DSP and Tube Amplification

Drugolf

Site Supporter
Site Supporter
This is probably meant as a question for Pat, but would love to hear what others have determined with the emerging DSP tech as it relates to integration of tube amplification.
Will there be a place for tube amps with all of this? Does there need to be a place for tubes to exist in the chain or does DSP make up for what we all otherwise feel tubes bring to the current passive system equation particularly with high impedence/efficiency speakers such as big Altec, JBL and Klipsch speakers?

Pat is designing speakers with the DSP control and filtering ahead of the amplification but the amplification is still integrated. Is it as simple as separating those two aspects, or does that defeat too much of the benefit?

Pat, what would you say if a potential client came to you and said "I want you to build me a set of your fine speakers, but I want to be able to use tube amplification?

I am wondering about all of this because I like the big ol Altecs, Jbl's, Klipsch etc and am fully invested in the notion that these almost always do best with tube amplification. (plus it's fun)
 
This is probably meant as a question for Pat, but would love to hear what others have determined with the emerging DSP tech as it relates to integration of tube amplification.
Will there be a place for tube amps with all of this? Does there need to be a place for tubes to exist in the chain or does DSP make up for what we all otherwise feel tubes bring to the current passive system equation particularly with high impedence/efficiency speakers such as big Altec, JBL and Klipsch speakers?

Pat is designing speakers with the DSP control and filtering ahead of the amplification but the amplification is still integrated. Is it as simple as separating those two aspects, or does that defeat too much of the benefit?

Pat, what would you say if a potential client came to you and said "I want you to build me a set of your fine speakers, but I want to be able to use tube amplification?

I am wondering about all of this because I like the big ol Altecs, Jbl's, Klipsch etc and am fully invested in the notion that these almost always do best with tube amplification. (plus it's fun)

The DSP that is present in Pat's designs is from minidsp I believe. Those units are also available as stand alone line level DSPs. I think one of the advantages of what Pat's doing is that he is taking the amplification of the equation when it comes to the end user. If you were to use tube amps on the tail end of an active speaker with a DSP you'd need a very reliable way to set the amplifier gains to match.
 
If you're wanting to neuvo-fit an existing two way, you could use a processor to drive two amps, then ditch your passive filters. minidsp makes a few models that would get you there.

For my designs the power of DSP has been really taken advantage of in the LF system alignment. It's a new approach that is kind of stupid-simple, but works beautifully. We set a CB for optimum transient response, then "buy" back the bass droop with EQ and an HP filer, then add GOBS of boost to hit the bandwidth target. So: the best of both worlds: extension and speed. That idea was not do-able before DSP and, to me, is one of the most compelling reasons for building next gen. It makes so much sense, now that the $/W ratio for amps has fallen so much with Class D, and 500W is a common denomination.

So, anyway, it could be arranged in a way that lets you run filtered signal to tube amps for the mids and tweets, but retain the brute power needed by the woofers. We'd probably want to put a 2x125 or a 2x250 on each, bridged to 250 or 500W. 1KW is need be.

That would be an interesting project, for sure.

Our standard setup uses Hypex Fusion plates, one channel per band per speaker, fed AES by a minidsp SHD Studio. The SHD Studio functions as a streamer, works as a Roon endpoint, can filter four channels (although it's not necessary since we filter in the Hypexes) and does Dirac on the aggregate of whatever's downstream. Plus the usual gain, delay, compressor, EQ stuff and source selection business.

The SHD Studio is digital only but another model called just SHD has both analog ins and outs. It has four channels of output - which would suffice for a two way. You'd just run two analogs to your tubeamp and two digis to Hypex LF amps driving just the woofers. Dirac automatically dials in the relative gain and timing.

If this idea wanted to be a three way, you'd need to either drive the midranges with Hypex or cascade a second processor - like a 2x4 - to feed a second tube amp. Cascading processors is a common practice, done all the time. Unlike an analog signal, you can take as many swipes at the data as you like.

Either way, putting Dirac ahead of a tube amps could be a kick in the pants.

Or you could just run your existing digital source into the LF sections' Hypex amps and insert a 2x4 DAC processor to run two pair of tube amps. I'll bet there are a dozen other ways around the problem.

The thing to watch when buying processors is their bitrate spec. Some are limited to 96KHz, which just won't cut it anymore.

Just me, I'll bet that the resolution we're getting out of the NCores is miles ahead of what could be achieved by anything with an output transformer. If we want a little rosy glow, we just dial it in in the EQ.
 
I have dived far into it yet but I wonder what versatility I am going to find with my MiniDSP 4 x 10HD in this regard. Solid state on a couple channels and tune on a couple other....

Some of this Pat is extending from my current set of speakers here in my office are 5cf vented cabs loaded with Stephens Tru Sonic 120 FR full range drivers, but helped on top with a Fostex super tweeter. If this is a design I want to make permanent I would go the next step to dial in the tweeter crossover etc. The Stephens love tube amps and they don't need much at that. So, all that extra class D power is kinda wasted (not that it costs much/watt).
 
While I don't have tube amplification with my Kites, I do run tube and SS preamps with them. For what it's worth, it's super easy to hear the difference between them. Just like it is when I swap styli on my cartridge. The Kites are super transparent sounding. You can put any flavour you want in front of them, and you will hear the difference.
 
I have dived far into it yet but I wonder what versatility I am going to find with my MiniDSP 4 x 10HD in this regard. Solid state on a couple channels and tune on a couple other....

Some of this Pat is extending from my current set of speakers here in my office are 5cf vented cabs loaded with Stephens Tru Sonic 120 FR full range drivers, but helped on top with a Fostex super tweeter. If this is a design I want to make permanent I would go the next step to dial in the tweeter crossover etc. The Stephens love tube amps and they don't need much at that. So, all that extra class D power is kinda wasted (not that it costs much/watt).
The 4x10 won't do better than 24/96, so something to consider. Deal breaker for me.

You could, though, test the concept.
 
I have a minidsp 2x4 that I used for my speakers active before. I had issues with noise and hiss from the DSP. This is something to keep in mind if your biamping very efficient drivers. Things might be better now and adjustable amp gains could have cured my system, but my wife absolutely couldn't stand the noise. I am somewhat interested in going back to an active setup as I do miss EQ.
 
With a digital source it should be dead quiet.

For analog, all bets are off. Applying lots of gain to a noisy source isn't a DSP problem. It's a source problem.
 
A question I have always had with bi-amps is how to keep two amps with different gain balanced?

I turn up my mid/high amp to mid 90dB, then use my RS meter a few inches from the drivers and adjust the bass plate amps a few dBs lower.

Then the wife comes in and I have to turn it down.
 
If the two amps have different gains, just put a variable attenuator in front of the hotter amp. With DSP, same idea except now you can boost the lower gain amp. Either way, as long as you're aware of the noise issues.

You can try to measure, but if you use pink noise you'll get a useless result. White.

There's a great android app analyzer:Audiotool. And Parts Express sells a calibrated mic for your phone. For 30 bucks total you can take half way decent FA measurements. The app also has a great RT60 tool.

But, really, that's the sort of job that you want to own the skill to do subjectively. In a two way, voice presentation oughta tell you the whole story pretty darn quickly. Vocals should sound natural and realistic at 'real person actually singing in my room' levels. Not sibilant, not nasal, not chesty or thick.

I'll recommend some tracks if you like.
 
Back
Top