Simple Attenuators - Design And Testing

[peefnik]

There doesn't seem to be any noticeable difference when listening to the clips. Since the resonant hump is recreated at the spkr this may be enough ... from a listeners perspective?

Having the resonance at the amp would, in theory, recreate the reflected impendance seen by the output stage and make the closed loop behavior of the system closer to normal. Again, would the player notice this difference. And mayhap it is not enough of a difference in the end?

In your clip the attack is pretty soft ... I wonder if a more percussive attack on the bottom end would feel different in the two scenarios? Does it chug, for instance. I can't really answer these questions as I don't have a unit built.

I added the PI to the DR output section only sim. Also created an M2 block. I have the input level set to get 5.3Vpp at grid of U4 (top, inverting tube in PI). This results in about 22 W into the load.

 

[peefnik]

Mayhap, since this is a simple attenuator, all of this resonance circuit stuff is pushing it out of that classification. The M2 sounds good as is and adding a resonance peak may not be worth the effort or $. Not to mention, if the peak were ever to be too high it could present a reflected imedance to the output transformer, tube stage which could potentially damage the amp. If it was designed correctly that wouldn't be an issue ... it just adds a level of uncomfortability if something were to go wrong. I guess that is true in any case if parts failed and resulted in a high impedance load.

 

[JohnH]

I agree with all of that, and its why I haven't pushed to build an M3 myself. The resonant parts can be relatively expensive, particularly the cap if its as high a spec as it probably should be. If the cap and L2 are there, then from low mids upwards, almost the whole power of the amp has to pass through the cap.

But, if the resonant circuit is used, I've tried to figure out what its implications would be and what benefits it might offer. Here's what I reckon:

1. If the unit is being used as a load box, with a line-out taken at the amp side, then the line out signal misses out on the bass peak if its not there. But a few have done this and not found it a problem.

2. While the attenuator has been designed based on an effective amp output Z of 20 Ohms (for the 8 ohm version), real amps vary from much less, even 5 ohms, up to much more, almost infinite if there's no NFB. This affects the voltage vs frequency seen by the speaker, which has a peak and rise matching the speaker impedance if amp Z is very high, down to a flat response with a zero amp Z like an ss amp. Somehow, the first reactive stage compensates for this and you can see it in analysis. You should see it too with your NFB model. Zero NFB youll get high amp Z and a rising signal at the speaker, high NFB, low amp Z and the response is flattened. It keeps in match with the full volume version, except that with no resonant circuit, the bass peak does not adapt and if the amp is showing less that an output Z of 18-20 ohms, the bass peak stays higher

3. Last thing I can think of: When the load has a higher impedance, with a tube amp of moderate of high Z, it will actually draw more power out of the amp, and so at those frequencies, and when the signal is high with the amp working hard, there will be more distortion, dynamics and sizzle. This adds the realism and feel which we found added when we moved from the early resistive designs on this thread. Its relevant if the amp is running not clean in the power stage, at least for high signals. Designs M and M2 get that right from about 150-200 hz up. But if there is anything like that happening in the bass, only M3 will find it. It would be heard as some compression and maybe harmonics of the bass notes. But I cant hear any so far. I think the implication of not have this effect for the bass end is that the bass stays a bit more solid, less distorted flub like its got more behind it, not a bad thing really!

 

[peefnik]

You can see the amp output Z, spkr output, vs NFB in this image. The lower green line in outputz is with standard 820 ohm NFB resistor. I stepped it 820, 2.2k, 1T (e.g. no NFB). The changes in amp Z are small but you can definitely see the effect how it reacts less to the spkr impedance changes and smooths it out.

 

[Oliver Gardiner]

I may be missing something but I'm struggling to understand why R5 is only 4R7 as this means the load presented to the preceding stage is only 4R. By the same token, R7 seems high at 33R and I would have expected something like 20R which would get you back down to 8R in parallel with the 13R6 you get with the 8R load and the 5R6 of R8.

As a hybrid approach, would there be anything intrinsically bad about using a Monacor AT-62H L-Pad after the 1st reactive stage of M2?

Thanks

 

[JohnH]

Hi @Oliver Gardiner , thanks for your message.

Those resistor values are fine. Although its right to note that the third stage does present a low resistance to the previous stage, that value is never passed on to the amp because it is always separated from the amp by Stage 1, and also Stage 2 if its on. So the effect on ohms to the amp is quite small. Similarly, the high value of R7 is never changing the ohms seen by the amp excessively.

What is more important is output impedance of these stages, which needs to be kept quite high like a tube amp output stage.

The key thing with the resistive aspects of these designs is to keep reasonably consistent ohms at the amp, and also as seen by the speaker, and then to set it a up so even attenuation steps can be made. It works for every combination of switch settings.

The idea with an Lpad may have a problem shared by many attenuators, in that the output impedances is not consistent, and gets very low at high attenuation. This damps the speaker, making it sound dull.

 

[Oliver Gardiner]

Many thanks - that makes sense. I've not really looked at attenuators before - I can see that the first stage will clearly dominate the impedance seen by the amp but had not considered the effect of output impedance. Have been playing with a variety of amp designs and have always used a master volume set up so far - am now thinking that an M2 would be a better approach. I've got a volume/clean boost last in the FX loop anyway so can tweak levels there.

Thanks again!

 

[Baby Thomas]

Hi guys!
Last night I was watching the last Tone-Talk show (ep. 77) and John Suhr was talking about his Suhr Reactive Load and some other similar products. He shared his thoughts about these devices, impedance, frequency response, resonant peak and so on. Maybe it will be interesting for some of you.
Have a nice day!

 

[peefnik]

So, after more analysis, I've come to the conclusion that trying to recreate the resonance hump as an inline attenuator is not practical/possible. This is due to the way the cab/spkr load interacts with the reactive/resonance load. If they are not matched, exactly, you get multiple resonant humps at the amp output that are surely not desirable. You would have to make an different attenuator based on the cab/spkr you were using. This is with a speaker model that is most likely not a 'cab' model but the end result is, I think, the same. The response of the actual cab/spkr is really an unknown for a generic attenuator.

You could have a switchable option for attenuator/load that would switch in the resonance components ... but that adds a bit more complexity.

The only reactive loads creating the low frequency resonant hump, AFAIK, are loads only, not attenuators (Suhr, TAE, etc). I'm not sure exactly what the TAE is doing but given the fact it has 4 settings for resonance implies the they are creating/shifting the low frequency response. I don't know if this is a hump or just a general low frequency bump. The TAE is definitely a load only device as it reamps the amplifier signal via a SS amp ... it does not pass it directly thru.

The multiple low frequency hump effect can be seen in the following images. First is with the stock M3 values, then adjusted to more closely match the speaker load.

 

[JohnH]

Mike Lind makes passive fully reactive attenuators, with the bass resonance. He posts often on TGP:

https://www.thegearpage.net/board/i...rplant-a2xr-reactive-attenuator-load.2100557/

He's also been very generous with comments and also his cab measurements, which have been a great help here.

He has a switch to tune the resonance, and I think the key thing is whether its to be based on open-back or closed-back. With most 12" speakers, this seems to make a shift from around 80hz to around 110hz.

My hunch is its not too significant in practice, and I still haven't found a compelling reason to it in my ones, since M and M2 seem to work fine just letting the speaker develop the bass peak.

 

[Gene Ballzz]

Mike Lind makes passive fully reactive attenuators, with the bass resonance. He posts often on TGP:

https://www.thegearpage.net/board/i...rplant-a2xr-reactive-attenuator-load.2100557/

He's also been very generous with comments and also his cab measurements, which have been a great help here.

He has a switch to tune the resonance, and I think the key thing is whether its to be based on open-back or closed-back. With most 12" speakers, this seems to make a shift from around 80hz to around 110hz.

My hunch is its not too significant in practice, and I still haven't found a compelling reason to it in my ones, since M and M2 seem to work fine just letting the speaker develop the bass peak.

So true sir! In real life practice, these JohnH units will usually be used for light (-7db to -14db) volume cuts and once it gets much lower than that the response of the attenuator becomes somewhat of a moot issue, when considering all the other aspects of such low sound pressure level! And sure, the resonance can be felt to be a tiny tad different at each setting, but is easily compensated for through a knob and/or technique tweak.

These attenuators are the greatest thing since sliced bread!

Thanks Again John!
Gene

 

[BrokenBones]

I've never worked with inductors. Are there any 'must follow' procedures as far as mounting or housing? I did buy some stainless washers and bolts. Can the bare inductor wire be touching the chassis?


***UPDATE***

I was able to find some pointers on mounting air coil inductors over on the audio karma forum. They give a lot of good tips on proper placement and materials.

https://audiokarma.org/forums/index.php?threads/mounting-air-core-inductors.639281/

 

[JohnH]

I've never worked with inductors. Are there any 'must follow' procedures as far as mounting or housing? I did buy some stainless washers and bolts. Can the bare inductor wire be touching the chassis?

Any secure mounting is fine, and I note you picked upon not using normal steel bolts, which can greatly increase the inductance. The insulation on the wires is quite thin and so the wires should not touch the chassis. Depending what the situation is, maybe some spacers, or thicker insulation sleeve, or mount the coil the other way up? If you scan through the thread you'll see a few versions.

 

[Esa Martikainen]

I put this together fast but it works so good that it deserves a neater housing. It is basically a reactive load which I first saw on Aiken site. In the middle is an aluminium 10 ohm 200W series resistor. On left is 1,8mH and 100ohm high frequenzy slope circuit. Top right there are 100ohm, 15mH and four blue capacitors total of 220uF resonant circuit.

Attenuation is done using a line transformer which is on bottom right and it is parallel with this reactive unit and loudspeaker is connected to it. It has three inputs and three outputs but in practise there are five different attenuations possible. I bought it after I tested how normal output transformer attenuates signal and it sound good but it did attenuate too much because of higher winding ratios.

Picture link did turn it 180 decrees so slope curcuit is top right, resonant circuit is on bottom left and output transformer on top left corner.

Esa

 

[Esa Martikainen]

That line transformer belongs to "100V standard" and can be used both ways. It has 12,5W, 25W and 50W pins and 4, 8 and 16 ohm pins and when I did put 100VAC to 25W pin there was 7.0VAC on 4 ohm, 10,0V on 8 ohm and 14.0V on 16 ohm pins

When 100V to 25W pin out came 10,0V 14,2V and 20,0V
And 100V to 50W pin out came 14,3V, 20,0V and 27,8V

I have not calculated attenuations yet because I am not used to them well enough

Edit: If I understand it right, and assuming it is 16 ohm load where 50W is inputted there comes 3,7W from smallest attenuation output to 16 ohm load and 0,24W out of biggest attenuated output without losses the transformer cause.

Esa

 

[JohnH]

Hi Esa, thanks for showing us those, it looks nice and simple to wire up. Sound clips would be interesting to hear,

 

[Esa Martikainen]

Thanks! I try to arrange some but it takes a while... Compared go resistive voltage divider load with a L-pad I used before this one seems to cause way less high frequencies drop. I did not build Your circuit yet when my resistors shipment did not include all I ordered and need.

Transformer winding ratios allow five different attenuations between 11dB...22dB.

This transformer/loudspeaker impedance must have some effect when its least winding ratio (11dB attenuation) is only 3.6 but without more study I can not say more. Using 16 ohm loudspeaker smallest input impedance is 207 ohms. Parallel with reactive unit the total mpedance is 14.8 ohms if reactive unit is assumed 16 ohms which seems reasonable? I don't know this line transformer inductance but it is rated for 50W and its -3dB frequency range is 25-20 000Hz.

If someone of You do have an output transformer and want a "bedroon quiet amp" you should test it as attenuator with a reactive load. And if you buy one I think you need only 10W or less transformer which are quite cheap.

This kind of "line transformers" also could be found cheap from various sellers but shipping might be expensive because they might be heavy? They were very common in buildings which did have a centralized annunciation loudspeaker systems like for example schools but they are still made new as well.

Because this mine attenuates at least 11dB next I will test mains transformers hoping that they do not restrict bandwidth too much...

Esa

 

[What?]

Excuse my excursion here. I'm wondering what JohnH thinks of Eminence's FDM approach to attenuation:



It appears that knob on the back of the speaker is threaded and pulls the slug part way out out of the voice coil as it is turned so that the speaker becomes less efficient but still gets the full signal from the amp.

There is an image posted with thiele small parameters halfway down the page here: https://www.tdpri.com/threads/looking-at-the-eminence-maverick-fdm.717638/

 

[JohnH]

Hj @What? thanks for posting that, which is very interesting.

I like the idea of the FDM speakers, nice simple direct method acting on the speaker, that anyone can understand.

I can see it should have its place, but I can't see it as being a general solution to attenuation, because:
1. The tone changes significantly as you dial it down
2. There's only 9db attenuation available. That's enough for dialing down for a gig or rehearsal but not enough for home studio use
3. You have to reach to the back of the speaker to adjust it, so except for cases where you can set it once and then leave it, its only adjustable on open-backed speakers with enough access, which would mostly rule out Marshall cabs and many combos too.
4 You can only use it with the cab or amp its built into
5. It doesn't have the potential for adding line-out functions, which are becoming popular on the wish lists of users
6. We can get more consistent results over a wider range, working with any amp or speaker, using a coil and a few switches and resistors.

 

[aceofbones]

Hey John,
This is a fantastic thread and an amazing project! Thanks for all your work and thanks to everyone else who has contributed! I‘ve spent the last week reading it in its entirety.
I will be building the M2 and had a question. Has anyone built an fx loop into their attenuator? I saw it asked once way back in the thread and the answer was use a line out for re-amping. Does it make sense to have a powered loop in the same unit?

I ask because I am building a Trainwreck Express clone. Apparently they don’t play nice with a built in loop, and since I’ll be building an attenuator, I figured I should look into it. I know some units such as the Fryette Power Station have this ability, so I wondered what the feasibility of such an option was.
I like how the M2 works as it is, so I don’t need to alter it, I was just curious.
Thanks very much,
Nick