300b lovers

I also bought 4 schottky diodes. What type of rectifier will sound better: a central tapped full wave rectifier or a bridge rectifier?

@alexberger You might consider HEXFRED rectifiers. They are ultra fast, ultra soft recovery and so are less prone to 'diode noise' (which is actually an interaction between the capacity of the diode junction and the inductance of the power transformer), lower than even Schottky diodes.

One tip - unlike silicon rectifiers, the maximum current rating of a HEXFRED cannot be exceeded even for a few milliseconds- there's no 'surge' rating. But they come in some pretty robust current ratings- the smaller ones are typically 8 Amps. There are 1200V versions too.

DHTs like a 300b do not have a cathode so cathode stripping during warmup isn't a thing.

@alexberger Since an SET is a constant current load insofar as the power supply is concerned, it won't make much difference if you use tube or solid state rectification. But you might want to consider what happens when the AC line voltage changes. The filament of a tube rectifier cools off when the AC line goes down- so the B+ voltage drops more than one might expect since the rectifier gets less efficient. . Plus you'll find that they each have a 'sound'. If you set up HEXFRED rectifiers properly they are about as neutral as it gets.

FWIW we run a separate power transformer for the driver section in our OTLs. We did this so as to prevent any modulation in the output section power supply from affecting the driver. This reduces IMD. Our OTLs run class A2 and have dual output section power supplies so modulation of the supply could be a concern.

In an SET a separate power transformer for the driver isn't going to have the same effect since the power supply for the B+ should be nice and quiet anyway. The separate transformer would be useful if you planned to direct couple to the power tube with a cathode follower though.

.... and those treasure globes look SO COOL!!

@1markr They look neat, but the globe is highly resonant. If the tube develops microphonics the globe exacerbates the problem. The regular Linlai tube does not have this problem.

It doesn't mean you cannot build a nice amp that uses NFB, but that "air" and sense of "realism" that you treasure is hindered by NFB.

That depends on how the feedback is implemented!! If the feedback is sent to a non-linear point in the input of the amplifier which is used as a feedback node (such as the cathode of an input tube) then you can expect it to be problematic, as Crowhurst pointed out 60 years ago, and Baxandall 'rediscovered' 15 years later.

In other areas of electronic design, feedback is known as 'control theory' and is very well understood. But in audio, it seems to get misapplied (and so gets a bad name) on a regular basis, then everyone points at feedback being the problem when its really just design flaws.

In addition, if the comparator is saturated, or if the power supply sags or is discharged, then recovery time can be quite long (tens or hundreds of milliseconds), much longer than the original clipping or slewing event.

During this settling time, amplifier distortion can be quite high, since feedback is only partially effective. This will not appear in FFT harmonic distortion or multitone IM distortion measurements, which are taken over several seconds and then averaged.

The settling time referred to above is a process of many amplifiers with feedback, but not so much opamps (unless overloading, which is easily avoided). In a nutshell, the reason you run into the problem described above is that part of the amplifier circuit is not in the feedback loop. So it can behave as described and as pointed out, lots of test equipment ignores this phenomena, although it can be measured if you have advanced gear. There is more at this link:

https://linearaudio.net/sites/linearaudio.net/files/volume1bp.pdf

If you don't want to read the whole thing, start at page 11, where the math is a bit lighter- but stay with it till the end of the article- its all relevant to this conversation.

I wouldn't try mixing and matching Ralph's approach with ours. Ralph has his way of doing things, and his own unique taste in sonics, and we have ours. Most designers in this biz have a distinct "house sound" that they aim for, which results from design approaches and parts selection.

FWIW Dept.: We've never had a house sound nor 'voiced' our circuits. 

My comments thus far have simply been based in sound engineering practice. Engineering after all made audio products possible, has kept airplanes in the air and provided reliable power when you want light in your house.  

It will always work to apply sound engineering practice to circuit design, plain and simple.

But there are choices that affect the sound.  If you need a 1 watt 1K resistor you can use any type on a cathode, but different types have different sounds.  The cathode bypass cap might need to be 100 uF.  Different 100 uF caps sound very different.  All will satisfy engineering standards, but parts and layout choices have profound effect on the final sound. 

When you run zero feedback, the circuit has no ability to reject things like this. So everything makes a difference. However, for something like an electrolytic bypass, I think you'll find that as long as the part is good quality, the big differences you hear will be more about the part forming up over time: they will arrive at the same place sooner or later.

For reasons that are not clear, various brands of metallized polypropylene capacitors sound quite different from each other, and there is little correlation with DA and DF parameters. Based on measurements, they should all sound the same.

Sometimes you have to do your own measurements because the specs of the manufacturer don't always tell the whole story. If you use a precision differential amplifier to drive the caps in question, you can measure how they behave and differ from one another while in circuit.

@donsachs FWIW I was referring to an electrolytic cathode bypass cap. I wasn't very specific about that in my last post.

And the colorations from different part selections are not the same as SET. This makes sense when you reflect on it ... the balanced circuit is cancelling most, but not all, colorations, and the residue left over can be unwelcome and surprising.

SETs produce a quadratic non-linearity, which in turn makes for a fairly prodigious 2nd harmonic. If set up properly (if you see what I did there) the succeeding harmonics will fall off on an exponential curve.

A fully balanced amp will have even ordered harmonic cancellation, so the resulting non-linearity is cubic in nature. So the 3rd harmonic will be dominant, but at an amplitude slightly less than the 3rd is when seen in an SET. Succeeding harmonics should also fall off on an exponential curve, but it will be one with a different exponent- the harmonics will decrease in amplitude faster as the order of the harmonic is increased. The reason for this is distortion is compounded less from stage to stage throughout the circuit. Since distortion obscures detail (in addition to altering the tone colors of instruments) this makes for a more detailed presentation, with less harshness and brightness than an SET can manage, which is saying something. In either case of SET or fully differential, the lower ordered harmonics will mask the higher orders.

The ear treats the 2nd and 3rd in much the same way- in that it finds them innocuous. FWIW, a properly functioning tape recorder will produce a 3rd harmonic as its primary distortion component also, so we have a pretty good indication on that alone that the 3rd isn’t a problem.

Its also worth mentioning that a fully balanced circuit, running zero feedback, will produce a greater percentage of usable power- close to or exceeding 95% of full power, while an SET is doing well to make 25% usable power.

Of course setting the correct operating point is critical in either circuit. In a balanced differential circuit, the best operating bias point will usually be just above the maximum gain that the differential circuit can do, with symmetrical clipping. If this rule of thumb is followed, there will be no unwelcome ’left over residue’. To achieve this, a proper Constant Current Source circuit should be used- a simple resistor to B- is inadequate owing to the rather low mu that most tubes have. It will be found that the current sensing resistor that is tied to B- is quite critical. I usually set it slightly high to allow for variations in the tubes themselves.

A good quality CCS cannot be made using a single tube or transistor- you’ll need at least two devices. Semiconductor CCS circuits can work exceptionally well and offer the benefit of no likelihood of tube damage as the tube warms up due to cathode filament arc-over concerns. If you work with differential circuits you find out quickly how important the CCS really is. In most solid state amps I’ve seen the CCS leaves performance on the table. If it is not optimized, the differential circuit will not achieve its best gain, distortion or Common Mode Rejection. So its important to get this bit right.

With 5687 you can get even wider bandwidth especially on bass.

@alexberger All tubes have response to DC. If you hear a difference with the ability to play bass, its not the tube type that's causing it. The reason we like to use 6SN7s is there are several that are new manufacture and most of the NOS types are great so there's a lot out there to support the design. 5687s OTOH aren't being made AFAIK. I like the plate curves of the 6SN7 better too- a bit more linear which is helpful in a zero feedback design.

Here is an example of direct coupling:

http://www.single-ended.com/Lagarto/shishido/811A.png

Is 6f6 grid load easy enough for SRPP?

ECC82 can be changed for 6sn7.

@alexberger It needs grid stop resistors. Its usually a good practice to bypass the output of a regulator like the LM317 with some kind of capacitance to improve transient response; 1uf is a good typical value. You might want a bit more capacitance after the 180 Ohm resistor since any voltage variation where the regulator meets the driver transformer can cause intermodulations. The 6F6 grid is no problem for the SRPP and you should be able to sub a 6SN7 since their characteristics are so similar.

Here's something to think about: many tubes perform better when a cathode bypass capacitor is employed. Yet there isn't one in the output stage. To install one you would need two parts, one for each side of the filament and its 30Ohm resistor (plus pot). It would not have to be a high voltage part but it would need to be a fairly high capacitive value- perhaps about 2,200uf on each side. 10Volt parts should work nicely since the voltage across the resistors and pot might only be 2 Volts.

Nice article! Too bad direct coupling wasn't tried. That would have been fun to see.

They now dominate high-end audio, which was not true thirty years ago. I remember going to some CES shows with hardly any tube amps at all, never mind DHT’s, and now they are everywhere.

While it is correct that there are more tube amplifier producers (many of whom show at audio shows) than there was in 1958, its incorrect to say that tubes dominate high end audio (most of these producers are very low volume). The larger players who produce solid state products are still the bigger sellers.

I should have been more clear when I mentioned direct coupling as a method. Specifically, I meant using a cathode follower which is direct coupled to the power tube so that the bias of the power tube is obtained from the driver tube. For this one would need a B- supply, but if the best reproduction is a goal then that cost isn’t important. IMO using a driver tube in this manner does not result in higher ordered harmonic generation, at least insofar as our OTLs seem to measure out. In them the higher orders fall off at a faster rate than any SET I’ve seen. They sound smoother too. One thing we sorted out from doing this over the years is that the plate needs to be heavily bypassed to prevent IMD and harmonic generation. In our amps we found we had to use bypass capacity values an order of magnitude higher than one would typically expect; after trying lesser amounts we found the seemingly excess amount made a difference to the distortion. IMO this aspect of cathode followers is poorly understood.

If you do direct-coupling correctly, it can be very reliable. I don’t think it wise to direct couple throughout the circuit though!

The advantage of a direct-coupled cathode follower driver tube are several. First, there is more current available to provide linearity if grid current is produced by the output tube(s) (which allows for class A2 or A3 operation since both of those classes produce grid current), and no worries about Miller Effect, so very wide bandwidth (+30MHz!) is very easy (we bandwidth limit our OTL amps in the voltage amplifier section). The stability of the circuit is very high- in our OTLs (where one driver tube is controlling many triode power tube grids), bias adjustment only needs checking once every 6 months or so; often no adjustment needed. Distortion is kept in check in a way not possible if a cathode follower with coupling cap were used. We went to this circuit because its more reliable and prior to its introduction to the marketplace, OTLs generally had a reputation for being unreliable. This approach was key to making OTLs as reliable as any other tube amplifier; one of the reasons we are still in business after more than 45 years.

How this might work in an SET is a different story- yes, you need a B- supply to really pull it off, but if you want to do it with transformers you’ll have to pay that price too, since decent interstage transformers are not cheap- in fact, several times more expensive than a B- supply.

The other advantages of using a direct coupled cathode follower driver are that the driver prevents output tube conduction until the driver has warmed up and stabilized. This is likely not a concern when DHTs are used, but it is if indirectly heated, since cathode stripping can occur if the B+ is applied while the power tube is warming up.

Another advantage, probably the big one, is that small value coupling caps can be used between the voltage amplifier and the driver tube. Since coupling caps have inductance (since they are wound) no matter what materials are used they will always introduce some coloration on this account. By minimizing their value while still allowing for very wide bandwidth, the coloration they might cause is minimized and the inductive influence octaves above the audio band.

If course you could use an interstage transformer for this latter function as well; the advantage being that’s the least expensive place one could be used and most likely to win the most performance from the transformer.

At any rate, by using this approach you wind up with a very simple circuit that offers excellent linearity. One idea I’ve had for a while for an SET is taking advantage of the B- supply and building a differential amplifier for the input voltage amplifier. This would allow a proper balanced input as well as reducing distortion in that gain stage due to harmonic cancellation.

Of course if you already have a fully differential/balanced amplifier design from input to output, the advantages multiply.

We’ve been building a balanced preamp with a direct-coupled output (for which we have several patents) that uses servo control to prevent DC offsets at its output. The inputs of the preamp are direct coupled as well, but coupling caps are used between stages. The DC servo has proven to be one of the most reliable aspects of the circuit, which has been in production since 1989. During that time we simply have not seen any servo failures! The output of the preamp uses a Circlotron, like our OTLs, so there are no excessive voltages produced as the preamp warms up.

The obvious advantage is a lack of coloration at the output of the preamp (most tube preamps use a large coupling cap, from which, as pointed out earlier, colorations are unavoidable) and very wide bandwidth; to the latter point the direct coupling allows for reproduction of extremely low bass that most tube preamps won’t even acknowledge. Since the preamp uses no feedback, controlling phase shift can only be done by having bandwidth down to 1/10th the lowest frequency to be reproduced, so its good to 2Hz. If phase shift is present, the ear perceives it as a lack of impact, even though the preamp might otherwise be perfectly flat at 20Hz. By direct coupling that problem and the often tubby bass that can occur are both avoided.

So while direct coupling (with or without a servo) might seem tricky, it offers great advantages and can be extremely reliable if properly designed.

I changed the load resistor to 120KOhm, and the overshoot decreased by amplitude and attenuation time. But still, there is a notable overshoot. I measured a frequency response and there is a hump +1.7dB at 35KHz. There is -3db at 19Hz and 47KHz.

Should I decrease the load resistor more to remove overshoot completely?

If yes, in which value range should be this resistor? For example, if I take a resistor less than 50K it can increase distortions.

@alexberger As you have noticed, if you are using a coupling (interstage) transformer, it will be needing proper loading to prevent ringing (distorting). You are nearly there with your technique so far; put a potentiometer across the output of the transformer, run a square wave through the active circuit prior (6SN7) and adjust the pot for minimum ringing (critical damping). IME its probably best if you leave a very slight amount of overshoot as opposed to rounding the square wave.

Its important that the driver to the transformer be active, since transformers transform impedance: Whatever impedance on the primary side, if it varies, will affect the critical damping value on the output side. Conversely, whatever is loading the output side will also load the input thru the ratio of the transformer. So you want to feed the squarewave to the active 6SN7 circuit so your loading value will be correct. Best to have the 6F6 running also for this very same reason, although the loading resistor will likely dominate that side of the transformer equation.

Once that is sorted out, you might find it interesting to measure the impedance at 1KHz on the primary side (you won’t need the 6SN7 in the circuit for that, but it would be a good idea to have the 6F6 in place and active) just to see what the load on the 6SN7 actually is. You may find that you have to adjust the operating point of the 6SN7 to obtain greater linearity (by adjusting the cathode resistor value if you have one); if you do that then you may have to readjust the loading value of the transformer since the source impedance of the 6SN7 varies a little with the operating point; in this way zeroing in on the optimal values.

Obviously you can stop at any point (call it ’good enough’), but in a zero feedback design I’ve found that the more you pay attention to refining things like this, the more it pays off in the end!

@lynn_olson I agree overshoot in a circuit using feedback is bad!

But if the feedback is applied properly you’ll get no overshoot at all. Our OTLs don’t exhibit any squarewave overshoot, being zero feedback and free of inductors in the signal path. The 10KHz waveforms are pretty good since they have wide bandwidth; our class D, which has less bandwidth owing to the output filter, nevertheless has a very similar 10KHz waveform, despite (well, actually because of) running 37dB of feedback; the difference being the residual sine waveform imposed by the switching frequency.

@charles1dad My point was addressing a comment made earlier by Lynn about overshoot in amps employing feedback; simply that if you do it right its not a problem. The OP mentioned using Spatial Audio Triode Masters who were a dealer of ours and have used our OTLs and class D on their speakers. It didn’t seem that off topic, especially if we discuss the issues of signal coupling, operating points and the use or lack of use of feedback.

As I understand it, you are particularly enamored of SETs; perhaps this thread might have convinced you there is more than one way to reach audio Nirvana 😉 I’m sure the Blackbird is well worth hearing.

How does excessive transformer ringing can influence on sound?

Does it make it too sharp or bright?

@alexberger Ringing contains higher ordered harmonics which can be heard as brightness and harshness. You also get lower orders which contribute to richness. Both are colorations and will obscure low level detail.

A current source could replace the resistor, but in practice, the performance is very similar to a current source, so it’s rarely done even in modern amps.

Usually in a differential amplifier, the plate resistors are matched if both halves are driven. In the case circuit of the above description, only one side is driven. So if a CCS is not used, the plate resistors can't be matched; one side must have a slightly higher plate resistance to compensate for the mu (gain) of the tube of the un-driven side, so as to get equal outputs from each half.

A good CCS eliminates this problem (which is nice since in the real world you can't count on the mu of each section to be equal or matching that of the tube specs on paper). A good CCS is both inexpensive and reliable, allowing the tube to be removed from the circuit while active (hot plugged) without damage.

Further benefit can be had from placing a CCS in the output tube cathode circuit, if you control the output tube(s) bias using fixed grid bias. The cathodes are tied together and the CCS feeds them; thus improving the differential effect of the output section, which reduces distortion and makes it slightly easier to drive due to increased gain.

Of course, if you have the input tube be a differential amplifier too, it can accept a balanced or single-ended input and can have excellent performance if a CCS is used for this stage as well. But its not a good idea to direct couple both plates to the succeeding driver stage; its OK to do one but the other should be capacitively coupled so as to prevent DC offsets of the first stage of gain from causing distortion in the driver.

The original BAT VK-60 of the 1990s used a differential input direct coupled to a differential driver; to deal with the DC offsets a potentiometer in the cathode circuit of the input tube allowed the plate voltages to be equalized. I found this approach to be problematic (we had tried that back in the early 1980s; one obvious problem is that it requires the user to make this fairly critical adjustment...) and often causes more problems than it solves.

So in the quest to keep the number of coupling capacitors down but retain easy operation, we started using a differential cascode voltage amplifier. The advantage of this was that all the gain of the amplifier was in a single gain stage, consisting of three dual-section triode tubes, one for the input differential amplifier, one for the top of the cascode, being plate loads for the bottom tube sections, and finally a 2-stage Constant Current Source for the circuit, tied to a B- supply of equal potential to the B+ supply. The CCS prevented changes in the AC line voltage from affecting performance of the voltage amplifier from 107VAC to 126VAC the difference was only 17 parts per million. So you couldn't see any performance change on an oscilloscope over that range!

So that allowed for enough gain, low distortion (once the correct operating point was set up), and only one pair of matched coupling caps (of a small value, in our case only 0.1uf, further minimizing the sonic impact of the coupling caps). They drive a pair of cathode followers which are direct coupled to the output tubes. So the power tubes obtain their bias voltage from the driver; therefore the bias and DC Offset controls are in the grid circuit of the driver tube. This allows for instantaneous overload recovery and rock solid bias control of multiple high-capacitance triode grids, with low frequency response to 1 or 2Hz no problem at all.

If you use a coupling cap in the critical area of the grids of the output tubes, it must be large so as to get good bass response since the grid bias network must be of relatively low impedance to properly control the power tubes. This means that the driver tube has a difficult load to drive and the large coupling cap can cause blocking distortion and slow the overload recovery. While this really isn't much of a problem driving pentodes, using this topology to drive triodes is a bad idea IMO/IME.

I've been describing how our OTLs work but obviously this would work well with a 300b too. We've managed to get our OTLs to 0.5% THD which is pretty low distortion for a zero feedback circuit! SETs by contrast tend to be about 10% THD at clipping which might be only 7 Watts. Since the OTLs tend to be much higher power capacity, the tendency is, for any power level the SET might have, the OTL has distortion that might be 2 orders of magnitude lower or more at the same power level. This is why they tend to be so much more transparent than SETs.  I've no reason to think this cannot be applied to a 300b circuit with similar results; SETs have the distortion they do out of the topology rather than the power tube that is used. So a pair of 300bs could be used to much greater advantage!

For those that might want to see more about how our OTLs work (and how this might be a topology for a 300b amplifier), there is a DIYaudio.com thread from several years ago that has a schematic and discussion. A lot of this would work very nicely with a 300b; for example the Circlotron output can be transformer coupled of course and have all the advantages (such as zero DC saturation of the output transformer) it offers.

The Dynaco, introduced in the mid-Fifties, took the drastic step of deleting the driver stage and its associated RC coupling, and driving the output tubes from the RC-coupled phase inverter. Although the open-loop performance was quite poor, rolling off around 100 Hz and 7 kHz

Actually the Dynaco has ~ 6Hz LF cutoff (-3dB) running open loop. Its distortion rivaled that of the Marantz 8B which was and is well respected, for a lot less money. You can reduce the distortion easily by obtaining a socket adapter off of eBay, which allows you to replace the 7199 driver tube (which is rare) with the much more common (and cheaper) 6GH8A. No other changes are required.

But many of these classic PP still sound fine! Why? Because feedback or because PP has PS noise cancelation?  

Feedback allows the amplifier rejection of that which is not the signal, so generally speaking, yes.

I really liked the 3 6sn7 tube version that tubes4hifi used to sell and perhaps still does.   It makes the amp considerably better. 

The problem with any mod that adds tubes to the circuit is the extra load on a power transformer that might already be 65 years old. Dynaco strikes me as being pretty precious about their transformer ratings- I don't like to take chances with them, especially in light of their age.

This is just me of course but if I'm going to modify a vintage piece I follow two simple rules. The first is don't add any extra load to the power transformer. The second is don't do anything that does not fit very easily into the existing chassis. Violate these and you're likely better off doing the whole thing from scratch.

The really glaring weakness in the ST70 is it should have been designed with dual rectifiers; as a result the 5AR4 is the most likely tube to fail in the amp. Triode Electronics of Chicago has a beefed up power transformer that is a drop in replacement that allows you to add a second 5AR4, thereby keeping the correct B+ operating point and so not stressing the output transformers as well. But you have to find room beneath the chassis for some 500V filter caps. It starts to get a bit ridiculous- at that point why not just do your own chassis so you can lay out things properly?

If an ST70 is properly refurbished but pretty well the stock circuit, it can be surprisingly good against a lot of modern PP and SET amps. Since it really does not have enough feedback, you have to help it along with good quality coupling caps and resistors in the voltage amplifier and driver circuit. The second thing to understand about this amp is because of its power supply weakness, you really should not push it hard (which is better for sound but also keeping that 5AR4 alive). CE Distribution in Arizona makes a drop in replacement filter can that features an 80uf section, which should be deployed after the choke, for the plates of the power tubes. That's about as much extra capacity as you can safely add to this amp without stressing the 5AR4.

A Mullard PP 300B works as follows: input tube direct-coupled to a long-tail pair (or CCS) of triode-connected 6V6 drivers. These in turn are connected to a PP interstage transformer with a modest step-up ratio, between 1:1.4 and 1:2. The interstage then drives the PP 300B grids. This would be a non-feedback amplifier, so good power supplies are required. I would imagine a number of the PP 300B amplifiers already on the market use this topology.

I suggested this earlier in this thread; and added to it that with the input circuit also being differential there is additional benefit. A fully differential circuit has harmonic cancellation at every stage of gain; not just at the output. This results in the 3rd harmonic being dominant; it is treated much like the 2nd by the human ear in that its innocuous. But compared to an amp that does not use this topology, the 3rd is at a lower amplitude, and succeeding harmonics fall off at a faster rate (than seen in an SET) on an exponential curve. So the 3rd is thus more effective at masking higher orders and since distortion is lower, the circuit can be smoother and more transparent.

The problem with two successive stages that are balanced and DC-coupled to each other is that DC drift is a big deal. A 1 volt shift on a 150 volt plate is normally inconsequential, but becomes a serious concern when the grids of the following stage have a 1 volt offset between them ... which is what DC coupling does.

In the post in which I suggested a differential topology, I also suggested several cures for this problem, which I pointed out in that post.

Try removing the 56K load resistor entirely and give it a listen. You might like it better. Remember, with conventional signal sources, there is almost no ultrasonic content, so the ringing in the transformer is never stimulated. And getting rid the load resistor has no effect on circuit stability or DC stability, since the grid sees the DCR of the transformer, which is a few K at most. All the grid resistor does is make square waves look pretty. It has no other function.

Its been my experience that allowing an audio coupling transformer to ring will result in brightness, since any signal presented to the transformer can cause ringing (this is easy to demonstrate- try it!). I'd not be surprised if some people mess with the damping to compensate for a weakness in the circuit elsewhere; IME/IMO you're far better off finding those weaknesses, sorting them out and making sure your transformers perform as good as they are able!

The overshoot, as these things go, isn't bad. You might be able to zero in on it a little bit more. The thing is, the more energy the overshoot has, the brighter/livelier the presentation so if you're going for a warmer sound this is something to avoid (you can see here how easily distortion can influence the tonality of the circuit). I would not reduce the loading resistance to the point it rounds the leading edge. A slight bit of overshoot is OK when trying to hit that critical damping value.

In a zero feedback circuit you have no correction to deal with this sort of thing, so you have to sort out details like this and get them right. The reward is greater detail since distortion and detail really don't go hand in hand. You can see by doing this sort of measurement how different power tubes and different speaker loads affect how well the output transformer can perform- and why people might have contradictory observations about how the same OPT and power tube sounds, because the way the transformer behaves changes depending on the speaker load.

Don, myself, and the team at Spatial intend to be on the straight and narrow when it comes to regulations ... we are not Tesla, Apple, or Microsoft, with armies of lawyers to smooth the path into new markets.

To gain the CE mark (which says that the product meets EU Directives) you can self-certify and no lawyers needed. Its not intended as a trade barrier.

If there are no digital or switching circuits involved, you have exemption from a good number of directives. Beyond that, if good practice for the AC wiring is observed (with a proper AC ground if a metal chassis is employed) and there are no exposed voltages, then you are 95% of the way there. It isn't required that you have a tube cage but if its not shipped with the equipment it must be available as an option. Finally the CE mark must appear somewhere: either on the shipping box, the associated owner's manual or the equipment itself.

Its a good idea to test the equipment for RFI generation, even if there are no RF sources such as switching technology. If you paid attention to Ps and Qs regarding layout and grid stoppers this is likely not an issue. 

RoHS must also be observed. For the most part this means lead free solder, but there's a percentage involved so it is possible to use leaded solder. You'll have to review the regulations on that.

I have a thorough dislike of working with lead-free solder.  Nothing flows in a point to point build like a good 2% silver solder. 

@donsachs That is why its worth it to work out the math. The lead content in your product is a percentage allowable. Since you have a lot of transformer weight and otherwise the circuit is fairly simple, my surmise is you can use leaded solder with no worries. But you'll want to look at the regulations and do the math to know for sure.

Pin 1 is shield and grounded to chassis, pin 2 is XLR + phase and pin 3 is XLR - phase.  It will drive any power amp with XLR connections, and will happily drive a 10K amp input impedance.

@donsachs How do you deal with the variable loads which the preamp will be driving? With no feedback, the transformer can ring or express the inter-winding capacitance if its not loaded correctly.

The Raven is not very sensitive to loads. In practice, power amps range from 10K (typical solid-state) to as high as 470K for a handful of vintage tube amps. Most modern tube amps are 100K. Plus whatever cable capacitance is there, along with the Miller capacitance of the input section of the power amp. So 100 to 400 pF is typical. The range of loads is predictable and well known.

What dominates the transformer performance is the source impedance, not the load. The source impedance from the preamp tube is much lower than the load, so it heavily dominates the transformer performance. Transformers don’t much care where the low impedance is, primary or secondary, so long as it is there.

Transformers are aptly named as they transform impedance. They do not isolate impedance. If the source impedance is low that will change the correct loading on the output (as opposed to a high source impedance), which will be found to be exactly one value. Above that value the transformer will ring; below that value it will roll off highs.

This phenomena is well known and is why Jensen Transformers specifies the loading to be used with their SUTs depending on what cartridge is used with them.

If the load is too high impedance, the inter-winding capacitance will come into play as well, causing the FR to no longer be flat, and in extreme cases the transformer may fail to express its turns ratio. The proper termination will yield the widest bandwidth out of the transformer.

This why in the old days when balanced lines were used, the termination standard was always 600 Ohms so the designer would have a pretty good idea of what to shoot for.

A rheostat, placed across the output (between pins 2 and 3) can be used to load the transformer for optimal operation. Or the transformer can be designed for 600 Ohms with a termination switch using a 600 Ohm resistor built into the equipment (this is how Ampex solved this issue on their 351 tape electronics). That way any higher impedance load such as 10K or 100K is negligible and would not affect the transformer performance. The latter approach is why we did with our P-2 balanced line preamp 30 years ago.

Ralph, I see you introduced a PP 300B amplifier at one of the recent hifi shows. How do you think it compares to your Class D amplifier, in subjective terms?

The prototypes at the show sound very similar to our class D amps; the big difference is the bass- the prototypes roll off at a higher frequency, and owing to no feedback and the OPT design don't throttle back their power when dealing with higher impedances, such as seen in the CALs at the show, which are bass reflex. So there was a bit of extra energy in the double impedance peak of the speaker. Its fun, but not correct.

Essentially that OPT loads the tubes at their power peak rather than a bit higher where the amp would have had a better chance at being more like a Voltage source. If the next iteration of the OPTs had been available, the bass would have been a lot more like the OTLs but ultimately lacking the extension that up until now has been something that OTLs do better than any other kind of tube amplifier.

The class D has the same relaxed mids and highs but is a little more transparent- its easier to hear what's going on in the rear of the sound stage if you have a 2-mic recording.

I think we can do better with the 300bs by using some of the tech we use in our OTLs. So the next PP version of the amp will be quite a bit different- with LF bandwidth similar to our OTLs with no sacrifice in high frequency bandwidth.

Wow, is that the first Atma-Sphere with an output transformer?

No, nor the first to be shown in public or sold. We make a small 5Watt/channel integrated amplifier that is suitable for desktop, bedroom, headphones or a main system if your speakers have the efficiency.

That one uses EL95s which are a little brother to the 6AQ5, running class A1 ultralinear. The tube is very easy to drive so the voltage amplifier is a differential amplifier using a 12AT7 with a constant current source. The amp can sit on a single sheet of notebook paper with room left over but its built with good parts throughout and an overbuilt power supply. The chassis has its corners welded and ground so it can be polished and chromed. We've been having fun with a variety of finishes and color schemes; blue with black (the transformers be the contrast finish), blue with chrome, black with chrome, chrome with black, grey with red and so on. They are entirely handwired point to point. 

It easily keeps up with SETs of the same power; it has a greater amount of usable power owing to greater linearity.  It also has wider bandwidth both on the bottom end and the top end (goes out to 100KHz).

This amp is one of the projects we did which aptly demonstrate why SETs are impractical and obsolete.

That's not to say we're not also having fun with SETs. We have a 300b SET project as well. It uses some techniques we use in out OTLs to minimize distortion driving the power tube.

Sounds like the 300B amp is well on track, and I imagine the next version will have current sources in the appropriate positions. I’m curious what your experience with a high-power current source in the output stage turns out to be. It didn’t work for us, but it did work for Allan Wright back when he visited me in the late Nineties.

Constant Current Sources are the key to getting differential amplifiers to perform. A PP output section is often wired as a differential amplifier so a CCS can work really well if output section is class A.

Sounds like @atmasphere is early on their journey with a PP 300b.  Certainly far enough along to demonstrate a prototype at a show, much as we did in Seattle last summer. 

We started prototyping with DHTs seriously about 20 years ago. Back in the 1990s we built a 300b-based OTL which we showed at CES. 

The trick with the CCS is doing a good one. A lot of the circuits that you see on the web leaver performance on the table. We've had CCS circuits in our amps for the last 35 years.

'A lack of solid state coloration' is matter of avoiding the distortion signature that is endemic in so many solid state amps. This has to do with proper design of the feedback loop and a whole lotta loop gain in an amplifier design. The 'solid state' sound is just bad feedback application and is part of why feedback has garnered a bad rep in high end audio. 

Feedback has been known to give tube amps a 'solid state' sound too; this is because the feedback signal has been distorted prior to mixing with the audio signal its supposed to correct. Its like having an incorrect map to guide you through town. Norman Crowhurst wrote about this problem nearly 70 years ago but did not suggest a solution, despite it being rather simple. Peter Baxandall 'rediscovered' the same problem in solid state amps about 20 years later but he suggested more feedback, which doesn't solve the problem.

Its only been in the last few years where we've seen self-oscillating class D amps where this problem has been solved pretty consistently. You can do it in class A/AB solid state amps too if you're willing to work out the 3rd or 4th order feedback loops that need to be involved. Most designers just use a resistor which won't cut it...

In my next phonostage I want to make an external power supply with a separate filament transformer. But I haven't decided which RIAA schematics to use. There are basically two types of tube phonostages. In one type RIAA is implemented in feedback and another type is passive RIAA.

@alexberger You might consider that since the cartridge is a balanced source that you could have a balanced phono section too, or at least a balanced input. If you run EQ via feedback, you run into the same problem that Norman Crowhurst wrote about nearly 70 years ago. You could avoid it by applying the feedback to the grid of the tube rather than the cathode (you'll need a series resistor with the input to allow the mixing to occur, similar to an opamp circuit). You'll have to recalculate all the EQ values since feedback to the grid behaves quite differently (higher impedance).

Passive feedback can work quite well. Just because you have passive EQ does not mean that you can't run feedback in the associated gain stages. H/K did this with their Citation 1 back in 1958.

The advantage tubes have over solid state in a phono section has to do with the fact that most cartridges are inductors; when in parallel with the capacitance of the tonearm cable, an electrical resonance is formed. That resonance can overload the input of the phono section causing ticks and pops that sound like they are on the LP surface. If your tube phono section is properly designed (easier because the operating voltages are higher), this won't happen; you may discover that LP surfaces are actually a lot quieter than the digital crowd would have you believe.

@alexberger , @lynn_olson mentioned earlier:

Ralph brings up a very good point about feedback: the underlying theory assumes a distortionless summing point. (The summing point is the comparator input between signal input and the sampled output.) Any distortion introduced at this point of the circuit will be amplified without correction, and there is a real possibility of introducing new, higher-order terms that are not present in the forward path of the physical amplifier. Norman Crowhurst mentions this in passing in his Audio magazine articles in the late Fifties.

Actually I read about this in one of his books. The point is that feedback applied to a cathode is going to generate higher ordered harmonics and IMD because the cathode is non-linear, even on a 12AX7. If you can, the thing to do is apply the feedback to the grid of the tube rather than the cathode. This gets tricky if you have two stages of gain as you see in the schematic above! It might also mean you have to have a feedback capacitor to block DC, which isn’t likely to treat the feedback signal very well. You see this technique being used in the line section of the Citation right after the tone controls.

You can do this in an amplifier too, wrapping the feedback around the entire amp circuit. Admittedly tube circuits are often lacking in the Gain Bandwidth Product to prevent distortion rising with frequency, but if the feedback is handled properly to start with overall its a better chance of it working right.

But I think to make balanced first amplification stage of the phonostage can be very helpful.

SUTs can have a balanced output if you like- they don't care. Transformers are very good at going back and forth between balanced and unbalanced. You will have to be careful about loading the SUT properly to maximize its performance. Why stop with a balanced input- balanced (differential) throughout gets you greater power supply immunity and lower distortion overall, as well as lower noise if the gain stages are properly executed.

Still out in the weeds dept.;

Many phono preamp designers over the last 70 years have ignored the significance of a magnetic cartridge whose output is in parallel with the capacitance of the tonearm cable.

Any time an inductor is in parallel with a capacitance there is an associated resonance whose frequency is set by the values of the cap and the inductor. With high output MM cartridges this peak is generally about 20dB and centered just inside or just above the audio band. If the phono section does not have good HF overload characteristics, this peak will overload the input of the phono section resulting in ticks and pops.

Just to boggle your mind a bit more, cutting a lacquer master from the two-track master tape meant the engineer "riding the controls" as the cutterhead neared the center of the record.

We never had to do this with any of the LPs we cut. We used the Westerex 3D cutter with 1700 series electronics, all stock. We found that the old myth about loss of bandwidth towards the end of the LP was just that. We cut 30KHz tones in the inner grooves that played back fine on our Technics SL1200 with Grado Gold (we used that setup to know if the groove we'd cut was playable on an average machine). I think a lot of that myth got started in the 1960s when stereo cartridges really just weren't all that good.

Both were used to play the lacquer once for quality control, then off to the plating plant.

Usually you don't want to play the lacquer at all- any playing degrades it. But you can cut outside the 12" diameter since the lacquer is 14"; in this way if there is a troublesome cut you can test that part of the recording by cutting it there and then playing it to see how it went. If OK then you can proceed with the regular cut.

I found that if we spent enough time with a project there was usually no need for processing of any kind; there is usually a way to cut a playable groove, even with out of phase bass (if its not too bad). Compression is really only used to speed up the mastering time; mastering time is expensive.

With use of RC cathode bias won't have such issue but we need to watch out when using fixed bias.

@kmtang When setting the bias for any fixed bias amplifier, its good practice to check the bias after an hour of operation. IOW all power tubes regardless of type will see higher bias current over time as the tube warms up. So usually after the amp has been on for about a minute or so you set the bias to about 85-90% of the bias current spec. That way it has less of a chance of damaging the tube as the tube heats.

The takeaway is that is impossible to "overdrive" the 300B. By contrast, a charmer like an EL84 can be driven with a whisper ... even a 12AX7 biased at 1 mA will sound good as a driver (which doesn’t work with any other power tube). A classic Mullard circuit is ideal for a pair of EL84’s since they are so easy to drive. 6L6's take a bit more muscle, so 6SN7's are a better choice.

The 300B is the opposite. A high voltage, high current, and ultra low distortion driver is mandatory, otherwise you never hear the 300B.

A simple and very effective way to drive 300bs is to use a cathode follower driver, direct-coupled to the grid of the 300b as it fulfills the requirements listed above. This requires a B- supply but you can control the grid so well that it can be driven class A2 (grid current) and you can easily overdrive the tube using a single 6SN7 section. This also allows for a much smaller coupling capacitor; 0.1uf (at the grid of the 6SN7) will allow -3dB bandwidth at 2Hz. This frees up the Voltage amplifier/driver of conventional design from a highly capacitive load. The downside might be that the power tube has to have its bias set correctly (so a provision to measure current is needed), which is done by setting the bias of the 6SN7. 

Doing this I've been able to overdrive 300bs (even multiples!) quite easily. The CF circuit, without the typically large coupling cap that often gives CF circuits a bad rep, has a tight grip on the grid of the 300b; so much so that driving class A2 with the substantial grid current that tube needs is no problem. You can easily drive the grid +15V WRT to the cathode with good linearity. 

The cost of a B- power supply is insubstantial when compared to the cost of a good inter-stage transformer and you get less distortion with greater bandwidth. You also don't have to introduce a power tetrode or pentode into the circuit.

A nice feature of this approach is the bias setting is very stable so might only need checking once or twice a year.

@alexberger Yes, that's how its done. Note that the EF86 input tube is wired in triode mode. Its quite linear that way; my Neumann U67 microphones use the same tube in the same manner sans feedback.