[AdamM]

Ran into an interesting thing today while i was doing some room correction / frequency analysis tests....

I was wondering if anyone might be able to shed some expert advice on this?

It was a two part test - pumping white noise through my monitors to see how my mid/mildly treated room is performing, and to compare some DIY SET KT88 tube amps with a big old Adcom SS.  A flat analysis mic was used with some frequency graphing stuff on the computer.

Very interesting results...!

I'll cut to the chase:  The tube amps have a big dip around 2-4k.  A shockingly large dip there, around 10-12db!  Some upper sizzle >10Khz of 3-4db...
Switch in the beefy Adcom, and the curves go almost flat.  There's a few lumps which i know are the speaker itself - but i was shocked at the differences between the amps! ?


Tube amp monoblocks:  Note big dip at 2-4KHz, and extra rise above 10KHz


Now here's the SS Adcom: Note essentially flat through the mid and upper ranges


Here's the impedance curve for the RAW HT2 speakers. Note the impedance curve is suspiciously similar



I modded/built the tube amps, and admit that i'm dangerous like a pilot with a few hundred hours. I sorta know what's going on, enough to get me in trouble, anyways.

Is my frequency dip because the somewhat wimpy 18 watt tube amp has a hard time with this speakers load?  The stronger Adcom just barges on through and doesn't notice the load variance?  I could tweak the output trafo...

Is my tube amp design / mods the likely culprit?  Testing them on another set of speakers is on the very near agenda!

Are tube amps this touchy with speaker impedance?

Cheers,
/Adam

[ctviggen]

First, make sure there are no differences other than amplifiers when you perform this test.  In other words, if you're standing in one location for one test and another location for the other test, then you can't really be sure of the results.  Between these pictures, I can see multiple differences, which may or may not affect the results.  Once you're sure the results are correct (i.e., not influenced by room conditions and only the amplifiers are different), then begin a further analsis. 

Assuming your results are correct, I agree with you.  Your tube amp appears to have a hard time driving 4 ohm loads.  The curves look too similar not to reach this conclusion. 

[AdamM]

Thanks ctviggen,

Yes, there's some long fiber wool i was using in the ports to see if i could tame that peak at 120Hz (didn't work).  There's some on the desk and atop of the speakers, in different positions between 'takes'

I stood in the same place and apart from the amp switch, they're the same test environment.  I bumped the camera on the tripod, so the shots have a slightly different FOV and position which makes it look like the speakers moved... they didn't.  You can see the same lens distortion differences on the little mixer.

I'd like to try the amps on another pair of speakers, to see if the curves are similar.

It could be so many things on the amp... where to start!   It's just when i looked at the speaker impedance curve, the similarities seemed to be more than coincidental.

Hmmmm

[MJK]

How hard were you driving the amps? If the volume was too high, maybe the tube amp ran out of "gas" while the big Adcom was still coasting. Try different lower volume levels to see of the tube amp behaves the same at all volume levels.

[AdamM]

How hard were you driving the amps? If the volume was too high, maybe the tube amp ran out of "gas" while the big Adcom was still coasting.

Good thought MJK.  The amps were up, but not anywhere near clipping.  I'll do another test tomorrow at a variety of levels to see if that's even a possibility.   The speakers are 89db 1w/m - a little thirsty - and it does suck the 18watts out of those amps soon enough, but i wasn't nearly that loud.

I've got some other taps on the output transformer... Going to try some experimentation there also.


/A

[Scott F.]

It's unfortunate but it looks like you have just experienced an amp/speaker mismatch. Tube amps can be very finicky when it comes to speakers. You have to remember that tube amps do a great job of providing voltage but when it comes to current delivery, often times they fall short. Little things like complicated crossovers and Zobel networks that your typical SS amp doesn't even notice and absolutely kill the sound of a tube amp. This is exactly what happened in your case.

There is a strong likelihood that the impedance dip at your crossover point is the major culprit. If your output transformers have 4 ohm taps, I'd try using them, that might help. On the other hand, it may not. You may want to try a quickie Zobel network. This will help some but I guarantee it will suck the life right out of your tube amp.

Harvey Rosenberg (God love him) did a great article on Speaker Matching to tube amps. Once you read past all his eccentricities, there is tons of great information there. He does a far better job of explaining it that I can.
Into The Heart Of Speaker Darkness

Bottom line is that you need tube friendly speakers to get the most out of your amp. Remember this when you are thinking about speakers to mate, just because a speaker claims "high(er)" efficiency DOES NOT mean it is a good mate to a tube amp. You HAVE TO look at the speakers reactive load. When it comes to speaker matching, less is best in the crossover department and watch out for high (and low) impedance humps (and dips).

[AdamM]

Wow! Great article, thanks Scott

Luckily, this wasn't an arranged wedding between the tubes and those speakers.  Looks like this was just a one night stand 

[Steve Eddy]

Ran into an interesting thing today while i was doing some room correction / frequency analysis tests....

I was wondering if anyone might be able to shed some expert advice on this?

It's just Ohm's Law at work.

Tube amps, and particularly SET amps, and especially those with little or no negative feedback, tend to have a rather high output impedance. Some as high as several ohms, and others a bit higher still.

The output impedance of your amp and the loudspeaker's impedance form a voltage divider. Therefore, the higher the output impedance of the amplifier, the more the frequency response will start resembling the loudspeaker's impedance curve.

Solid state amplifiers can have vanishingly low output impedances, from a few tenths of an ohm to a few hundredths of an ohm. Because of this, the frequency response variations into a loudspeaker load are proportionately smaller.

Check out the reviews of tube amps versus solid state amps in Stereophile sometime. John does frequency response measurements into a simulated loudspeaker load (a 2uF cap in parallel with an 8 ohm resistor) which presents the amps with a non-flat impedance. You'll see that while the frequency response of the solid state amps is typically ruler flat, tube amps show varying degrees of peaks and dips in the frequency response due to the amp's output impedance.

se

[JoshK]

Speaker designers assume an infinitely low output impedance when designing their speakers, with your SET amp that isn't the case.  Steve's explaination is succinct.  If you want to mate speakers to your SET amps, you need higher sensitivity and flat impedance. 

Not all tube amps vary like that, but your SET amp like others said has no feedback (other than local degenerative) so the output impedance is relatively high.  Trade offs for a particular design goal.   I think you will see now why SET lovers choose many single driver speakers and other eccentric designs.   For the record I am not poo-poo'ing SETs as I am planning to build a pair myself.

[Steve Eddy]

If you want to mate speakers to your SET amps, you need higher sensitivity and flat impedance.

Yeah. But name a speaker out there that fits that description.

I think you will see now why SET lovers choose many single driver speakers and other eccentric designs.

But not even single driver speakers have a flat impedance. You still have the large peak at the driver's resonant point along with a rising impedance above that due to voice coil inductance. Though the latter may not be such a bad thing for a single driver speaker.

se

[JoshK]

Steve, I agree with your points...just saying that those are the goals.  I think many SET lovers deal with the tubby bass from the driver's resonance or xo the driver they are running to avoid it.  I don't like tubby bass myself so my intention in trying SETs is to use them in an active setup in a driver's bandpass.

[Steve]

I also agree with Steve and Josh. Many SETs can have 3 or more ohms output Z. But it is not impossible to have output Z lessened by a factor of 10 with SETs (without feedback); if designed for that goal. Just have to find the right SET and speaker selection will be easier, or at least less FR deviation. That or, as mentioned above, somehow find a high efficiency speaker that has a flatter Z.

Good luck.

[Steve Eddy]

Steve, I agree with your points...just saying that those are the goals.

Oh no, I know what you were saying. Just that the way you said it left the impression that there are speakers out there which met those goals.

I think many SET lovers deal with the tubby bass from the driver's resonance or xo the driver they are running to avoid it.

I'm not so sure. Given the wide variety of speakers used with SET amps out there, I think the frequency response aberrations may be part of the SET "character" that many seem to prefer.

se

[AdamM]

Thanks for the info Steve, Josh  (two birds with one stone there)

So i guess it would be safe to assume that this is a pretty typical response given this gear combination.

Most people probably aren't graphing responses with analysis mics. I'm surprised how few reviewers do -  not wanting to air dirty laundry?  It's funny... Like in the article by Harvey Rosenberg that Scott sent out where he mentions that when people buy cars, they expect all sorts of analysis and data, but for speakers that info is often amiss, or highly massaged.

Doing a quick search of speakers commonly used with tube amps reveals interesting information. Many impedance curves look like profile shots of tsunamis! What kind of crazy non-flat responses are many people experiencing?  enjoying?  Clearly, a ruler flat graph isn't the only valid pursuit but if you are searching for an accurate and uncoloured reproduction, it is an important consideration.

One does have to chuckle a bit when people start waxing poetic over mojo snake oil doo-dads, and you know a test might show the system to have +-15db lumps in it.  In my setup those two amps sounded very different, especially with female vocals.

Everyone gets to draw their own personal line between accurate and enjoyable, that's part of the fun...but, is your system a transparent conveyor, or a 'Fifth Beatle'?

Cheers,
/A

PS  I'm tempted to mess a bit with the screen resistor....

[Daygloworange]

In my setup those two amps sounded very different, especially with female vocals.

Actually, I was just going to ask you Adam, if you did any back to back listening between the 2 amps?

We've seen the graphs. Can you do a review of how this correlates to a subjective comparison in a listening session?

Maybe pick a couple of tracks you know well, and give us an comparison.

I know some people might poo poo it, saying that now that you "know" what the response differences are, you'll give a biased review. But, I want to hear what your subjective impressions are anyways.

I'm sure you can figure out a way to conduct a listening test to not heavily polarize the outcome. 

Cheers

[AdamM]

Actually, I was just going to ask you Adam, if you did any back to back listening between the 2 amps?

Hi DGO,

Yes, i did listen to both quite a bit.  As you may know, i'm a fan of RAW products   Having the HT8's in my living room, and mucking about with various amps between them.  The tubes work better on the HT8's for enjoyment listening at low levels, than my SS amps seem to.  That room is really wet and bouncy, so i have to just shrug and enjoy what happens.  Not a critical environment whatsoever, but nice in the big open lots of sound everywhere kind of way.

The project studio is a different affair, i'm going for ruler flat.  In that environment the tube amps revealed themselves right away. 

I'll make some brief comments now, but will do another test and take notes for a more detailed opinion.  I personally don't care either way, with another DIY amp on the way (GB 150D's) and two other SS amps in storage, i'm not partial or biased whatsoever.  It's the technology i find interesting about it.

Adcom SS:
Dry, tight bass, razor sharp center image, effortless and sterile.   Note that these comments are in comparison with the tube amp.   In this case 'dry' and 'sterile' are probably synonymous for  'accurate' - they're just doing their job and not adding a pile of even order harmonics and bloom and all that other SET tube 'character'

DIY SET tubes:
Rich, warm, open.  Noticeable sizzle up top.  Female vocals sounded 'scooped'.  Percussion was really forward. They felt to me like a 'happy face' EQ was put on them, and the data supports that.  While i liked the sound for casual stuff, they came across as exaggerated and over the top.  Too much of a good thing?  In the context of a studio, they drove me a bit crazy.  I didn't know what to trust and was internally lunging for the EQ.  It would be very hard to do a clean mix on them.

I think it's fine to have a double standard.  I like the tubes in the living room, they're friendly to the material.  When you sit down and really listen hard though, you can tell that they put a sexy evening dress and some smoky make-up on whatever you're playing.

More info soon.  Off to the beach!

/A

[Scott F.]

Doing a quick search of speakers commonly used with tube amps reveals interesting information. Many impedance curves look like profile shots of tsunamis! What kind of crazy non-flat responses are many people experiencing?  enjoying?  Clearly, a ruler flat graph isn't the only valid pursuit but if you are searching for an accurate and uncoloured reproduction, it is an important consideration.

I really hate to bring this up because it will likely cause this thread to spiral out of control but....frequency response based on an anechoic measurements doesn't reveal the entire story about how a speaker truly sounds. When you look at a Lowther PM2A (my favorite driver) from 200Hz and up the FR graph looks like the Swiss alps. When you listen to it in your (or at least my) system you encounter a completely different world. There is less coloration with this driver than with nearly any other driver or speaker I've played with. Don't ask me why because I can't explain it. Maybe its room interaction, maybe its the reactive nature to the SET amps I use, maybe its my placement of the driver on its OB,.... who knows, maybe I'm just deaf and haven't realized it yet but this single driver (150Hz and up) sounds so much better than any two way I've heard that I will never go back. It does roll off around 12-14k and yes unless you use a really expensive cabinet (like the Teresonics) or augment with a sub (as I do) you don't get any bass, I'll give you those points. What it does within its super wide bandwidth can seldom be matched (IMO) in a conventional design. And no, I don't hear any coloration at all. In fact I find that it is quite the opposite. I find some of the single (wide range) drivers far more natural sounding than most traditional designs.

When it comes to bass and being flabby, I have heard a few designs that have had 'bad' bass. On the opposite hand, I have heard a number of designs that have had excellent bass. One of the easiest and most common of single driver designs the Voight Pipe (ie Cain and Cain Abbey) has rather decent bass. It isn't unnecessarily flabby by any stretch. Some of the vintage designs do bass quite well like the big Altecs and Tannoys. With a decent SET, those sound marvelous. I really think quality bass from a speaker designed for low(er) powered tube amps has as much to do with the output iron as it does anything else. No 'proof' there, no 'science' there, just shitloads of hands on experience talking there.

Oh, if you are really worried about the rising impedance on one of these drivers, make sure you by the 16 ohm version of the driver and put a swamping resistor on it. Properly sized, that will nearly halve your spikes.

Bottom line is you do need to pay attention to FR and impedance graphs but they aren't necessarily the last word.

[MJK]

I really hate to bring this up because it will likely cause this thread to spiral out of control but....frequency response based on an anechoic measurements doesn't reveal the entire story about how a speaker truly sounds. When you look at a Lowther PM2A (my favorite driver) from 200Hz and up the FR graph looks like the Swiss alps. When you listen to it in your (or at least my) system you encounter a completely different world. There is less coloration with this driver than with nearly any other driver or speaker I've played with. Don't ask me why because I can't explain it. Maybe its room interaction, maybe its the reactive nature to the SET amps I use, maybe its my placement of the driver on its OB,....

It is not the amp, I am using my PM2A's on an OB powered by a 200 watt SS amp and the set-up sounds great. A SET amp, or any tube amp for that matter, is not a requirement for the PM2A. Good system design by including the influence of the type of amp on speaker's response is all that is required.

[Bob Reynolds]

I posted about this a few weeks ago. It was new to me and quite interesting.

http://www.audiocircle.com/index.php?topic=43938.0

When an amp has a high output impedance its frequency response tracks the impedance curve of the connected loudspeaker. This is why tube amps can sound so different from solid state, why speaker cables with tube amps will have a sonic impact and why picking a speaker for a tube amp is so important. And, unfortunately, without some accurate measurements picking a speaker is just trial and error.

[Bob Reynolds]

... You have to remember that tube amps do a great job of providing voltage but when it comes to current delivery, often times they fall short. ...

For typical tube amps with a relatively high output impedance, it should be just the opposite. Solid state amps with low output impedance are considered true voltage sources. However, as the output impedance rises above 1 ohm, as with many tube amps, the amplifier becomes a current source. Basically, the voltage drop across the output impedance increases enough to limit the voltage applied to the load.

A good example of this is Bob Carver's Sunfire amps. He provided two sets of output posts, one labeled voltage (intended for a woofer) and one labeled current (intended for a mid/tweeter). The only difference between the two outputs was a 1 ohm resistor in series with the current output posts.