Tube amp & electronics newbie here. Does a tube amp’s output impedance in itself have any effect on the sound, or on how hard the amp is working etc? Like, does 16 ohm amp + 16 ohm cab sound any different than 8 ohm amp + 8 ohm cab? Does the amp work any harder at higher impedance values? Any benefit to choosing 8 over 16, or the other way around?
I have been meaning to watch this for ages, but it may help…
1 Like
I can’t answer the nuances re: different ratings of matched amplifier and load, but I’ll give a reminder of why matching is important:
The amplifier is designed with the expectation that a load with a certain impedance will get attached to it. The amplifier applies a particular voltage to the load, and the amount of current that will flow will be inversely proportional to the impedance of the load.
If the impedance of the load is lower than what the amplifier is designed for, more current will flow through the amplifier than it was designed for, and the amplifier will likely be damaged.
If the impedance of the load is higher than what the amplifier is designed for, less current will flow through the amplifier than expected, but there will be no harm to the amplifier. All other things being equal, in the case where the impedance of the load is higher than expected, the sound from the speakers will be quieter than through speakers whose impedance matches the impedance of the amplifier. **** EDIT: This does NOT mean it’s safe to operate the amp with no load connected at all, it just means it’s safe to have a load with a higher impedance rating than the amplifier. ****
Sometimes novices will make the mistake of attaching lower impedance speakers to the amplifier, because they want them to sound louder, but the amplifier will end up getting damaged.
3 Likes
Good explanation @Frylock, thanks!
Actually the output section of my Road King has always kind of confused me since I don’t want to blow anything up. I know the 8-16 output is correct, since I have an 8 ohm cab and I was advised to use that one, but I was wondering about the 4 ohm options. (And no worries, I am using the shorting plug on the B side and I have all the channels set to use A side only.)
Each side has two 4 ohm outputs, so that’s probably the default way you’d connect two 8 ohm cabs, like it says down there. One to A1 and the other to A2.
But the “Use with two cabs” labeling is confusing to me. Is there a relationship between A1 and A2, like are they shared somehow? If I plug something into A1, do I have to take A2 into account? Is it safe to connect one 8 ohm cab to one of those, and then that just produces the high resistance low volume scenario, and nothing breaks?
Yeah, the Road King manual says that too. Apparently also produces less attack.
That’s my understanding, but please confirm with the manufacturer’s documentation rather than take my word for it. Per your photo, as I understand it, schematically, A1 and A2 are parallel branches that connect to the same place inside the amplifier. A 4ohm cab in parallel with an “infinite” cab nets to 4ohms, just as an 8ohm cab in parallel with an 8ohm cab nets to 4ohms. An 8ohm cab in parallel with an “infinite” cab nets to 8ohms, which would be a “safe mismatch”. Danger would be someone plugging a 4ohm cab into A1, and a 4ohm cab into A2: the net load of the two in parallel would be 2ohms, an UNSAFE mismatch (any parallel pair of cabs that nets less than 4ohms would be an UNSAFE mismatch, e.g. an 8ohm cab in parallel with any cab less than 8ohms).
Article below may help. Sorry if it’s a duplicate of the documentation you already have:
https://mesaboogie.zendesk.com/hc/en-us/articles/218267437-What-is-the-best-way-to-connect-speakers-or-cabinets-
The impedances of the amplifier and speakers do have an effect on the sound, but it’s hard to generalize too much, for several reasons.
First, speakers don’t really have single-valued impedances because impedances are by definition frequency-dependent. In other words, a ‘16 ohm’ greenback has an impedance vs frequency curve that looks something like this:
Notice that impedance is only 16 ohms at a few frequencies (~20 Hz, ~150Hz, and ~600Hz). That’s because the rated impedance value is an average over the range of frequencies in the graph. The lowest impedance on this curve is actually about 8 ohms, even though conventional wisdom would say never to match a higher impedance output with a lower impedance speaker. Contrary to popular belief, a little mismatch is not dangerous. It is in fact inevitable, as the graph demonstrates. That said, some amps are more robust to mismatch than others, though instances of damage in mismatch ratios of 1:2 are so rare I’ve never heard of one.
Every speaker type will have its own impedance curve, the shapes are generally pretty similar, but the specific values are highly variable. They won’t generally match each other except in broad terms:
Notice that there are no numbers on the vertical axis of this graph.
Second, the effect of impedance is probably subtle compared to things like cabinet construction and amplifier design. Speakers are generally chosen by amp designers to complement and enhance the sound intrinsic to the overall design of the amplifier. They are a fine-tuning factor, not a dominant one. E.g. If the amp has an overwhelming high end, you might choose speakers that minimize that. If it’s a bit muddy as designed, you might find speakers that shift the lowest impedance to a lower frequency to compensate. Etc. And these considerations can’t always be reduced to simple statements like ‘more highs’, or ‘less mud’. There are lots of frequencies to consider on both the amplifier and speaker side of the equation.
So overall, the answer to your question is, try it and see. You might be able to make some overly general statements about tone response, but you won’t know whether you’ll like it better or worse until you try it.
As for safety, my rules of thumb are these:
- Never operate a tube amp with a load impedance of more than double the amplifier output impedance.
Tube amps are actually in more danger when you overdo the speaker impedance than when you underdo it because of voltage reflection entering the amp from the speaker. Note that running a tube amp with no speaker attached is actually a load of nearly infinite impedance, which definitely violates the rule. I personally have left a 100 watt tube head running overnight with no speaker attached and had no problems, but I wouldn’t make it a habit.
- Never operate a solid state power amp with less than 4 ohms attached.
Solid state power amps will overheat if they provide too much current. They will generally be just fine with no speaker attached (infinite impedance means no current).
- Never operate any amplifier with less than 4 ohms attached. (Just to be safe).
The general idea here is that higher impedance is a lower load because less current is required to maintain a given voltage. With a higher impedance source, the difference in the amount of current between available for mid range (low speaker impedance=high current) vs bass and treble (high speaker impedance=low current) is more drastic than in the case of a lower source impedance. With zero source impedance, there would be no difference at all, and the response would be flat. This is actually nearly achievable in solid state power amps, and is why they are known for flat response. But guitars generally don’t sound good with flat response, and solid state amps are generally tuned to give ‘less ideal’ response that is more similar to tube amps and less suitable to hi-fidelity sound reproduction. The guitar amplifier is an effect in and of itself.
Some would say that the history of guitar amplification is generally one of getting fantastic sound with suboptimal engineering. The magic is often in the limitations. When those limitations are engineered away, so is the magic. Hi-fi sounds really bad in guitar amplifiers.
Sorry for the dissertation.
3 Likes
Thank you for your immensely informative post @induction! I really appreciate the diligence. Thank you.
I’m gonna need more experience with tube amps to really digest everything, but a few things came to my mind:
This is interesting, because by reading my amp’s manual and watching Youtube about amps in general I thought it was the other way around. So that in practice, the thing you should be really watching out for is too small of a load, not too high.
The Road King manual repeatedly reminds to always have the dummy shorting plug plugged in to avoid accidental transformer damage etc, and that damage will happen if there’s a no-load scenario. But it never exactly warns you about having too much load. In fact it says “You can always have a higher resistance (16 ohms, for example) without damaging results” (at page 34). I’m sure this is probably just a matter of emphasis and wording really, but at least it’s good to be aware of these things.
I think I’m missing out on some fundamental electronics knowledge… So connecting speakers increases the load (makes sense), but unplugging completely increases the load to infinite, instead of bringing it back down to zero? Sorry, like I said I’m very unfamiliar with how electronics work.
Loads can be counterintuitive because conceptually, the degree of loading is inversely proportional to the impedance of the load.
The output signal is a voltage. The amplifier has to supply enough electrons (or electron holes) at the output nodes to generate this voltage. When current flows through the speaker, those electrons don’t build up quite as easily, and the amp has to work harder. The lower the speaker impedance, the more current flows.
Think of it like trying to fill up a sink, and the output voltage is the level of the water in the sink. If the plug is in place (i.e. no speaker is attached), then even a small trickle of water will eventually fill the sink. But if the bottom of the sink is cut out (i.e. a short between the output nodes), no amount of water flow will raise the water level in the sink at all. A high impedance speaker (say 16 ohms) is like having the plug open a little bit, so the water can run more slowly to fill the sink. A low impedance speaker (say 4 ohms) is like having the plug much more open, but not all the way, so a stronger water flow would be necessary to fill the sink to the specified level. The output impedance of the amplifier or transformer is represented by how fast the water can come out of the tap, with lower source impedance correlating to a stronger water flow. The sink analogy is imperfect because the water only flows one direction (i.e. it doesn’t flow backwards into the tap), but it’s a decent starting place for your intuition.
The math behind this, if that’s your thing, is just Ohm’s Law: Voltage = Current * Impedance (aka V=I*Z). ‘Load’ specifically refers to the amount of current that the amplifier is required to generate to maintain the output voltage. It would more accurately be called ‘current load’. Since I = V/Z, as the speaker impedance drops to zero (i.e. the output is shorted directly to ground), the amplifier will try to generate an infinite amount of current, which is a heavy load indeed. When the speaker impedance is very high, it is easy to maintain the output voltage with very little current flow, because the electrons that are responsible for creating that output voltage can’t complete the circuit, so they just sit at the output nodes creating voltage until the signal reverses.
In light of the above, this should make more sense now, though some of what’s printed in the manual might be worded misleadingly. A no-load scenario is actually a higher resistance than 16 ohms, and is precisely what they are warning will lead to damaging results.
On a solid state amplifier with no output transformer, an unloaded amp is hardly working at all, so there is little risk of overheating, which is usually the relevant danger with most solid state amps. For a tube amp like the Road King, the output transformer provides some protection against the infinite current hazard. Instead the danger is that the signal current will not have anywhere to go (no load) and thus be reflected backward into the output transformer. The output transformer would then work in reverse, and generate large voltage spikes inside the power amplifier and possibly blow up your power tubes. The danger in both cases, is exacerbated by the existence of a signal. So if you just let the amp sit there in a hazardous setting, the risk is lower than if you are actively playing through the amp, especially at a high volume setting.
Note that the relevant distinction between amplifier types is actually whether there is an output transformer, not whether the amp is tubes or solid state. But in practice, most solid state amps don’t have output transformers (because they are expensive and solid state systems will work without them), and most tube amplifiers do have them (because the amplifier won’t work correctly without them).
Finally, it is possible to use protection diodes to reduce the potential for damaging voltage reflections at the output transformer. Whether those diodes are used or not depends on the design of the amplifier. Some amp manufacturers/models use them, some don’t.
As you can see, it’s easier to just give people simple rules to follow rather than write as much as I have to explain the details of loading. (Not to mention that I’ve just scratched the surface, and have simplified things to a degree that approaches inaccuracy. A competent electrical engineer would probably be tempted to punch me in the face, but a layman would have to put in years of study to understand their objections.) Bottom line: follow your amp manufacturer’s advice, even if you don’t fully understand all of the details. Their simplified explanations are often inaccurate in detail, but the cookbook-style operating advice in the manual is usually reliable.
Sorry for yet another super-long post.
4 Likes
No, thank you for yet another super-long post!
The sink analogy helped me a lot. Now I understand those perhaps misleadingly worded parts in the RK manual that you pointed out. Thanks.
2 Likes
I’ve seen the ‘reflected voltage’ thing quite a bit regarding this issue. Always struck me as a confusing way to understand the issue. A transformer is more/less two inductors coupled together. When a speaker is connected to the secondary, it’s kind of in parallel with those inductors, so the primary impedance will basically be the stepped up/down impedance of the speaker, because it’s smaller than the impedance of the transformer coils.
When no speaker is connected, no current is flowing through the secondary, so it has no (or at least a crazy small amount at audio frequencies) impact on the impedance of the primary. So you’re left with an inductor connecting the tube ‘output’ to the high side voltage. But a, like, huge inductor. The tube is trying to control the current through that inductor. The voltage across that inductor is proportional to how fast that current changes, and the bigger the inductor, the larger the voltage. If the amp is turned up and has signal going to it, that voltage can get really big, and something is probably going to fail, which may lead to a cascading failure of multiple components. It’s not terribly difficult to design in protection circuitry, depending on how much protection you need. Honestly, it baffles me that this isn’t just in every tube amp. Best practice is to not deviate from the owners manual.
The reflected voltage thing is kind of true, but then it’s also true for any inductor circuit, and it would be considered unusual to use that language. dI/dt is good enough.
ps, you put together a really nice summary here, well done.
1 Like
Off-topic: Protectors, good attenuators, DI outs, line outs, IR loaders. I’m really hoping that in the future all modern-styled tube amps could get these things as standard built-in features. Mesa’s Badlander is already showing some promise, and TubeMeister is another one.
Less OT, but I have an old Fender Stage 100 SS amp. The distortion is kinda funky but the clean is damn good. I was doing some work on it and got the schematic for it. I seemed to have combination voltage and current feedback scheme, which I always figured was to get matched output impedance with the speaker. It also had a DI out that was a stepped down version of the speaker signal. The DI signal is hot garbage, but the combo sounded great. I came to the conclusion that the frequency response of the speaker/amp combo is key. Go figure…
Were you using a cabinet simulator with the DI signal?
I was not, and I’m sure that would help. It’s a bit of a crux-y problem though. I could take an IR of that speaker and I’m pretty sure it would sound dead on. I’m sure a different IR would still sound good, but it wouldn’t sound the way it would if it were plugged into the amp, since a different speaker would result in a different DI signal if the impedances weren’t the same between speakers. Same goes for a cab sim, probably would sound good, just interesting that without a physical load it’s difficult to characterize/model what it would actually do as you go between components.
I think I understand what you mean. For dummy loading, physical modeling is a reasonable approach since the free-air response of a speaker in free can be pretty well-modeled with a sufficiently complex LRC circuit that can be replicated with physical LRCs or software models.
But for cabinet simulation, physical modeling is much more complex since it needs to take a multi-physics approach, including both electronics and acoustic modeling (at a minimum). There are effects of air pressure, spider stiffness, physical deviation limits of the speaker, how the speaker is mounted, and so on. Speaker distortion would probably be very difficult to physically model, not to mention things like Fletcher-Munson effects.
The most noticeable effect of a guitar speaker cabinet is to remove the high and low frequency content that guitar speakers can’t physically reproduce. For specifically this reason, the effect of cabinet simulation drastically outweighs the effects of impedance simulation on the resulting tone. Without cabinet simulation, DI signals sound straight-up awful. (‘Hot garbage’ is exactly right.)
I approach cab sims the same way as dummy loading: build a custom EQ that captures the effect and tune it by ear. IRs are also great if you’ll be working in a DAW, especially if you want to capture the sound of a specific cabinet (with specific mike placement in a specific room, etc.). I’ve never seen cabsims done through physical modeling, but I’d be surprised if it’s never been attempted.
TBH, the most distressing thing is that there is still no consensus on what model corresponds to ‘good’ tone. I spend a lot of time thinking about what would happen if electric guitar were invented today. Super linear pre amps and power amps, flat speaker response. I don’t think anybody would bother playing. I guess we can feel fortunate that even when it seems overwhelming to try and find your tone, you can always listen to the music of the past and know that there is a way to do it right, even when it’s not immediately obvious what that is.
Necro-thread I know, but it’s more likely Fender’s addition of current feedback to that output stage was to do the opposite, and attempt to “tubify” it by undamping the speaker a bit by raising the output impedance of the ss poweramp. They weren’t likely trying to match anything. To give a dumbed downed idea why, Tubes in general are inherently high impedance devices especially pentodes and tetrodes/beam tetrodes used for the typical guitar amp output stages. Even with an output transformer and quite a bit of local negative feedback applied back to the input of the poweramp, they still have a very hard time driving a reactive element like a loud speaker, particularly in the areas of lower impedance, and have an easier time where impedance is higher like (using the plots above) in the areas of speaker resonance and rising voice coil impedance. So much so, the load a loudspeaker presents to a valve output stage is very elliptical even with a great deal of -fb (voltage feedback) applied. The amount of negative feedback typically used in most guitar amps is usually not enough to completely dampen the effect of that, but that interaction is part of what makes a tube output stage sound and react like a tube output stage anyway, and a very big part of what most people like about them.
Most solid state devices by contrast, are intrinsically very capable of driving very low impedances on their own, so much so, one of the most expensive elements in an amplifier can be eliminated. However because of this, they typically also have a much easier time with something that presents a variable load ie driving a reactive element like a speaker, which subjectively may lead adjectives from the guitar community like “sterile” “plain” “hi-fi” etc… (note these are actually desirable attributes in other communities). Adding some current feedback to the mix or selectively doing so, is likely an attempt to change this interaction and make it a bit more tube like or at the least more guitar friendly.
This isn’t exactly true. The rated impedance value of the loud speaker isn’t the average of what’s there, It’s the value of what is always expected to be there, which is termed the Nominal impedance. Also this is more nitpicking, but the speaker with the nominal 16 ohm impedance in your graph above also doesn’t dip down to 8 ohm, it only slightly dips down to around 14R or so, in the 300hz-400hz region, This is where you typically find the nominal impedance to be in a lot of loud speakers used for MI. Note that 14-15R value could be the considered the true nominal impedance of this speaker, but manufacturers will often quote the 16R instead especially in the MI world to avoid misunderstanding and confusion regarding what that really means or implies. Celestion used to quote the true nominal impedance years and years ago.
This is generally true with one big caveat - it’s not so cut and dry as just taking the reactive aspect of speaker into account, and the term “mismatch” is very misleading, you can’t just define that with speaker load alone. You also do have to know what the other working parameters are in the amplifier to truly be able to define what that is. The reflected primary z (turns ratio) of the ot in relation to the other operating parameters such as plate and screen voltage come into play here, and there is a wide grey area of what is truly considered a mismatch.
Again this mostly true, but very generalized without the context as to why, but it does get the point across and most people on this site likely won’t need to or want to dig deeper.
Well yes and no. The impedance of a speaker should not be confused with the frequency response of the speaker by any means, and the cabinet construction etc… actually has an effect on what that impedance curve will inevitably look like particularly at resonance. Also the impedance curve of the speaker/cabinet is really important in how certain tone controls in the amplifier will react, particularly those that work by decoupling the local -fb around the power amplifier like the typical presence or resonance/depth control. Those controls only work the way they do because of those non linearities in speaker impedance curve and how they reflect back. If you were to drive a purely resistive load, those controls would stop working altogether, this is also one of the very reasons why people claim resistive attenuators and loads sound flat. It’s also the reasons controls like that will react somewhat differently to different speaker cabs.
I would disagree here and say they are one of the biggest. They are the final shaping tool you have , and a change of cab can be quite profound. They are also one of the biggest that are user interchangeable.
For the typical user this generally considered okay, but there may be some exceptions depending on some of those other parameters mentioned above. So for the typical user, if there ever is a question as to wether or not to do this or if it’s okay to do, and no manufacturer clarification available, just don’t.
So this is mostly true, however needs a bit more explanation. typically what happens in this scenario isn’t tube failure or over dissipation of the output tubes, that tends to be the case in a “closed” load scenario. In the case of too large of a load, or open load the typical failure is that of insulation puncture of the OT windings due to a large inductive flyback voltage that can get generated across them in this scenario.
With a closed load, you will get tube stress and failure due to over-dissipation of the output tubes, as this closed load makes the load line slope look vertical, and sure to breach the maximum dissipation curve of the valves even at idle, but this is the more preferred of the two, as it’s easier and cheaper to replace and fix, and the hope is that you will notice it before plates of the tubes melt.
So this a bit problematic because you first have to define what this mismatch is and what makes it a mismatch. In reality there is no perfect reflected load for a tube output stage. The general consensus since the dawn of tube output stage design is that there is a load that is best for power transfer, and one that best for lowest distortion figures and the two are often incongruent with each other so there is often a compromise associated here. Guitar amps in general are notoriously badly designed though and are horrible audio systems by audio and engineering standards, but that’s part of what makes them sound neat. One famous example of this, is marshall using an OT for most of its amplifiers that will reflect back an A-A load of 1.7k when it’s corresponding secondary taps are “appropriately” loaded. This is a bit “low” for EL34’s running at the typical 470V plate and screen supply marshall typically uses in their higher wattage amps. You could even say that this in itself is somewhat of a mismatch by design, even when the taps are “correctly” loaded.
Blockquote
Some would say that the history of guitar amplification is generally one of getting fantastic sound with suboptimal engineering. The magic is often in the limitations. When those limitations are engineered away, so is the magic. Hi-fi sounds really bad in guitar amplifiers. Sorry for the dissertation.
This pretty much sums it up.