[jrebman]

While I'm still learning this stuff step by step, I'm still not completely comfortable evaluating possible candidates for a driver tube, so what about the 6SL7, or 6SN7 (or better yet their 12 volt filament versions, which are much cheaper), or the one mentioned earlier that Don Garber uses in his Fi amps?

I could see the driver tube debate going on for much more time than it will probably prove to be worthwhile.

Another option would be circuits for a couple of different drivers that people could choose among -- the power supply, and OPTs are still where most of the money is going to be spent (though it would be nice to not have to spend a fortune on a good driver tube, especially one you have to pay a premium for matched sections and low noise.)

I suppose a 12B4A will be overkill here too?  ECC99?

-- Jim

[floobydust]

 Jim, et al,

 Nice to get some initial feedback... I purposely put two text-book stages together to get some conversation going. I have everything needed sitting here to breadboard an amp based on the posted schematics and James transformers. There are quite a few good choices out there... and I have a very large stash of tubes here so optimizing a driver stage will not be a problem... stay tuned for some follow-up details and maybe a pic or two.

 Regards, KM

[aerius]

To quote the late Dr. Gizmo, "Don’t use your Porsche for plowing fields", he illustrating the point of proper speaker & amp matching but the same thing applies with tubes & amplifiers, use the right tube for the job. 

A 6BL7 is a pretty nice driver tube for 45's & similar DHT's, but it's not going to work in a phono amp since it doesn't have enough gain.  A 6SN7 will drive pentodes such as EL34's quite nicely but it's going to struggle with a 300B.  Sure you can pound a square peg into a round hole, but it works a lot better if you use a round peg with a round hole.

[aerius]

While I'm still learning this stuff step by step, I'm still not completely comfortable evaluating possible candidates for a driver tube, so what about the 6SL7, or 6SN7 (or better yet their 12 volt filament versions, which are much cheaper), or the one mentioned earlier that Don Garber uses in his Fi amps?

The 6SL7 suffers much the same problems as the 12AX7, it just doesn't have the juice to get the job done.  The 6SN7 can be made to work though I'm not entirely happy with it either.  The plate impedance is a bit high and it doesn't pass enough current for my tastes.  The 6SF5 will suffer the same issues as the 6SL7 & 12AX7 as it has similar electrical characteristics.

I suppose a 12B4A will be overkill here too?  ECC99?

-- Jim

A 12B4A should work nicely provided your preamp can put out a healthy 5-10V or so, it's got loads of current and a low plate impedance but not much gain, thus the need for a bigger input signal.  The ECC99 should also work quite nicely, it can also handle a fair amount of current and has a low Rp as well along with more gain than the 12B4A.  Just to be safe, I'd slap a CCS load on the ECC99 to make sure it has enough gain even with a weaker CD player or preamp output.

[JoshK]

I'd think an Aikido gain stage would be the cat's PJ's.  6SL7 for the gain (~35x) and ECC99-oids for the driver. EH makes a nice octal 6H30pi.  It still isn't expensive to make either.   And you solve gain and drive in one swing, along with low noise, low distortion.

[JoshK]

My previous post not posted....

This is good stuff... There are lots of opinions out there on how to go about it.  I'll just share some of my perspective for "seasoning" choices. 

James looks to be a really good choice for iron, since iron makes or breaks a amp.  If you really want to go budget Edcor's OPT's are suppose to be quite a bit better than the Hammonds for about the same dough.  I have gotten some iron custom made from Bud Purvine of Onetics (same Bud as in EnAbl) at very reasonable prices and then there is Jack Eliano at ElectraPrint.  Lundahl of Sweden has good iron too.  All good choices.

Some other choices for drivers:
6SL7
1579
6AQ8
6922 cascode
D3A (triode or pentode)
12HG7/12GN7A (triode or pentode)
C3G
5842/417A


Topology choices...keeping it simple
Aikido SET amp (aikido gain + SET output...do search, plenty of examples with really good performance, can buy aikido PCBs)
SimpleSE (http://www.tubelab.com/ can buy PCBs making it a snap)
Thoersten Loech's Legacy amp. (www.thunderstoneaudio.com or something like that)
JELabs amp (more of a traditional SET amp, with various iterations and lots of fans)

Maybe not quite as simple, but still not hard...
Loftin-White (zen simple in parts count, but IMHO, not simple in error-proof-ness...DC coupling can bite you if you mess up.  But follow proven schematic to a T and it is super simple)
DRD amp (maybe not quite as easy, but still pretty simple IMO to follow the proven examples out there)

Anyway, just a few good examples of SET amps out there beyond the plain vanilla that aren't hard to build and/or inexpensive (DRD is arguably not inexpensive but can be).

45's are very good sounding but not very powerful.  1.5 watts won't fill a big room or play loud/complex music on most speakers.  2A3 is an option for 3 watts and there are numerous modern remakes.  6B4G is an octal base version of the 2A3 with 6volt filatments.  300B is 8watts.
But one can get good results with a triode strapped KT88, EL34 for less money.  If it were my money and I was tight on funds I'd rather opt for a KT88 and good iron, than a 300B and so-so iron.

[floobydust]

 Aerius, JoshK,

 I'm not completely sure what you are basing much of your recommendations on for the various tubes and driver topologies. Are you using known formulas for calculating required slew rates, current drives, etc? Again, the scope of the project is to have a very simple 45 based SET amplifier and do this at a low cost. I did start with a textbook gain stage and a reasonable output stage with hum balance for the filament. But it's just a starting point, but there is some math behind it. I don't expect this to be a final circuit of course but will post updates as the design progresses.

 One can calculate the required slew rate and drive current required to meet the required voltage swing at the required bandwidth. You may be surprised at the required current to drive the 45 grid input to 55 volts with a bandwidth of 20KHz.

 First, calculate the slew rate for a 20KHz bandwidth and a voltage swing of 57 volts (45 bias point) as:
 2*pi*Bandwidth*voltage change = slew rate
 2*pi*20000*57 = 7.2 V/microsecond.

 Note.... Per Borberly (and others apparently) multiply the slew rate by a factor of 5 to fully overcome all input capacitances.

 Next, calculate the combined input capacitance (of the 45 tube) as:
 (gain+1)*Cgp+Cgk = Ctotal (use the actual in-circuit gain, not the mu)
 (2.75+1)*7pf+4pf = 30.25pf

 Now, calculate the require current for slew rate to drive the 45 input to 55V at 20KHz as:
 Ctotal*5*slew rate = I (driver stage current)
 30.25pf*5*7.2V/usec = 1.09ma

 So, you do need around 1.1ma of current in the driver to push the 45 grid 20KHz. Of course this is a starting point, and will probably move as the circuit is prototyped. I followed a similar design method when designing my 45SET amps. I used a 12AU7 input/driver. The driver stage is biased at 2.8ma of current which is correct for around 60 volts swing and 50KHz bandwidth, a point I opted on based on the transformers -3db point at 100KHz. In operation, the amplifier is flat within 1dB from 25Hz - 45KHz at 1-watt output (no feedback).

 I started putting together the breadboard for the low-cost 45 amp.... should have something function in the next couple nights depending on free time. Will post more once I have some initial measurement data.

 Regards, KM

[jrebman]

Anything to report, KM?

Just curious.

-- Jim

[floobydust]

 Jim,

 Got most of the breadboard assembled... but I'm now in Tampa for work for the next few days... so lack of playtime will most likely prevent firing it up until the weekend. I'll get some pics once the breadboard is done and do some initial tests to report.

 Regards, KM

[aerius]

Aerius, JoshK,

 I'm not completely sure what you are basing much of your recommendations on for the various tubes and driver topologies. Are you using known formulas for calculating required slew rates, current drives, etc? Again, the scope of the project is to have a very simple 45 based SET amplifier and do this at a low cost. I did start with a textbook gain stage and a reasonable output stage with hum balance for the filament. But it's just a starting point, but there is some math behind it. I don't expect this to be a final circuit of course but will post updates as the design progresses.

The problem is the textbook ideals do not happen in the real world.  Textbook theory assumes that the grid of the DHT is essentially a load with infinitely high impedance, and that it never draws grid current until it crosses over 0V.  Unfortunately, neither is true in the real world, power triodes such as 45's, 2A3's, 300B, and so forth will start drawing grid current well before the grid gets anywhere close to 0V with respect to the cathode.  The grid impedance can also be shockingly low, especially when it starts nearing the 0V.

To make a long story short, it means a lot more current is needed than the textbook formulas would suggest, and that RC coupling is not a good idea if you want to push the amp anywhere near full output.

[floobydust]

Aerius, JoshK,

 I'm not completely sure what you are basing much of your recommendations on for the various tubes and driver topologies. Are you using known formulas for calculating required slew rates, current drives, etc? Again, the scope of the project is to have a very simple 45 based SET amplifier and do this at a low cost. I did start with a textbook gain stage and a reasonable output stage with hum balance for the filament. But it's just a starting point, but there is some math behind it. I don't expect this to be a final circuit of course but will post updates as the design progresses.

The problem is the textbook ideals do not happen in the real world.  Textbook theory assumes that the grid of the DHT is essentially a load with infinitely high impedance, and that it never draws grid current until it crosses over 0V.  Unfortunately, neither is true in the real world, power triodes such as 45's, 2A3's, 300B, and so forth will start drawing grid current well before the grid gets anywhere close to 0V with respect to the cathode.  The grid impedance can also be shockingly low, especially when it starts nearing the 0V.

To make a long story short, it means a lot more current is needed than the textbook formulas would suggest, and that RC coupling is not a good idea if you want to push the amp anywhere near full output.

 I think you're missing the point... again, this is a starting point, a simple 45 amp designed on the forum which is almost certain to change (in fact, it will change) before completion. One goal is to use more commonly available parts and keep it simple. In most cases you either start with some well known circuits, ie, the textbook ones which are actual measured circuits by the tube manufacturer (this is real world), or you start with plate curves and work it out yourself. I'm very aware of how a DHT triode responds, I've designed and built many amps and preamps over many decades. My last 45 amp outputs 2.2 watts and the 2A3 amp outputs 4.5 watts which is about 10% beyond what most designs do.

 I guess my real issue with your posts is you show no backing data... you state that the grid impedance can be "shockingly low"... define what this is in a real world situation using real world data. You also state that RC coupling is flawed and can not achieve full output, what real world data do you have to back up this claim? I have amplifiers that are RC coupled to the grid of a 45/2A3 and they achieve full output and do so at very low distortion, excellent linearity, wide-bandwidth and very low output noise. I have the amplifiers and the data to back it up, all measured on properly calibrated equipment.

 I've no problem with constructive feedback as this is an open forum and this thread is intentional for gathering that due to it's nature. However, I would prefer some real facts and backup data to such broad statements about what works and what doesn't work in the real world. If you've designed and built some amplifiers and have factual measured data, that's different.

 Regards, KM

[tubesguy2]

KM -

I think what aerius and the others who have posted in this thread are trying to say is that the 12AX7 is not an "excellent choice" as you indicated back on page one of the thread, and that it is just as easy to design from the start with a better tube than it is to start with an inferior one.  That's all.  What we can't figure out is why you're so  wedded to that tube, but my guess is that most folks who peruse the thread, without prior knowledge, are now likely to use something else. If your intent was to provide a blueprint for a great amp, at a minimum you've pointed folks to some good transformer choices, and that's appreciated. - Pat

[aerius]

Starting from the beginning;

Stray capacitance from wiring, tube sockets, and other sources also need to be factored in.  Figure on 5-15pF depending on how good your layout is.  Call it 10pF.
This gives total input capacitance as 40pF.

Next, we calculate the capacitive reactance of 40pF at 20kHz.
1/(2*Pi*f*C) gives 199k Ohm for the grid, which is in parallel with the 300k grid leak resistor, giving 120k total assuming an ideal triode.

Which brings us to the first problem.  Low current high plate impedance tubes such as the 12AX7 or 6AV6 will not drive a 120k load, especially since the capacitive reactance component gives a poor phase angle for the current.  The load line will be steep and it'll open up into a wide ellipse, which means the tube will be going into the cut-off or saturation regions, possibly both.

The next problem, which is also the killer, was shown to me by a local electrical engineer who's helping me with the build of my 2A3 amps.  I refer of course to the grid impedance drop of power triodes as they near the 0V point.  For a 2A3 it starts taking a steep dive around the -10V mark and by the time 0V is reached it was only around 6-7k  For a 300B it was even worse, it started falling around -17V or so and bottomed out at under 5k by 0V.  The 45 & 50 start falling at -10V, going under 10k by 0V.  Push the grid on any of them positive and the impedance drops to the 1-5k range.

No 12A*7 nor 6SN7 is going to be happy operating into that kind of load, they just don't pass enough current and the Rp is too high.  The frequency response plot may look fine, but the distortion spectra is ugly, lots of higher order harmonics which should not be there at all.  That's why he uses WE437A and 6C45P tubes in his DHT amps, I'm not nearly as good as him at circuit layout and I don't have his tools so I currently use the 7044.  These tube can pass at least a couple dozen mA, have an Rp of 1-2k, and in short they just don't care if the load impedance has taken a dive below 10k since they have the brute power to overcome it.

Basically, avoid high Rp low current tubes when driving power triodes, it's not a job that they do well.  There's lots of better tubes for the job such as the 5687 family, ECC99, 6H30P, 6BL7, and 5842 among others.

[floobydust]

 I finally got a bit of free time and started the breadboard prototype. No driver stage yet but have the power supply and output stage mounted, wired and running. I used some old iron from an ancient HP vacuum-tube frequency counter that was dismantled years ago... I wound up with the iron 

 The power transformer has a dual-primary and has two outputs: 125VAC @ 100ma and 6.3VAC @ 1.0 amp. The choke is a 6Hy @ 100ma. HP always sourced nice parts... run cool and quiet. Under load I get about 335 volts DC from the voltage doubler and around 325 volts at the output of the pi filter. I lose another 10 volts in the output transformer (James H3115 - 5K) and bias is about -51 volts using a 1.5K resistor. Effective plate voltage is 265 volts. This works out to 34ma plate current and 9 watts dissipation. This should perform well and yield long tube life.

 I've not had any more time to play but plan on getting an input stage added shortly. I will use a 9-pin socket which will allow good flexibility for trying various tube types from many twin triodes which will include 12AX7, 12AY7, 12AU7, 12AT7 and 5751 as well. Will post more news once I get more time.



 Regards, KM

[floobydust]

 So I played some hooky last night and added a driver stage. I biased a few variations but I did make an alteration to the circuit.... I opted for adding a cathode follower to ensure better linearity with higher voltage swings. I first biased for a 12AX7, which functions well for cathode follower operation. Output power was pushing clean to 1.8 watts with just under 2% distortion. Bandwidth with the James output was flat within 1dB from 20Hz-30KHz at 1-watt output. Not bad but more tubes to play with... 5751 wasn't much different albeit less gain. The 12AY7, while a nice tube, just doesn't have enough gain to run with.

 So.... the ECC81, aka 6201, aka 12AT7... went to the plate curves and drew a few load lines for each stage and have come up with a good driver setup. Gain is still adequate but the rest of the specs have improved quite a bit. Square-wave response is very clean scoped at the 45 grid even up to 50KHz. The James outputs have a tendancy to ring at 10KHz however... but there is better iron out there, should probably try the MQ RH-40. Output power before clipping is a solid 2-watts output, bandwidth has improved being within 1dB from 20Hz-40KHz. Distortion has dropped below 1% at 2-watts output and below 0.4% at 1-watt output.

 Other circuit changes are the output stage, I employed split balancing on the AC filament. With good quality NOS tubes, unweighted output noise is below 250 microvolts. This equates to a signal-to-noise ratio of better than 80dB referenced to 1-watt (it's quiet). I'll be doing some additional testing (and listening) during the week as time allows. I like to check gain linearity over the useful output range. I'll get an updated schematic posted soon.

 Regards, KM

[jrebman]

Kevin,

Very nice, and thank you for this.  Despite the fact that I'm currently committed to building the SimpleSE, I still plan to get to this one, which really looks like it should be quite easily and cheaply built.

We have a group build going on here in town and I'm working with two other guys to build 3 SimpleSEs concurrently, and currently have most of the parts ordered and in hand, and after I finish up two smaller projects, this 45 amp will be next.  Which is to say to not worry that I've abandoned this, just had to shift priorities a bit.  The good news is that I've been forced to wake up a few dormant brain cells and I'm feeling much more confident with the design/analysis process (but still have a long way to go . )

Thanks again for your continued work on this,

Jim

[floobydust]

 Jim,

 Was a fun little project. The actual amps will be built during the next few weeks as my son has some time off from school and these are for him. The Simple SE is a nice little amp... I would still recommend the split balance approach for the 45 output. I designed this when I did my design which uses the Hashimoto iron and it's works out better than either fixed or single hum balance adjustments. You could easily add this in and the driver stage change could also be done easily. This would give you yet another "simple se" amp to play with. Do post some pics of the amps you and others are building.

 Regards, KM

[floobydust]

 Below are a few pics and an updated schematic. First is a pic of the completed prototype which the circuit was validated and tested on. Second is a scope picture. What is shows is simple. Top trace is the output directly from my signal generator which is set at 10KHz square wave at 2 volts peak-to-peak output driving the amplifier input. The lower trace is directly at the 45 grid using a 10X scope probe. This is 100 volts peak-to-peak output, ie, driving the 45 to it's maximum. Granted, I need to clean up my sig gen a bit but you can see it's basically unaltered except being 180 degrees out of phase and a gain of 50. Last pic... schematic, shows amp and supply.







 Regards, KM

[aerius]

Nice choice on the ECC81, which I think is one of the better tube in the 12A*7 family.  Here's a little something fun for you to try, which I've had a lot of fun with in the input stages of my amps.



LED biasing, simple, and it works.  It allows you to get rid of the cathode bypass cap without losing any gain as the AC impedance of the LED is only a couple dozen ohms at most.  Use a red, orange, or yellow LED rated for 5-10mA, and the dimmer it is the better.  Something like this.

[floobydust]

 Thanks Aerius,

 I've actually done similar things back in the 70's and 80's... current sources, current sinks, SRPP, Moscode (aka George Kaye) etc. and the bias battery too (that was popular in the 50's) Over the years, I've mostly abandoned hybrid approaches. I still think a resistor and cap of good quality works fine, plus you can tailor the bias operating point (LEDs have a narrower voltage range) and low-frequency roll-off which is important. Depending on the bandwidth of the output transformer and the intended frequency range used, allowing the amp to run down to sub-sonic frequencies is really not a great idea and can overload the amp for normal listening levels from something as simple as a warped record. Plus, I really don't see anyone using a 2-watt amp to reproduce a 16Hz pedal note from a pipe organ. I generally tend to roll off below 10Hz but prefer not to stack the time constants together between the stages.

 The one (hybrid design) that worked out extremely well was a highly accurate and temperature compensated current sink for a long-tail phase inverter (replaced the large common cathode resistor). This worked wonders for phase inversion accuracy, linearity, reduced distortion and high voltage swing, but the parts cost to build the little guy was pretty high due to the parts used which included a high-voltage MosFET, precision voltage reference, and DC-DC convertor with isolation and regulation. It went into a custom amp design based on Dyna Mk-II/III iron and metalwork. I did this design back in the 80's and still have the amps... used a 6FQ7 phase-inverter/driver and a 6072 for the input amp and a pair of real KT88 outputs. Very nice amplifier overall and a solid 50-watt output with wide bandwidth and very quiet. I did employ about 6dB of negative feedback however, but the amp was exceptionally stable in open-loop operation.

 Overall I think the (low cost 45) design worked out pretty well and it does keep a low price tag with good quality parts. I was a bit surprised that I was able to get such low-distortion from the circuit... but I'll take it. The power supply could be improved using a more typical, albeit more expensive, plate/filament transformer and a vacuum-tube rectifier (and better caps). I'll draw one of those up shortly and post it as well for those who prefer an upgrade or want to build a single stereo chassis. Ultimately I prefer a monoblock design as it provides total isolation of channels and you can use shorter cable runs to the speakers for less loss. When you build your 2A3 amps, try the split balancing approach. I've used this with NOS 2A3 tubes and have had excellent results with low output noise.

 Regards, KM