Why Such Variety in Small Signal Tube Sound?

[watchnerd]

I've been playing around with tube rolling (insanity, I know, but I recently retired) in my Fi Yph phono stage.

I'm no EE, but from what I understand it's a pretty simple circuit. Goofy subjective review here, but you can at least see some pictures and a bit of the components inside:

6moons audio reviews: Fi Yph

Note, I don't actually need to use this as my phono stage, as my amp has a perfectly decent solid state phono section, especially for MM carts. This is all for amusement.

Over the past couple of weeks, I've been swapping in 18 different tubes, following a uniform warm up procedure, and listening to the same record for consistency, and the same volume setting on the amp.

What I'm not doing: trying to measure the electrical output between tube swaps on the amp, which is clearly a methodology blind spot. The hope is that, as they're all 6DJ8-compatible tubes, the error bars on output voltage wouldn't be grossly wide.

So here is the list of tube. A few are current production, but most are NOS:

Brand	Model
Amperex (Orange Globe)	6DJ8
Amperex	7308
BEL	E88CC
Brimar	ECC88
EH	6922
GE	6DJ8
JJ	E88CC
Mullard	E188CC/7308
National	7DJ8/PCC88
Philips SQ	E88CC
Reflektor	6H23P-EB
Siemens	E88CC
Siemens	E188CC/7308
Sylvania JAN	6922
Telefunken	PCC88/7DJ8
Tesla	E88CC
Toshiba	PCC88/7DJ8
Tungsram	E88CC

Without lapsing into audiophile review slop, I do hear differences. Some are clearly defects, but others are hard for me to explain:

1. Differences in noise/hiss (not surprising, some are in worse conditions than others, or simply lower quality)

2. Some have more surface noise from the LP than others.

3. Differences in pops and ticks. For some tubes, the pops and ticks stand our more than others.

4. Differences in soundstage width / stereo separation

5. Spectral differences. Differences in bass vs mids vs treble emphasis, almost like an EQ has been applied.

6. Differences in dynamics. Some are more compressed sounding, others are closer to closer to solid state.

Given the circuit is pretty simple, what electrical qualities of the tubes could account for this?

Or could some of this be do to changes in the loading of the cartridge caused by a change in tube? (I'd think not, but...)

 

[EERecordist]

There are threads on phono and preamplifier testing. Have you considered using something like REW feeding in in place of the cartridge, and measuring the out? There are also threads on phono test records if you want to take the cartridge into account, but test records are very limited in frequency and have unknown distortion in the cutting. Once you have some reproducible testing, you could then try ear training on your reference track music.

The impedance of a vacuum tube grid is very high. That is not going to significantly vary between tubes. If there is not a transformer input, the cartridge is going to see the 47K input terminating resistor. The grid is going to be a Megohm. There are many similarities between tubes and FETs.

Pretty wild industrial design. It would be helpful if you could post the schematic of that preamp, which should be doable taking it apart, there are tube experts on ASR.

 

[DVDdoug]

It must be a terrible design!!!

Transistors, MOSFETs, and ICs vary too but you don't have to swap or select parts to get good performance. Since tubes age, the amp should be designed to be stable and stay in-spec as long as the tube is within spec.

I'd like to see some measurements to see what's REALLY happening.

2. Some have more surface noise from the LP than others.

More gain would amplify the signal and gain without affecting the signal-to-noise ratio (with regard to surface noise) but of course, amplifying the noise makes it more noticeable unless you turn the volume down to where it was with the other tube.

Since most of this is higher frequency, frequency response changes can change that too.

3. Differences in pops and ticks. For some tubes, the pops and ticks stand our more than others.

Possibly frequency response, or again maybe gain.

4. Differences in soundstage width / stereo separation

Frequency response can sometimes affect the soundstage illusion. But I've never seen an amplifier with separation as "bad" as a phono pickup. 20dB or so from a pickup is good-enough for regular stereo and about as good as it gets. Any amplifier should be MUCH better.

"Soundstage" mostly comes from the recording and from your brain. Speakers and the room can make a difference but the electronics shouldn't affect it at all (except some headphone amps have a blend control to intentionally reduce separation).

6. Differences in dynamics. Some are more compressed sounding, others are closer to closer to solid state.

That shouldn't happen. Amplifiers are linear unless over-driven into clipping/saturation. Clipping is a "bad kind" of dynamic compression and it's heard as distortion rather than a loss of dynamics.

P.S.
It's not rocket science to make a phono preamp that's better than the records themselves. Using tubes just makes it more difficult and more expensive.

 

[watchnerd]

There are threads on phono and preamplifier testing. Have you considered using something like REW feeding in in place of the cartridge, and measuring the out? There are also threads on phono test records if you want to take the cartridge into account, but test records are very limited in frequency and have unknown distortion in the cutting. Once you have some reproducible testing, you could then try ear training on your reference track music.

The impedance of a vacuum tube grid is very high. That is not going to significantly vary between tubes. If there is not a transformer input, the cartridge is going to see the 47K input terminating resistor. The grid is going to be a Megohm. There are many similarities between tubes and FETs.

Pretty wild industrial design. It would be helpful if you could post the schematic of that preamp, which should be doable taking it apart, there are tube experts on ASR.

It's not obvious how I would use REW to do that. Feed it into a D/A and then into the amp?

I've never seen a circuit schematic of the amp anywhere.

I could try to take it apart, but it's such a hand-made thing with less than precise tolerances, I'm a little afraid I might not be able to get it back together.

 

[watchnerd]

It must be a terrible design!!!

Transistors, MOSFETs, and ICs vary too but you don't have to swap or select parts to get good performance. Since tubes age, the amp should be designed to be stable and stay in-spec as long as the tube is within spec.

I'd like to see some measurements to see what's REALLY happening.


More gain would amplify the signal and gain without affecting the signal-to-noise ratio (with regard to surface noise) but of course, amplifying the noise makes it more noticeable unless you turn the volume down to where it was with the other tube.

Since most of this is higher frequency, frequency response changes can change that too.


Possibly frequency response, or again maybe gain.


Frequency response can sometimes affect the soundstage illusion. But I've never seen an amplifier with separation as "bad" as a phono pickup. 20dB or so from a pickup is good-enough for regular stereo and about as good as it gets. Any amplifier should be MUCH better.

"Soundstage" mostly comes from the recording and from your brain. Speakers and the room can make a difference but the electronics shouldn't affect it at all (except some headphone amps have a blend control to intentionally reduce separation).


That shouldn't happen. Amplifiers are linear unless over-driven into clipping/saturation. Clipping is a "bad kind" of dynamic compression and it's heard as distortion rather than a loss of dynamics.

The tubes with 'meh' dynamics are the ones that were not NOS, just used. So maybe they're just worn out?

It might be a terrible design.

The designer, Don Garber, was known for taking an 'artisan' approach and doing stuff by ear and looks.

In the review above, this is what he had to say about it:

As stated earlier, when asked about his design goals for the Yph, Don said 'I'm not sure I have design goals other than something that sounds really good, something that when I'm done with it just makes me want to play records."

 

[EERecordist]

It's not obvious how I would use REW to do that. Feed it into a D/A and then into the amp?

Yes. There is someone here doing measurements with a low cost interface, the Focusrite. It has the D-A and A-D in one box

I've never seen a circuit schematic of the amp anywhere.

I could try to take it apart, but it's such a hand-made thing with less than precise tolerances, I'm a little afraid I might not be able to get it back together.

You don't have to take it much further apart than you have. It looks like the power supply is in one wing - you can ignore that. Then each channel gets a wing which are probably identical. The tube is a dual triode, so the circuit has 2 stages of gain which is common in phono preamps. You could probably even start a drawing, tracing connections, reading the markings on the capacitors and resistors, take some photos, it is very easy, the preamp wing is point to point without a PCB, and start a discussion on a DIY electronics forum for help. From your drawing, the DIY community can redraw it in a standard way, and analyze the design.

Some say that designer had several designs. Those capacitors may be vintage in their performance by now.

It would be a valuable service to the vintage DIY community to spend the time on it. It's probably about an hour or two.

If you really want to get into it, most towns have a hobbyist with a tube tester that has been calibrated for transconductance and other tube parameters in a number, not just pass/fail. Tube curve tracers are another rare vintage tool. There are even some very nice designs for modern tube testers driven by ADCs and DACs and controlled by a computer.

 

[watchnerd]

Yes. There is someone here doing measurements with a low cost interface, the Focusrite. It has the D-A and A-D in one box

I have an RME ADI-2 Pro.

I could also just use my regular Schiit Modius.

You don't have to take it much further apart than you have.

To clarify, those photos aren't mine or from my unit.

Those are from the review link.

If you really want to get into it, most towns have a hobbyist with a tube tester that has been calibrated for transconductance and other tube parameters in a number, not just pass/fail. Tube curve tracers are another rare vintage tool. There are even some very nice designs for modern tube testers driven by ADCs and DACs and controlled by a computer.

Good idea, although a number of the tubes I have are NOS and have measurements.

For the Matsushitas, for example, I have E Tracer labels that cover I (ma), rp (ohms), gm (umhos), mu (V/V).

 

[EERecordist]

I have an RME ADI-2 Pro.

I could also just use my regular Schiit Modius.

You are set. That would work well with REW.

To clarify, those photos aren't mine or from my unit.

Those are from the review link.

Hahaha, your mission, should you decide to accept, is to take apart yours! There are probably local techs who would be competent to do so. That kind of construction is easy and safe to disassemble, though it is a good idea to discharge the power supply capacitors. Just find a tube tech who feels comfortable with it based on the photos you showed.

With REW and deriving the schematic with help from the DIY forums, you could write an official ASR review.

 

[SIY]

Spectral differences. Differences in bass vs mids vs treble emphasis, almost like an EQ has been applied.

That’s a clue to the design. Just looking at the photos and from parts of your description, this is likely a two tube (section) gain stage with a passive EQ between the first and second tube. Resistor plate load snd probably bypassed cathode resistor ( this circuit will need all the gain it can get, that family of tube will only give you 25-28dB of gain per stage). No cathode follower (though maybe there’s a transistor hidden inside to be an emitter or source follower, but I’ll bet against that).

So the output impedance will be high and the EQ will have a strong dependence on the plate resistance of the first tube, which varies from tube to tube. Ditto the gain. Noise variations between tubes is also normal.

 

[Dustyc]

Sweet! I always wanted to grab a Garber set ( X amp, Y preamp and the phono stage), hook them up to whatever horn loaded speaker tickled my fancy and let it rip! Since the circuit is simple, not surprised that "tube rolling" has an impact as you noted.

 

[watchnerd]

That’s a clue to the design. Just looking at the photos and from parts of your description, this is likely a two tube (section) gain stage with a passive EQ between the first and second tube. Resistor plate load snd probably bypassed cathode resistor ( this circuit will need all the gain it can get, that family of tube will only give you 25-28dB of gain per stage). No cathode follower (though maybe there’s a transistor hidden inside to be an emitter or source follower, but I’ll bet against that).

So the output impedance will be high and the EQ will have a strong dependence on the plate resistance of the first tube, which varies from tube to tube. Ditto the gain. Noise variations between tubes is also normal.

Yes, the review says it's a passive design.

Does the plate resistance vary by tube model or just random production quality control variances?

Would swapping tubes 1 and 2 (i.e. same model, just move sockets) possibly make a difference?

 

[SIY]

Does the plate resistance vary by tube model or just random production quality control variances?

The latter. Yes, swapping tubes between sockets may change things. It's a major defect in the design.