Hello everybody,
I read in an old google groups thread that it is possible to use, say, 2 taps from the OT at the same time (ask if you want the link). For instance, a OT with 16 and 8 ohm taps can be safely and correctly used if a 32 ohm speaker is connected to the 16 ohm tap, and 16 ohm speaker connected to the 8 ohm tap.
I'm not 100% convinced that it is safe and correct because of some qualifying remarks in the thread but generally it seems to suggest it is ok. So, is it safe and correct usage?
If yes, is it possible to implement a simple attenuator by replacing the 32 ohm speaker in the above example with a, say, 50w 50 ohm rheostat set at approx 40 ohms? This would be a simple way to get attenuation but would it also avoid all the tonal changes (ie "tone suck" but not the SPL psychoacoustics) associated with attenuators?
Thanks in advance for any all comments.
I read in an old google groups thread that it is possible to use, say, 2 taps from the OT at the same time (ask if you want the link). For instance, a OT with 16 and 8 ohm taps can be safely and correctly used if a 32 ohm speaker is connected to the 16 ohm tap, and 16 ohm speaker connected to the 8 ohm tap.
I'm not 100% convinced that it is safe and correct because of some qualifying remarks in the thread but generally it seems to suggest it is ok. So, is it safe and correct usage?
If yes, is it possible to implement a simple attenuator by replacing the 32 ohm speaker in the above example with a, say, 50w 50 ohm rheostat set at approx 40 ohms? This would be a simple way to get attenuation but would it also avoid all the tonal changes (ie "tone suck" but not the SPL psychoacoustics) associated with attenuators?
Thanks in advance for any all comments.
1) such connection can* be done but I regard it as clumsy, and in any case, where would you get a 32 ohm guitar speaker?
Only known ones are Ampeg 32 ohm 10" and those are for Bass, very different sound.
2) you would get very little attenuation because all you do is replace one speaker with a resistor, leaving the other run full throttle.
Yes, it is safe, but you would definitely get tone suck because you are killing the varying impedance present in any Guitar speaker, which helps get some tasty EQ when fed from a highish impedance Guitar amp (tube or mixed feedback SS) by paralleling it with a flat impedance resistor.
I suggest if you want reduced power, you use one tap, and connect one speaker there, through an attenuator.
You will be able to fully throttle sound down to 0 or very low "bedroom" values if you wish, and some sound quite good.
Only known ones are Ampeg 32 ohm 10" and those are for Bass, very different sound.
2) you would get very little attenuation because all you do is replace one speaker with a resistor, leaving the other run full throttle.
Yes, it is safe, but you would definitely get tone suck because you are killing the varying impedance present in any Guitar speaker, which helps get some tasty EQ when fed from a highish impedance Guitar amp (tube or mixed feedback SS) by paralleling it with a flat impedance resistor.
I suggest if you want reduced power, you use one tap, and connect one speaker there, through an attenuator.
You will be able to fully throttle sound down to 0 or very low "bedroom" values if you wish, and some sound quite good.
In other words, no matter what I do to the resistor (eg. add voice coils/inductors) to make it appear as having dynamic response (reactance hence non-flat impedance), electrically I'm just constructing an attenuator and it would be equivalent to the 1 tap -> attenuator -> speaker configuration you suggest., correct?
Thanks for the reply.
You didn´t mention this before.
And "adding voice coils/inductors" is not "doing something to a resistor" but building a different circuit.
You only mentioned "a rheostat" which is a "plain" resistor, only adjustable.
Again:
1) the proper way to attenute is amp > attenuator > speaker
2) there are simple "boring" attenuators, very good ones, and anything in between,you choose.
3) in my view, others may disagree, the best reactive load is Randall Aiken´s , google it.
Based on that, adding a "regular" resistive attenuator, you can build a device which both provides a realistic load, and attenuates.
Best of both Worlds.
I've looked around and there are some implementations that resemble Aiken's circuit. For instance, J. Hewitt (user johnh on some sites) has a reactive load attenuator where his initial designs had 2 inductors but not the bipolar cap that Aiken's has. Does the omission of the cap make it significantly different? If so, could I use his circuit (https://www.marshallforum.com/threads/simple-attenuators-design-and-testing.98285/) and replace the front end with Aiken's circuit?
I imagine something like taking Hewitt's stages 2, 3, and 4 (for his 16ohm version since Aikens' is 16 ohm) and tacking them on after Aiken's R3 (68 ohm 5W). I assume that the resistor values of the resistive part can stay the same even though the inductor values in the reactive part are significantly different, especially L2 (1.1 mH vs 50 mH; Aiken's site has a typo where L2 is also marked as L1).
I assume that the bipolar cap does make the Aiken's circuit different/better but at the moment, my "ghetto" sensibilities tell me that a 1.1mH inductor is way cheaper than a 50mH one. Hewitt's M2 version will likely be cheaper still with only 1 inductor. "Bedroom" level, no one's recording career depends on this 🙂
Thanks for the pointers.
Aiken´s is better, period, because it emulates not only the rising mids-highs impedance (easier, only 1 small inductor) but the much harder (so almost nobody else does) low frequency resonance peak, which requires a large cap and a large inductor.
Note: don´t just "mention" "John XXXX´s attenuator" or send me to a 170 page thread, YOU find the relevant schematics and post them here, all together.
Thanks.
Note: don´t just "mention" "John XXXX´s attenuator" or send me to a 170 page thread, YOU find the relevant schematics and post them here, all together.
Thanks.
Are you expecting some sort of night and day difference just by changing a cap value or leaving it out?
The 1mH is the rise above 1kHz.
The 50mH with the hundreds-uFd cap is the Bass Resonance.
Both are probably important?
Maybe the 50mH more so because you can fake the treble rise with studio tone control, but few knobs emulate a speaker's bass resonance well.
I couldn't form any expectation actually. I didn't know what role the capacitor in Aikens circuit played. JMFahey points out that its for the low frequency resonance peak speaker simulation. The (large) capacitor and large inductor is what other reactive load implementations typically leave out, and judging from JMFahey's reply, it contributes a lot to the realism of the simulation.
I'm proceeding based on that information. As well, J Hewitt's latest version has returned to two inductors and added a large cap, it seems that he implicitly recognizes the low frequency resonance peaks significance. So at this point, what I'm trying to figure out is 1) if J Hewitt's latest reactive load attenuator (M3) is suitability close to an Aiken's reactive load circuit with a resistive attenuator tacked on, and 2) if so, can I possibly cheap out a bit by lowering inductor L2.
Earlier today (while looking at this thread), I realized that it's possible to distribute power between two loads (say, a good quality reactive dummy load and a loudspeaker) in any desired proportion using an autotransformer. Has anyone ever tried this? I worked out the math and whipped up an LTspice simulation. It seems to work exactly as I expected. In the ideal case, with identical loads, the impedance presented to the amplifier is identical at any attenuation setting.
That's what puzzled me. Hewitt's first versions has 2 inductors (no cap) but the largest inductor value is 1.1mH. I mistakenly compared apples to oranges, thinking that it was similar to Aiken's (2 inductors) circuit just because of the number of inductors.
So looking at Hewitt's M3 version, the inductor values are still no higher that 2mH. He's not designing for low frequency despite the inclusion of the cap.
Guess I'll have to look else where for a easy solution.
Note: in post #10, question #1 about Hewitt's circuit's similarity to Aiken's has been determined. After PRR's post pointing out the significance of the values of the inductors, it is clear that Hewitt is not concerned with low frequency.
Last edited:
I guess the much maligned Marshall Power Brake used the basic approach I described. A lot of people seem to think the Power Brake sounds terrible, yet in comparisons I've found demonstrating the (normalized) sound at different attenuation levels, it sounds pretty neutral. Hmm....
What PRR said.
It´s not just number of inductors alone, but only Aiken (that I know of) loads amp the same way as a speaker does, including the very important Bass resonance peak.
Math 101: you NEED a large (expensive) inductor and a large capacitor there simply because you are getting (not "simulating" but actually "getting", the real deal) a low frequency resonance peak, as what a real speaker provides, and that at speaker impedance level.
LC (inductor capacitor) filters are cheap(ish) to get at higher impedances, as in old style (Mesa Boogie or Acoustic) graphic equalizers , those use pinky nail sized inductors wound out of hair thin wire around tiny cores, handling milliampere level current, milliwatt power, but when we are speaking about 50-100 watt, full amperes of current, 8 or 16 ohm impedance, we need a big fat coil wound around a large core: the perfect recipe for $$$$$$
And it takes what it takes to get that.
As of the multi-tapped transformer Marshall attenuator, it works fine, it "reflects" speaker impedance towards input, but that transformer is expensive too.
I commercially wind transformers and yet when I saw the myriad taps they use, my first reaction was WTF????
To put some numbers into it, and from https://goodcalculators.com/resonant-frequency-calculator/
for 50mH and 100uF we get:
71Hz is even lower than the lowest frequency produced by a Guitar (which is typically around 82 Hz), I guess it was chosen to cover even dropped tuning ones, and give a deep chest thumping sound in spades.
In a nutshell: it is cheap and easy to simulate/emulate curves and response using Op Amps, and that´s what most "Compensated Line Out" circuits do, but when you want to get (not simulate) the interaction between an overdriven Tube amp and the speaker it´s driving ... there is only one way.
It´s not just number of inductors alone, but only Aiken (that I know of) loads amp the same way as a speaker does, including the very important Bass resonance peak.
Math 101: you NEED a large (expensive) inductor and a large capacitor there simply because you are getting (not "simulating" but actually "getting", the real deal) a low frequency resonance peak, as what a real speaker provides, and that at speaker impedance level.
LC (inductor capacitor) filters are cheap(ish) to get at higher impedances, as in old style (Mesa Boogie or Acoustic) graphic equalizers , those use pinky nail sized inductors wound out of hair thin wire around tiny cores, handling milliampere level current, milliwatt power, but when we are speaking about 50-100 watt, full amperes of current, 8 or 16 ohm impedance, we need a big fat coil wound around a large core: the perfect recipe for $$$$$$
And it takes what it takes to get that.
As of the multi-tapped transformer Marshall attenuator, it works fine, it "reflects" speaker impedance towards input, but that transformer is expensive too.
I commercially wind transformers and yet when I saw the myriad taps they use, my first reaction was WTF????
To put some numbers into it, and from https://goodcalculators.com/resonant-frequency-calculator/
for 50mH and 100uF we get:
71Hz is even lower than the lowest frequency produced by a Guitar (which is typically around 82 Hz), I guess it was chosen to cover even dropped tuning ones, and give a deep chest thumping sound in spades.
In a nutshell: it is cheap and easy to simulate/emulate curves and response using Op Amps, and that´s what most "Compensated Line Out" circuits do, but when you want to get (not simulate) the interaction between an overdriven Tube amp and the speaker it´s driving ... there is only one way.
A note about J. Hewitt's M3 version. I missed the details noted in his schematic. L2 is larger the 2mH (I looked too quickly at the lower charts), it's actually 18mH for the 16 ohm version. Much bigger than his initial versions, but not close to Aiken's 50. I'm sure he has his design goals. No disrespect intended.
There are a few others: Suhr Reactive Load, Fryette Power Station/Load, and Fractal Audio X-load. There's some data here and a few more curves here.
So I guess my autotransformer idea would probably be prohibitively expensive to make.
I just noticed this thread, which has some discussion about my attenuator designs. So I thought I'd join in.
Aiken is a key reference and I learnt a lot from his designs. His circuit is a load box, from which you take a signal for active reamping. It needs to replicate both the treble rise and the bass resonance if it is to be accurate. But to do that and then follow through to a passive output instead of active while also maintaining tone needs some further insights. I found that the key was to put the reactive impedance partly in series and partly in parallel with the load. It's also important to maintain not only accurate impedance as seen by the amp, but also output impedance as seen by the speaker. Most passive designs don't do this.
I have 3 designs that are referred to in this tgread:
M, with two coils, together replicating just the treble rise. Two coils are used so the inductance can be part in series and part in parallel to ongoing resistive stages.
M2. With a single treble coil, now feeding a delta configuration to get the series-parallel arrangement. The performance is virtually the same as M, but the maths to design it was a tad trickier.
Neither M or M2 creates a specific bass resonance for the amp to see. But, because the attenuator has a high but controlled output impedance, the resonance is created by the real speaker. This is different to the typical reamping unit, where the speaker is fed by a low impedance active stage, so is highly damped and all tone must be generated before.
Many dozens of M2 attenuators have now been built and are well liked. Recorded and listening tests have confirmed that the absence of an explicit bass resonance circuit makes negligible difference. The speaker makes it itself.
But, the amp doesn't see it, so in principle, there can be certain overdriven bass harmonics that are not generated, even though the basic tone is good.
If should be noted that bass resonance is only significant if very low notes are played, around low A. Its not nearly as important amp to tone as the treble rise.
But, I wanted to explore how bass resonance can be added to my design. Once I'd gone from M with two coils, to M2 with just 1, I could add a bass resonance LC branch and get that built in. This is design M3. I haven't built it myself, but a few others have done so. The concesus is that the considerable extra bulk and expense of the additional big cap and inductor is not really of benefit when driving a guitar speaker, but it may he so if the unit is used as a load box instead, more like Aikens and Suhrs units.
Our thread:
https://www.marshallforum.com/threads/simple-attenuators-design-and-testing.98285/
Its now a huge thread but Post 1 has most of the key info
Aiken is a key reference and I learnt a lot from his designs. His circuit is a load box, from which you take a signal for active reamping. It needs to replicate both the treble rise and the bass resonance if it is to be accurate. But to do that and then follow through to a passive output instead of active while also maintaining tone needs some further insights. I found that the key was to put the reactive impedance partly in series and partly in parallel with the load. It's also important to maintain not only accurate impedance as seen by the amp, but also output impedance as seen by the speaker. Most passive designs don't do this.
I have 3 designs that are referred to in this tgread:
M, with two coils, together replicating just the treble rise. Two coils are used so the inductance can be part in series and part in parallel to ongoing resistive stages.
M2. With a single treble coil, now feeding a delta configuration to get the series-parallel arrangement. The performance is virtually the same as M, but the maths to design it was a tad trickier.
Neither M or M2 creates a specific bass resonance for the amp to see. But, because the attenuator has a high but controlled output impedance, the resonance is created by the real speaker. This is different to the typical reamping unit, where the speaker is fed by a low impedance active stage, so is highly damped and all tone must be generated before.
Many dozens of M2 attenuators have now been built and are well liked. Recorded and listening tests have confirmed that the absence of an explicit bass resonance circuit makes negligible difference. The speaker makes it itself.
But, the amp doesn't see it, so in principle, there can be certain overdriven bass harmonics that are not generated, even though the basic tone is good.
If should be noted that bass resonance is only significant if very low notes are played, around low A. Its not nearly as important amp to tone as the treble rise.
But, I wanted to explore how bass resonance can be added to my design. Once I'd gone from M with two coils, to M2 with just 1, I could add a bass resonance LC branch and get that built in. This is design M3. I haven't built it myself, but a few others have done so. The concesus is that the considerable extra bulk and expense of the additional big cap and inductor is not really of benefit when driving a guitar speaker, but it may he so if the unit is used as a load box instead, more like Aikens and Suhrs units.
Our thread:
https://www.marshallforum.com/threads/simple-attenuators-design-and-testing.98285/
Its now a huge thread but Post 1 has most of the key info
A belated thanks to JohnDH for the original plans and to all the others for their comments. Since my last post, I have been experimenting with non-attenuator solutions. I should note that I was leaning towards an M2 based design but since I won't be performing live, the easiest (and adequate so far) solution has been to build an isolation box and mic the sound to an audio interface into my stereo system.