotary for Stage 3 is 2 pole, 6 position with shorting contacts.

With an 8 ohm build and a 50W amp, the Stage 2 switch needs to have contact ratings of at least 2.5A at 125V ac (allowing a factor of a bit more than x2). The rotary can be rated at 2A. See how in position 1, where it is at zero, it's two poles are in parallel, sharing the current, which takes a bit of pressure off its loudest setting.

The Stage 3 has two chains of resistors to gradually increase attenuation. For each amp Ohms, I've given the best values I can figure out using either the standard range of resistors (like 10,12, 15, 18, 22 etc), or the slightly different range that I've being buying from an online seller in China:

https://www.ebay.com.au/itm/25-50-1...cc9d:m:mDimLUo3wNHFIEXvjElODhA&frcectupt=true

That's an Aus eBay link, but I'm sure you can find it.

All the above is untested so far, but follows tested principles for its design.
 
Last edited:

Skorpio

New Member
Joined
Dec 22, 2019
Messages
10
Reaction score
2
Hi, new to the forum. I will introduce myself in another thread but right now I have quick question because I had some thoughts about attenuation myself for the past few days and found this.

I run DSL15h as a pedal platform into 1x12 8 Ohm G12M-25 Greenback and since nowadays i play mostly at home, I wired simple resistors into the cab some time ago.

John, could you explain main DIFFERENCE that your design make, so that it's superior to classic wirings that i use ? Yes, I read thread from the first page but I couldn't find simple answer

I have 8Ohm parallel to the +/- on the 8Ohm speaker (so i have 4Ohm) and 12 Ohm in series to back in up to 16Ohm, then 16Ohm parallel to have ~8,3 Ohm measured at the cable. I want it modify it, not some switches/bypass, simple hardwired resistors bacause i don't gig/don't jam with it anymore... and i can change it later if I do ;)

I read that you:
1. Wire resistors, so that resistor that is wired in series is closer to the speaker, then parallel compensates for increased resistance and is closer to the amp and grounded
2. Ground connection - this means that the speaker and parallel resistors additionaly have to be connected to the ground/common ground to "see" proper impedance that thus work correctly without tone suck ?
I undersantd schematics and I calculate and additionaly check all resistances by myself if i need to, if you could explain where your design differs from l-pad/what i do.
Thanks!
 
Last edited:

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
hi @Skorpio , thanks for your interest in these designs.

What you describe, with two parallel resistors (one across speaker, one across amp), plus a series one, is a decent design , called a 'pi' attenuator. Your numbers check out and its safe. It will take -12db off the volume.

Speakers have an impedance that varies with frequency. It rises at high frequency due to coil inductance, plus there is a resonance in the low bass. Tube amps react to this. Plus, tube amps have quite a high output impedance themselves, so the changing speaker impedance is reflected in the voltages across the speaker, and hence the frequency response that it creates. Solid state anps, and the dullest form of L-pad attenuators when at high attenuation, damp this natural resonpse down to a flat line. We get a muffled sound, and less dynamics in the tone. But yours is actually not bad. I fed it into one of my spreadsheets and produced this:


Pi-12db.gif

Red represents the signal as a frequency response in db, across a full speaker (a G12M), fully connected to a typical tube amp (mine), with no attenuation.

Blue represents the same speaker, with your attenuator, with the overall -12db taken out so the curves can be compared. You can see the same rise and peaks, but lower relatively. Id expect about 2 or 3 db less in the high treble and low bass. Actually, not a huge difference. Does it sound good? A response for a simple 2 part l-pad would be flatter.

The green line is what the amp is seeing - almost a pure flat resistive load. So when you dig in, it cant react in the same way as to a natural speaker.

My design adds a natural reactance for the amp to see, which in the M and M2 design comes into play above about 150hz. Also, the different resistor arrangement maintains the high output impedance to the speaker, so its response in maintained. here is an equivalent plot:

reactive191224.gif

Now the response at the attenuated speaker (blue) is virtually identical to that of an unattenuated speaker (red), with the attenuation removed form the plot, and the amp is seeing a load that is much closer to that of a pure speaker, and reacting accordingly (green)

Does that help?
 

Attachments

  • Pi-12db.gif
    Pi-12db.gif
    22.2 KB · Views: 3
Last edited:

Skorpio

New Member
Joined
Dec 22, 2019
Messages
10
Reaction score
2
Wow, I didn't expect a frequency response for my 'design' :D Subjectively, some highs may be taken out, but since I can increase master and my amp, and speaker (broken in) are bright (and honestly, presence control is for something, right?) it actually does sound better than I expected (not muffled and not fizzy)

And yes, this explains a lot, especially with graphs - i wonder if more flat green line with my (and any similar) design actually put more strain on the power amp, I heard some bad things about resistive load

However, mainly I was refering to this quote:
Also, the design works with the series resistors (eg R8) on the speaker side of the grounded resistors (eg R7). Its the key trick by which the right impedances are maintained. As drawn its shown the other way around, similar to most l-pads, which suffer tone problems in amp attenuators.

So, I suppose that wiring 8 Ohm in series to the '+' of a speaker for total 16, then parallel 16 Ohm to reduce to 8 and doing that again is gonna improve frequency response by itself and it's going to be basically yours 2-stage hardwired design ? Something along page 3 design with no switches. I was also using 2 adjustable resistors per l-pad calculator site double checking values, with different wiring however... (parallel on the speaker side) and values have to be way more precise so i'm more with this.
 

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
Yes that proposal should work well. Its also -12db, but it would put your blue line right up to where the red line is.

But adding the inductor or inductors does make it better again.

Ive also read reports about damage from resistive attenuators, but they don't make sense and Ive read better experts debunk them.
 

Skorpio

New Member
Joined
Dec 22, 2019
Messages
10
Reaction score
2
Due to resistors I have at my disposal right now, I ended up wiring (modifying) something like this without adding inductor
circuit-gigapixel.png

I guess it's about -18db, tested it and i'm quite happy with it. I cannot crank the amp at 4 or 5 without neighbours going crazy so I can't really A/B how much response I lose with no attenuation. With my G12M it's doesn't sound too much though, many of todays reissue speakers (especially H or M type Greenbacks) are brighter anyway, so small teble loss is something I can live with for now - it's main reason right now I just sligthly modified original design with more power reduction. There's still plenty of bass but that could amp as well (I have deep switch which works like resonance at 100%)
Thanks for you input, I'll incorporate inductor and change wiring later on.
 
Last edited:

JohnH

Well-Known Member
Joined
May 31, 2008
Messages
4,992
Reaction score
3,192
Location
Wilton NSW
Due to resistors I have at my disposal right now, I ended up wiring (modifying) something like this without adding inductor
View attachment 65480

I guess it's about -18db, tested it and i'm quite happy with it. I cannot crank the amp at 4 or 5 without neighbours going crazy so I can't really A/B how much response I lose with no attenuation. With my G12M it's doesn't sound too much though, many of todays reissue speakers (especially H or M type Greenbacks) are brighter anyway, so small teble loss is something I can live with for now - it's main reason right now I just sligthly modified original design with more power reduction. There's still plenty of bass but that could amp as well (I have deep switch which works like resonance at 100%)
Thanks for you input, I'll incorporate inductor and change wiring later on.

A good option with your parts, a bit better in balance with regard to ohms, would be based on your last diagram above, left to right, 16, 8, 16, 8 then omit that last resistor at right. That should give-12db and the amp sees 8 and the speaker sees 16 which is close to optimum.

EDIT: further on that, if you do in a similar arrangement, left to right, 16, 8, 16, 12, you get -13.5 db and the amp sees 8.2 ohm, with balanced tone.
 
Last edited:


Top