[hagtech]

Recently, Stereophile reviewed a $350,000 tube amp.  Sure, maybe it sounded great.  Maybe it is worth every penny.  Personally, I was a bit disappointed in the measurements.  I expected better.  Have I been missing the boat here?  Is this stuff not important?

I put a lot of time designing and tweaking my new Cymbal amp.  My mantra is: "not only must it sound great, but it must also measure great".  I'm not convinced you can have one without the other and still have a marketable product.  So I was curious to see how these two amplifiers compared, at least in the measurement arena.  $350,000 versus $2,500.

Gain is 34dB versus 15dB.  High gain means you are stuck using too much attenuation on your volume control.  SNR for both is decent, 88dBA versus 95dBA.  Output impedance is 5 ohms versus 1.6 ohms.  A higher number means lower damping factor and softer bass.  The following graphs tell the rest of the story.


Frequency response into simulated 8 ohm speaker load.  Note the 5dB rise in bass.


Cymbal frequency response into 8 ohms.  -3dB points are 10Hz and 95kHz.


Distortion vs. power.  1% limit is less than 1W.  Lots of distorted headroom.


Cymbal distortion vs. power.  1% limit is at 6W.  Only 0.18% at 1W, almost a magnitude lower.


1kHz square wave response.  Note bass boost tilts upward.


Cymbal 1kHz square wave.


10kHz square wave response.  Note ringing and transient problems.


Cymbal 10kHz response.  Very clean, phase linear bandwidth.

In my opinion, the Cymbal really stands up well.  Especially for a design that uses no negative feedback!

jh

[ohenry]

So, when do we get to see more about this new product?  

Or maybe I've missed some prior discussion regarding this one.  What valves does it use, topology, stereo or mono, etc., etc?

[hagtech]

The Cymbal will be available sometime next month (October '04).  I have already placed the order for production quantity circuit boards.  Yet to write assembly manual (which takes a lot of time).

Presently getting machines ready for the RMAF show next month.  Once I ship everything out, I'll have a few days time to hopefully write.  Meanwhile, new website in the works and a bunch of marketing stuff (I need pamphlets).

Temporary web page: www.hagtech.com/cymbal.html

jh [/url]

[Steve Eddy]

Temporary web page: www.hagtech.com/cymbal.html

Wow! Only $190 per pair? Such a deal! I'll take two!

se

[penance]

The Halcro amps seem to be held in high regard, very good measurments. But many people who have heard them (myself included) seemed under welmed to say the least. Maybe just an exercise in pure numbers!

[Tonto Yoder]

$190/ pair is a rather misleading figure --it'll just get you a pair of PCB's.
The "real" price is mentioned a little lower on the page:
"The Cymbal is not an "el-cheapo" amplifier, rather a fully tricked out and substantial product. The DIY half-kit version will cost you $95 (pcb) + $95 (electronics) + $120 (iron) + $190 (tubes) + $85 (chassis) for each monoblock."
So the DIY price would be $1170/ pair.

I'm not quite sure why penance brings up Halcro when it's Wavac that is the amp in Stereophile??? (Unless just to compare sonics to measurements.)

[JohnR]

Well, Jim did say $2,500 in his original post...

Jim, I'm guessing that four of the 6H30s are parallel push-pull and the other as a high-gain input stage?

What are the requirements for the output transformers? Just wondering if an 8k primary will work...

[penance]

Tonto, I mentioned Halcro purely as another make that flaunts figures but is un exciting:)

[Grover]

This amp looks really nice and is a great addition to the lineup.  I'm confused about the power ratings stated as 8 watts into 8 ohms and 5 watts into 4 ohms.  Most other specs I've looked at will have a higher power rating into a 4 ohm load.  Why is the Cymbal lower power into 4 ohms?

[gurevise]

Measurements for $350000 amp were done with simulated speaker load. Measurements for Cymbal were done with pure 8 Ohm resistor. This will make measurements for Cymbal look unfairly much better. Simulated speaker load has number of LCR circuits to simulate complex speaker behavior. It should explain some HF ringing and non-linearity for expensive amp.

General point about poor measurements and "great" sound as an indication of some coloration or hidden problem is valid one.

[hagtech]

Thanks for all of the comments!  

Yes, the simulated speaker load is obviously a more difficult situation.  That wasn't meant to be hidden, only that it was the only graph stereophile supplied.  And I don't have their simulator.  Yes, all tube amps with relatively high output impedances will get pushed around a bit, including the Cymbal.  That's why I also added the Ro numbers.  What you can get from the plot is that there is indeed an artificial bass boost and some ringing/instability in the ultrasonic region.  

"5W into 4 ohms."  That's a typo!  Shoot, where did I put my notes...  Aha, it remained at 8W.  Typical for transformer-coupled tube output stages.  Using the 4 ohm tap you'd still get 8W.  Solid state high-feedback amps usually double the power into 4 ohms.  With a tube design, the load impedance gets reflected back to the tubes and shifts the operating conditions.  Basically, it is due to a finite output impedance.  To be sure, the Cymbal works best into a kind 8 ohm load.

Great point about the Halcro.  This happens all the time with some high feedback designs.  I fully agree.  Measurements are not everything.  

jh

[dado5]

Thanks for the comparison Jim.  Very interesting stuff.

Personally I find little correlation between standard measurements and sound. I have heard good sounding gear that showed good specs and I have heard good sounding gear that had less than ideal metrics. The reverse is true as well (obviously as most bottom line Radical Shock all in one HT sets have vanishingly low THD figures).

As a consumer, it is relatively simple for me to use my ears in my decision making process.  For a designer however, I imagine it would be a nightmare to use the same process in evaluating every step of his product's development. I find it hard to believe that every change in plate resistor value or bias voltage level or cathode bypass capacitor value could be checked by ear in direct comparison to the previous circuit configuration without the designer ending up either broke or in psychiatric treatment.  Maybe this can be done with Zenish op amp boxes or cheap tube SETs that use 4 parts, but otherwise I can't see it happening in the real world.  So I can fully understand why designers would use measurements at least in conjunction with listening to evaluate their design.  

In the end, as audiophiles/music lovers, it is the sound quality that is important, so measurements are really not of much use in and of themselves. Honestly, would any one of us choose a product whose sound we did not care for over one that sounded better just because it had better metrics?

Thanks,
Rob

[gurevise]

I think, measurements are important. They do not guarantee of great sound but they a good foundation for it. They also provide some proof of clams for people with tech background.

Some manufacturers like to clam much more then they can deliver.
Obviously, measurements used to fool people without good technical knowledge.

As we know, standard set of existing measurements, like THD, do not correlate well with subjective performance. But, they still very useful. You can't go wrong with good power output, flat frequency response, low distortions and high slew rate. Products that do not measure well but sound "good" usually have considerable limitations with speakers used or types of music they can play well or many hidden for inexperienced user colorations.

Just my opinion.

Thanks
Sergey
Dayton, OH

[hagtech]

Found another example.  Highly acclaimed manufacturer of $29,000 pair of monoblocks.  Keep in mind it's that all-important first watt that really counts.  I am still quite pleased to see the Cymbal stack up well against the heavyweights.


Distortion vs. power, 8-ohm tap into 8 ohms (second from bottom) shows 5 watts at 1% distortion.  On par with the Cymbal.


10kHz square wave.  Looks really good, slight undershoot on rising edge (dual slopes).

Overall, pretty decent.  But at over ten times the cost, no significant advantage except for more power.

jh

[eweitzman]

As a consumer, it is relatively simple for me to use my ears in my decision making process. For a designer however, I imagine it would be a nightmare to use the same process in evaluating every step of his product's development. I find it hard to believe that every change in plate resistor value or bias voltage level or cathode bypass capacitor value could be checked by ear in direct comparison to the previous circuit configuration without the designer ending up either broke or in psychiatric treatment. Maybe this can be done with Zenish op amp boxes or cheap tube SETs that use 4 parts, but otherwise I can't see it happening in the real world.

Herb Reichert's Flesh and Blood 300B amp was designed exactly this way. It was published in Sound Practices, so it's not exactly a commercially designed product. Yet Herb made and sold about a dozen of them. Every part must be as he specified, most are boutique parts, and if you change one, he says it's your amp design, not his. I recommend the article as a fun read if nothing else.

- Eric

[MarinRider]

an astoundingly low 1.6 ohm output impedance  

[eweitzman]

an astoundingly low 1.6 ohm output impedance  

I don't understand your point. Can you please explain?

[hagtech]

>> 1.6 ohms

I understand the point.  Transistor amplifiers using tons of negative feedback routinely achieve numbers ten to a hundred times better than this.  Some people like to compare apples and oranges.  Interesting for perspective, but somewhat irrelevant and immature.

In the realm of tube amplifiers not using feedback, the figure is quite good.  Most SETs, for example, pump out power at 4 to 8 ohms, which is a big contributor to their "soft bass" characteristic.  The low output impedance was one of my main goals when designing the Cymbal.  But not at the expense of feedback.

To each his own.  

jh

[dado5]

WTBD?

1.6ohm should work fine with most speakers of appropriate sensitivity (except maybe electrostatics).  With some 4ohm speakers, it may give up a degree of bass control, but this may be subjectively tolerable, even desirable.

I have heard plenty of amps with worse damping sound just fine.   Room placement will likely have far more impact on bass performance.

Rob

[MarinRider]

an astoundingly low 1.6 ohm output impedance

the figure is quite good

I now agree!