Announcement

Collapse
No announcement yet.

Ultimate Attenuator

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Ultimate Attenuator

    I thought UA was just line-out signal (made with a voltage divider) from the load resistors into a standard poweramp, but it isn't, as I've found the hard way (i made a device on that principle and it severely influenced the dynamics and rolled off highs). After some poking around the net, I came across some info given by the distributor (in a forum discussion) - the device is a voltage follower (an amplifier topology which has no voltage gain, which is why UA's maximum volume equals the volume on the input).

    So, yep, we have some info: it's a voltage follower (I assume with BJTs), and a number of gutshots in this album i made
    ImageShack Album - 18 images

    I've tried building a work-a-like with a load resistor (30ohm), line out (facilitated by a pot wired as a voltage divider) into a standard poweramp, but it destroys the highs and dynamics, so that's a no-go.

    I don't currently have enough electronics knowledge to design a voltage (emitter/source) follower circuit from ground up, and I believe there are people here that can. UA doesn't seem difficult to implement, there aren't a lot of components.

    Of course, if anyone has one, gutshots with wiring visible, the underside of the pcb and transistor types would be greatly appreciated.

    My (failed) attempt at making one:
    ImageShack Album - 4 images

  • #2
    I've tried building a work-a-like with a load resistor (30ohm), line out (facilitated by a pot wired as a voltage divider) into a standard poweramp, but it destroys the highs and dynamics, so that's a no-go.
    Well, the plain voltage follower circuit is really no different in that regard. First and foremost, using a plain resistive attenuator will load the amp differently than a reactive speaker load, causing the amp to have a different type of, much flatter, frequency response. The voltage follower then just buffers the attenuated signal so that enough current can be pushed to low impedance loads. But the voltage follower - in its basic form will not colour the signal in any way worth of mentioning.

    If you didn't like a resistive attenuator coupled to a basic power amp, a good chance is that you won't like it coupled to plain voltage follower either because aside from a bit of voltage amplification in the first they really have no difference.

    On that note, the UA's attenuator isn't a plain resistive one, based on the photos there seems to be an inductor in the circuit so the attenuator is likely more or less a simulating the complex reactive load of a loudspeaker.

    The UA's voltage follower design in its basic form is pretty simple: It's just a push-pull emitter follower, biased by two diodes that are thermally coupled to the output transistors. It's pretty much the crudest example of such circuit. Such circuit topology also tolerates quite a lot of variation in component values. You can likely find some optimum of them but all in all, to get it working you don't really need to be a rocket scientist. Just putting it together with some proper layout will get it working.



    Basic design is pretty much just two power transistors in emitter follower push-pull configuration. Emitter resistors R1 and R2 provide a bit of feedback to keep the transistors thermally stable. D1 and D2 provide the 0.6V voltage difference for emitter and base when the amp idles or approaches the crossover region, additionally they adjust this "bias" in interaction with temperature; hence you see them thermally coupled to the output transistors, basically by placing them to very close vicinity. R3 and R4 draw base current from the power supply, and again a wide tolerance for these components works. C1 and C2 are capacitive AC coupling from the attenuator network. Rail-to-rail voltage can be anything from about 40VDC to 120VDC, sometimes even more or less and basically depending on the pverall magnitude of attenuated signals and the power dissipation rating of the output transistors. It's a very basic design that you could improve and make more complex and better performer, but that is not something the UA would seem to do so let's just leave it at that.

    I have naturally excluded the attenuator network from the schematic. It doesn't really matter in the overall picture: The emitter follower circuit is just a voltage follower for whatever input signal it gets.

    I haven't depicted the rectifier/filtering circuit either but that's pretty basic too, as can fior example be seen in the following picture: Bridge rectifier configured for a dual rail supply, two filter caps. Center node is grounded.

    Based on the photos, there seems to be a lot of variation in the UA versions, though. The one in the attached picture looks to be pretty much identical to my schematic. Others seem to include somekind of protection or at least a start up delaying circuit that drives a relay in the speaker line. Naturally the intention of such circuit is to prevent the startup "pop" you'd be hearing, or to disconnect the speaker line if there's DC in the output or if the current through the output transistors exceeds some threashold at which damage would begin to occur. Featuring some control for the attached fan is also a very possible scenario. Examples of such circuits can be found with Google.



    In addition, if I interprete the photos correctly, some UA versions seem to feature a series resistor in the output coupled to the speaker; this resistor (typically having a value ranging from 1 to 4 ohms) would increase the voltage follower's output impedance, resulting into weaker damping and skewing the response like tube amplifiers with similar high output impedance tend to do. It would attenuate the overall signal a bit but additionally also enhance the response at speaker's resonant frequency and at higher frequencies. It might be the difference you are hearing. Not all UA design would portray that characteristic, though.

    Some designs seem to also feature tone controls and one seems to hold something that could be a complete power amp board. Anyway, I'm not too keen on starting to guess and evaluate all what's inside them. Goes beyound the point. The basic version of the UA is indeed something very basic and pretty much follows the simple schema included in this post. But as stated earlier, it won't provide any different tone than just coupling the attenuator network to any linear power amp would.
    Last edited by teemuk; 06-14-2011, 11:46 AM.

    Comment


    • #3
      that was the most helpful reply i ever got! you explained it completely!

      now, it is i who has some explaining to do. the relay in some of the UAs is for 'channel switching' (switching between two volume pots) the one that looked like it had an eq to you actually has an effects loop (which i can't imagine being done with an emitter follower).

      as for the inductor, i believe it's wired parallel to the resistors and acts as a high-pass filter. this has been indicated in many posts on the internet. there is an inductor in every UA, but you have to pay for the switch, which puts it out of circuit and is called 'plexi switch'. that's supposed to tame the highs in bright amps, but it shouldn't be a problem for me to leave the inductor out completely, as i have an ac30. alternatively, i can have my favorite local transformer winder wind one inductor, or i can even have like 11 of the smaller ones wired in series and switched with a 12-position rotary switch, with a switch lug between each inductor. but that's only if it proves to be necessary.

      that said, my ac30 outputs around 26V (ac on my cheap dmm, and that's supposed to be rms, probably; it's not really reliable) when full on through the resistor network.

      i drew the schematic you were suggesting in proteus (a program for schematic drawing, simulation and layout drawing), but i still have some questions.

      do you have any transistors to recommend (to-247 an to-220 preferred)? are there any other design considerations when using integrated darlington pairs, such as tip140-tip145 (the simulation shows they work well)?

      C1 and C2 are electros? what is their polarity? they seem awfully large. are their values correct? does that influence the sound? should i go for bipolars?

      what about the volume control? is a pot wired as shown alright? what's a good value? 25K seems alright, there is only 0.1A across it.

      what are the power ratings on the components?

      can the current gain be regulated? getting the power gain to 1 on max volume would be perfect. also, how to minimize the rail-to-rail voltage?

      here's a snapshot of the schematic. the project file is attached, so you could try to simulate the circuit, if you wanted to. the program is proteus 7.8sp2. please comment.

      Click image for larger version

Name:	ua.jpg
Views:	1
Size:	482.1 KB
ID:	820929

      as you can see, i can't contribute much as far as electronics is involved, but i will make the pcb and make a photo-tutorial on it, if it works well.
      Attached Files

      Comment


      • #4
        Agree with Teemu's post and add: I *think* that any regular SS power amp is essentially the same as the "basic" UA (thus saving you *a lot* of work); after all both end up driving a unity gain emitter follower stage (or a Sziklai pair, which behaves as one) with the very important caveat that you must be very careful with the attenuation/gain balancing.
        I mean that given its high gain (typically 15X to 30 or 40X) it is way too easy to overdrive a regular SS power amp, and *that* would absolutely destroy any "tubey" characteristics you may be trying to reproduce through the speaker cabinet.
        I'm talking about the kinky waveform overdriven tube power amps produce.
        *If* you overdrive the final SS power amp (which as I said is very easy to do) this turns into a flat, boring, farty SS squarewave.
        In the real UA, since you have an emitter follower which at best provides 1X gain, and most of the time *less* than that (because of the attenuator) , it's much easier to reproduce the exact original waveform, only at a lower peak to peak value (which is the attenuator's intention, anyway).
        So, in a nutshell, I think that if you watch carefully the waveform produced by your AC30, measure its peak to peak value (graphically, by counting squares in your scope screen), then couple it to an SS power amp with at least the same rail voltages (or somewhat higher) and the set the attenuation so it *never* surpasses that peak to peak value, then that SS amp will be physically able to reproduce the original tube amp at the same or lower level.
        But if you have not enough rail voltage, it will be physically impossible to do so.
        This is how I see this problem, others may very well disagree, of course.
        Juan Manuel Fahey

        Comment


        • #5
          ...the one that looked like it had an eq to you actually has an effects loop (which i can't imagine being done with an emitter follower.
          Then it makes sense why one of the boards in that particular versions was a complete amplifier with voltage amplifier input.
          do you have any transistors to recommend ... are there any other design considerations when using integrated darlington pairs, such as tip140-tip145
          Not really. Ones with high gain, such as the aforementioned Darlingtons, are more optimal, given that that the base current of the circuit can be awfully low. They will naturally need a higher voltage bias source and reliable thermal tracking gets more complex with them. But I'd just ask whats around locally in the power handling range you'll be needing.
          C1 and C2 are electros? what is their polarity? they seem awfully large. are their values correct? does that influence the sound? should i go for bipolars?
          Yes. They are quite large because the impedance of the base circuit is somewhat low and you don't want the RC circuit to roll down low frequencies too much. The value again varies based on design and the circuit simulator can portray the responses with different capacitor values, allowing you too choose the lowest one. Yes, I'd go for bipolar if possible.
          what are the power ratings on the components?
          Depends on your design, really. The emitter resistors likely fall down to 3 - 5W, R3 & R4 to around 2W, possibly lower, possibly higher. Power transistor rating depends very much on the rail voltage and the lowest minimum load impedance you intent to pick. Again, you may find the circuit simulator extremely helpful.
          can the current gain be regulated? getting the power gain to 1 on max volume would be perfect.
          What do you mean by that?
          also, how to minimize the rail-to-rail voltage?
          Well, uh, by decreasing it. It will naturally result to earlier clipping threshold so you likely have to compensate by adding more attenuation.

          The UA is just a dummy load consisting of two 16-ohm resistors connected in series and an optional low value inductor, also connected in series. The amp's output signal is then fed straight - I think, at least based on some reading I did - to a 1Kohm potentiometer that is the actual signal "attenuator" part and naturally in parallel with the dummy load. Finally signal from potentiometer's wiper is fed to the emitter follower. In some versions, I believe, they substituted the pot with switched resistive attenuator.

          As UA is pretty much a plain resistive load by nature the amps hooked to it won't be portraying any of the "peakyness" commonly associated with high impedance output amps (typical tube amps) driving reactive loads. Highly overdriven signals from amps loaded by UA are in fact also those "boring square waves", though some tube amps may portray some pass-band characteristics that alter the wave making it appear a bit less "harsh".

          Comment


          • #6
            Something we didn't talk about, and is important in the real world, is that the "simple" emitter follower, specially the bipolar version, is *not* short proof by any means, if you short it you kill it.
            I'm sure that the "real" UA has some means to be safe.
            At least that unknown value series resistor.
            Juan Manuel Fahey

            Comment


            • #7
              I'm sure that the "real" UA has some means to be safe.
              I wouldn't count on that. All UA's I've seen have been built incredibly shoddily: poor layouts, thin and somewhat randomly done wiring schemes, cheapest components one can find, and overall very amateurishly put together units. I can really look past that in one-off DIY projects but I'd be embarrassed to sell anything like that as a legit business. Actually, I'd be deeply concerned about how such units would hold together and about them being safe to the user himself. Passing UL or CE? Most likely would just end up being the laughingstock of the inspectors and returned with a big REJECTED sticker. :-)

              Seeing how they've been built doesn't really provide high hopes of them caring too much whether the circuit is short proof or not. At least that one basic version I posted the gutshot of is pretty much just an emitter follower and nothing else, no fancy short circuit protections or anything else of similar category. It will die a poor death if short circuited.

              Overall, the idea of that circuit is not weak but if I would build anything like that I sure as hell wouldn't regard the original UA's / Ho Attenuator's as any kind of reference concerning quality of the build.

              I mean... seriously... their competitors build stuff like this:
              Last edited by teemuk; 06-15-2011, 02:00 PM.

              Comment


              • #8
                Why are the attachments blank? Why even bother to have this info on this page without the diagram?

                Comment


                • #9
                  The server crashed some time ago and many attachments were lost. If you're looking for something in particular, there are many here willing to help out.
                  "I took a photo of my ohm meter... It didn't help." Enzo 8/20/22

                  Comment


                  • #10
                    Thanks Dude.

                    In particular, I'm looking for the attachment in post #3... the UA.zip file.

                    Comment


                    • #11
                      Well, andrej hasn't logged in here since 2015. Maybe someone here will see this and saved the file previously. I did not. Sorry.
                      "I took a photo of my ohm meter... It didn't help." Enzo 8/20/22

                      Comment


                      • #12
                        Also the first attachments were links to images or albums on Imageshack which around 2015 they went commercial and started charging for storage, no big deal if they applied that from a certain date on, but they NUKED millions of images which had been posted earlier, leaving millions of posts such as these without images.
                        If at least they had kept them "hostage" and offered liberating them for a small reasonable fee it wouldnīt be *that* bad , but they erased them and now you canīt get them back even paying.
                        Despicable guys.
                        Juan Manuel Fahey

                        Comment


                        • #13
                          I have an Ultimate that I have done lots of mods to. The reason I was looking here was to get an idea of what transistor to use for this. All of them that I have seen have the label sanded off of them. Just trying to figure out that part of the puzzle.

                          Tony

                          Comment


                          • #14
                            Ok,you modded it.
                            Which means you have been inside, adding and pulling parts.
                            Being such a simple circuit, please hand draw it and upload it here.

                            Write the values you can see or measure, including rail voltages, drop across emitter resistors if present, etc, and label unknown transistors as Q1 , Q2 , etc.

                            Looking at what is around them, we can probably make a good guess about what they are.
                            Juan Manuel Fahey

                            Comment

                            Working...
                            X