[floobydust]

 Okay,

 Now that everyone has their favorite single driver speaker, seems some folks need something in the DIY camp to drive it. Cost is always an issue... hence DIY, but I would suggest to start from the end of the amp and work your way back, ie, start with a good quality output transformer and pick a known performer for the output tube which you can still find. IMHO, the 45 triode is still a performer and good quality NOS pieces can still be found at fair prices. If you're interested in going this route, read on... if not, exit here.

 So, 45 triode is in... which also means you'll need a nice pair of 5K primary trannies to pair them up with. There are numerous choices out there from both the US, Japan, Europe and Asia. Budget iron in the US is Hammond with basic iron under $30 each to bigger stuff at $100 followed by Magnequest (Mike offers the Robin Hood series at $99 each). Taiwan has the James Transformers at $145 a pair and then you'll find all sorts of brand x-y-z all over the board on quality/price. I built some amps with the Magnequest RH-40 outputs and the 45 triode... nice. I also have the James outputs in black but they are still in the box. They have received good reviews above Hammonds. So you should look at spending $300 for the output iron and tubes. This is the bulk of the cost and really where the quality pays off.

 If anyone is still interested, I'll finish up with two power supply circuits which are affordable and the input driver circuitry. This is a project that my Son will finish up... all the parts are on hand and cost was quite low.

 Regards, KM

[jrebman]

Ok, I'm interested .  I'll have to look into those more affordable MQs -- I didn't see them mentioned on the web site, but then again, I didn't look all that hard either.  I've definitely had my eyes on the James and Hashimoto for my other project though.

-- Jim

[floobydust]

 Okay, I think the first thing to do is showcase some good iron to pick from... my overall favorite is Hashimoto, but they are expensive. So I'll start with that and work our way down.

http://www.tube-amps.net/Hashimoto_Products.htm - the H-507S is what I use in my personal amps... superb but also $172 each, or $344/pair/ They also offer bigger iron at 20- and 30-watt (I have both here) but they are far beyond the costs implied here.

http://www.magnequest.com/robin_hood_series.htm - The RH-40 is what I used in some entry amps, nice at the price but under chassis mounting seems more natural and safe as no wires and terminals stick up from the bobbin. Nice at $100 each, or $200/pair

http://euphoniaaudio.netfirms.com/ea/nfoscomm/catalog/product_info.php?products_id=142&osCsid=7462650e5cdabcd82c0f690f5ec050cb - Only sold in pairs.... and the bargain so it seems at $145/pair. I have these but not built up yet. Nice quality for the price and terminals underneath.

http://www.angela.com/catalog/transformers/Hammond_Transformers.html - Hammond 125ESE is the budget performer at $44 each, or $88/pair but are universal types so not really optimized. They offer less expensive ones as well, but no real point.

 Based on keeping this a budget-conscious effort, I would recommend either James or MQ as your budget requires. Remember to find a decent NOS pair of 45 tubes... expect to pay anywhere from $80 - $150 for a matched-pair. I recently picked up a trio of 45 NOS in the boxes under the Zenith label for $120... all are superb and perfectly matched, truly a bargain find. Look beyond ebay too...

 The 45 is pretty easy to drive and a single high-gain triode will fit the bill. Most universally agree that the 12AX7 is just plain great for an application like this... albeit prices are just high for good NOS samples. The 6AV6 is readily available and dirt cheap in NOS samples and is essentially a 12AX7 single triode in a 7-pin miniature. In any case, use what you have... 12AX7, 5751 or a 6SL7.

 I would suggest using Hammond iron in the supply unless you can find some NOS pieces at good prices. There are two options here... build a single chassis stereo amp (and common power supply) or separate monoblock amps. I can show simple and effective power supply designs for both.

 There are some good surplus places that have some NOS tube sockets, wirewound potentiometers, metal-film resistors, capacitors, trannies and chokes and you can probably score most of the parts on the cheap. I'll put together a simple schematic for the amplifier and two for the power supply. Sprague Atom capacitors will work fine for filtering and bypass and are cost-effective.

 Regards, KM

[jrebman]

Great way to get started -- thanks.  I am going to explore a couple of other options that I know about to see how they might stack up.  A couple of questions first:

How much filament current would we need to be able to cover all bases -- monoblock or single chassis, twin single triodes or single dual package, and of course the 45s (I just don't have this data in an easy to access format right now.)

What about the possibility of tube rectification as an option?  That would typically add a 5v filament winding at 2 - 2.5 amps.

A couple other transformer makers I'm going to look at are: Electra-print, Transmatic, Antech, Transcendor, and possibly Plitron.

What kind of wattage rating should we be figuring for  a minimum, and ideal for the OPTs?

PM coming your way on another subject...

-- Jim

[Dmason]

There are really good opportunities for outputs for not alot of money, and this component is second only to power supply in importance. Jack Eliano and Mikey LeFevre both have really nice irons, the Robin Hood MQ, and also Jack will wind single Zout 8 which goes by some name like Arthur, but he will know what it is, in the 2.5K/3.5K flavours, I believe they are ~$108 ea. The Robins are ~$100 ea, and both have a nice, soulful sound.

Transcendar and One Electron are also really nice irons, and many have raved about the stout 3.5K Transcendars with 300B, so there are lots of choices.

I like the idea of a single? 12AX7 driver for 45. Mo simpla is mo betta. This sounds good already!

[floobydust]

 Jim,

 Yes, there are many options on OPTs... Jack at E-P makes nice iron too.... you just need to pick your poison acquire a pair. Personally I wouldn't drop much below the James level.... otherwise just buy one of the less expensive kits and build it.

 Three supply options I'm thinking are:

 1- Stereo amp - uses proper plate/filament transformer and a 5V4G rectifier. It also uses a pair of 2.5VCT@3.0A filament transformers for the 45 triodes and a common filter choke.

 2- Mono amp - uses generic 125VAC and 6.3VAC power transformer. Solid-state full-wave voltage doubler yields proper plate voltage/current. Also uses a single 2.5VCT@3.0A filament transformer and a single choke.

 3- Mono amp - uses proper plate/filament transformer and a 5V4G rectifier. It also uses a 2.5VCT@3.0A filament transformer and a single choke.

 In general, you should consider having approximately 340 volts available... you'll have around 275V on the 45 itself (cathode to plate), self bias will suck up another 56V on average and you can usually count on around 5-10V loss in the OPT.

Filament voltage/current for various tubes:

5V4G full-wave rectifier 5.0V @ 2.0A

45 Power Triode 2.5V @ 1.5A

2A3 Power Triode 2.5V @ 2.5A

12AX7/12AU7 12.6V @ 0.15A or 6.3V @ 0.30A

6AV6A 6.3V @ 0.30A

 You can always get a solid 2-watts RMS output from a good 45 in a good circuit with adequate supply. A 5K load seems optimum in most cases.
Even if you get a bit more or less it doesn't amount to more than a fraction of a dB or so. The transformers I listed are adequate for all but low frequency extremes. Most are rated at 5-watts or so and show -3dB points at 30-50Hz. The Hashimoto H-507S in my 45SET amps are within 1dB from 25Hz - 45KHz at 1-watt output.

 I was also looking at other input tubes... the 6SF5 is a good option.... actually Don Garber uses this in his Fi SET amps.

 Regards, KM

[jrebman]

Yup, one rectifier, one driver, two outputs -- the same things that drew me to the Vaughn Carina, and which I'm just waiting for my number to come up on.

I've got some nice 5751s and a pair of GE 12ax7wa in Tektronix boxes, and for those who know what Tek did to qualify their tubes, you'll know why that's interesting.

Only potential downside to 12ax7s are pretty slow slew rate, and maybe a bit too much gain.  We should also look into whether the 12AT7 will work.

-- Jim

[jrebman]

Ah, you posted while I was typing.

Thanks for the filament data -- I need to find an online source of tube data that I can work with.  Think I'll work on that this afternoon.

All good stuff.

-- Jim

[floobydust]

 Jim, et al,

 Here's an online source for vacuum tube data: http://tdsl.duncanamps.com/tubesearch.php

 I'm thinking a very simple design... let's face it, when you get bare bones, there's not much variation from classic textbook design. As this is a DIY project I want to avoid direct-coupling... biasing can be touchy and there's no good reason to risk toasting a nice pair of 45 triodes. Besides, there is only 1 coupling capacitor... you can make it a nice one.

 I'll get a schematic posted before end of day for the basic amplifier circuit. 12AX7 and 5751 are excellent choices. As this is a 2-stage amp you'll need some decent gain in the first stage, hence the 12AX7 or 5751. If you go mono you can parallel the triode sections but I don't really like any paralleled tube configurations, then again, don't like SRPP stages either.

 Regards, KM

[floobydust]

 Hope these are legible schematics.
First the base amplifier circuit:



 Then the first power supply version:



 Regards, KM

[aerius]

A few things.  The plate impedance of a 12AX7 or 6AV6 is somewhere in the region of 70-80k ohms, and the plate resistor is only 100k which is far too low.  Tubes like to see a plate resistor of around 4-6 times their plate impedance in order to run with minimal distortion, however this isn't really possible with a 12AX7 unless a CCS is used.  As is, the 12AX7 or 6AV6 is going to distort, a lot. 

Also, a given that a 45 is biased around 50V or so and typical sources these days put out a healthy 2V or so, the high gain from the 12AX7 or 6AV6 is mostly wasted since all it does is overload the output stage, and with RC coupling the overload recovery is slow and ugly.  A 12AT7 or 12AY7 will give you enough gain, along with the bonus of having a much lower plate impedance so that they'll actually run fairly distortion free with a 100k plate load.  You could even drop the plate resistor to around 65-80k with the 'AT & AY7, respectively.

Or if you really want to have some fun, redesign the input stage to use a 5842, or if you're somewhat insane, the 6C45P.  Much different sound from the typical audio tubes, much faster, much more wide-band, and more dynamic.

[jrebman]

Or if you really want to have some fun, redesign the input stage to use a 5842, or if you're somewhat insane, the 6C45P.  Much different sound from the typical audio tubes, much faster, much more wide-band, and more dynamic.

And possibly spend the rest of your life trying to get it quiet .  No, actually, very nice tubes, as is the WE417, but need very clean filament supplies and good shielding, and very short leads  to keep them from oscilating.

[aerius]

I learned that the hard way experimenting with the 6C45P in a headphone amp, but once I got it working it was pretty amazing.  It was all peachy once I slapped gridstoppers on it, put in a nice fat ground bus, made the layout as compact as I could and terminated the unused pins.  It took me a week or two to sort it all out, fortunately, I'm still young enough that I didn't get any grey hair or bald patches.

[JoshK]

I wrote a big long post at work just before leaving and forgot to post it... 

I wouldn't use a 12ax7 unless there were really no other alternatives, but that is me. Its high rp and low Ip make it a pretty poor single driver tube IMO.  It is going to have high distortion, but also it is going to suffer from blocking distortion when clipping making it not clip nicely.

[floobydust]

A few things.  The plate impedance of a 12AX7 or 6AV6 is somewhere in the region of 70-80k ohms, and the plate resistor is only 100k which is far too low.  Tubes like to see a plate resistor of around 4-6 times their plate impedance in order to run with minimal distortion, however this isn't really possible with a 12AX7 unless a CCS is used.  As is, the 12AX7 or 6AV6 is going to distort, a lot. 

Also, a given that a 45 is biased around 50V or so and typical sources these days put out a healthy 2V or so, the high gain from the 12AX7 or 6AV6 is mostly wasted since all it does is overload the output stage, and with RC coupling the overload recovery is slow and ugly.  A 12AT7 or 12AY7 will give you enough gain, along with the bonus of having a much lower plate impedance so that they'll actually run fairly distortion free with a 100k plate load.  You could even drop the plate resistor to around 65-80k with the 'AT & AY7, respectively.

Or if you really want to have some fun, redesign the input stage to use a 5842, or if you're somewhat insane, the 6C45P.  Much different sound from the typical audio tubes, much faster, much more wide-band, and more dynamic.

 Actually, the design values were taken directly from Philips' published data dated January 1970. According to said data, the ECC83 (aka 12AX7) will have a gain of 57 v/v, have a maximum output of 30 VRMS (which is around 85 p-p) at 2.7% distortion maximum... all based on the operating parameters per the schematic I posted. If this was a big problem I think someone would have found it before now. You can also find lots of Modified Dynaco PAS preamps with near identical stages in them. Also, the 45 requires more voltage drive than a 2A3... so the amp will not have that much overall gain.

 Regards, KM

[JoshK]

But it has no current output so its going to roll off from the miller C of the following tube.  The 45 doesn't have a great deal of miller C but it has some.  Then since this is cap coupled and there is no Ip to speak of you are going to run into blocking distortion, since you don't have the I to charge up the capacitance quickly enough.  I think this is why there is a huge multitude of designs out there that try other means of getting there.

I read some experience from a few experienced DIY'ers such as SY who claim that the 12ax7 *can* have good linearity and distortion figures but that it is *extremely* finicky and hard to actually obtain in a real circuit.

[floobydust]

But it has no current output so its going to roll off from the miller C of the following tube.  The 45 doesn't have a great deal of miller C but it has some.  Then since this is cap coupled and there is no Ip to speak of you are going to run into blocking distortion, since you don't have the I to charge up the capacitance quickly enough.  I think this is why there is a huge multitude of designs out there that try other means of getting there.

I read some experience from a few experienced DIY'ers such as SY who claim that the 12ax7 *can* have good linearity and distortion figures but that it is *extremely* finicky and hard to actually obtain in a real circuit.

 Josh, I think time has proven that the 12AX7 has been one of the most used audio tubes ever... the list of top-end manufacturers over the decades from the real heyday of audiois endless.... McIntosh, Marantz, Harmon-Kardon, Dynaco, the list goes on. The 12AX7 was successfully used in phono preamps, high-level stages, tone control and EQ circuits, McIntosh used them to drive output tubes as well. I can't possibly consider it a "finicky" tube by any means.

 Regards, KM

[JoshK]

Well I don't really care to argue about it, especially when your retort is what it is.  All those names you mentioned used that tube in a very different way, namely gain over linearity because they used gobs of feedback.  The tubes were designed for gain first, distortion/linearity a distant second.  Here in this setting we aren't using gobs of feedback so linearity is crucial and there are much better tubes for this type of design IMO.  If you choose to use it that is fine, I just stated why I think it isn't a great choice.  If you get good results I'll cheer for you and be interested to see how. 

There is plenty already written about why the 12a*7 family is a poor choice for these kind of circuits so I won't rehash it all here. 

[aerius]

Actually, the design values were taken directly from Philips' published data dated January 1970. According to said data, the ECC83 (aka 12AX7) will have a gain of 57 v/v, have a maximum output of 30 VRMS (which is around 85 p-p) at 2.7% distortion maximum... all based on the operating parameters per the schematic I posted. If this was a big problem I think someone would have found it before now.

Umm...yeah...that's the problem right there.  Gain of around 57, maximum output of 30V rms, and the output of a typical CD player is around 2V rms.  Ok, let's say you have a weak CD player that puts out 1V rms, even then you'll run out of voltage headroom and clip the signal before the volume knob gets cranked to full, in other words, too much gain in the system.

Also, there's no way a 12AX7 with a 100k plate load will have a single-digit distortion figure while putting out 30V rms into the grid of a 45 using RC coupling.  The 2.7% figure assumes that the tube is connected to a high-value strictly resistive load, unfortunately, the grid of a DHT isn't one.  As JoshK mentioned, it's got Miller C (not to be confused with beer) and worse yet it starts drawing grid current well before the grid reaches 0V with respect to the cathode.  Which means distortion is going to be a lot higher than the ideal datasheet values.

I'm not nearly as experienced as SY when it comes to DIY stuff so I haven't been able to make a 12AX7 sound good and measure well, but I know it can be done since the Artemis Labs PH-1 phono amp manages to do that.  Thus I go after tubes which more or less use the brute force method of solving problems.  For instance, Miller C isn't a problem anymore when I have an Rp of 1-2k and 30mA to play with.

[JoshK]

As aerius so keenly points out, swinging 10's of volts is different then swining mV's.  For phono stages its a better choice, but still Allen Wright, Arthur Loesch and many others have written about its slewing distortion there and why there are much better choices still.