Simple Attenuators - Design And Testing

[Freddy G]

Hi @Freddy G There's a couple of things to try ahead of rebuilding the amps!

We are talking if fairly small changes in level and small attenuations. Also your application has arrived at the combination of two features which were both add-on extras, kind of lucky work-arounds, being the use of the 3.5 resistive stage on its own, and the use of 16 Ohm speakers with the 8 Ohm build. Together they give about a 16 Ohm load to the amp, and also when you use the 16 Ohm speaker in the 8 ohm build, there is about another 1.5 to 2db of attenuation, so compared to what a 16 Ohm speaker coming out of a 16 Ohm amp might have seen, you might have been experiencing about -5db with how you had it wired.

(Side note: All of that is also within range of the kind of variations you get from different speakers. eg, my V30 1x12 is supposedly about 2 or 3 db louder than the G12M's in my combo)

I want to show you more numbers and plots from 'under the hood' that I don't normally post, to avoid excess WTF? effects from readers:

This one shows calculated results from the 8 ohm build that you have, running either an 8 Ohm speaker into the standard 8 ohm outputs (lower blue chart), or the 16 Ohm speaker into Output 3 (upper yellow chart):

View attachment 76694
In this table you can see results for a full volume speaker, and attenuated settings 1 to 9, with 1 being the nominally -3.5db setting and 2 being stage 1 nominally -7 db.

8 ohm Speaker into Output 1

Left to right on the lower chart, we have the calculated attenuation, the increments (all pretty close to 3.5 db), impedance seen by the amp at 440hz (labeled 'R440' all not too much above 8 Ohms), then rises in the bass peak and at 5khz in db, for tonal comparison. The graphs show attenuation vs frequency, with red being full volume.

16 Ohm speaker into Output 3

Now move to the top, with an 16 ohm speaker in Output 3.

The tones are reasonably consistent with full volume but all the levels are dropped a bit, -5db at setting 1 and -9db at setting 2. Highlighted green, the impedance seen by the amp in Setting 1 is 16.18 Ohm, but at settings 2 to 9 its still close enough to 8.

16 Ohm Speaker into Output 1

Here's what happens if you put your 16 Ohm speaker into Output 1, ie without the tone tweaks that Output 3 has

View attachment 76693

Have a look at the graphs and the bass and 5khz rise values. You can see that they are a bit more suppressed compared to full volume which is why I added Output 3. In my amps you can hear this but it does depend on the amp and also how you set it. You might find it is not a problem with your amp, or that you can adjust for it with presence.

See how the attenuation levels (2nd column) are restored. Setting 2 has gone from -9 db to -6.1 db. If the tone is acceptable, then Setting 2 using output 1 might do the job for you.

Also see Setting 1, louder again but its impedance is at 13.12 Ohms, kind of in between values, your call if its OK to use.

Parallel attenuator as load box

This was the new thought from a page or two ago, and I don't have a plot for it. Set the attenuator to Setting 2 ie the stage 1 reactive is on and all others are off. (this would nominally be -7db for normal use). This is now so far as I can tell, a decent 16 Ohm reactive load box. Run it in parallel with your 16 Ohm speaker for a -3db reduction, which should have good tone too.

When used in this parallel way as a load box, I think its best either as above in setting 2, or as Id originally thought, put all attenuation to max to make an 8 Ohm load for running with a speaker into a 4 Ohm tap. I don't think there's any thing useful in between as increments.

Thanks John. Everything is getting clearer. I will try the parallel set-up and see how that goes!

 

[Gene Ballzz]

@Freddy G
Just want to confirm that you caught the memo concerning the fact that this attenuator is "in effect" another 8 ohm load and when you use it in parallel with your 16 ohm speaker you get "nominally" 5.33 ohms that the amp is seeing. Of course I'm sure you are aware that a multimeter will read a bit lower than those numbers, as DC ohms! An 8 ohm speaker typically measures ast between 6 & 7 ohms DC.
Just Remindin'
Gene

 

[JohnH]

@Freddy G
Just want to confirm that you caught the memo concerning the fact that this attenuator is "in effect" another 8 ohm load and when you use it in parallel with your 16 ohm speaker you get "nominally" 5.33 ohms that the amp is seeing. Of course I'm sure you are aware that a multimeter will read a bit lower than those numbers, as DC ohms! An 8 ohm speaker typically measures ast between 6 & 7 ohms DC.
Just Remindin'
Gene

Hi Gene, It should be OK, using the new 'wrinkle'. When the attenuator is set to max attenuation, then with no speaker in it, it does indeed make an 8 Ohm load. But if instead its set to just stage 1 attenuating (usual-7db) , other stages off, then it happens to make a pretty accurate 16 Ohm load. Basically, its R1 (15 Ohm) plus the two coils in Design M in series and if there's no speaker plugged into the box, it doesn't see anything else. This would also work in design M2 too.

 

[Gene Ballzz]

WOW! @JohnH
I must have missed that "wrinkle" in the details you provided earlier, or possibly discounted its relevance! Still a bigger fan of the 16 ohm M-lite and the multiple options it can provide, as long as the intended amp is not 8 ohm only! I'm especially fond of it not being as dependent on speaker impedance for optimum performance as the 8 ohm version.
Just Sayin'
Gene

 

[littlewyan]

Just wanted to say a quick thanks for starting this thread. I've been working my way through the posts and have learnt a LOT! Very informative and incredibly valuable information in here! I plan on building myself one of your M2 attenuators when I get time. In the meantime I've got my custom Airbrake which I've now modded to sound better at high levels of attenuation. The bedroom control on these often sounds a bit fizzy because the impedance the speaker sees is too high (188Ohms at the highest setting!!), so I've modified it to stick a 40R resistor (two resistors in series in my case) across the speaker when the bedroom control is in use and also removed the 0.1uF bypass cap on the rheostat. Huge improvement.

Thanks again! I'll post when I start building my attenuator.

 

[_Steve]

Hi again!

So after a very long boat from China and some sitting around on my workbench for a few weeks I finally built an M2.

Just wondering however if there is a safe way to test it before plugging it into an expensive amp?

 

[JohnH]

sure yes. Plug the speaker in but not the amp, then measure resistance between hot and ground at the input. It should be between 7 and 10 in all settings (or x2 if its a 16 ohm version)

 

[_Steve]

sure yes. Plug the speaker in but not the amp, then measure resistance between hot and ground at the input. It should be between 7 and 10 in all settings (or x2 if its a 16 ohm version)

Got it. Thanks!!!

 

[telesto]

Hi again!

So after a very long boat from China and some sitting around on my workbench for a few weeks I finally built an M2.

Just wondering however if there is a safe way to test it before plugging it into an expensive amp?

Sure: plug it into a cheap amp first

 

[nicnac]

Hey guys!

Thanks to John for this great forum and awesome design and hard work and to the guys for all the testing and thoughts! I have enjoyed the journey.

I took the plunge and built the multi ohm (4/8/16) in/out M2 version (decided to build 5 and sell 4 off at cost to get mine free basically) and the tests have provided very big smiles so thanks again. This is my perfect studio solution as i run multiple cabs and heads in search of sounds.

I went for a bigger alu case 190 x 190 and split my initial resistors to 2 x 50w as I noted a post that someone mentioned they are roughly the same size as the 100w versions and my thinking is a little more heat sinking can't hurt in the long run and the bigger box has plenty space. I didn't cut any corners on parts so name brand res and coils from coilsnstuff (Steve) eBay. I got most of my bits from switch electronics uk and farnell uk. Over rated switches and branded wire too. Came to around 150 quid per unit (saved a bit on the bulk buy).

It is a tad tedious drilling etc but the step drill bit is your friend for switches and jack inputs. I also bought a 60w soldering iron as the resistors will suck up the heat as you solder so it makes life a little easier.

I quite like the line out after the R2 point in the circuit as it sees the inductors and benefits from some resistance. Will also reiterate with all the others that it will require a cab sim (Don't expect it to sound like the amp in your room).

I have run my 5e3 tweed deluxe, champ clone, ac30 and silver face 100w bassman all have sounded brilliant and no noticeable heat issues.

I 'll post pics soon.

Thanks again, especially JohnH.

 

[JohnH]

hi @nicnac
That's great and thanks for posting. I'll look forward to the pix in due course. Id also be very interested in your impression of the multi-ohm features, particularly at the input using the 4 or 16 inputs given the core of the unit is for 8 ohms. Id expect a small volume step down, but hopefully the tone stays reasonably consistent?

Line outs and cab-sims

The raw signal from a line out on any such unit is very bright indeed, but these days people seem to send it off into an IR setup to provide the cab sim. But I'm now working up an analogue cab sim, just to see how good I can get it. I'm hoping I can get good results just as a passive circuit.

So far, Ive been recording miced samples off of the speaker, and simultaneously a direct output. Then I normalise them and make frequency plots of each, and I subtract them from each other in excel. This gives me a guide to the ideal transfer function for a cab sim that will take the raw signal and twist it to approximately match the miced signal.

Then, I'm concocting an LCR network to try to follow that ideal shape, analysing that in SPICE and also spreadsheets to find out its response, and feeding that response back into my Audacity recordings to see if it will bring the raw signal down to one close enough to the miced signal for use in PA, silent headphone playing or recording.

Im finding you can get an acceptable sound, you can definitely tell from listening that it is not quite the same, but it still sounds reasonably close to a miced cab, at least for some uses.

This plot has the wiggly line based on the theoretical change in frequency response derived from the samples (based on various played samples and frequency sweeps), and the theoretical response of my LRC network:



By adjusting component values, I can shift the peaks and dips up and down and sideways.

Clearly, the trace based on real tests has lots of ups and downs as different resonances occur, and you can see these also in the plots provided by speaker makers. But its interesting to discover how important or unimportant these variations are in getting a good tone.

All work in progress, not physically built yet.

 

[telesto]

Hi John,

I've been playing around with various configurations of attenuators, and am now going into the area of the load-box, so since your M3 with resonance circuit effectively uses similar Caps+Coils, I can give one a try for you and let you know how it sounds (I've spec'd mine with 0.85mH treble and 8mH bass, and a cap in the 180 to 220uF range).

I'm searching for parts now, and found a reasonably priced coil 8mH with 1mm (18 gauge) wire with 0.5ohm resistance for 9EUR I think for the coil about 1mm/18AWG was agreed to be ok for a 50W amp (for 100W maybe a little thicker wire?)


..and for the cap, I'm a little confused about the specs. I read several times what Aiken wrote about the ripple current, but not all caps have the ripple current in their specs. Some give ripple voltage (VAC rating), and I guess from there the ripple current can be derived by ohms law (divide VAC by calculated Z of cap). For example, this 220uF cap has a 63VDC/23VAC rating, and Z = 1 /(2*(pi)*f*C), a low E-string is 88Hz, so plugging it all in:

Z = 1 /2*(3.14)*88*.00022
Z = 8.22

23VAC/8.22 = 2.8A

...so am I correctly calculating the ripple current to be 2.8A? Aiken calculated ripple current of 3.5A for a 100W (28VAC) amp, so if I divide 28VAC by (Z) 8.22 I get 3.4A so I guess it's correct.

I was looking at this 220uF rated at 23VAC cap that costs only 4EUR. A 50W amp puts out about 20VAC, so the 23VAC rating may be a little close, but maybe still OK (altho Aiken reccomended x1.5 the min value for extra headroom). Or what capacitor specs did you decide on for a 50W amp? I guess any with a rating over 20VAC or ripple current of 2.8A?


I was looking at this one, which actually specs the amp output rating at different ohm loads, which is interesting/nice (less math for me ). These are sold at 2 pieces for 20EUR by me, so not too bad and should be well within spec for a 100W amp (@8ohm out). So with 10EUR for a 10mH coil and 10EUR or less for the cap, it's only 20EUR to put the bass circuit in, which isn't too bad really.

On a side note, the schematic/build I see at TGP is a little strange, the cap and bass coil looks to me way too big and over spec (not that that it's bad, just raises cost unnecessarily) and then they use a 200W power resistor, that is actually rated at only 80W, and then is below spec for a 100W amp. Strange. I should maybe ask about it on that thread, but I'm too busy/lazy right now to make an account there

 

[JohnH]

Hi @telesto

I also get unsure about those cspacitor specs. The one you found looks ok though. If i built an M3 I was going to try big polypropylene caps with very low losses, but they are a lot more $. So I think your selections look good.

In comparison to Aikens design and the one derived from it on TGP, our box only sends half the current down through R1, while in the Aiken design, everything has to go through the main resistor and the cap.

 

[telesto]

In comparison to Aikens design and the one derived from it on TGP, our box only sends half the current down through R1, while in the Aiken design, everything has to go through the main resistor and the cap.

Yea, that was actually something I was thinking about and forgot to mention. Actually, my attenuator design reduces the signal even more, which means the signal hitting the RLC circuit will be kind of low. While this means I can actually have lower ripple tolerances for RLC components, I'm wondering about the effect Aiken wrote about that low signals hitting relatively high resistance in the LC will "kill the Q". The resistance of the cap+coil I was looking at was about 1 ohm or less, so I don't think it's too bad. The Spice SIM looked OK. But I guess I'll never know for sure till "the rubber hits the road"...or "the signal hits the coil" Anyway, I'll let you know in a week or two once I get the parts and the rubber and coils start to fly

 

[Grouville]

I first came across this thread a month or more ago, and it seems just what I need to tame my Egnater Rebel 30 for home use. It is the combo with a 16ohm speaker. Love the tones from this amp, but the neighbours are tired of listening to me practise the same Pink Floyd number over and over!

Am planning to build the M2 design posted in July? I ordered the resistors from China and 1 metre of 18AWG copper wire to hand build the inductor coil. I sat back and awaited the deliveries...
And now I cannot find the post which detailed the diameter and turns I would need to build the 1.8mH coil!
I have searched and searched - perhaps I had originally found instructions on another forum?

Either way - could someone please enlighten me? I have used some online calculators, but they all seem to indicate I will need a coil 25 inches in length with 5000 turns! (I exaggerate, but not not by much!).

I was sure that 1m of 18AWG would be enough - was I wrong?

 

[JohnH]

Hi @Grouville , thanks for your interest. Im sorry but we haven't posted any turns/wire length numbers for the inductors on this thread. Generally we just order a coil of the required inductance and wire gage from a supplier, and there are a few of them depending where you are.

But I have seen calculators for multilayer coils. I'll have a look but id expect a lot longer than 1m is needed.

 

[JohnH]

this one looks useful:
https://www.teslascientific.com/products/coil-inductance-calculator/

I threw in a guess of 300 turns in a coil with 15 layers, diameter 20 internal to 50 external, and 20mm long. I got a bit over at 1.9mH. That's about 33m of wire.

 

[Grouville]

Hmm. I obviously completely misunderstood the other resource I read, or it was wildly off the mark.
Could anybody recommend a UK supplier for a ready built 0.18mH inductor coil?

Have not used such a thing before...

Found a number of coils on eBay, some are wire, some are copper tape?.
Is the AWG significant? Some quote resistance?

Any recommended products?

 

[JohnH]

Im in Australia, but we have had a few builds in the UK. Maybe someone will chip in.

But for the 16Ohm M2 attenuator, you need a 1.8mH air core inductor with 18 gage wire, which means the wire is 1mm diameter. (in the US, there's Madisound that make a range with 19 gage wire, they seem good too for up to 50w amps and Gene has been using them)

The resistance is not a spec that we need to specify directly, but if you get that wire gage then the resistance is reasonably low.

Here's a couple of UK links that look OK to me.

https://www.falconacoustics.co.uk/a...-air-core-audio-151-180mh-audio-inductor.html

2nd from the bottom on this page:
https://www.qtasystems.co.uk/loudspeaker-components/inductors.htm

There were quite a few others too. They tend to be sold by people who sell loudspeaker cross-over parts, often wound in-house. Ive looked on ebay but could never find ones that had all the specs clearly.

 

[Grouville]

Thanks John. Have gone with the second of those two (min qty 2 for the first). Looking forward to building this - will let you know how it turns out.