I’m wanting to build a simple fixed resistive attenuator, to bring two of my tube guitar amps down to about 5W, ie -10db reduced from 50W. The reason is for quiet home playing, but still with the tone , features and speakers of a full sized amp and without buying another low power amp. *Its also just a nice project to try out.
Im hoping for some comments on the options for how to do this.
Both the amps (Marshall VM226c and DSL401) have good master volume controls, and sound good at very low volume but they sound great if turned up at least a bit (and wonderful at high volume but I can save that for other times).
I’ve already rigged up a -6db box, and it works fine, tone is still good, but I want to try a few more db reduction. I figure -10db is a good target. It will give me some more spread on the volume control, reduce noise floor and let the tubes start to run a little more.
There are lots of variations on combinations of resistor values to match this attenuation while maintaining 8 ohm load on the amp. Here are three, being a conventional L-pad, a reversed L pad and a ‘Pi’ configuration.* Once all are calculated, they all have the same load to the amp and attenuation level.
The difference is in the output impedance as seen by the speaker - see sketch below., ranging from 2.5Ohms up to 18Ohms asuming 1 ohm amp Zo.
As I understand it, a tube amp has considerably greater output impedance than a solid state amp, and in a guitar amp, this contributes to the tone from the speakers, reducing damping allowing bass resonance to develop, and relatively adding to highs on account of the speaker impedance rising with frequency.
Some commercial attenuators have reactive components to maintain better tone as volume is further reduced. I’m not planning to go there yet, but what do you think of these arrangements, for my application? Any other comments? Once I settle on a design, Ill make each branch out of a group of wire wound resistors so that power dissipation is adequate.
With thanks for any comments.
J
Im hoping for some comments on the options for how to do this.
Both the amps (Marshall VM226c and DSL401) have good master volume controls, and sound good at very low volume but they sound great if turned up at least a bit (and wonderful at high volume but I can save that for other times).
I’ve already rigged up a -6db box, and it works fine, tone is still good, but I want to try a few more db reduction. I figure -10db is a good target. It will give me some more spread on the volume control, reduce noise floor and let the tubes start to run a little more.
There are lots of variations on combinations of resistor values to match this attenuation while maintaining 8 ohm load on the amp. Here are three, being a conventional L-pad, a reversed L pad and a ‘Pi’ configuration.* Once all are calculated, they all have the same load to the amp and attenuation level.
The difference is in the output impedance as seen by the speaker - see sketch below., ranging from 2.5Ohms up to 18Ohms asuming 1 ohm amp Zo.
As I understand it, a tube amp has considerably greater output impedance than a solid state amp, and in a guitar amp, this contributes to the tone from the speakers, reducing damping allowing bass resonance to develop, and relatively adding to highs on account of the speaker impedance rising with frequency.
Some commercial attenuators have reactive components to maintain better tone as volume is further reduced. I’m not planning to go there yet, but what do you think of these arrangements, for my application? Any other comments? Once I settle on a design, Ill make each branch out of a group of wire wound resistors so that power dissipation is adequate.
With thanks for any comments.
J
Attachments
This definitely fall into the category of 'what will be subjectively the best' and as such, and given the low cost of parts to try this out, well I would say try all three over a decent time period and see if you can come to a preference that suits you.
Technical specs and aiming for the best theoretical approach count for little if you don't actually like the result.
Technical specs and aiming for the best theoretical approach count for little if you don't actually like the result.
Thanks Andrew
The 1ohm Zo value is just an example, but Ive seen several references to a range of around 1 to 3 ohms for tube amps. The higher values apply to non negative feedbacj designs and lower to NFB circuits. Mine are one of each.
http://www.diyaudio.com/forums/tubes-valves/165634-typical-output-impedance-tube-amps.html
This is for home use, the amps sound fine as they are at rehearsal or gig volume.
The 1ohm Zo value is just an example, but Ive seen several references to a range of around 1 to 3 ohms for tube amps. The higher values apply to non negative feedbacj designs and lower to NFB circuits. Mine are one of each.
http://www.diyaudio.com/forums/tubes-valves/165634-typical-output-impedance-tube-amps.html
This is for home use, the amps sound fine as they are at rehearsal or gig volume.
Mooly. Yes thats a fair comment about trying them out. Just trying to get some more technical insight too by calling in here. The -6db is sounding fine, so at least I know I like that recipe.
Theres lots of reports about tone with attenuators getting dull/flat/harsh etc at high levels of attenuation, but being better at moderate levels. Just exploring. ..
Theres lots of reports about tone with attenuators getting dull/flat/harsh etc at high levels of attenuation, but being better at moderate levels. Just exploring. ..
What Mooly said.
There's a few things going on here.
The first order thing is that (with lots of attenuation) the amp is seeing a resistive load NOT the reactive load that a speaker presents. This is amplified (sorry) if you've got an amp that uses feedback.
The second order thing is that at lower power levels the speaker non-linearities are going to be reduced (e.g. voice coil heating and non-linear magnetic gaps) or removed (e.g. cone breakup)
Thousands of electrons have died discussing this and there's many clever schemes around
I built up a -7.5 db attenuator, using a 'Pi' arrangement with resistors only. The amp sees 8 ohm and the speaker sees about 5.5Ohms.
It does what I wanted at very low volume so I'll probably pause here for a while. I tested it with my Vintage Modern combo, and in case anyone is interested, here is what happened:
After playing with it and deciding I liked it, I miced up the amp with a Rode M1, as if it was for a PA, then fed that via a neutral mixer into Audacity, to get some more analytical insight.
I set up a short riff in a looping pedal, and set the amp for my usual dirty-clean slightly crunchy tone.
Then I recorded, without attenuator at Volume 1, 3 and 7, then at 7 with the attenuator. Then I looked at the frequency response of each clip, after normalizing them in the software to the same max amplitude, and exporting to a spreadsheet. I also looked at the differences in response of each trace relative to un-attenuated at volume 7.
At Volume 1, (unattenuated), you can hear and see a definite harshness compared to higher volumes, in the form of a rises in treble above about 2 khz. THis is why I started looking at this.
At Volume 3 (unattenuated), the harshness is gone and the tone is the nearest of all tests to the Volume 7 tone - deviations at each frequency are small (within 2db, with no general trend with frequency) and it sounds good.
Volume 7 attenuated shows a very slight general reduction in treble > 2khz, maybe 1.5db max. it also sounds good, and this change is within range of the amps EQ controls.
Out of all that, I think that, for this amp, the attenuator is a great help to get out of the very low ranges to a level of about 2.5 or 3 on the Volume, useful for quiet practice. Above that level, the post-phase-invertor MV on this amp seems to do a great job, so the attenuator is not needed at least for how I am using the amp.
It does what I wanted at very low volume so I'll probably pause here for a while. I tested it with my Vintage Modern combo, and in case anyone is interested, here is what happened:
After playing with it and deciding I liked it, I miced up the amp with a Rode M1, as if it was for a PA, then fed that via a neutral mixer into Audacity, to get some more analytical insight.
I set up a short riff in a looping pedal, and set the amp for my usual dirty-clean slightly crunchy tone.
Then I recorded, without attenuator at Volume 1, 3 and 7, then at 7 with the attenuator. Then I looked at the frequency response of each clip, after normalizing them in the software to the same max amplitude, and exporting to a spreadsheet. I also looked at the differences in response of each trace relative to un-attenuated at volume 7.
At Volume 1, (unattenuated), you can hear and see a definite harshness compared to higher volumes, in the form of a rises in treble above about 2 khz. THis is why I started looking at this.
At Volume 3 (unattenuated), the harshness is gone and the tone is the nearest of all tests to the Volume 7 tone - deviations at each frequency are small (within 2db, with no general trend with frequency) and it sounds good.
Volume 7 attenuated shows a very slight general reduction in treble > 2khz, maybe 1.5db max. it also sounds good, and this change is within range of the amps EQ controls.
Out of all that, I think that, for this amp, the attenuator is a great help to get out of the very low ranges to a level of about 2.5 or 3 on the Volume, useful for quiet practice. Above that level, the post-phase-invertor MV on this amp seems to do a great job, so the attenuator is not needed at least for how I am using the amp.
Keep in mind the article is written by someone with a vested commercial interest in promoting power scaling.
Sent from my phone. Please excuse any typpos.
For home use, I have had some success using a simple resistive 8 ohm or 8.2 ohm load on the amp output in place of a speaker, putting a resistive voltage divider across that to cut the signal down to instrument level (say 100 mV), and then running that signal through a graphic EQ pedal, and then into a (solid-state) small P.A. or powered speaker.
It's a somewhat convoluted route to output attenuation, but, particularly when low volumes are needed, I like it better than a simple resistive attenuator, which I've also tried.
I feel the graphic EQ plays a fairly key role here, as it lets you implement some of the frequency response shaping that otherwise occurs from the loudspeaker impedance curve, the loudspeaker acoustic response, and the speaker enclosure (cab).
I'm not convinced that the bass impedance peak at around 100 Hz is of any real significance, but I do find that shaping the midrange (to mimic the acoustic peak many guitar speakers have), and cutting the very high treble (to mimic the steep roll-off above 5-6 kHz most guitar speakers have) can help a lot in achieving a good sound.
-Gnobuddy
Resistive attenuators are electrically flat by themselves and can be designed so damping factor as seen by speaker matches the one seen by directly connecting it to amplifier; main problem is that human ear loses high and low frequencies *a lot* when listening at low volume, check Fletcher Munson curve to see by yourself how gross it actually is, so in practice simple attenuators tend to sound thin - buzzy - muddy .
Yet when recorded (with an "air" microphone) amp + speaker sounds fine at any volume.
Microphones do not suffer from Fletcher Munson loss.
Yet when recorded (with an "air" microphone) amp + speaker sounds fine at any volume.
Microphones do not suffer from Fletcher Munson loss.
Agreed, but our ears still do suffer from the Fletcher-Munson curves, when we listen to that low-level playback of the (microphone) recorded sound!
And yet, as you say, we can get very enjoyable recorded guitar sounds like this, played back at fairly low SPL.
So: it's possible to get good guitar tone at low volume, played back through a conventional flat-response speaker. The key seems to be that during the recording and mixdown, the recording engineers applied EQ, compression, reverb, etc, and used their ears to tweak everything in such a way that the final low-volume playback sounds good.
The question now is, how can we do the same thing for live sound, at similar living-room-friendly levels, without ever going through a microphone? I think the answer is to try to recreate the same process the studio did with their recorded guitar sound. Or, at least, the most essential parts of it.
To my ears, EQ is the single most essential part, which is why I follow the resistive attenuator with a graphic EQ. I've also tried adding reverb after the EQ, and that also improves the sound. I haven't tried adding a bit of compression as well, but I bet that would be a further small improvement, if the compressor is set tastefully.
We are still missing "cone cry" and other things that a good guitar speaker would otherwise have added to the sound. So it's not quite the same. And nothing will compensate for the emotional impact of a loud guitar amp - our brains react differently to loud sounds. Think kitten meow versus roaring lion.
But I do think the flat resistive attenuator / graphic EQ / electronic reverb / flat-response amp and speaker chain works quite a bit better than all my attempts to directly attenuate a guitar amp down to living-room levels.
-Gnobuddy
No, I have not. Three hundred and fifty buck (USD) is too rich for my blood ( BRAND NEW THD Hot Plate Power Attenuator 2.7-Ohms, Direct from THD Full Warranty | eBay ). I've built entire (valve) guitar amps for less than that. Heck, I've bought entire guitar amps for less than that (including the hybrid Fender Superchamp XD). There better be solid-gold components inside the Hot Plate to make it worth that kind of price for the simple passive device that it is.
What I have tried is use an L-pad between amp output and speaker cab. I also tried a couple of home-made attenuators made with fixed-value power resistors, and used in the same way.
My experience is that all of these work for small amounts of attenuation, but, just as everyone says, the more attenuation you use, the less good they sound.
I also think that L-pads and resistive attenuators work better for clean tones than for overdriven ones. Overdriven tones lose all their subtlety, and start to sound like a cheap fuzz-box once attenuated down to low volume. This probably has a lot to do with the Fletcher-Munson effect, as our ears start to filter out both the thunderous bass and the rich treble overtones, leaving only a thin, reedy-sounding buzz.
There is at least one expensive commercial "attenuator" that actually consists of a fixed load resistor, followed by a complimentary (PNP/NPN) pair of output transistors wired as emitter-followers, just like your typical transistor power amp. There is no gain stage, so voltage "gain" is just under unity. There is also a power supply in the box, obviously, to supply power to those output transistors.
You can bet there is a fair amount of crossover distortion in this device, as there is no negative feedback at all, and I don't think the biasing scheme for those output transistors was particularly sophisticated, either (i.e., they are biased a bit cold to avoid thermal tracking issues, so there will be excessive crossover distortion.)
I don't remember the brand name of this attenuator (might have been "Ultimate Attenuator", actually.) I think it was a small-volume "boutique" device. This thing was dissected and discussed on the Aussie Guitar Gearheads forum some years ago.
To use it, you run your valve guitar amp into the internal fixed load resistor, and connect your speaker cab to the output of the transistor emitter-follower. It's basically a resistive attenuator followed by a (crude) solid-state amplifier!
-Gnobuddy
Sorry, haven't tried those either. I just designed and built a push-pull 2-watt valve guitar amp instead!
Even that was still too loud for my apartment. So I added an L-pad, with the usual problems we've been discussing.
Then I modified the preamp to include a small-signal pentode as the last stage. Basically, the preamp itself is now a bit like a tiny, milliwatt-power, single-ended, output transformerless, guitar amp. It feeds the tone-control, then a master volume, then the thunderous two-watt valve "power" amp.
So now I can get pentode distortion at any volume I want, since it comes from the pentode in the preamp, rather than from the output pentodes. It's not perfect, and doesn't sound as good as a good 6V6 amp, but it's not too bad, either.
Next step (long delayed) is to add an FX loop between preamp and power amp. And once that's done, for all practical purposes, I'll back to the "guitar amp, resistive load, EQ, FX, flat-response power amp" concept I was using before with more powerful guitar amps!
-Gnobuddy
this isnt really the response you want, but i've spent ALOT of time and money on various attenuation schemes over the years, from simple lpads and the like to most major commercial offerings. Even to the point of having a Weber Mass custom built to my specs.
and i'll i've achieved is the knowledge that they all effect the sound quality to such a degree that I find them useless. most of my thousands of dollars worth of tube amps sit collecting dust, becasue of my living situation, i can't open up my big marshall like i could when i lived out in the country with no neighbors. 99% percent of my low volume playing is done through amp sim software now. it's not the perfect solution, but it meets my needs currently, without bothering anyone else. there are numerous cheap or free options available, if you're interested in checking them out.
now if i could just find a way to play my drums silently....
and i'll i've achieved is the knowledge that they all effect the sound quality to such a degree that I find them useless. most of my thousands of dollars worth of tube amps sit collecting dust, becasue of my living situation, i can't open up my big marshall like i could when i lived out in the country with no neighbors. 99% percent of my low volume playing is done through amp sim software now. it's not the perfect solution, but it meets my needs currently, without bothering anyone else. there are numerous cheap or free options available, if you're interested in checking them out.
now if i could just find a way to play my drums silently....
The comments and experiences above are very interesting. For myself, Im happy with the passive resistive design for what Im doing, I like the results I get and Id like to undestand deeper. Im not worried about the Fletcher-Munchen effects and the ideal is if a good 50W amp can be run with an attenuator as a good say 5W amp. I also think that guitar sounds are best heard through guitar speakers, even at low volume.
The effects of damping are interesting. An SS amp or a simple Lpad with large attenuation could add a lot of damping to a speaker response. That will reduce bass and treble, I think, for two different reasons. At the high end, the rise of inductive speaker impedance, combined with a higher amp output impedance, results in a higher voltage across the speaker than at lower frequencies, so lifting the highs relatively (ok so thats not a damping effect but an impedance/frequency/voltage divider effect). At bass frequencies, the free resonance has a peak, seen sharply in speaker impedance. If amp impedance is very low, this resonance is damped down greatly. But with higher imepedance, damping is less. The effect of less but still significant damping on response is to reduce the peak but also spread it wider. So a sharp peak in free resonance becomes a smaller general lift across bass in a hi z amp and is squashed flatter in a low-z amp or high -db attenuator.
But, coming back to resistive attenuators, using a 3rd resistor, we can achieve control of Z as seen by the speaker, and reduce damping, if this is what we want. But is this what we want in a guitar system?
My other question applies to Z as seen by the amp, and rekates to the load-box /active systems too. If the amp is then driving a pure resistive load, how differently does the amp output signal differ from when it's driving a real speaker? Are there effects there that cant be dealt with by EQ?
Aiken amps have a page showing a complete LRC network as a load-box, that creates impedance rise with frequency and also a low resonance:
Designing a Reactive Speaker Load Emulator
Maybe a factored version of that could be used to create a more complex Lpad that keeps all effects in scaled proportion while attenuating, sending a small % to the real speaker.
The effects of damping are interesting. An SS amp or a simple Lpad with large attenuation could add a lot of damping to a speaker response. That will reduce bass and treble, I think, for two different reasons. At the high end, the rise of inductive speaker impedance, combined with a higher amp output impedance, results in a higher voltage across the speaker than at lower frequencies, so lifting the highs relatively (ok so thats not a damping effect but an impedance/frequency/voltage divider effect). At bass frequencies, the free resonance has a peak, seen sharply in speaker impedance. If amp impedance is very low, this resonance is damped down greatly. But with higher imepedance, damping is less. The effect of less but still significant damping on response is to reduce the peak but also spread it wider. So a sharp peak in free resonance becomes a smaller general lift across bass in a hi z amp and is squashed flatter in a low-z amp or high -db attenuator.
But, coming back to resistive attenuators, using a 3rd resistor, we can achieve control of Z as seen by the speaker, and reduce damping, if this is what we want. But is this what we want in a guitar system?
My other question applies to Z as seen by the amp, and rekates to the load-box /active systems too. If the amp is then driving a pure resistive load, how differently does the amp output signal differ from when it's driving a real speaker? Are there effects there that cant be dealt with by EQ?
Aiken amps have a page showing a complete LRC network as a load-box, that creates impedance rise with frequency and also a low resonance:
Designing a Reactive Speaker Load Emulator
Maybe a factored version of that could be used to create a more complex Lpad that keeps all effects in scaled proportion while attenuating, sending a small % to the real speaker.
I have read this hypothesis (about the major role played by the guitar speaker fundamental resonance interacting with the amp's output impedance) many, many times. It's all over the Internet.
Personally, I'm quite skeptical, mainly for one quite simple reason: a typical electric guitar has a frequency range of somewhere around 45 semitones, while the speaker's resonant peak rarely spans more than a few semitones, and that too, usually around the 3rd-fret "G" on the 6th string.
So, unless you spend much of your guitar time playing six-string F#, G#, and A chords, I can't see how this can play a major role in the sound of a guitar amp! At best it could be a minor effect that makes a tiny handful of chord voicings sound "boomy" - not a particularly desirable outcome to my way of thinking!
I think the other factor (voice coil reactance rising with frequency) does play a role. It's easily simulated with a little treble boost in the graphic EQ pedal I follow the resistive attenuator with.
That question could probably be answered with some frequency/phase measurements on a guitar amp, with resistive and real speaker loads.
If you try it, I hope you discuss your results here. (I'm skeptical about it making a difference, but also curious to know if I'm wrong.
)I have tinkered with the post-resistive-attenuator graphic EQ, putting in a peak around 100 Hz, and never found that it mattered to me. But that's just me.
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.