Watt Sucking Fireball Series

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Aaaaaaaaand

Again the principle is to lower primary impedance and to drive output tubes in counter-phase independent grid-cathode drive.
 

Attachments

  • Oct 1951 p5.jpg
    Oct 1951 p5.jpg
    105.3 KB · Views: 712
  • Oct 1951 p2.jpg
    Oct 1951 p2.jpg
    78 KB · Views: 682
best method?

And Yamaha also considered diff. SRPP stage (in their pat. 4241313 with a bunch of SIT amps circuits)

The Yamaha B1 amplifier was an output structure not too dissimilar to the WSF-4. The WSF-4 style of design was derived from the Sinclair-Peterson concept long before I ever serviced a Yamaha. The B1 uses a differential current source from a far away minus rail called "C" as well, however Yamaha uses another pair of "SITs" to drive the output pair. This was due to a need to reduce the gate drive Z because of a fundamental flaw in the device geometry that could not be economically solved at that time. That flaw eventually caused the cessation of production as well as the damning advent of vertical MOS in the late 70's. The B2 version was an entirely different animal using N and P channel "SITs" in a complimentary output stage but not as elegant in design and not as clean in my estimation.
 

Attachments

  • Yxistors.JPG
    Yxistors.JPG
    529.2 KB · Views: 696
Last edited:
PS

This illustrates the fundamental concept as originally promoted and as I prefer to configure it. It should be noted that the filament winding capacitance for the upper tube should be in a guard shield coupled to the output. You can just imagine where the displacement currents at high frequency would go otherwise.
 

Attachments

  • Sinclair-Peterson-Simple.jpg
    Sinclair-Peterson-Simple.jpg
    290.9 KB · Views: 675
Oh, you like differential SRPP, do you?
But try to make a power stage of this!
Differential White Cathode Follower...

If the pic is a little confusing, cause of
5VPP CM @ 400Hz + 5VPP PSR @120Hz
just showing how it rejects all that...

I got a bunch of General Radio stuff,
masterpiece of my collection one of
their slotted lines. Never figured why
GR874 connectors didn't catch on...
 

Attachments

  • wakko3.gif
    wakko3.gif
    34 KB · Views: 619
Last edited:
I got a bunch of General Radio stuff,
masterpiece of my collection one of
their slotted lines. Never figured why
GR874 connectors didn't catch on...

They were 14 mm and 7mm was the next venue represented by Hewlett Packard APC-7 the ones used in microwave after that were 3.5mm but I agree with Tektronix that the GR-874 was hands down the most versitile connector overall through 8 GHz. Even the 900 was not as generally useful though more precise for what I use them for. All my lab setups are dominated by 874 interconnects from audio to RF.
 
6DJ8 model? Probably Duncan Amps' if I recall...
Or you asking which brand real 6DJ8 do I use???

Amperex mostly, from my dad's collection of HP
and Tektronix pulls. I prolly got 50 of those...
Sylvania and Telefunken too, but not so many.

"NH" No Heater... Yeah, definitely Duncan..

But back to the point. You are about the only
other person I've ever seen abusing a split tail
CCS with cap bridged across. And doesn't look
like the first time you have played with that
particular configuration?
 
Last edited:
Hmmmm

"NH" No Heater... Yeah, definitely Duncan..

Thank you for that

But back to the point. You are about the only
other person I've ever seen abusing a split tail
CCS with cap bridged across. And doesn't look
like the first time you have played with that
particular configuration?

The reason for using an impedance coupled source pair is manyfold. It first allows me to insert a complex impedance if I want to control bandwidth or rolloff in any direction. This method also allows me to fiddle with DC conditions that would otherwise ruin the common mode rejection on the first order, when for instance, as in the WSF-4 driver stage, an output servo loop is integrated. You can see how bloody simple that worked out. This is a technique that I used frequently years ago in servo controls when tailoring the response and drift characteristics of high power motor control amps. Since audio IS AC and only AC I have no druthers about liberally employing stabilized AC gain throughout where needed in an amp design. This mechanism also allowed me to produce a three phase differential amp once for a three phase signal balancer in an energy converter. For other reasons it also allows you to mix different current loaded circuit elements and not have to worry about otherwise complex DC balacing tedium. Of course phase considerations in an audio application need to always be carefully managed particularly when any feedback is used. Much simpler when the amp is a zero feedback design, which overall I prefer but needs more time in the design bucket to make worthy. Note also the DC biasing for the coupling caps to maintain polarization and keep leakage low. Other than that, I have little opinion on the matter:D

K-wood

PS: Ran your model...........looks great even using 12AT7's
 
Last edited:
yet there is more

One of the clever things you can do is to use different device types for a differential stage, like using one PNP and one NPN with independent current sources in the emitters and cross coupling with a cap or suitable complex impedance of the order needed. Then folding the pair by mirrors to a summing node of the next stage loaded with a resistor to control gain and loading impedance for the first pole rolloff. Check out some of the stuff 4QD-TEC does with complimentary differentials. A very clever mind. In addition, differentials using different types of devices like a (Bipolar and J-Fet) or (MOS fet and J-fet) or (bipolar and triode) differential lends a bit of circuit finesse out of the habitual box we reside in and may make a design possible that would otherwise be overlooked. Only imagination limits. The only thing that really matters is CONCEPT. As long as it has definable merit, who cares what the circuit looks like. As my father used to say, "The only problem is, defining the problem, that definition is everything you need to know".

With all regard .......................K-wood
 
Last edited:
I've done the Triode/Pentode and or Triode/IGBT dance on top the split tail pair.
The higher GM device (as the "top end" of the folded cascode) forces itself to
become anti-compliment. Overall much like SE sounding, but twice the power.
And with forced DC balance, no need for oversized gapped SE OPT...

Wierd thing: You can drive into either side or both, changes nothing. The Big
GM device behavior patterns after the smaller GM (usually the Triode). But too
similar GM, you get blended behavior that is folded cascode both ways at the
same time. And if the devices are the same, dumbs itself down to push-pull.

Bridging PNP with NPN to make a differential never occurred to me??? But I'll
remember now to abuse that trick somewhere down the road.
 
Again, looks familiar... But maybe just the choice of components up front?
I never considered this one as differential in any way, just a pair of offset
followers. Why my obsession with followers???

I tried numerous times to SRPP this drive into the MOSFET gates, but the
sims kept coming up oscillator. So it never went any further... Bandwidth
limited by how fast one can symetrically slew the gates...

Like for instance, drive the gates from the emitters Q1 Q2, seems like a
plausible thing to do... But LTSpice keeps tellin me otherwise...
 

Attachments

  • JFork1.gif
    JFork1.gif
    55.8 KB · Views: 662
This was one variant with SRPP, I don't recall if it was a "working" sim or not.
Several in my series of thought experiments too easily broke into oscillations...

I've just re-run this sim to see if it works. Yeah, but nothing to be proud of.
Too many bugs and squirrels. Dioded for Class AB, but still biased for Class A.
Unfortunately, I can't locate the more evolved version to ask your opinion.

Mainly I wanna know why my SRPP drive makes the MOSFETs go ape?
I'm assuming your drive SRPPs of Post#1 are a bit more stable?
 

Attachments

  • jfets99.zip
    1.7 KB · Views: 115
Last edited:
Well.............

Sometimes, what appears to be a good idea in the beginning will have need to be chucked. My circuits professor Dr. Slottow kept encouraging us to take whatever idea we had to solve, a problem we thought was clearly defined, and thow it out and start over as though the previous idea had never happened. We were to keep doing this until only one simplest approach remained. A discipline that takes some practice because pet ideas are exactly that. If you examine the node currents in each leg of the circuit and they bear little resemblance to the through signal, then a great deal of correction must be going on to reassemble them into the final waveform. Those corrections, that I see, are so many, I'd start over. :( A linear amplifier usually looks like a linear transfer function everywhere and coupling impedances throughout should generally be constant to maintain lowest intermod performance. Loop gain may be cheap but it sure sounds like hell.
 
Last edited:
Futsing

Ya know the 99 version really doesn't have a lot of problems until the limits are being reached and the rail crash becomes problematic.........so since a 3 tone shows a several decade down intermod floor, and if it never reaches that range, what's the big deal? Most of the oscillations are when an internal source impedance takes a wild swing to compensate for the combined nonlinearities of some circuit elements, like a bipolar square law combined with an exponential of an insulated gate device. Even worse is a phenomenon with MosFets at the edge of turn on, right at the threshold voltage edge, thier effective high frequency gate input resistance goes negative and causes any parasitic elements in the circuit layout to form an incidental oscillator that can spin at several hundred megahertz. While the MOS device transits through these high frequency bursts near threshold, GM of the device will drop at the lower frequencies and the resulting shift in impedance causes a distinct distortion notch at that point. It's one of the reasons to insert a critical input resistance in series with the gates of paralleled MosFets to kill Q and prevent cross oscillations. Not due to the same reasons but similar in effect is the HF oscillation of new 12A-7 anything of the new shorter plate construction. For instance, a Fender guitar amp that was designed and "finessed" to work with the older long plate 12AX7's often will sound like junk when a Shuguang short plate triode is substituted. The curve trace of it and the older RCA/Sylvania/Etc. look the same but the electron transit time for this new construction is shorter and it's higher fT can interact with the distributed capacitance and residual inductances of the wiring and produce a 200-300 MHz oscillator. While the tube is wailing away at high frequency the low frequency gain suffers and becomes nonlinear. It takes a wideband sampling scope to see the oscillation. Best way to fix it and still use the newer construction triode is to put a couple of lossy ferrite beads in the cathode to kill the high frequency gain. Suddenly, the amp sounds good again and those Chinese triodes aren't so bad after all. We often are only looking at an amp's low frequency performance and forget to identify a high frequency parasitic that makes an audio amp uncompensatable or unstable. Many discrete parts have bandwidths out to the hundreds of Mhz's and often are committed to a layout that will guarantee instability during some transient condition and the design gets prematurely tossed for lack of a single HF rolloff cap or a HF gain killing ferrite bead in a source or emitter. Models are getting better and some of this appears in the design model, mostly however, not. Spice seldom sees what your layout residuals are. A good test to see if the circuit is oscillating is to use your finger to probe around the circuit often loading a node just enough to change or stop an oscillation, moreover to show that one is there when performance changes.

K-wood
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.