|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
24.22.235.48
In Reply to: RE: Baby Ongaku 2A3 posted by drlowmu on October 11, 2014 at 08:44:02
Among other things, it would need a 500v power supply - a minor consideration I suppose if you are following his advice to throw out the entire power supply and replace it with much larger parts! But that is not relevant to someone who already has an existing amp, and it defeats the design purpose of making a smaller cheaper amp that still sounds "like" the Ongaku.
I personally agree with him about the driver; in my experience the difference is small but real. You can easily connect the interstage capacitor to the lower tube plate instead of the upper tube cathode, and hear the difference for yourself.
Follow Ups:
> > You can easily connect the interstage capacitor to the lower tube plate instead of the upper tube cathode, and hear the difference for yourself. < <
Sure can. It won't have the same drive but it's an easy choice for me. This makes it a different circuit of course ... tube CCS, which although is less than ideal still sounds better than SS CCS to my ears.
Naz
> > connect the interstage capacitor to the lower tube plate instead of the upper tube cathode, and hear the difference for yourself. < <"This makes it a different circuit of course ... tube CCS"
AC currents through the two sections in the schematic linked by Paul are nearly identical. The upper tube is in no way acting as a CCS, regardless of takeoff point. I am highly suspicious of any claim to a sonic difference relative to the cited circuit considering that only 333 ohms separates the two points being discussed. That resistance is virtually a dead short in comparison to the AC resistance of the circuitry in which it is operating.
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/16/14 10/16/14 10/16/14
True that the AC currents are almost identical but that's about where the likeness begins and ends.
Taking drive from the top tube makes the circuit an SRPP. The top tube is effectively a cathode follower with a less than ideal CCS in its tail. OP impedance is low, all good things on paper. Taking drive from the bottom tube is essentially equivalent to a current sourced common cathode circuit, higher OP impedance but...
I'm sure I said that the CCS is less than ideal but it is, nonetheless a CCS of sorts because it multiplies the effective plate load R of the bottom tube by mu+1.
To act a more ideal CCS a change in configuration is required, allowing the use of a much larger plate R and hence a more horizontal load line, picture a mu follower with the OP taken from the bottom tube's plate.
Regardless of the apparent likeness of the circuits I'm with Paul on the difference in sound and if you couldn't hear the difference I'd be very surprised. Try it for yourself, don't take our word for it.
Naz
"Taking drive from the top tube makes the circuit an SRPP."
No it doesn't. This circuit *looks* like SRPP, but the values are wrong, so it doesn't create SRPP functionality. There is NO push-pull action.
"I'm sure I said that the CCS is less than ideal but it is, nonetheless a CCS of sorts because it multiplies the effective plate load R of the bottom tube by mu+1"
It functions as a simple resistance. It's not a CCS to any degree.
"I'm with Paul on the difference in sound"
Fine, but the conclusions you've reached don't logically translate to the negative generalities being voiced on this forum. SRPP is a highly useful and very transparent topology when done right.
--------------------------
Buy Chinese. Bury freedom.
The circuit configuration IS recognised as an SRPP regardless of whether the circuit is in balance with proper PP operation or not. I never referred to its operation nor attacked it, that's a whole other discussion.
I understand what you are saying about balance and PP but that begs the question why you would use the circuit in this particular application considering the load characteristics and what load would you balance it for?
Sure the CCS performance aspect is far from ideal and could easily be improved so if it makes you happier thinking of it as simply a higher value resistance I would not argue the point.
Taking your point, what name would you give the circuit?
Naz
First, this is absolutely *not* about semantics. If we refer to a capacitor as a resistor, or a transformer as a battery, that's not semantics. It's simply incorrect. The same holds true if we refer to the upper triode of the pseudo-SRPP as a CCS."The circuit configuration IS recognised as an SRPP regardless of whether the circuit is in balance"
It hasn't been "recognised," it's been misidentified. I've referred to it as SRPP myself in some discussions, but only to establish a mutual understanding as to the construction. When the term is used in discussions having to do with functionality - and a high-Z load is involved - the term "SRPP" is being misapplied. That invariably leads to misconceptions regarding the sonic qualities and suitability of the circuit.
"I never referred to its operation nor attacked it"
Didn't you say that it sounds better when audio is extracted from the lower triode? That means - in your opinion - the SRPP configuration is inferior.
"I understand what you are saying about balance and PP but that begs the question why you would use the circuit in this particular application considering the load characteristics..."
I wouldn't even consider using SRPP in this application (Baby Ongaku). SRPP was originally designed to drive low Z loads. In that mode, the two sections operate in push-pull as intended. With a high-Z load, it's no longer SRPP. The circuit operates as a SE amplifier, and the benefits originally ascribed to it evaporate. In addition, because the two sections are operating at one-half the B+ (and the voltage is typically not raised to accommodate this fact), the tubes are frequently being used in a region where they perform poorly. I believe this might be the reason you find the SRPP to be inferior.
"...and what load would you balance it for?"
I have little use for this circuit in its balanced, push-pull mode. That would require designing it for a fixed load impedance, one that is very low, probably below 200 ohms. Where I find the circuit useful (and exceedingly transparent) is in high Z applications where a high output voltage swing is needed. In this use, B+ is increased substantially, simultaneously raising the available voltage swing and reestablishing the normal operating conditions for the tubes. For example, a 6SN7 or 6BL7 could be configured in pseudo-SRPP and used with a B+ of 500-600V. This voltage can't be safely used on most single driver triodes, so "SRPP" construction is essential to meeting the design goals.
"if it makes you happier thinking of it as simply a higher value resistance I would not argue the point."
The upper triode in the Baby Ongaku can be replaced with a fixed resistance of approximately 36K for virtually identical performance. If you want to call a 36K ohm resistor a CCS, you *will* get an argument. :)
"What name would you give the circuit?"
Anything other than SRPP. Some people call it a totem pole. That's OK with me. Incidentally, these concepts regarding the SRPP are not something I invented. Discussion of these points can be found in the Tubecad journals, Valley and Wallman’s *Vacuum Tube Amplifiers*, and probably many other sources.
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/18/14 10/18/14 10/18/14
Other than pointing out how this circuit works, which many of us already know you seem to be contradicting yourself in terms of the merits of the circuit and where you would even use it. In the first place you said you've used the circuit yourself and implied it sounds good. Then, in your last post you've basically attacked it. I only said that I preferred the sound when OP was taken from the plate of the bottom tube, nothing more about about its sound quality.
The thumbnail breakdown I gave of the circuit's operation is accurate in this application. I don't know why you get so fired up about terminology when these are relative terms in the context of discussion. But I accept that we are all different and that doesn't make my POV right, it's just my opinion.
To be honest it doesn't bother me what the circuit is called, it's the operation that counts and I've read the references you provided long ago. As you've pointed out it has many names but SRPP is by far the most common IMO.
Speaking semantics, even CCS is a relative term ... at what point do you call it just a high impedance versus a CCS? To take your point to the extreme I say there is no such thing as a CCS, period, because none are perfect no matter how well designed!
The point you made about replacing the top tube with a resistor in this circuit is totally erroneous. Only the load resistor is multiplied by mu+1 when the tube is in circuit but the top tube still functions as a CF with a much lower OP impedance relative to swapping it out with a resistor. If you want to speculate about the bum rap the circuit gets for sound, perhaps the CF output should be on the list of possibilities.
Naz
"you said you've used the circuit yourself and implied it sounds good. Then, in your last post you've basically attacked it."Not at all. I've only attacked the form of the circuit that is being misused. I've pointed out exactly what constitutes that misuse, and the two ways in which the construction we're referring to as SRPP works well.
"the top tube still functions as a CF with a much lower OP impedance relative to swapping it out with a resistor"
It's neither a CCS nor a cathode follower. Take a look at the two circuits below, and tell me if you think they function the same:
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/18/14 10/18/14
I don't see any circuits but if you substitute the top triode with a resistor the bottom tube obviously functions as a grounded cathode.
I will concede however that in order for the top tube to function as a CF the resistor value needs to increase significantly. Can we agree that as the value of R increases the circuit tends more towards a CF?
Ultimately, to make the circuit something I would use the R value would need to increase significantly requiring a circuit change such as swapping the R with an L or CCS or adding a cap. I like such variations of the circuit and use them a lot but never use the circuit under discussion exactly as it. Even so I still prefer the sound with the OP taken from the bottom tube's plate.
Back to the circuit with the current values, even swapping the top tube with an R will still sound different regardless of the technicalities IMO. To believe otherwise is to believe that there is no difference in the sound of components including tubes.
Naz
"Can we agree that as the value of R increases the circuit tends more towards a CF?"
No, I don't agree. You're off the point, though. We're discussing the Baby Ongaku circuit as-is. You know, it's ironic that you're the same person who chastised me for needlessly describing this in such detail when people here supposedly already understand it. Clearly, that statement wasn't meant to be autobiographical.
"I still prefer the sound with the OP taken from the bottom tube's plate."
Does it matter? The circuit under discussion is incorrectly designed and has been misapplied. That's what you and Paul are listening to, a dysfunctional circuit with values no competent designer would use. How it sounds is irrelevant. Perform a scientific test with a legitimate SRPP, or a properly constructed totem pole running at favorable operating points, and your statement would be of interest.
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/19/14
It seems that every word must be dissected. For starters, I DON'T use this circuit and I'd be pretty damn sure Paul wouldn't either. Partly for the reasons you are pointing out but mainly because the circuit can be greatly improved (IMO).
You took a statement that is absolutely true, (our opinion of sound preference with a very minor change) to mean that it's impossible to sound different and that we somehow endorse the circuit over others. Then you complicated (or confused) things by defending it.
Where it suits your argument you speak of the circuit as is, with exactly the same values but then go on to imply that its rubbish and can be improved. At least we agree that it can be improved.
If we are going to discuss this exact circuit with the values shown then stick to it. If you want to expand to a broader argument of improvement and the like then let us know before assuming what we are thinking.
So, what we agree on is that neither of us would use the circuit. I have my own ways of doing things but I refrain from telling others how to suck eggs unless I'm asked for my opinion. Otherwise I'd have spelled out many possible changes I would make to this amp DEPENDING ON MY GOALS.
One thing I have learned though is that any amp can be made to sound good using a multitude of different topologies, even some that are technically inferior. There are so many variables. I have a technical background but experience tells me that we still can't fully explain all the reasons why some things sound different purely through engineering or measurement, at least as we know them. I'm sure this will provide more fuel for argument but it's just my opinion and anyone is entitled to disagree!
Naz
Wow - you wander off for a few days and when you come back there's a whole half-page of argument! You guys have been taking this pretty seriously ...
"... I DON'T use this circuit and I'd be pretty damn sure Paul wouldn't either. "
Just to be accurate, I DID use this driver topology (except for an unbypassed cathode resistor) in my first commercial design some 15 years ago. I did so largely on the basis that Gordon Rankin was using it a lot (still does I think) and he liked the sound. I am a believer in the sound as heard by myself and other people whose ears I trust! That's the engineer as opposed to the scientist I guess. But I'm also enough of an engineer to be wary of extrapolation - just because it sounds good in a particular amplifier does not mean it will sound good in a different amplifier.
My memory is that we later decided we liked it better with the output taken from the lower triode's plate, but for an unrelated reason I replaced the upper triode (I think "active load" is a good term that skirts the question of what exactly is a CCS) with a solid-state cascode current source originally developed by John Camille.
We had earlier found that a mu-follower was an improvement over paralleled driver triodes (in a different design), and also decided that taking the output from the lower triode plate sounded best. I have since privately suggested to a friend the cathode choke in the active load and he has been using it for around a decade with very satisfying results, again in a 2-stage 2A3 amp.
Incidentally, since it has not come up yet, the SRPP and other totem-pole configurations have the large-signal advantage of putting the active devices in effective parallel, unlike normal push-pull where they are in series at AC. Hence their advantage in driving low impedance loads.
Perhaps too seriously Paul:)
Anyway thanks for your input. I can say that technicalities aside your experience mirrors mine exactly! For some time my gain stage of choice has been the equivalent of a mu follower but with drive taken from the plate of the bottom tube. I like your term so let's call it grounded cathode with tube active load.
I found little benefit in substituting the plate load R with a choke for small signals, however adding the choke provides a decided advantage in driver circuits where large swing is required. The better (although slightly elliptical) load line provides excellent linearity, even up to the hundreds of volts P-P needed to drive an 845.
Naz
"adding the choke provides a decided advantage in driver circuits where large swing is required."
That's not correct, at least not as it's implemented in the schematic you posted here. My earlier statement on this point probably wasn't clear, but the output swing of that circuit is in fact more limited than the "standard" totem pole. That's due in part to the fact that the choke forces the two sections to operate in push pull, and that currents within each triode are extremely non-linear. If you like the sound, great, but I wouldn't even consider using this circuit in a high-end amplifier unless the SPICE analysis can be disproved.
--------------------------
Buy Chinese. Bury freedom.
First, let's clear up that the suggestions I made went somewhat further than a minor change to the operating point in the existing driver circuit. I also eliminated the coupling cap but as I pointed out this compromises OP power, which would ideally be rectified by increasing HT for the best of both worlds. The bonus is a considerable improvement in headroom afforded by A2 transition. Or, one could choose to sacrifice it by taking drive from the plate of the bottom tube, a better sounding option IMO and that of others I respect.
For clarity, I've done what you suggested earlier and concentrated on the driver differences. I've re-Simmed both circuits and went as far as optimising for the best output swing and even tweaked the operating points to be close enough to identical for fairness.
Then I set identical output swing, to the max in the case of your circuit, which yielded 240V PP. My circuit comfortably goes to 300V PP swing but I've shown the OP waveforms and ffts at 240V PP. The non-linearity in your circuit becomes visible at this OP whereas the waveform still looks very clean in mine. The ffts magnify the differences and even if they aren't completely accurate they serve to provide a compelling comparison.
First your circuit. Note that it requires a high 5.5V input signal to drive it to maximum output. Then look at the distortion level, particularly the damaging high order harmonics.
Now, my circuit. Only 2.5V input is required for the equivalent output. More importantly harmonic distortion beyond the 4thH is insignificant by comparison.
Is your model for the 12AT7 freely available? I'd like to compare it to mine.
--------------------------
Buy Chinese. Bury freedom.
Here it is but any tube will exhibit similar characteristics after optimising and equalising the operating points. Incidentally, I would have liked to see higher idle current through the 12AT7 in both scenarios but I didn't want to throw another variable into the mix for the purpose of the exercise.
.SUBCKT 12AT7_ECC81 A G K
XV1 A G K TRIODENH
+PARAMS: LIP= 1 LIF= 0.0037 RAF= 0.09869 RAS= 1 CDO=-0.5
+ RAP= 0.1 ERP= 1.4
+ MU0= 45.093 MUR= 0.012937 EMC= 0.00000863
+ GCO=-0.5 GCF= 0.00012
+ CGA=1.60E-12 CGK=2.30E-12 CAK=4.00E-13
.ENDS
Naz
OK, we're using the same tube model, it's the one from the Duncan Amps Website.Being as this discussion seems to be centering on distortion, gain and output swing, I'll elaborate on my previous points in this regard. However, I want to stress that I consider the choke-loaded totem pole that you've introduced into the discussion a kludge. I've stated why already, but you've failed to acknowledge the points I made in any meaningful way, even after I posted a SPICE analysis showing the horrendously distorted currents within the tubes.
After creating the image below, I took a last look at your latest post and realized it's not quite the same as what we were discussing. I regret not noticing that sooner, but I'm already spending too much time on this. I won't sim it again for the sake of a minor variation, so we'll have to make do. Below are the circuits I tested, with text under each describing the results I saw, plus FFT plots. I've also posted the internal currents again, including those of the standard totem pole, for comparison.
Clearly, the choke loaded circuit isn't responding for me in the way you describe it, other than exhibiting more gain. Nevertheless, it is what it is. Given these results, I'll never use it, and I have no intention of discussing the circuit ad nauseum for no purpose. Of course, if you can address the non-linear currents within the triodes in a way that demonstrates this characteristic to be a trick of the simulation, a ghost, an anomaly that doesn't truly exist, I will reconsider. Absent that, I don't see any reason to continue this discussion.
Choke-loaded totem pole at 1V RMS input:
Standard totem pole at 1V RMS input:
Internal Currents - Choke-Loaded Totem Pole:
Internal Currents - Standard Totem Pole:
--------------------------
Buy Chinese. Bury freedom.
Edits: 10/23/14
TK, you keep ignoring what I'm saying.
Also, for a guy who attacks others for their beliefs on what they hear Vs what can be verified technically I find your "kludge" statement attacking a circuit which is both technically superior and better sounding extraordinary. You haven't built or listened to the choke loaded variation whereas I have built, tested and Simmed most possible variations of the "Totem Pole" circuit.
I will summarise one last time.
The choke loaded circuit has more than twice the gain so comparing both on the basis of an identical input signal as opposed to output is clearly ridiculous, especially since the input signal is on the point of overloading the choke loaded circuit.
The compromises in the simple mods I made to the original circuit provided other benefits. However, since this became an argument over topology I Simmed both circuits and levelled the playing field by optimising the values for a fair and direct comparison.
You are not the only one who hasn't the time for an endless and pointless argument over this but I want to ensure that the message would not be lost on anyone following this thread as to which variation of the circuit is technically best. As for sound, anyone can form their own view, I can only provide my opinion.
Finally, can you not see the glaring error in your SIM depicting my mods? Does the total current through the tubes not provide some clue? It's obvious that you have not included any resistance in your choke. Sorry, my bad in using the resistance of a real world Hammond choke of 1666 ohms without pointing it out.
Frankly, I don't care what you personally think of the circuit as long as it's for good and real reasons but you've shown neither so far! Further, to suggest that I've somehow tampered with the Sim to provide the result I want is a personal affront to my integrity ... I thought you were better than that.
My last word on the subject is an attached link to the ASC files so you and others can verify my Sims. Dave Slagle was always big on posting the ASCs and made it easy for us on his Web site. I should have headed his good advice and posted the damn things in the first place.
Naz
It's clear that you want validation for this circuit, and because I'm not giving it to you, I am now your target for personal attacks. I don't disagree with everything you've said, but we're not going to discuss this further. I'm done with it.
--------------------------
Buy Chinese. Bury freedom.
So as not to simply abandon this discussion to the theoretical, here's what I suggest to remedy the shortcomings of the driver circuit of the Baby Ongaku. Everything is derived from SPICE simulations, so tweaking will be necessary to optimize the circuit with live components. Also, these changes assume that one desires to keep the topology in its current form (pseudo-SRPP), and that a low distortion amplifier is the goal. If the warm sound of abundant even-order harmonics is preferred, don't do this.
According to simulations, the circuit clips at a somewhat low 2.8V RMS input. In addition, while I don't rely on SPICE for distortion analysis, deformation of the output sine wave is visible well under clipping, beginning somewhere around 2V. These results imply that the onset of audible distortion probably occurs at 1V RMS or less.
The solution is to A) remove the 220uF bypass cap at the lower triode, and B) increase both cathode resistors to approximately 2K ohm. When this is done, distortion should remain inaudible until just below the threshold of clipping, and clipping will occur at approximately 5V RMS. The tradeoff to these changes is a loss of 3dB in gain.
There are alternatives to this, but nothing easier, more reversible, or less expensive to try.
--------------------------
Buy Chinese. Bury freedom.
Here are my suggestions. Note, this is only a half way point, simple modifications that would provide VERY tangible benefits despite the compromise of type of tube used and lower than ideal HT voltage. Two large improvements are a result of swapping out the bypassed cathode R in the driver for Schottky diodes and DC coupling.
A little power is given up by increasing the R value on the cathode of the OP tube but the ability to transition well into A2 will more than make it up.
The driver is far more linear with a choke instead of an R and I find that a choke in this position is not detrimental to sound in the way a typical plate choke is.
I still prefer the sound with drive taken from the bottom tube's plate but you lose any worthwhile A2 transition (call it headroom).
Naz
Edits: 10/19/14
Did you analyze this circuit before you posted it? It's not Class A, and it clips much earlier than the totem pole we were discussing. Currents within each of the sections are also extremely non-linear. Let's let this go now. I've said what needed to be said about SRPP. If you want to discuss something else, it might be better to start a new thread. Just so you understand that I did take the time to analyze what you posted, the result is below. It's shown with a drive level of 1V RMS. Again, this isn't a circuit I'm interested in discussing at this time.
--------------------------
Buy Chinese. Bury freedom.
You've made some erroneous statements so before you take your bat and ball they need to be corrected.
1) Yes I've analyzed the circuit and yours too.
2) I have no idea where you got your currents from but if it's not class A what is it?
3) Mine handles less input voltage swing because it has more than twice the gain of yours for equivalent output swing due to a far better load line. Hence it's more linear, not less.
4) Say what you like about the currents it's the result at the speaker terminals that counts and overall the amp is much more linear than the mods you have put forward. At comparable output (0.5V in on my circuit), 2ndH -60db for yours compared to -90db for mine. 3rdH, -80db for yours Vs -90db for mine. Higher order H, -85db for yours, Vs negligible for mine. I can post the ffts if you like but you don't trust them.
5) Finally, my circuit is not SRPP. As I said I am putting forward simple changes to the original circuit that I would make (in line with the original poster's request before the thread was hijacked). With the proposed changes this circuit would be better served with higher HT as I've already stated but even as is I guarantee it will sound better than original or with the mods you proposed. Anyone out there willing to try compare and report back, I'd be most grateful.
Now I'm done too!
Naz
It's not just the (simulated) output that matters; it's how you get there. If that wasn't true, all our amps would have 30dB of NFB.
"I can post the ffts if you like but you don't trust them."
And for good reason. None of these topologies will produce harmonic energy at -90 dBc (~.003%) in real life.
--------------------------
Buy Chinese. Bury freedom.
The baby Ongaku in Sound Practices, by Reps, Frank.
JM
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: