What is Zinky Thinking?

[octal]

regarding speaker Q:

I don't have the specs for the Jensen 10" alnico speakers but I remember a little bit about speaker Q in general.

Qms is a measurement of the control coming from the speaker’s mechanical suspension system (the surround and spider). I don't recall the Jensen 10" alnicos to have much in the way of a stiff mechanical suspension. I've always thought them to be loose and floppy with flimsy cones.

Qes is a measurement of the control coming from the speaker’s electrical suspension system (the voice coil and magnet). I don't recall the J 10" alnicos having a whole lot of voice coil/magent strength. Certainly nothing in comparison to what modern ceramic magnet speakers have.

Qts is the ‘Total Q’ of the driver.

Qts = (Qes * Qms) / (Qes + Qms)

That said, I can't imagine how a floppy alnico Jensen in an open backed cab would make this stuff relevant. This stuff would really matter in a sealed or ported cab, but for open cabs nobody even bothers with Thiele-Small math.

Of course, I'm no expert. Anyone?

thanks everyone for posting.

I'm not an expert either, but Qts should have an effect on low end frequency response, even in an open-backed cabinet. You are totally right that it's more critical in a sealed or ported enclosure though. The reason no one does the Thiele Small math for open backed cabinets is that the Qts (referring to the 'system' Q) in the cabinet will be just about the same as the Qts of the raw driver. The normal "Thiele-Small" parameters for speaker design are based on the assumption that an amp with a reasonable damping factor/output impedance is being used. Just looked up a Jensen alnico 10"= Qts = 1.44. Driven with a high damping factor SS amp, there's going to be a peak in the bass response at the Fs resonant frequency. Drive it with a non NFB amp and there's going to be a significant peak, as Qts will increase due to the increase in Qes. Somewhere there's graphs showing low end response vs. Qts. once you start getting over .9-1.0, there will be a peak at Fs.

[pdf64]

There's quite a lot of gain after the reverb tank, with little high pass filtering; perhaps there was a tendency for excessive rumble / very low frequency instability without that extra roll off in place?

[bob p]

The speaker is going to "flop" and try to reproduce the LF content regardless of cabinet- of course, less efficiently in an open back cabinet. In fact, I would think filtering would be more relevant in an open back box where there is no cabinet effect on maximum cone excursion. In a closed cabinet, cone movement can be limited by cabinet volume and porting. I'm thinking that the filtering in this amp is to tighten low end and make it less "mushy" by using a HPF to limit low frequency. It would also increase amp efficiency by cutting those unwanted lows.

i guess that's an effort at "pseudo-damping".

I was wondering if it might have something to do with the location being immediately after the effects return, as if they might be trying to deal with an abnormal LF signal coming in through the return. i can't imagine what that might be, though.

[Gnobuddy]

In those cases the amp sounds like it doesn't have NFB even though it has a small residual amount. How does that compare as far as speaker damping is concerned?

We can get at at least part of the answer:

Output resistance of an amp with feedback = output resistance (without feedback) divided by loop gain.

So if you apply enough feedback to halve the open-loop gain (6 dB), you also halve the amps output resistance.

If you only have negligible feedback - say half a dB - the gain is only being lowered by 6% compared to zero feedback. Output resistance is also hardly changed, only 6% less than it was with no feedback at all.

Considering that the output resistance of most pentode (or beam tetrode) output stages is way, way higher than the actual speaker resistance, its probably safe to say that a few percent change in amp output resistance will have very little effect on the speaker damping. So I think it is fair to say that your ears are being entirely accurate when they tell you that there just isn't any real difference between a teensy-weensy amount of negative feedback, and none at all.

Is the the near-zero NFB condition with the high value resistor going to cause speaker damping to be improved significantly over the open loop condition?

I don't think so, no. The amps output resistance and the speaker voice coil resistance are in series as far as the back-emf from the speaker voice coil is concerned, so it is the combination of these two that determines how much electrical damping the speaker experiences.

If you imagine the amp starting out with infinite output resistance (current source), the speaker has no electrical damping at all, and its Q will basically be Qms (assuming no enclosure, speaker in free air).

Now if the amps output resistance is progressively reduced, there will come a point when it equals the speaker voice coil resistance; at this point, amp and voice coil resistances have equal effect on the speaker damping. Continue to reduce the amps output resistance, and it will have less and less effect on Q, because the damping currents in the loop are now limited mostly by the resistance of the speakers own voice coil, and not the amp. If you get the amps output resistance all the way to zero, the speaker now settles down at Q = Qts.

Putting all this together, it seems reasonable to say that in a valve amp with output pentodes or beam tetrodes, and no negative feedback, the Q of the speaker is probably pretty close to Qms (which is usually high enough to cause an absolutely horrifically huge bass peak near resonance).

According to the Aiken Amps website, typical feedback in valve guitar amps is maybe 6 dB to 10 dB. That would lower the output resistance of the amp by roughly a factor of two or three respectively, compared to no feedback at all. Considering how high the data-sheet anode resistances of pentodes and beam tetrodes are, and even with the step-down ratio of the output transformer in between, my guess is that this would still leave the amps output resistance much higher than the speaker voice coil resistance - and, therefore, the speaker Q still stuck somewhere very near Qms, i.e., way too high for a flat response.

As an example, an old Tung Sol 6V6 datasheet lists an anode resistance of 77 kilo ohms at 315 volts. Suppose you have the usual 8 ohms to 8 kilo ohms output transformer; that's actually 8 ohms to 2 kilo ohms on each half-primary, so an impedance ratio of 250 times. That would lower a 77k anode impedance down to 308 ohms on the speaker side of the OT.

Now, three hundred ohms is still way higher than the roughly 6-ohm DC resistance of a typical 8-ohm speaker. If we added the maximum 10 dB negative feedback that Aiken mentions, we could get that three hundred ohms down to around one hundred ohms. And guess what, one hundred ohms is still a heck of a lot larger than 6 ohms! So the speaker is, for all intents and purposes, being driven by a current source, and its Q should still be hovering right around Qms.

This may also explain why nobody bothers to design guitar speaker cabs (for valve amps) using the Thiele-Small parameters. Why even bother, when the free-air speaker Q starts off at a value of four or five or worse (Qms)? You know you will have a huge hump in the bass, and the worst transient response you can imagine, and it can only get worse when you put it in a box (cab) and the Q goes even higher. The nice engineering value of Qts is completely out of reach - no way to get there unless you have a solid-stage guitar amp!

I think I now also know why you virtually never find Thiele-Small parameters for guitar speakers. Once again, why bother, when the high output of your guitar amp is going to push speaker Q virtually all the way up to Qms?

I have never actually measured the output resistance of a no-feedback valve amp, so this is a bit of guesswork (based on reasonable data, though) on my part. Please feel free to correct me if you know better.

-Gnobuddy

[Chuck H]

Zinky hasn't worked directly with Fender for a long time as far as I know. Some of their amps are based on Zinky designs. Also, Zinky worked a lot of prototypes and I've suspected before that some prototype circuits that might have later been simplified were simply executed rote. In other words, the guys designing the final board might have simply included a circuit tacked onto a circuit because that's what was there.?.

I'm also game for Pete's suggestion of rumble control.

EDIT: Ok... On closer inspection I think the amp is just weird. WTF is up with the effects loop send straight off the mixing pot? And if you're going to use a roach trem, why suffer two triodes to do it? And what's with the dual triodes at the input with one dedicated to the reverb? Not very good economy of design IMO. So, as to why the dual HP filters, no clue. Maybe Bruce just thought a 30* phase shift was needed.?. But really, who knows?

[Justin Thomas]

I won't pretend to understand "why" Bruce Zinky did anything any particular way, but having had my hands on a real one and being able to jam it out a little, it's a damn nice amp. Very crisp, bright, clear, very hard to dial in Fender Fart (hey, the Bass control was actually USEFUL above 2!). That trem defijitely sounds a little more complex than the usual Fender roach approach. Probably as close as I've ever played to my 62 Concert. I'd say the VK & possibly the Dual Professional of the 90s (moreso than the Tone Master & Prosonic) were Zinky's best efforts to "out-Fender Fender." If only I had the $1500 to buy yet another amp... I also like the silk-screened eyelet board, with component designators and values printed on the black fiber.

I let my buddy who usually uses a last-British-issue AC30 borrow the VK while his Vox was in for service, and he said he'd rather use the house Twin Reverb RI - the VK was "TOO LOUD."

So, whatever Zinky did, he did right...

Justin

[bob p]

We can get at at least part of the answer:...

Thanks for the very thoughtful post. I have a question though:

...
As an example, an old Tung Sol 6V6 datasheet lists an anode resistance of 77 kilo ohms at 315 volts. Suppose you have the usual 8 ohms to 8 kilo ohms output transformer; that's actually 8 ohms to 2 kilo ohms on each half-primary, so an impedance ratio of 250 times. That would lower a 77k anode impedance down to 308 ohms on the speaker side of the OT.

Now, three hundred ohms is still way higher than the roughly 6-ohm DC resistance of a typical 8-ohm speaker. If we added the maximum 10 dB negative feedback that Aiken mentions, we could get that three hundred ohms down to around one hundred ohms. And guess what, one hundred ohms is still a heck of a lot larger than 6 ohms! So the speaker is, for all intents and purposes, being driven by a current source, and its Q should still be hovering right around Qms.
...

I'm thinking that an 8 ohms to 8 kilo ohms OT would have 4 kilo ohms on each half-primary rather than 2 kilo ohms.

This causes a ripple in the numbers that doesn't seem large enough to change the conclusion, though.

[pdf64]

...Output resistance of an amp with feedback = output resistance (without feedback) divided by loop gain.

So if you apply enough feedback to halve the open-loop gain (6 dB), you also halve the amps output resistance....

Bear in mind that the loop gain at speaker resonance is likely to be somewhat higher than at nominal, as the impedance may be >10x nominal; so even a bit of NFB may have more of a damping effect than might otherwise be apparent.

From memory, at high signal levels, OL output impedance of regular tube guitar amps tends to be ~3x nominal load impedance; at low signal levels is a fair bit higher, maybe 10x.

[darkfenriz]

Yes you could achieve an identical enough transfer function with just one RC highpass. What is different is the load line of V4a and thus overdrive characteristics. With two RC filters the load is somewhat lower up to say 120-150Hz, maiking it less prone to farty bass overdrive (not sure if overdrive is supposed to happen here).
Nothing to do with a speaker

[bob p]

Zinky hasn't worked directly with Fender for a long time as far as I know. Some of their amps are based on Zinky designs. Also, Zinky worked a lot of prototypes and I've suspected before that some prototype circuits that might have later been simplified were simply executed rote. In other words, the guys designing the final board might have simply included a circuit tacked onto a circuit because that's what was there.?.

I'm with you -- I was thinking the same thing about something just being mindlessly carried over. But when I looked at the 4 different schematics, it looks like Zinky is credited for drawing and signing-off on revisions A, D, E and F.

The schematic drawing date on Rev. F is 9-20-95 but the engineering change note for Rev. F is dated 5-17-97. I can't explain the date discrepancy, and I don't know the date when Zinky left Fender, but it looks like he's getting credit for the schematic revisions. This makes me think that everything we're looking at was done intentionally.

EDIT: Ok... On closer inspection I think the amp is just weird. WTF is up with the effects loop send straight off the mixing pot?

I think I can help to put some of this into perspective.

I also thought that the idea of a CF driving the mixing pot, which fed the dry signal to the FX loop was odd, though it seems useful to have a buffered FX drive level control. Upon closer inspection, it looks like this amp has taken the 6G13 reverb unit (or more accurately, the 6V6-driven "63 Reissue" reverb unit) and cut-and-pasted it at the front end.

Back in the days when people would run their guitar into a reverb unit and the reverb unit into their no-reverb amp, they effectively had the CF driving the mix pot, and the mix pot going to the output. It looks like Zinky's intent was to preserve the existence of a cable connection between the reverb output and the amp's input and to call that an FX loop.

And if you're going to use a roach trem, why suffer two triodes to do it?

Two triode stages seems to be de rigeur for a roach trem -- first triode does the oscillating, second one drives the lamp. Looking at this amp, I get the impression that he just cut and pasted an AB763 trem circuit.

And what's with the dual triodes at the input with one dedicated to the reverb? Not very good economy of design IMO.

The idea for a cathode drive buffer for the clean signal is taken from the external reverb unit. This VK implementation actually has more economy of design than the reverb box, as it only uses a single stage of a 12AX7 to drive the 6V6, while the original 6G13 used two stages of a 12AT7 to drive the 6K6 and the 63 Reissue used two stages of a 12AT7 to drive the 6V6. So it seems that in going to the 12AX7 for drive, Zinky eliminated one triode. That was probably done to avoid sticking another two-triode tube in the amp while utilizing only half of it.

(Maybe the bean counters are responsible for this. I had heard stories that after Zinky left Fender, he produced and amp called the "Electroverb" [or something like that] allegedly because that's the amp that he wanted the VK to be. I don't know if that's truth or marketing fable. Looking at this schematic I get the impression that Zinky had free reign to do whatever he wanted. The only place where I can see a compromise is in the conservation of a triode by changing the reverb driver to use 12AX7 instead of 12AT7.)

So, as to why the dual HP filters, no clue. Maybe Bruce just thought a 30* phase shift was needed.?. But really, who knows?

I'm thinking that Pete and Gnobuddy have plausible suggestions. Looking at the reverb section of this amp I see that there's been some gain-shaping in the cathode bias circuit to shave away LF response of the signal going into the reverb driver. That makes sense as LF into a reverb tank produces a lot of mud. Maybe the double filter thing on the FX recovery stage is to try to clean up spring artifacts?

[bob p]

I won't pretend to understand "why" Bruce Zinky did anything any particular way, but having had my hands on a real one and being able to jam it out a little, it's a damn nice amp. Very crisp, bright, clear, very hard to dial in Fender Fart (hey, the Bass control was actually USEFUL above 2!). That trem defijitely sounds a little more complex than the usual Fender roach approach. Probably as close as I've ever played to my 62 Concert.

Interesting. My recollection of the concert is that it had the "harmonic vibrato" tremolo system. Looking at the trem circuit in the VK, it's nothing more than a cut-and-pasted AB763 bug circuit. So I'm surprised that it sounds different than the typical bug circuit, especially given your trained harmonic-vibrato ear. But I can say that to my ears the plain-Jane bug circuit does sound different in that amp, and I can't explain why.

I'd say the VK & possibly the Dual Professional of the 90s (moreso than the Tone Master & Prosonic) were Zinky's best efforts to "out-Fender Fender."

I've never heard the Dual Pro or the Prosonic though the Prosonic gets a lot of love on the forums.

I have heard the VK played live, back in 1995 when it was a new product, and that was a transcendental experience. By chance I happened to be in Bloomington, IN and hit a local nightspot to catch two folk rock bands touring together, 16 Horsepower and The Innocence Mission. I had always thought that IM's Glow CD had very tastefully recorded reverb and tremolo on the guitar parts that was captivating in a haunting Mazzy Star sort of way (this was the mid-90s). I got an earful of clean Vibro King that night. All I can say is, "WOW!". In the right hands (where it wasn't pushed into gratuitous distortion) it was quite an amp. The clean Gretsch tones were amazing even at high volumes, in spite of the amp having no NFB loop. Zinky definitely did something right.

My only other experience hearing a Zinky amp in an optimal environment was at the House of Blues in Chicago. Another chance event, I happened to worm my way into the booth at the last minute for a sold-out Midnight Oil show. One guitarist was playing a Strat into a pair of AC30 and had the characteristic jangly tone in spades. The other guitarist was playing a P90 Goldtop into a Tonemaster half stack. I saw God that night. The mix was beyond outstanding. The combination of 56 LP and the Tonemaster 4x12 with V30s that night was truly magical -- to this day I think it has to be the best rock tone I've ever heard live. The Tonemaster was voiced for that 412 V30 cab, and that combination is another case where Zinky definitely did something right.

If only I had the $1500 to buy yet another amp...

My problem is that I just can't find the room to squeeze one more amp into the house. I can't justify owning a Tonemaster stack just for my P90 LP.

[bob p]

Yes you could achieve an identical enough transfer function with just one RC highpass. What is different is the load line of V4a and thus overdrive characteristics. With two RC filters the load is somewhat lower up to say 120-150Hz, maiking it less prone to farty bass overdrive (not sure if overdrive is supposed to happen here).
Nothing to do with a speaker

i don't doubt that the load impedance has significant effects on the load line. what i'm trying to figure out is why they used two series RC circuits instead of just using a different resistor in a single RC filter to establish the same load impedance.

[tubeswell]

The two switchable vol pots after V4b in the Dual Professional are weird as well. I can't really see why this makes any difference in a guitar amp.

So are the multitude of separate ground return wires in his layouts. And the weird location of the bias pot underneath the main board in the tonemaster (and all the parallel filter caps in the same amp).

[Gnobuddy]

The first filter, comprised of C5 and R21, looks like it amounts to a subsonic filter.

The second filter, comprised of C6 and R15, looks like it has a musically relevant corner frequency in order to throw away some LF content.

I simulated this in LTSpice just to see if there really was anything clever going on. I estimated a 40 kilo ohm source resistance from the 12AX7 (100k anode load in parallel with the internal anode resistance rp), and tacked that onto a voltage source to simulate the triode driving the filter.

Nope. Nothing clever going on that I can see. From 10 Hz up, it just acts like a normal first-order high pass filter. Shorting out C5 (in the simulator!) barely changes the frequency response. The two filters have the same resistance, so they load each other very heavily, and never really have a chance to behave like cascaded filters. The pot has nearly the same resistance as well, so it loads the heck out of the filters too.

Wanting to extend Zinke the benefit of the doubt, I extended the simulation down to 1 Hz. Nope, nothing clever happening down there, either.

I have no idea what Zinke was thinking, but, thanks to LTSpice, I do know what his circuit does: it wastes one R and one C, pretty much as Bob P. surmised from the start.

There are two files attached; a screenshot of LTSpice simulating the circuit (in this case, with C5 shorted out), and a Zip file which unzips to the LTSpice .asc file, in case anyone else wants to try running the simulation too.

-Gnobuddy

[J M Fahey]

The second RC filter has a 93 Hz crossover and can be seen as a subsonic/mud killing network.
The first one has a 1.5Hz cutoff and is an absolute waste of components.
Why is it there?
I have seen many no-clue "designers" switch parts at random, until they hear something they like, then they superstitiously freeze the design at that point, fearing to lose what they achieved, so they freeze it errors and all.