5-6 dB is the worst case reported in the Sound on Sound article I link to in the Crest Factor section of my Taming the LM3886 article series.
From the data above, it looks like 27-30 dB is a probable best case, though I would never design an amp for that. My recommendations are based on the average CF of 14 dB reported by SOS. I also design for a max heat sink temperature of 60 ºC with an ambient temperature of 25 ºC. That means that a design which is right up against the limit will get quite hot (but survive) with sine wave operation. It may trip the thermal protection in the LM3886 after an hour of operation with a continuous sine wave into a 4 Ω load at the worst case dissipation, but will barely get lukewarm in operation with a music signal at normally loud critical listening levels with speaker loads and music signal.
I'd be interested in the source for the 1:20 recommendation as well. I'd also be interested in the perceived effects of a lower DF (scientific results, not subjective so-and-so-on-the-internet-said opinion).
Tom
I'd never design thermals for best case CF, but it's useful for one of the upper bounds on how much power you need, even if the number of albums that require it are small. Telarc 1812 has a (badly clipped) 38Hz fundamental on the cannon shot on the CD version. And it's interesting to have a collection of oddities in one's music library 
Of course not, but when the cannon blast on the telarc 1812 is 26dB* above the orchestra if you have the orchestra at 75dBa then you need to have enough power for approx 101dBa* from whatever produces your bass at 38Hz. Assuming 85dB/W and nearfield listening that's 40W. BUT if you have the orchestra at 85dBa, which some do although I would suggest a tad louder than I listen you need 400W. Suddenly a plurality of powered subs seems a good idea! (or listen at more sensible levels).
Not realistic cannons, but for a hifi torture track does the job. And a data point on how much power you need, even if sizing your system for 90 seconds of music is a level of daft beyond that which most audiophiles will go
*There is evidence for the CD release that they wimped out and cut 6dB from the cannons. I need to prove this.
**arm wavy calculation well aware that there are several places where reality could tip things 3dB one way or the other.
Not realistic cannons, but for a hifi torture track does the job. And a data point on how much power you need, even if sizing your system for 90 seconds of music is a level of daft beyond that which most audiophiles will go
*There is evidence for the CD release that they wimped out and cut 6dB from the cannons. I need to prove this.
**arm wavy calculation well aware that there are several places where reality could tip things 3dB one way or the other.
There are much more related with good cannon sound reproduction, since the cannons actual have a quite wide spectrum, the phase/group delay is important and much more difficult to achieve in a small room. The size of the drivers to produce the SPL necessary is also critical, the Linkwitz site has some quick calculations of what surface area that might be necessary to accomplish this. The driver data I have seen on Klippel measurements suggest an 18” driver reach max excursion around 25~30 volts, another consideration when working with high SPL low frequency reproduction.
luckily there is only one classical piece that calls for cannons and I have several recordings that use real cannons. It's an edge case and not one I would optimise my personal system for. I'm only allowed one 18" black cube in the living room so I am limited to around 100dBa peaks in the bass. That will do me and envelope says that, in combination with the apogees running actively off mod-86 I am good for all but wild parties, which I don't have
Attachments
Yes and no. It's from some discussions I read a few years ago, when I was interested in the subject. So it might have been a magazine interview with someone who was designing amps in the 60's and 70's, maybe a Rane Note, there were a few other places I dug up and read.
I did a search of my current HD, nothing there (well, some, but related to turntables). If I remember, next time I read some of my archived drives, maybe I saved something.
The more interesting part about DF was how it became an issue in the first place. In the early and mid 1960's, when solid state amps began to enter the HiFi market (versus the "Teen Beach Party" FM/AM radio market), naturally there was a competitive nature to the marketing, and SS amps had these very low output impedances. So it was proposed as an inherent quality advantage over transformer-coupled vacuum and SS (the McIntosh patent) amps; the ad copy of the day touted it as a feature.
Somewhere in all that (not withstanding the ad copy) it seemed there was a consensus amongst those who dismissed the high DF argument that "a DF of 20 was adequate". I remember those exact words. A lot of the reasoning behind that is the combined impedances of everything after the binding posts, making the SS amps DF a little less impressive.
I also took note of the "lunatic fringe" car audio guys, running 2 and sometimes 1 ohm bass drivers, and no-one is complaining about lack of woofer control, at least amongst the Sound Quality (SQ) guys (there are a lot of bad car audio installations out there). In that crowd they almost without exception prefer high power and ported LF drivers, so you have to wonder what all the fuss is about. Plus you see ads for car audio amps rearing the DF "advantage" once more, which I find suspicious since these ads are not touting the "best sounding" power amps according to the car audio guys. That's a weird market, to be sure.
My own personal experience led me to the conclusion that high power did a better job of controlling the woofer's back-EMF than a high damping factor seemed to. I'm just lucky I was in a position to be exposed to so much gear that my suspicion was raised. For example, when I first heard the Bryston 4B, although it's a flawed amplifier, especially on the top end, it has stunning bass control.
One of those things where the audio reviewers might start talking about "huge differences". And on almost anything you threw at it ... big woofers, dual big woofers in huge boxes, horn loaded woofers, electrostatics. There is a lot of variation in impedances there. But they would measure out to 250wpc or 480w or therabouts in mono, yet the output impedance wasn't different than other amps using sand, and that was a time when people were just starting to talk about cable, so the combined impedances probably weren't anything to rave about.
It's not a subject I really want to go back to, unless I come across new information that challenges my current position, which is I'm buying what the DF:20 guys are selling, for now at least. Maybe Toole is right, and DF:10 is the magic number, but 20 isn't too hard to achieve with SS, and I also own a transformer-coupled vacuum state power amp (Anthem Amp1; 40wpc, EL34's, and it sounds best on the 4 ohm taps) and I see no reason to fault it on woofer control.
Last edited:
Correction:
Bryston 4B (1970's version) 200wpc 8 ohms stereo, 800 wpc 8 ohms bridged into mono. And we used them in mono for a bunch of DIY subwoofers, plus a lot of full-range speakers. Bryston warned us about using them in mono with 4 ohm loads, but we never had problems with things like big Infinity's, three-panel Magnepans, Dayton-Wright XG8's, and the usual conventional woofers of the day. I can't remember breaking even a single one, and I can't say that for a lot of highly touted amps of the day on some of those loads.
Maybe Tom can chime in (or anyone else who knows) but is the output impedance halved or doubled when you bridge the output transistors?
Bryston 4B (1970's version) 200wpc 8 ohms stereo, 800 wpc 8 ohms bridged into mono. And we used them in mono for a bunch of DIY subwoofers, plus a lot of full-range speakers. Bryston warned us about using them in mono with 4 ohm loads, but we never had problems with things like big Infinity's, three-panel Magnepans, Dayton-Wright XG8's, and the usual conventional woofers of the day. I can't remember breaking even a single one, and I can't say that for a lot of highly touted amps of the day on some of those loads.
Maybe Tom can chime in (or anyone else who knows) but is the output impedance halved or doubled when you bridge the output transistors?
Last edited:
There is of course this on damping Impedance, and how it affects audio equipment . For me it's a system consideration.
Hi Johnny2Bad,
The amplifier will see the load resistance as 1/2 it's actual value in bridged mode. The damping factor rises to 1/2 it's normal value.
I can't recommend the practice of bridging amplifiers. To many negatives compared to a little positive. I think you would be further ahead running each channel into it's own speaker.
The old 4B was a piece of junk. They were right in telling you not to attempt running them in bridged mode. THe fact you didn't have any problems says one of two things.
Either you didn't need to put them in bridged mode, or
You were so very, very lucky.
My advice to anyone today is the same. Don't do it.
-Chris
The amplifier will see the load resistance as 1/2 it's actual value in bridged mode. The damping factor rises to 1/2 it's normal value.
I can't recommend the practice of bridging amplifiers. To many negatives compared to a little positive. I think you would be further ahead running each channel into it's own speaker.
The old 4B was a piece of junk. They were right in telling you not to attempt running them in bridged mode. THe fact you didn't have any problems says one of two things.
Either you didn't need to put them in bridged mode, or
You were so very, very lucky.
My advice to anyone today is the same. Don't do it.
-Chris
Last edited:
The maximum output of a pair of amplifiers in bridged format is exactly double the power into double the load impedance of one single amplifier.
i.e. two 200W into 4ohms amplifiers will give 400W into 8ohms.
EXACTLY the same total power, nothing gained and nothing lost. No magic !
Hi Andrew,
Theoretical power increase is 4X the normal power of one channel. Practically speaking, you may see 3X-ish, and that depends on current limiting.
Why don't you see the full 4X? Losses due to supply voltage droop and double the I_R losses in each amplifier. It would be similar to going from an 8R load to a 4R load, you never seem to get exactly 2X the output power.
-Chris
Theoretical power increase is 4X the normal power of one channel. Practically speaking, you may see 3X-ish, and that depends on current limiting.
Why don't you see the full 4X? Losses due to supply voltage droop and double the I_R losses in each amplifier. It would be similar to going from an 8R load to a 4R load, you never seem to get exactly 2X the output power.
-Chris
Bryston should warn about using their amps bridged with low-impedance loads. In a bridge, each amp half "sees" half the load impedance, so each amp half will be driving a 2 Ω load if the speaker is 4 Ω.
A 200W/8Ω amp will need to produce 56 V peak to provide the needed power into the 8 Ω load. This will result in a load current of 56/8 = 7 A. If you connect two such amps in a bridge, you will get a total of 2*56 = 112 V peak swing across the load. With a 8 Ω load, that's 112/8 = 14 A, i.e. twice the current, hence the notion that each amp half "sees" half the load impedance. With a 4 Ω load, this load current doubles again to 28 A. So now you'll have an amp that's designed for 7 A output current being commanded to deliver 28 A. That might not go well.
Just because you can do it a few times and get away with it, doesn't mean it's a good idea or good design. The laws of physics are pretty unforgiving to lawbreakers...
I suggest drawing a schematic or block diagram that includes the amp and its output impedance. You'll quickly see that in a bridge, the output impedance doubles, thus the damping factor is halved.
That's another way of wording it.
For commercial designs, I can't recommend the practice either. The only exception is if the amp is specifically designed for bridging or has a bridging mode (such as the Parasound A-23) that'll let you bridge the two channels of a stereo amp. Even then, stay within the manufacturer's recommendations for load impedance.
The other drawback of BTL amps is that both speaker terminals should be considered hot. Don't connect any ground (such as your scope probe ground clip) to any of the outputs.
Tom
to achieve those you have to use a loading that is more severe than the rated loading.
a pair of 4ohms rated amplifiers when bridged becomes an 8ohms rated amplifier.
The rated output of the two single amplifiers is doubled when the load on the bridged pair is also doubled.
There are no extra losses to take account of. There is no guessing at what protection circuits might step in because you have applied a load that is more severe than the rated loading.
Let's look again at the way you want to assess the bridged amplifier.
Take any 4ohms rated amplifier. Let's say it can deliver 200W into 4r0 dummy load and is rated to drive 4ohms speakers.
If we connect a 2ohms speaker we know it might shut down or enter some limiting condition. So we don't usually attach a 2ohms speaker. it might drive 300W into 2r0, or even 350W into 2r0, but both of those when using a 2ohms speaker are outside the manufaturer's rating.
Now bridge two of those amplifiers and test them as a bridged pair into a 4r0 test load. We might get three times the power. But now attach a 4ohms reactive speaker. We now have the situation of trying to drive a load that is half the rated load of the amplifier.
It might work, it might survive the event. But you are taking it outside the envelope specified by the manufacturer, it is a pair of 4ohms amplifiers.
Last edited:
Hi Andrew,
Instead of dancing around the issue, look at it the way a standard user looks at it. Take a speaker (8R), and bridge the amplifier. Now the current and voltage are both doubled in a perfect world. That's 4X the rated power. In real life, that amplifier isn't perfect and will deliver somewhat less power than theory would suggest.
This is the normal situation.
Now, if you are going to suggest the use of two speakers, the average person will put them in parallel, not series. Oops!
-Chris
Instead of dancing around the issue, look at it the way a standard user looks at it. Take a speaker (8R), and bridge the amplifier. Now the current and voltage are both doubled in a perfect world. That's 4X the rated power. In real life, that amplifier isn't perfect and will deliver somewhat less power than theory would suggest.
This is the normal situation.
Now, if you are going to suggest the use of two speakers, the average person will put them in parallel, not series. Oops!
-Chris
The question was how does the *output impedance* change when you bridge an amplifier, or does it change? Maybe I should have gone in search of the answer myself, I was in a lazy mood.
RE: "Just because you've tried it a few times" ... we "tried it" hundreds of times, and in some cases continuously in daily operation for years.
I understand Bryston's concerns. I also know the particular version of the 4B that we used was, well, bulletproof. I can't comment on the current models.
Hi Johnny2Bad,
If you're talking about the original 4B, it was anything but bulletproof! Holy crap, it was made with the least expensive parts you could buy! I know because I worked on them. I reverse engineered that model because it was a secret at the time. I am more than a little familiar with those.
However, the current product looks good and sounds much, much better. One day I might see inside one of those.
-Chris
If you're talking about the original 4B, it was anything but bulletproof! Holy crap, it was made with the least expensive parts you could buy! I know because I worked on them. I reverse engineered that model because it was a secret at the time. I am more than a little familiar with those.
However, the current product looks good and sounds much, much better. One day I might see inside one of those.
-Chris
James Tanner knew what we were doing and was as interested as we were in how they were standing up. He also kept the engineers at Bryston in the loop. The conclusion was although the 4B could not handle a 4 ohm load @ 800w with the FTC rule one-third power @ 1 hour pre-conditioning test it could handle it with music signals without issue.
The Infinity IRS and similar, less expensive large Infinity's had a 4 ohm impedance (not common back in the 70's) which dipped to 3 ohms in places, but were extremely power hungry at the same time. 200w just wasn't enough. On the other hand the Dayton-Wrights ... full range electrostatics that actually had genuine bass down to 30 Hz and incorporated a transformer the size of a toaster oven which was a 2-man job to move, per speaker ... caused a couple of amps we tried to self-destruct, despite not being a low impedance load and despite those amps supposedly being able to drive that load. Only the Brystons and the Threshold 400 could drive them without releasing the "magic smoke". On the other hand, the Threshold and various tube amps we had around with the proper transformer taps just didn't have the power.
We also used 4Bs in mono with stacked paralleled Klipsch La Scalas in auditoriums. No problems. You could get the clipping indicators to light if you just wired them in stereo, but not in mono configuration. We also sold 4Bs to Rock Bands, who knows what loads they saw, but not one failed.
The 4B's have always incorporated robust safety circuits so no damage, they just shut down with the FTC rule bench tests into 2 ohm/stereo and 4 ohm/mono, but never quit with music. We had to find some way to drive a few speaker systems we had in stock or just give up on selling them. We tried a lot of amps ... bringing in things like BGW with Rock'n'Roll reputations for reliability to try. They weren't reliable on those speakers. Since we serviced everything we sold, we couldn't afford to be selling amps that didn't last, and we knew what lasted and what didn't.
The Infinity IRS and similar, less expensive large Infinity's had a 4 ohm impedance (not common back in the 70's) which dipped to 3 ohms in places, but were extremely power hungry at the same time. 200w just wasn't enough. On the other hand the Dayton-Wrights ... full range electrostatics that actually had genuine bass down to 30 Hz and incorporated a transformer the size of a toaster oven which was a 2-man job to move, per speaker ... caused a couple of amps we tried to self-destruct, despite not being a low impedance load and despite those amps supposedly being able to drive that load. Only the Brystons and the Threshold 400 could drive them without releasing the "magic smoke". On the other hand, the Threshold and various tube amps we had around with the proper transformer taps just didn't have the power.
We also used 4Bs in mono with stacked paralleled Klipsch La Scalas in auditoriums. No problems. You could get the clipping indicators to light if you just wired them in stereo, but not in mono configuration. We also sold 4Bs to Rock Bands, who knows what loads they saw, but not one failed.
The 4B's have always incorporated robust safety circuits so no damage, they just shut down with the FTC rule bench tests into 2 ohm/stereo and 4 ohm/mono, but never quit with music. We had to find some way to drive a few speaker systems we had in stock or just give up on selling them. We tried a lot of amps ... bringing in things like BGW with Rock'n'Roll reputations for reliability to try. They weren't reliable on those speakers. Since we serviced everything we sold, we couldn't afford to be selling amps that didn't last, and we knew what lasted and what didn't.
Last edited: