Manufacturer Specifications and Interpretation

[slipperybidness]

This is one thing I don't like about starting these threads, they go over my head too quickly. We haven't even made it to page 2 and I'm confused.

Yeah, but there was a time ~5 years ago that some of this was over my head too, or at least I would have to go look at reference materials.

There is no substitute for experience. But, EVERYTHING seems very intimidating when you don't know much about it. As you learn, you wonder why you ever felt so intimidated by it.

 

[Seth=L]

Yes, "good" transformer is ideal. You must start considering saturation and hysteresis when you are looking at transformers (as well as the amps that can be supplied). A larger hunk of iron tends to be more difficult to saturate (a good thing). A good transformer will also have better Power Supply Rejection Ratio, and that is a desirable trait too.

These are things I'm in the dark on.

Yes, Class D amps with linear supplies are not too common. The fact is that Class D topology already has a lot in common with SMPS tech, so mating the 2 together does make a lot of sense. Class D is more similar to SMPS topology than it is to classic A or A/B topology.

Not to mention SMPS is usually less costly to manufacture.

My general rule of thumb from my studies and from amp building: 10,000 uF per rail per 100Wrms. And, that is probably a little bit of a luxury too, could get by with less capacitance per rail.

My Niles SI-275 is a 75 watt per channel at 8 ohms 0.05 thd both channels driven and 120 watts @ 4 ohms both channels driven, same thd. The manual states the amp is stable to 2.66ohms. Inside there's a double wound toroid (3kva per if I recall) and 4x10,000uF (don't recall the voltage).

What was the WPC rating on that Onkyo?

60 watts x 6, I don't recall specifics beyond that without looking it up.

 

[slipperybidness]

This is one thing I don't like about starting these threads, they go over my head too quickly. We haven't even made it to page 2 and I'm confused.

Let me give you my simplified approach to reviewing amplifier specs. This info would have come from a couple of books about building HiFi amps.

1) The THD +N spec is the single most useful spec on the data sheet! I want 0.1% THD + N at rated power or better!

2) The CONTINUOUS power output with all channels driven and 20Hz to 20kHz is next on the list. Unfortunately, not many manufacturers will supply this info.

3) Ruler-flat frequency response. Actually, this is probably the 1st priority, but modern SS amps it is pretty much a given (unless the designer choose to add a non-linear response like the PS Sprout).

4) S/N > 100dB. However, the S/N is an even more critical parameter for the pre-amp

5) It is nice to know the input sensitivity to reach full output, and the dB of gain from input to output. This let's you know if your source will be able to drive the amp to full output, and can give you an idea of whether noise on the input may get amplified enough to be audible on the output. I typically design my amps for 29dB gain. The normal range of gain, depending on the manufacturer and the application is ~27dB to 32dB. Again, this is nice to know to anticipate possible problems, but not absolutely necessary.

6) Pay no attention to the damping factor, modern SS amps, it really is a moot point.

Unless I missed something, that's my approach.

 

[Seth=L]

5) It is nice to know the input sensitivity to reach full output, and the dB of gain from input to output. This let's you know if your source will be able to drive the amp to full output, and can give you an idea of whether noise on the input may get amplified enough to be audible on the output. I typically design my amps for 29dB gain. The normal range of gain, depending on the manufacturer and the application is ~27dB to 32dB. Again, this is nice to know to anticipate possible problems, but not absolutely necessary.

I knew this, just not the maths.

I knew the others as well.

My point is that we can't place too much faith into manufacturer specs.

Hypothetical - Two amps are rated 100 watts per channel both channels driven at 8 ohms @0.05% THD. What we don't know is what happens on both of these amps once they cross that threshold. Lets assume we increase the output on each amp to 120 watts if possible. Amp 1 increases in THD to 0.15% THD while they other jumps to 20% THD. This isn't a specification we're likely to see because as you said under 0.1% THD is ideal so why would the manufacturer condemn themselves by rating their wattage higher at the sacrifice of appear to have a low fidelity amplifier?

 

[slipperybidness]

These are things I'm in the dark on.



Not to mention SMPS is usually less costly to manufacture.



My Niles SI-275 is a 75 watt per channel at 8 ohms 0.05 thd both channels driven and 120 watts @ 4 ohms both channels driven, same thd. The manual states the amp is stable to 2.66ohms. Inside there's a double wound toroid (3kva per if I recall) and 4x10,000uF (don't recall the voltage).



60 watts x 6, I don't recall specifics beyond that without looking it up.

Here is the deal with SMPS:

They have inherent problems for audio applications. They produce incredible amounts of electronic noise and interference due to the rapid switching of the semiconductor devices!

However, electronic engineers are bright guys, so they have figured out ways to engineer the noise etc down into an acceptable level for audio applications. So, SMPS takes MUCH more time, money, and expertise to DESIGN than the good old simple linear unregulated power supply.

However, once you have spent all that time and money to get the design right, the SMPS can be mass produced and spit out rapidly from auto-fab machines! SMPS don't need the large transformers that linear designs require, so that is yet another cost savings item. You can't realistically shrink the transformer on linear supply.

The other item to consider is that now SMPS have all this engineering involved, and a much higher part count. More complex and more parts = more likely to fail (statistically speaking).

The de-facto best supply for a POWER AMP if performance is the end goal is the tried-and-true unregulated linear supply.

If efficiency, less heat, lighter weight is the end goal, then SMPS makes sense.

Edit: If we want a good laugh, ask TLS how he feels about SMPS

 

[slipperybidness]

I've mentioned in other threads, but there are 3 guys that have several great books on amp design. Any info from me would have almost certainly come from 1 of the 3, or more than likely all 3 agree or have similar philosophies about particular design aspects.

Douglas Self
Bob Cordell
G. Randy Slone

On a side note, I know that Nelson Pass is also well respected in that field. I haven't read much of his work. But Pass does have some other off-the-wall, borderline hair-brain designs to his credit too! For Pass, it would be more of these oddball designs that I tend to read up on. Somehow he manages to keep the odd designs from tainting the name of Pass Labs. Firstwatt.com is his page dedicated to those less traditional ideas.

 

[PENG]

You gotta have a good transformer right?

Again, they typically have excellent sustained overload capability but yes, like anything else there is always a range. Generous use of copper windings, good choice of core material and good designs are obviously key to the overall quality.

 

[PENG]

What is your opinion of Zoebel Networks for speakers? You know, to get back to PF = 1.

If I remember, I'm pretty sure my GE T2s have this, making them an easy load to drive.

I have no opinion on something I do not know about. I am also not aware of any brand name speakers that use the Zoebel network. I suspect like most things, there are typically pros and cons and if there is a net benefits, one would expect more speaker manufacturers would have embraced the technology enthusiastically.

 

[slipperybidness]

I have no opinion on something I do not know about. I am also not aware of any brand name speakers that use the Zoebel network. I suspect like most things, there are typically pros and cons and if there is a net benefits, one would expect more speaker manufacturers would have embraced the technology enthusiastically.

Well, just for reference, the stereophile review on the T2 says "Zobel Network Across the Tweeter", which isn't exactly what I remembered.
http://www.stereophile.com/content/goldenear-technology-triton-two-loudspeaker#hptL6i3WAAtEmsdI.97

It seems like a great idea, I'm just wondering what the catch is. "No free lunch"

 

[PENG]

Well, just for reference, the stereophile review on the T2 says "Zobel Network Across the Tweeter", which isn't exactly what I remembered.
http://www.stereophile.com/content/goldenear-technology-triton-two-loudspeaker#hptL6i3WAAtEmsdI.97

It seems like a great idea, I'm just wondering what the catch is. "No free lunch"

Agree, but I don't know what the catch is, yet.. I do have enough background in network synthesis but it will take a lot of time to read up on this thing before I can make any comment. You got me curious now for sure, I just need time..

 

[herbu]

That's why we need people like you, and herbu..

slippery, I'm not sure if that's a compliment or clever innuendo. I'd like to believe the former, but there is that laughing face. Think I'll just go fishing and contemplate the difference between Class D and SMPS.

 

[PENG]

slippery, I'm not sure if that's a compliment or clever innuendo. I'd like to believe the former, but there is that laughing face. Think I'll just go fishing and contemplate the difference between Class D and SMPS.

It is the former. When Slippery (me too, I know, sometimes) gets deep into the technical details, you are there to translate it into something all can understand. Serious I do think both are important.

 

[3db]

Herbu, You have an innate talent for putting techno speak into english. Like Peng and Slipper has been saying, that is a very good thing.

 

[herbu]

Herbu, You have an innate talent for putting techno speak into english. Like Peng and Slipper has been saying, that is a very good thing.

There are a lot of engineers here. But engineers do different work. Most focus on the technical. However my career was in the Quality organization. That means taking the technical reports/position from the very best and smartest engineers, and distilling it down to a business recommendation. Then presenting that recommendation to execs for their concurrence.

Many of the execs were also engineers, and enjoyed escaping the world of business and playing engineer again for a while... so they liked to poke and ask technical questions. But some of the execs did not have any kind of engineering background. My challenge was to explain the problem and the recommendation in such a way that all the execs could understand. Then if the old engineers wanted to poke, be able to talk to them.

My approach was to prepare a presentation that my wife, (an accountant), could understand. Then only dive into the more technical explanations if those kinds of questions came. Over the years I lost the ability to perform the more complicated engineering activities, but got better at reading the audience and explaining technical stuff to non-technical people. Sometimes it seems to come in handy here.

 

[TLS Guy]

I have no opinion on something I do not know about. I am also not aware of any brand name speakers that use the Zoebel network. I suspect like most things, there are typically pros and cons and if there is a net benefits, one would expect more speaker manufacturers would have embraced the technology enthusiastically.

How would you know if there is a Zoebel or not? It is a simple RC network to limit the rise of impedance of a driver so the series inductors cut off properly. Now in a two way you can often dispense with this and use an over sized inductor to handle BSC and cut the woofer off with the correct slope. This is not the sort of thing that is usually mentioned in the "glossies" or the spec. sheet.

In the T2 sited it does look by the impedance curve as if there is a Zoebel across the tweeter. Normally you would not do this as you don't cut a tweeter off. I suspect this was done to ameliorate the HF rise above 10 KHz, common in a lot of ribbon type HF units. This was only partially successful and not really what a Zoebel is for. Better would have been a small series choke with a resistor bypass, if the sensitivity of the driver allowed for it. If not then that was probably the only option.

 

[PENG]

How would you know if there is a Zoebel or not? It is a simple RC network to limit the rise of impedance of a driver so the series inductors cut off properly.

I did not mean to say there wasn't any in use, just that I wasn't aware of them being used in brand name speakers that I am familiar with. Besides, Slipper asked me about Zoebel network used to bring power factor back to 1 (see his post#18) so that's what I responded to, not just a network used across a tweeter or something, but specifically what Slipper asked me about.

I also assume he is talking about applications to conventional type of speakers. So far I have not seen any published phase angle versus frequency graphs of any conventional brand name speakers that has anything close to power factor=1, ie. zero degree flat graph. Again, I am not assuming there isn't any existing but I am definitely not aware of one. If you know of any, I am very curious to know which brand does such a thing. Not that I would buy one, just curious, that's all.

 

[Swerd]

The Zaph Audio SR71 is a DIY design, not a commercially available brand name speaker. I mention it because it offers an optional impedance flattening circuit, and shows impedance curves ± the circuit. It's an LRC circuit as opposed to a Zobel with only RC. See the green box and green text in the schematic:

The speaker is a 2-way with a Seas ER18RNX 6.5" woofer and a Seas 27TDFC 1" fabric dome tweeter, and uses a 4th order crossover at 1750 Hz. Zaph never discusses any audible difference with and without the impedance flattening circuit. He says this about it:

"Electrically, on the woofer we have a simple two component 2nd order network. Also shown is an optional impedance flattening circuit for those with tube amplifiers…"

"Most of the impedance curves shown in this design reflect an impedance flattening circuit. It's not required however and can be removed without any effect if you are using a solid state amplifier. Since this is a relatively sensitive speaker, usage of the impedance flattening circuit could be useful to those who are using tube amplifiers."

Impedance curves with (blue) and without (red) the impedance flattening circuit clearly show the speaker has a nearly constant impedance above ~100 Hz with the circuit:

 

[TLS Guy]

I did not mean to say there wasn't any in use, just that I wasn't aware of them being used in brand name speakers that I am familiar with. Besides, Slipper asked me about Zoebel network used to bring power factor back to 1 (see his post#18) so that's what I responded to, not just a network used across a tweeter or something, but specifically what Slipper asked me about.

I also assume he is talking about applications to conventional type of speakers. So far I have not seen any published phase angle versus frequency graphs of any conventional brand name speakers that has anything close to power factor=1, ie. zero degree flat graph. Again, I am not assuming there isn't any existing but I am definitely not aware of one. If you know of any, I am very curious to know which brand does such a thing. Not that I would buy one, just curious, that's all.

I see where you are coming from.

Getting a power factor of one would be next to impossible.

However Raymond Cooke founder and CEO of KEF before his much too early death, did a lot of work on trying to get speaker loads as close to resistive as possible. In his time with a lot of tube amps that made more sense, with solid state amps, the issue is mute. Trying to solve this makes for a much more complex crossover and really ups the part count and expense.

I did however want to give a nod in his direction with the design of my speakers downstairs.

The speakers use his KEF B139 bass drivers, among my favorite drivers of all time.

The look of the speakers was designed to evoke the memory of some of the styling of the Cooke era. In addition I used his precepts to get the load as resistive as possible.

Here is the result. Not perfect, but all but one of the tuning peaks is blunted and the crossover peaks at around 400 Hz and 4 KHz are significantly blunted.

The phase angles deviate only briefly from zero in three areas, but over a narrow frequency range.

The crossovers are on three boards, and the component cost was more than a lot of people would spend on speakers, although not of that quality.

My Raymond E. Cooke Memorial Speakers.

I have to say, that in addition to sounding very good, they are an easy load on the amp, which stays nice and cool pushing them hard.

Anyhow that was the best I could do to get the power factor as close to 1 as I could.

 

[TLS Guy]

The Zaph Audio SR71 is a DIY design, not a commercially available brand name speaker. I mention it because it offers an optional impedance flattening circuit, and shows impedance curves ± the circuit. It's an LRC circuit as opposed to a Zobel with only RC. See the green box and green text in the schematic:

The speaker is a 2-way with a Seas ER18RNX 6.5" woofer and a Seas 27TDFC 1" fabric dome tweeter, and uses a 4th order crossover at 1750 Hz. Zaph never discusses any audible difference with and without the impedance flattening circuit. He says this about it:

"Electrically, on the woofer we have a simple two component 2nd order network. Also shown is an optional impedance flattening circuit for those with tube amplifiers…"

"Most of the impedance curves shown in this design reflect an impedance flattening circuit. It's not required however and can be removed without any effect if you are using a solid state amplifier. Since this is a relatively sensitive speaker, usage of the impedance flattening circuit could be useful to those who are using tube amplifiers."

Impedance curves with (blue) and without (red) the impedance flattening circuit clearly show the speaker has a nearly constant impedance above ~100 Hz with the circuit:

Both sections are electrically second order. There is a zobel on the tweeter, for reasons not clear to me. May be to make the graph look pretty? Unfortunately the phase response is not shown.

 

[PENG]

The Zaph Audio SR71 is a DIY design, not a commercially available brand name speaker. I mention it because it offers an optional impedance flattening circuit, and shows impedance curves ± the circuit. It's an LRC circuit as opposed to a Zobel with only RC. See the green box and green text in the schematic:

The speaker is a 2-way with a Seas ER18RNX 6.5" woofer and a Seas 27TDFC 1" fabric dome tweeter, and uses a 4th order crossover at 1750 Hz. Zaph never discusses any audible difference with and without the impedance flattening circuit. He says this about it:

"Electrically, on the woofer we have a simple two component 2nd order network. Also shown is an optional impedance flattening circuit for those with tube amplifiers…"

"Most of the impedance curves shown in this design reflect an impedance flattening circuit. It's not required however and can be removed without any effect if you are using a solid state amplifier. Since this is a relatively sensitive speaker, usage of the impedance flattening circuit could be useful to those who are using tube amplifiers."

Impedance curves with (blue) and without (red) the impedance flattening circuit clearly show the speaker has a nearly constant impedance above ~100 Hz with the circuit:

Thanks Swerd, that's impedance flattening, but Slipper asked about the Zoebel network used to kind of flatten the phase angle, actually more, not just flattening but to zero degree. As TLS said, that could be next to impossible, but of course "next" is the key word. Even if that can be done, surely they would be some trade-off, or "catch".