Simple Attenuators - Design And Testing

Discussion in 'The Workbench' started by JohnH, Oct 21, 2017.

  1. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    hi @matttornado , @Gene Ballzz makes good points.

    Design M (see page 1 summary) is the most versatile because it does every 3.5db step and can run 4, 8, or 16 cabs in the 8 ohm version. But Gene's is the simplest and easiest to build and to use, if you don't need that first -3.5db stage on its own. Also, its absolutely essential to be able to follow the wiring and relate it to the schematic of whatever you build and, if the whole box starts with -7db always on, then there's nothing that can accidentally happen in use to the switches downstream that can hurt anything.

    The toggle switches needed typically have 6 lugs and are in fact two switches side by side. The centre lugs connect to the upper or the lower lugs. To do the simpler version, you only need one side of the switches but you can link the two sides together with three wire links to double it up to get more current rating and reliability if you wish. Such switches may be called on-on or on-off, but if its six lugs and two positions it would be right.

    Don't need switched jacks (I use one for the red-box section of my circuit though). Just basic mono jacks are fine for, but for the ones I bought I found the stereo version to grip the plug better.

    ps... this thread has now had over 30,000 views, with more than 150 in the last 12 hours!
     
    Last edited: Aug 22, 2019
  2. matttornado

    matttornado Well-Known Member

    Joined:
    Mar 14, 2012
    Messages:
    897
    Likes Received:
    390
    Thanks again JohnH!

    Yes this thread is fantastic! Thanks for sharing and answering everyone's questions too! I should be starting my build within the next few weeks.
    I still have a few parts to get and wait for all of my resistors to arrive.

    So you mentioned earlier that with the attenuation at its lowest setting, (the lowest volume), it can be used as a dummy load. Even if it normally would have sound coming out of it if a speaker were connected?

    The Marshall Powerbreak I had, when set to dummy load, would not put out any signal. That's why I'm asking.

    I'm going to add a line out to my attenuator and might try slaving for the fun of it.
     
  3. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    yes well, when the attenuator is set to max attenuation, less than 0.1% of the power gets to the speaker. 99.9% of what the amp sees is the attenuator. So it then makes no significant difference to the amp if the speaker wasn't there at all.
     
  4. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    I thought it might be helpful to redraw Attenuator M, stripped back to minimal fetaures and less complexity, as discussed with @Gene Ballzz above:


    This is exactly the same as the full version, but without a third output and without bypass and with the -7db setting being the minimum attenuation. It's a very simple practical design, still with the same full range down to -31.5db. I use this max setting frequently late at night.

    Also, maybe by looking at this diagram, it will be easier to see how the full featured version relates to it. Component references and all values are unchanged, except that the power rating for the -3.5db Stage 4 can be less in this 'lite' version. (also, Sw1 is deleted)

    I kept two outputs shown. I think that is useful for running extension cabs, and its also another 'safety' reminder when setting up, as to which jack is in and which two are out.

    One thing to note. I've drawn them with the order -7, -14, -3.5 in the three switched stages. There is a small electrical logic to this, and it is needed in the full version. But, that doesn't need to relate to the physical position of switches on the front panel - which is a free choice.
     
    Last edited: Aug 24, 2019
    Heph333 and Nik Henville like this.
  5. Gene Ballzz

    Gene Ballzz Well-Known Member

    Joined:
    Feb 19, 2013
    Messages:
    1,784
    Likes Received:
    1,180
    Location:
    Las Vegas, NV
    Hey @JohnH

    I've got what may be a dumb question, though I think there was brief discussion earlier in this lengthy thread. What happens if one of these attenuators gets used in parallel with the speaker load? Specific scenario, for discussion purposes: 16 ohm attenuator and a 16 ohm speaker plugged out of parallel speaker out jacks of the amp. This could be either two parallel jacks with a switch on the amp with switch set to 8 ohms or as on a DSL amp parallel out of the 8 ohm jacks or any other proper way to parallel the attenuator and the speaker, while having the amp set to the proper output tap for the type of use.

    I realize that there will likely still be some attenuation and possibly reactivity, but I'm guessing it won' be at all the same as running the attenuator in the more standard series manner! Would there need to be an added resistor at the output of the attenuator, to complete that circuit? Or.......?

    Thanks For Any Thoughts? Just Curious Here?
    Gene
     
  6. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    Hi @Gene Ballzz . I was just logging in to add something about exactly that scenario, and you went and raised it anyway!

    I think that could amount to a useful workaround if you do want to take just a few db off, and if the attenuator at hand doesn't have the -3.5db setting. Set the attenuator to max attenuation, so that it then can function as a reasonable load box with no speaker. So it will get half the power and the real parallel speaker the other half, being a -3 db attenuation. The amp will be perfectly happy with that. The tone should also be fine since this is a reactive box, It is designed to follow a real speaker impedance curve at least from low mids/high bass (say 150hz) upwards. The bass peak in the very low frequencies may be reduced by a few db.
     
    Nik Henville likes this.
  7. Gene Ballzz

    Gene Ballzz Well-Known Member

    Joined:
    Feb 19, 2013
    Messages:
    1,784
    Likes Received:
    1,180
    Location:
    Las Vegas, NV
    That, my good friend, is exactly where I was going with part of the question! :agreed: I was also considering the concept that the speaker is already, actually providing some reactance. My biggest concern was whether or not the output of the attenuator needed to be terminated in some way, or not?

    I am also making an "assumption" that the nominal impedance of the attenuator should be generally trated as another speaker? So 16 ohm attenuator + 16 speaker needs to be driven from an 8 ohm tap and an 8 ohm attenuator + an 8 ohm speaker needs a 4 ohm amplifier output tap?

    With that said, it is another example of why the 16 ohm version seems even more versatile, as many amps do not have a 4 ohm tap.

    Thank You Again Sir!
    Gene
     
  8. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    @Gene Ballzz , I dont think its necessary to plug anything extra into the attenuator when used as a load box like that, provided all stages are engaged. Without a speaker plugged in, it only makes a very small fraction if an ohm difference as seen by the amp.

    Thats all true either for the ones I draw, with a -31.5db max, and also for yours where the -14 stage is subbed for another -7.

    And I agree about the 16Ohm version, where that works with the rig. Its also good when uprated for 100W amps since it cuts current flowing through jacks and switches compared to the 8Ohm equivalent. In my case, I needed 8 Ohm build since my main amp doesn't have a 16Ohm tap unless I mod it. Also, an 8 Ohm version would work better with a pair of 8 Ohm cabs. So each user should pick their horses to suit their courses!
     
  9. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    Talking about analysis....

    ..so please bear with me here!

    I designed this attenuator specifically around my Vintage Modern, and once Id homed in on how to find the best values, it works great, as it should, assuming my maths are correct.

    But by now 8 months later, we have quite a growing collection of different builds made mostly by the good folks on this forum, and the design has been tested with multiple different amps at both 8 and 16 Ohms, including to our knowledge, several Fenders, and most of the key types of Marshall (including VM, DSL combos both old and new, Plexi's and now both new Studio amps. Different valves have been represented and different master volume arrangements NVM, MV and PPIMV, with and without negative feedback. All seem to work well. Why? Where are the limits? Its all good news but I've found this puzzling.

    On the main circuit diagram, there are curves of frequency response, stepping through attenuation stages. You can see that they are consistent. They apply to small signals and are based on an amp output impedance measured from my VM, with an 'Aiken' model to represent speaker impedance. The trick is to match frequency response at the speaker with attenuation, with that of a fully connected full volume speaker.

    But Im finding that having derived it for the measured output impedance of the VM of around 18 to 20 ohms (8 ohm tap), virtually the same consistency occurs if the amp is assumed to have either a lower or a higher output impedance, with the same component values. (This applies above about 150hz). This ability to adapt to different output impedances (and so different amps) and maintain consistency for that amp seems to be part of the key to this.

    But all of that is for small signals. When the amp is overdriven hard, its output changes enormously, tending to hit a hard limit and behave as a voltage source. This is hard to model fully. But an extreme analogy is to feed the circuit with transient square waves of low impedance. Here's my model, with some results, based on 2kz and -10.5db attenuation:


    The model is of the type Ive been using to do the design throughout. It has two Aiken speaker models, signal sources and the attenuator, with some attenuator components at very high or low values if they are to be taken out of circuit. It's made in 5Spice software, but LTSpice would also do it.

    I scaled the full volume amplitude so it overlays with the attenuated. Blue attenuated signal follows Red full volume very closely indeed. These are being fed out of a 1 Ohm amp impedance, like an amp so overdriven that its about to melt . But it matches just as well at different amp resistance too. The signal is getting through the attenuator and feeding the speaker, without changing its shape.

    This type of model, is very sensitive for finding the optimum for R9 (R34 = 27 Ohm in the above, for an 8 Ohm build). The value adopted is ideal for a 10.5db attenuation. At higher attenuation, slightly lower at 22ohm is a bit better. But that is very close, you'd never hear a difference.
     
  10. Gene Ballzz

    Gene Ballzz Well-Known Member

    Joined:
    Feb 19, 2013
    Messages:
    1,784
    Likes Received:
    1,180
    Location:
    Las Vegas, NV
    Great Stuff @JohnH

    I realize that "standard" frequency for testing is often 1K and that your thoughts for testing may be to look at how smooth that 2K range is and that is certainly a valid consideration, but that frequency doesn't use involve much power/energy. It may be interesting look at the higher powered/harder working part of the spectrum, say down around the 200hz to 400hz region? That's where the tubes and iron get a good workout! :naughty: Easy to forget that the "fundamental" of notes on a guitar ends a little above 1.1khz, although most of the stuff guitarists obsess over can be well above that! :wow:

    Just Sayin'
    Gene
     
  11. Nik Henville

    Nik Henville Well-Known Member

    Joined:
    Jul 18, 2018
    Messages:
    948
    Likes Received:
    1,442
    Location:
    Eastern Seaboard, the nook-shotten isle of Albion
    :applause:

    I
    think
    you just proved
    that a good design is a good design is a good design...

    :hippie::pirate::uk:
     
  12. matttornado

    matttornado Well-Known Member

    Joined:
    Mar 14, 2012
    Messages:
    897
    Likes Received:
    390
    So at this point, I'm going to presume that if I had to pull two tubes in an emergency and run on just two instead of four, that I could set my amp to 8 ohms instead of 16 and still use the 16 ohm attenuator. Correct?

    That was a good question too, Gene Ballzz! I ran my Powerbreak in parallel, set to dummy load, along with a 4x12. It knocked my volume down a little, nothing drastic though. I done that at few gigs when I still had my PPIMV installed.
     
  13. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    Hi Matt,
    Yes I think that follows the usual advice about impedances when tubes are pulled.

    If it was my amp, and in that situation due to some emergency, I think Id also take some pressure off the circuit by using a bit less attenuation so its not driven so hard. Nothing related to the attenuator, just being careful with the amp.
     
    matttornado likes this.
  14. matttornado

    matttornado Well-Known Member

    Joined:
    Mar 14, 2012
    Messages:
    897
    Likes Received:
    390
    Yep! I agree! I can't wait to get all of my parts together so I can build this thing! I'm sure I'll have more questions when I'm ready to test.... It'll be a little while though.
     
  15. Heph333

    Heph333 New Member

    Joined:
    Jul 4, 2019
    Messages:
    4
    Likes Received:
    0
    What I've found to work really well is a combination of tube adapter + attenuation. I'm a massive fan of Smicz adapters. They basically convert any power tube to 1w (single) or 3 watts (pair) using 6ak6 tubes. Cranked they sound like 6L6. Attenuating a pair of these will get power tube saturation, but with only a 5 watt peak load in your transformers.
    Even the better known Yellow Jackets could accomplish the same if you like the sizzle of el84. Could drop a 50-100watt amp down to 18-20. Then attenuate that. Obviously you're driving the power tubes hard, so they will wear. But it's much easier on the trannies.
     
  16. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    Analysis comparison: MiniMass and Attenuator M

    Ive been designing the attenuators mainly using a circuit analysis, as shown in previous posts above. It uses an electrical model of a speaker at full volume and then through the attenuator, to compare the shape of waveforms and also frequency responses.

    So I thought it would be interesting to do the same with another simple reactive design by way of a comparison.

    Weber make some great products, and in various threads we are seeing good reports about them, particularly the MiniMass model. This is listed as 50W rating, suitable for up to 35W amps to allow for overdriving. I don't have one myself but by all accounts it's good value, well built and well liked. So its well positioned for all the 20W Studio amps.

    Weber MiniMass - Analysis

    The Mass series work reactively by using a cone-less speaker motor as part of the attenuation network, together with a high-power wire-wound pot.

    Here is a link to a (non-official) schematic:

    https://encrypted-tbn0.gstatic.com/...T8xqgRA2Xg15Y2AsJRV6wpzpBvAIzJzDRr9flIdFMu1k6

    In the analysis, I'm representing the speaker motor with an Aiken speaker model, without the low resonance (which depends on the cone). Plots compare the frequency responses at full volume and attenuated -10db and -25db, with and without a +6db treble boost capacitor engaged (it also has a +3db treble boost):


    The signal seen by a full-volume speaker is Red, the lowest (Blue) line is as seen by a speaker at -25db, and the Green line is the signal at the amp, with -25db attenuation. The lower of the two charts shows the same with treble-boost switch engaged. Also, similar plots at -10db are shown.

    What can be seen is that, as the attenuation is increased, the usual bass peak and treble rise at the speaker, get flattened (although treble rise is partly maintained as seen by the amp). This is consistent with reports from users. Switching the treble boost does compensate at the speaker, but see how the signal seen by the amp (Green) drops right down, implying impedance to the amp is dropping at high frequency, due to a 'treble bleed' capacitor being added across the pot.

    The plots are all based on the 16 Ohm setting. The unit also has 8 and 4 Ohm options which are engaged by putting fixed resistors in parallel with the input. This will add extra damping, making the net load more resistive than reactive.

    Attenuator M - Analysis

    Here's the equivalent plot for the Attenuator M:


    Plots of speaker signals are all parallel at full volume, -10db and -25db, and signal at the amp is consistent above 200hz. This is suggesting a more consistent performance than the MiniMass. No treble compensation is needed.

    So this is suggesting a good consistent result, which is what we find in practice, and likley more consistent than the Weber.

    But of course, I may not be being fair to the Weber, since it all assumes that this analysis is actually a valid model, I don't have exact characteristic of the motor and in any case, Attenuator M has the benefit of having been designed using this model!
     
    Last edited: Aug 31, 2019
  17. Nik Henville

    Nik Henville Well-Known Member

    Joined:
    Jul 18, 2018
    Messages:
    948
    Likes Received:
    1,442
    Location:
    Eastern Seaboard, the nook-shotten isle of Albion
    More I read about 'em, more I want one.
    Can't do the build malarkey - so how about:-

    Anyone building one suitable for 100Watt RMS and 8.Ohm - build 2, yes... TWO, and charge ALL the parts to me, keeping one unit, sending one to me. Preferably Blighty based to avoid import duties and the like.

    :hippie::pirate::uk:
     
  18. Gene Ballzz

    Gene Ballzz Well-Known Member

    Joined:
    Feb 19, 2013
    Messages:
    1,784
    Likes Received:
    1,180
    Location:
    Las Vegas, NV
    @JohnH
    Great stuff! It's interesting that you chose -10db & -25db for your testing levels. Going by the positioning of the pot, I'm guessing that the range of -10db and less, along with -25db end up working out to a couple of the approximate "sweet spots" of operation that I've mentioned before. I'm going to guess that your numbers would look much "uglier" in the -15db and above -25db ranges! FWIW, with most amps there are only two or three of those "sweet spots" while a few amps seem to allow one or two more. And though I can't put my finger on exactly what the shortfalls are, in between those spots, the amps just seem lackluster and lifeless!

    I'm also gonna bet that the impedance "seen" by the amp's output varies "WILDLY" with the sweep of the pot! Your designs seem to keep a fairly even (within a few ohms) impedance from one setting to another, with your 16 ohm, "Design K" seeming to be the most consistent!

    While I don't have any sophisticated modeling/testing software, I do have a 50 watt MiniMass, various impedance speakers/dummy loads, a good Fluke meter and another OK EXTECH that I could take some readings with, at various settings? If this might be any help?

    Let Me Know?
    Gene
     
  19. Gene Ballzz

    Gene Ballzz Well-Known Member

    Joined:
    Feb 19, 2013
    Messages:
    1,784
    Likes Received:
    1,180
    Location:
    Las Vegas, NV
    Nik,

    While that sounds pretty OK, it should be noted that these builds are fairly time intensive and I'd guesstimate the labor to have at least double the value of the price of all materials/components. And then, even with calling it all a "friends/family" arrangement, shipping may be fairly steep! While you don't tell us where you are, I'm kinda assuming somewhere in the British Isles? I would say that a wide ballpark would be a cost to you of between $150 & $200 US $$$ to get one to your door. Possibly even a tad more, given the extra size/weight of a 100 watt capacity build!

    With all that said, used properly and as intended, these units are truly worth every penny of that!

    Just My $.02,
    Gene
     
    Nik Henville likes this.
  20. JohnH

    JohnH Well-Known Member

    Joined:
    May 31, 2008
    Messages:
    3,616
    Likes Received:
    1,562
    Location:
    Wilton NSW
    Hi @Gene Ballzz , thanks for the offer of some measurements in the interest of garage science!

    The real variations in response and impedance come with frequency which are hard to meter, and I expect that what we are building tracks the impedance at mid to high frequencies much better, although the MiniMass does show the amp a rise in impedance with frequency, unlike fully resistive designs. But this rise appears not to transfer through to the speaker output, except at low attenuation, hence the treble switches provided.

    But with a meter, you can at least measure the dc conditions. You could try the following, which should give a check on my MiniMass model:

    Set it to 16 ohms, and plug in a 16 ohm speaker of known measured ohms (usually about 13 Ohm). Then measure resistance at the input of the Minimass. Im expecting it will vary from about 10 Ohm at max and min, to about 16-17 Ohm max at around half turn? see what you get, and where on the pot turn is the max.

    Then repeat without the speaker plugged in. id expect it to vary from 50 Ohm at min attenuation down to 10 ohm at max attenuation.

    Thanks!
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice