Simple Attenuators - Design And Testing

[JohnH]

hi @RoadShow

Just to try to get to the bottom of the volume discussion, Im just wondering about the 18W amp. How loud is it usually, if not attenuated? and how compared to the JTM45?

So far as I know, a JTM45 puts out just under 40W RMS clean. If the 18W was doing 18W, then the difference in level between them is somewhere 3 to 3.5db if similarly cranked. This is a small step in volume, about the same as the smallest step available on the attenuator. Does that match up to what you experience?

You could also check by measuring AC voltage at the amp output, ie at the input to the attenuator or across the speaker if no attenuator. (needs some care to clip test leads securely). Nominally, at 18W into 8 Ohms, you expect 12V ac at the output. A measurement made by a meter using a guitar or other signal will be much more variable, but you should get 'several' volts there.

With the JTM45, which attenuation setting gets you down to about that nice conversation level of volume?

 

[RoadShow]

hi @RoadShow

Just to try to get to the bottom of the volume discussion, Im just wondering about the 18W amp. How loud is it usually, if not attenuated? and how compared to the JTM45?

So far as I know, a JTM45 puts out just under 40W RMS clean. If the 18W was doing 18W, then the difference in level between them is somewhere 3 to 3.5db if similarly cranked. This is a small step in volume, about the same as the smallest step available on the attenuator. Does that match up to what you experience?

You could also check by measuring AC voltage at the amp output, ie at the input to the attenuator or across the speaker if no attenuator. (needs some care to clip test leads securely). Nominally, at 18W into 8 Ohms, you expect 12V ac at the output. A measurement made by a meter using a guitar or other signal will be much more variable, but you should get 'several' volts there.

With the JTM45, which attenuation setting gets you down to about that nice conversation level of volume?

@JohnH

OK, this is interesting. After reading this reply I thought of something I didn't try before. I had tried the

• 18W-8ohm out to attenuator with 8ohm speaker in 8ohm attenuator jack (no worky)
• JTM45-16ohm out to attenuator to 16 ohm speaker in 16 ohm attenuator jack (worked)

So, I swapped heads around and ran the 18W-16ohm out to the attenuator to 16 ohm speaker in 16ohm jack, BINGO, works as expected. This may only add confusion, but the 16 ohm plug in works as expected with the 18W for initial attenuation followed by additional attenuation with each switch.

If all of my ohm readings from the other day were more or less correct, does the issue using 8ohms make sense? It doesn't to me.

So I inspect everything, and embarrassingly find the cable to the 8ohm cabinet went flakey. Why it worked straight from the amp to speaker beats me, must be it gets bent just right.

Embarrassed but relieved and happy!!

Thanks for bearing with me guys...

 

[JohnH]

i guess there must be some issue either with the 8 ohm output tap on the 18W amp, or with the 8 Ohm attenuator.

Heres a couple more tests:

This one is zero risk if you can do it:

Get a battery, any type, maybe a 9V one but a 1.5V will do fine, feed it through a resistance of about 100 Ohms - whatever you have, and feed that into the front of the 8 Ohm attenuation, with 8 ohm speaker plugged in. Now measure the voltage at the front end across teh attenuator input jack - with 9V and 100 Ohms it would be about 600 to 700 mV. At the attenuator output it would be about 18mV at max attenuation across the speaker, and increase by a factor of about x1.5 each time you reduce attenuation by 3.5db (eg 18, 27, 40 mV etc) up to about 300mV at -7db. Exact values depend on your battery and resistor but there should be a fairly smooth increase by approximately constant multiplication factors. This is basically a dc version of assessing the attenuation.

Do exactly the same with the same battery and resistor, in the 16 Ohm build and the voltages should all be greater, about 1.2V at the front, 33mV at the 16 Ohm speaker with max attenuation., etc

Second test:

Do you have an 8 Ohm out on the JTM45? can you try it with the 8 ohm attenuator and 8 cab?

 

[Gene Ballzz]

@JohnH

OK, this is interesting. After reading this reply I thought of something I didn't try before. I had tried the

• 18W-8ohm out to attenuator with 8ohm speaker in 8ohm attenuator jack (no worky)
• JTM45-16ohm out to attenuator to 16 ohm speaker in 16 ohm attenuator jack (worked)

So, I swapped heads around and ran the 18W-16ohm out to the attenuator to 16 ohm speaker in 16ohm jack, BINGO, works as expected. This may only add confusion, but the 16 ohm plug in works as expected with the 18W for initial attenuation followed by additional attenuation with each switch.

If all of my ohm readings from the other day were more or less correct, does the issue using 8ohms make sense? It doesn't to me.

So I inspect everything, and embarrassingly find the cable to the 8ohm cabinet went flakey. Why it worked straight from the amp to speaker beats me, must be it gets bent just right.

Embarrassed but relieved and happy!!

Thanks for bearing with me guys...

Are you using any shielded instrument cables as speaker cableS?
Just Askin'
Gene

 

[RoadShow]

@JohnH

I think you missed the line where I found it was the cable. Works great now.

i guess there must be some issue either with the 8 ohm output tap on the 18W amp, or with the 8 Ohm attenuator.

Heres a couple more tests:

This one is zero risk if you can do it:

Get a battery, any type, maybe a 9V one but a 1.5V will do fine, feed it through a resistance of about 100 Ohms - whatever you have, and feed that into the front of the 8 Ohm attenuation, with 8 ohm speaker plugged in. Now measure the voltage at the front end across teh attenuator input jack - with 9V and 100 Ohms it would be about 600 to 700 mV. At the attenuator output it would be about 18mV at max attenuation across the speaker, and increase by a factor of about x1.5 each time you reduce attenuation by 3.5db (eg 18, 27, 40 mV etc) up to about 300mV at -7db. Exact values depend on your battery and resistor but there should be a fairly smooth increase by approximately constant multiplication factors. This is basically a dc version of assessing the attenuation.

Do exactly the same with the same battery and resistor, in the 16 Ohm build and the voltages should all be greater, about 1.2V at the front, 33mV at the 16 Ohm speaker with max attenuation., etc

Second test:

Do you have an 8 Ohm out on the JTM45? can you try it with the 8 ohm attenuator and 8 cab?

 

[RoadShow]

Are you using any shielded instrument cables as speaker cableS?
Just Askin'
Gene

Nope, 16 gauge dual conductor cable.

 

[Yamariv]

@JohnH

Ok, so I've had all the parts to build 2- M2 16ohm Attenuators for a while now and finally got to soldering tonight and actually finished one of them! So I just wanted to double check the readings I got out in the shop before I plug in one of my beloved amps into it! I'm scared but also excited to finally have an attenuator

So I hooked up my test speakers which are 2-16ohm in Parallel which reads at 7ohms together (had a brain fart and forgot I was building a 16ohm setup and should have just plugged one of the 16ohm speakers direct..)
Anyway, so having basically an 8ohm load plugged into the back, I was getting between 15.8 and 18ohms at the input in any and all combinations of switching. I'm assuming that is good to go readings? Am I safe to test it on an amp?

Out of curiosity, if I have an 8 ohm speaker hooked up to an M2 16ohm, why does the amp still see 16ohms with this design? Kinda cool but I'm lost to why it does lol

 

[JohnH]

Hi @Yamariv

Great! those readings seem fine.

When you measure from the front end, like what the amp sees, its at least 80% the attenuator and hardly any of the speaker. And since your build is designed to use a 16 ohm amp tap, it stays within a couple of ohms of 16.

So should be ok. Just start with low amp volume and make sure its coming through to the speaker and that the switches are working to change the level as expected

 

[Yamariv]

Hi @Yamariv

Great! those readings seem fine.

When you measure from the front end, like what the amp sees, its at least 80% the attenuator and hardly any of the speaker. And since your build is designed to use a 16 ohm amp tap, it stays within a couple of ohms of 16.

So should be ok. Just start with low amp volume and make sure its coming through to the speaker and that the switches are working to change the level as expected

Awesome, thanks John! I'll update when I try it!

 

[Osman Qureshi]

ok great, I might try to rig something up equivalent here, then we'll have more confidence in values etc.

As an alternative, one thing I've been doing without having a dedicated line-out is to go from one of the attenuator speaker outs, into a line-in on a very small and cheap Behringer Xenix mixer to set levels, then from there to a laptop mic in.

Hi John, as I start building an 8-ohm M2 attenuator, I wanted to include “placeholder” line out for the future line out with cab sim that you are working on. When convenient, can you please check the attached drawing of the proposed placeholder line out to see if the connections are correct and the specifications of the resistor (R12) and the 10k potentiometer are also correct? And am I right to assume that given the inclusion of the potentiometer, the line out could also be used for headphones or even to connect into a mic in (like into an iRig Pro I/O)? Thanks, Osman

 

[Gene Ballzz]

@Osman Qureshi
Might I politely suggest that if and when you (or anyone else) post any personal revisions of JohnH's original drawings, you include some sort of obvious identifier on said drawing, like "Osram Version" ,"Modified by OQ" or similar, to help avoid any confusion, down the road? I will leave it to @JohnH to comment in detail on your proposed line out!

I do however, like the idea of deriving the line out "after" the attenuation, instead of from the input, "before" the attenuation. My only concern might be that if fully attenuated, it may not leave enough signal strength to properly drive headphones. Maybe a "PRE/POST" switch to toggle between the output and just after the initial, first attenuation stage (just prior to R4) may be in order? I'll make no other comments, because of not knowing if such a "line out" may or may not have some sort of "loading" or other affect on the attenuation process.
Keep Up The Good Ideas & Work!
Gene

 

[Osman Qureshi]

@Osman Qureshi
Might I politely suggest that if and when you (or anyone else) post any personal revisions of JohnH's original drawings, you include some sort of obvious identifier on said drawing, like "Osram Version" ,"Modified by OQ" or similar, to help avoid any confusion, down the road? I will leave it to @JohnH to comment in detail on your proposed line out!

I do however, like the idea of deriving the line out "after" the attenuation, instead of from the input, "before" the attenuation. My only concern might be that if fully attenuated, it may not leave enough signal strength to properly drive headphones. Maybe a "PRE/POST" switch to toggle between the output and just after the initial, first attenuation stage (just prior to R4) may be in order? I'll make no other comments, because of not knowing if such a "line out" may or may not have some sort of "loading" or other affect on the attenuation process.
Keep Up The Good Ideas & Work!
Gene

Makes a lot of sense is Gene. I should have put a warning. Thanks, Osman

 

[JohnH]

Hi @Osman Qureshi

Your line out could be OK, But the pot would have three connection lugs. The pot value is OK at 10K, but use a log pot for good control and use a 24mm diameter one for a bit more rating. The 2.2k resistor is fine but it needn't be more than 1/2W rating since there's very little current flowing through it.

Im still thinking about line-outs and plan to post soon - there are a lot of different uses and different levels needed so I'm trying to come up with recipes that would suit a wide range. Im thinking of post, pre, balanced, unbalanced, cabsim etc

What you show looks OK for a line out and should be adjustable down to a mic level with care. I dont think it will drive phones and in any case they will all sound bad until there is some form of cab sim, IR loader or post EQ is applied. But if you can get the signal into a PC, then all that is possible.

In your case I think you have a laptop setup with just a combined headphone/mic socket? If I was you id get a very simple small mixer with USB out, like a Behringer Xenix 502 USB or similar, which gives you a bunch of useful mic and line ins and outs, plus a headphone out, level controls and a bit of EQ. I have an old non-usb version and its great. Its about $100. With it, I can get a clear recording directly out of the attenuator output speaker jack, and use the mixer to set the level.

 

[Osman Qureshi]

Hi @Osman Qureshi

Your line out could be OK, But the pot would have three connection lugs. The pot value is OK at 10K, but use a log pot for good control and use a 24mm diameter one for a bit more rating. The 2.2k resistor is fine but it needn't be more than 1/2W rating since there's very little current flowing through it.

Im still thinking about line-outs and plan to post soon - there are a lot of different uses and different levels needed so I'm trying to come up with recipes that would suit a wide range. Im thinking of post, pre, balanced, unbalanced, cabsim etc

What you show looks OK for a line out and should be adjustable down to a mic level with care. I dont think it will drive phones and in any case they will all sound bad until there is some form of cab sim, IR loader or post EQ is applied. But if you can get the signal into a PC, then all that is possible.

In your case I think you have a laptop setup with just a combined headphone/mic socket? If I was you id get a very simple small mixer with USB out, like a Behringer Xenix 502 USB or similar, which gives you a bunch of useful mic and line ins and outs, plus a headphone out, level controls and a bit of EQ. I have an old non-usb version and its great. Its about $100. With it, I can get a clear recording directly out of the attenuator output speaker jack, and use the mixer to set the level.

Thanks John. Great idea about the Behringer Q502USB. It retails for less than £50 here. But I think I'll wait for your design and be the first beta tester--or even the alpha tester. On the three connection lugs on the pot, one is input, one is output, and the third is ground, which I guess just gets connected to the ground line going back into the amp? Regards, Osman

 

[acromarmot]

Hi JohnH and others,

First of all :thank you thank you thank you!!

I'm currently building my second tube amp, based on a Friedman BE-100 with some minor changes and a different power amp [~45W out of 2xEL34]. For space reasons I designed a pcb rather than having it on turret or "handwired" (BTW today's Friedman's are not exactly handwired). But I don't think a well design board is necessarily worse than hand wired, same goes for the type of caps, etc...

As I'm looking for a good but simple yet not too expensive attenuator I stumbled across this forum.

I want to include the attenuator into the amp (as I don't like to play around with several boxes and cables to draw, I'm more the keep-it-simple approach).
For obvious reasons I'd include a bypass.

Using a small headshell space is an issue, too. So I'd love to use the design with just one coil if it's the same result as the two-coils-design. Read your post about the whole process but couldn't find the one-coil schematic, can you point me to it?

Did you wind the coil yourself or did you buy an existing one? How about the power dissipation of the resistors? I've seen the chart with the calculations and max 30W dissipation (@50W in), however still you guys use twice or even three times the power rated components. Did I miss something?
I don't want to include a fan (unless I'm forced to) and case mounting is possible but I 'm not sure how effective (it's not a real heat spreader/sink!)... Do you think it is a problem to go with 35-40W rated resistors when case mounted (heat vents provided as I already have them for the tubes)....

Thanks for your thoughts,

Cheers
Gunnar

Edit N°1: found the latest M2 schematic, will include the bypass myself

Edit N°2: will try to get the inductor from a speaker parts supply.

Edit N°3: to fit into my amp I'll probably create a small pcb with TO-247 housing resistors and contact foil instead of paste as well as a heat sink. I could provide kiCad project files in case someone's interested.

 

[JohnH]

hi @acromarmot , welcome to our thread. That sounds like a great project!

The one-coil design is called M2. It's posted in a few places but it's most easily found near the top of post 1 on page 1. (July 2020 update). It's performance is virtually identical to the previous two-coil version, M, also described in post 1.

You can certainly have a bypass switch, just be careful to not use it to switch 'on the fly', only change it when the amp is off or on standby. Maybe it could be visually separated from the other switches, by moving it away or with some markings? Treat changing it like changing a loudspeaker cable. The other stage switches in M2 you can change while running.

The power ratings I've been recommending allow for a 50W amp, overdriven by a further 50%, then with a factor of 2 on that. The rated values on power resistors are only valid with a particular area of very effective and infeasibly large heatsink. For what you're doing I'd keep with the stated ratings (eg R1 is 100W, R2A and B 50W and the others are 25W), and maybe fix them to something substantial and aluminium in your chassis, but maybe open not enclosed in a full case. With that, and with a 45 W amp, it shouldn't need a fan.

You can wind the coils yourself on a bobbin but you'd need an inductance-measuring meter to get it right. There are online inductance vs winding calculators but it'd be very unlikely be exact. These air coils are easily available all complete from places that sell loudspeaker components, where they are used in cross-over networks. We know of places in several countries, depending where you are.

Best wishes for your project.

 

[acromarmot]

hi @acromarmot , welcome to our thread. That sounds like a great project!

The one-coil design is called M2. It's posted in a few places but it's most easily found near the top of post 1 on page 1. (July 2020 update). It's performance is virtually identical to the previous two-coil version, M, also described in post 1.

You can certainly have a bypass switch, just be careful to not use it to switch 'on the fly', only change it when the amp is off or on standby. Maybe it could be visually separated from the other switches, by moving it away or with some markings? Treat changing it like changing a loudspeaker cable. The other stage switches in M2 you can change while running.

The power ratings I've been recommending allow for a 50W amp, overdriven by a further 50%, then with a factor of 2 on that. The rated values on power resistors are only valid with a particular area of very effective and infeasibly large heatsink. For what you're doing I'd keep with the stated ratings (eg R1 is 100W, R2A and B 50W and the others are 25W), and maybe fix them to something substantial and aluminium in your chassis, but maybe open not enclosed in a full case. With that, and with a 45 W amp, it shouldn't need a fan.

You can wind the coils yourself on a bobbin but you'd need an inductance-measuring meter to get it right. There are online inductance vs winding calculators but it'd be very unlikely be exact. These air coils are easily available all complete from places that sell loudspeaker components, where they are used in cross-over networks. We know of places in several countries, depending where you are.

Best wishes for your project.

Hi @JohnH,

thanks for your thoughts. Yeah, we're on the same page. Inductive meter I could use from
my job but I'd buy the coils for laziness

No switching on the fly, that's for sure. Gonna put the switch to the rear close to the speakers selector.

In front of the tubes I have space left, ve ting directly to the top. I'll design the pcb to use a dedicated heat sink, as it's gonna pointing upwards like the tubes and transformers the heat ain't gonna ruin my amp caps. So that should work flawless...

As soon as the whole thing is done (currently I wait for the mainboard to arrive from etching) I'd maybe share the results here but up to now I don't have any recording equipment....

Cheers
Gunnar

... sitting in Germany...

 

[acromarmot]

Hi again @JohnH,

I read somewhere you were addressing a little loss in treble and wanted to add a parallel cap to the later or last attenuation stage... Has this come to a solution yet?

btw, the loss of high frequencies with lower volumes has not so much to do with the non-linearity of our ears. This kills low end, i. e. when volume is down we cannot sense low frequency as before.

I guess(!) the loss in treble is a loss in harmonic distortion plus maybe a non-linear speaker behaviour or simply the fact the speaker is not moving enough.

Cheers
acromarmot

 

[TheOtherEric]

Hi again @JohnH,

I read somewhere you were addressing a little loss in treble and wanted to add a parallel cap to the later or last attenuation stage... Has this come to a solution yet?
...
I guess(!) the loss in treble is a loss in harmonic distortion plus maybe a non-linear speaker behaviour or simply the fact the speaker is not moving enough.

Cheers
acromarmot

Unless I missed something, I don’t think folks are finding this attenuator (unlike some others) to cause a loss of treble at higher attenuation. I’ve found that my amps lose treble at higher output (which seems to be normal) but that’s an amp-specific issue so I’m not sure I’d want to try compensating by monkeying with the attenuator design, although I guess it’s feasible.

Maybe there was a case of using the attenuator outside its intended use that was causing lost highs?

 

[acromarmot]

@JohnH, @TheOtherEric

OK, then most likely I misread sth. Nevermind.

Cheers
acromarmot

Edit: I re-read and it was @TheOtherEric having trouble with treble but what he detected to be caused by his amps.

@TheOtherEric : odd behaviour of your amps. Mine has some (other) issues but an amp losing treble when cranked sounds a bit strange to me. Usually an amp would respond with more harmonic distortion (overall distortion increasing) and thus enhancing treble, some even being sharp/causing hiss [therefore the anti-hiss-cap in the PI in some designs or the need to reduce presence pot when cranked]. As with more volume our ears are more sensitive to low frequencies this would in theory also be an explanation... But then the bass needs to be brutal and washed to interpret this as muffled.
Just because I'm curious and play around with amp designs... What amps are you using ('cause I wanna learn sth)?

Cheers
acromarmot