Let's build a tube preamp

[moody]

For the masses. Consider a single 24 volt transformer with a voltage octupler attached for both heater and B+ voltage. 24 volt through an octupler would be about 270 volts (of the top of my head) Given the low current requirements this is a very doable.

 

[agedhorse]

An "octuples" might have a lot more losses than you might expect. I haven't done the calculations, but iirc the larger the number of "doublings" the less perfect each "doubling" becomes.

 

[moody]

An "octuples" might have a lot more losses than you might expect. I haven't done the calculations, but iirc the larger the number of "doublings" the less perfect each "doubling" becomes.

True there are losses, but not as bad as you may expect. Modern and larger cheap capacitors go a long way to minimising losses. I have made a few amps with quadruplers and a preamp with an octupler, no issues. I did use very big capacitors (ex computer power supplies). A 24 or 30 volt transformer is a lot more available than high voltage, especially only for preamp valves.

Bunch of food for thought here: Invalid Link Removed

 

[Martin Beer]

True there are losses, but not as bad as you may expect. Modern and larger cheap capacitors go a long way to minimising losses. I have made a few amps with quadruplers and a preamp with an octupler, no issues. I did use very big capacitors (ex computer power supplies). A 24 or 30 volt transformer is a lot more available than high voltage, especially only for preamp valves.

Bunch of food for thought here: Invalid Link Removed

This is a bit of a digression from the main thread, but for DIY preamps based around cheaply available transformers, it can work to use a pair of back-to-back transformers, where the mains voltage is stepped down to 6 or 12v (which can be rectified and used for DC heaters), then a second transformer is used to step back up to a higher voltage. This was used in Fred Nachbaur's Real McTube project - Nachbaur used 110v/12v transformers back to back for a B+ of around 140V, but if you used a transformer with a 230V winding you could achieve a B+ closer to 300V. Nachbaur's schematic is here: The Real McTube Article: How It Works
I'm sure nobody would design a commercial product that way, but as a way of using inexpensive or salvaged transformers for DIY it doesn't seem like a bad idea.

 

[Martin Beer]

I read that too. I'm not sure. The power switch turns on the amp, and the standby switch kills downstream B+ while allowing the heaters to get hot. Trying to keep it simple, I was going to stick to the old school with details like that.
Post what you come up with in regards and we can talk about it, I'd like to hear more.

I was thinking primarily of Merlin Blencowe's page on the subject; The Valve Wizard, although I have seen similar points made in other places; Link Removed

The gist of those arguments is that cathode stripping does not occur to any significant extent at the voltages used in MI amplifiers and the standby switch is a design idiosyncrasy of 50's Fenders which was copied by Marshall, and once those two big players used them, everyone else followed suit. Although I'm not sure where Ampeg fit into that as they also used Standby switches early on, IIRC.
The standby switch also tends to be based on a switch which is not rated for the voltage it sees (which can cause reliability issues), and it is argued that running the heaters with no other applied voltages can be harmful to your valves. I'm not an EE myself (so not best equipped to defend those points), but it seems persuasive to me and I have had 40 year old pieces of equipment with no standby switch which were still running their original valves, so it does appear that any consequences of leaving it out are not catastrophic. On a preamp I would be tempted to use a simple mute switch to short the audio signal to ground, which could also simplify your layout by keeping the power switch on the back panel.

 

[moody]

This is a bit of a digression from the main thread, but for DIY preamps based around cheaply available transformers, it can work to use a pair of back-to-back transformers, where the mains voltage is stepped down to 6 or 12v (which can be rectified and used for DC heaters), then a second transformer is used to step back up to a higher voltage. This was used in Fred Nachbaur's Real McTube project - Nachbaur used 110v/12v transformers back to back for a B+ of around 140V, but if you used a transformer with a 230V winding you could achieve a B+ closer to 300V. Nachbaur's schematic is here: The Real McTube Article: How It Works
I'm sure nobody would design a commercial product that way, but as a way of using inexpensive or salvaged transformers for DIY it doesn't seem like a bad idea.

That's the other option, depending on whether you want one or two transformers. An octupler needs a lot of capacitors and diodes.

 

[NKBassman]

Thinking about my Universal Audio tube mic preamp, and there's no standby switch. Food for thought.

 

[rufus.K]

Right on, everybody. Good food for thought.

 

[Mark Reccord]

@agedhorse beat me to it but a 2 diode full wave rectifier on a transformer with a CT is equivalent to a bridge on a non-CT transformer in terms of the rectified voltage. In fact, for a unipolar power supply like in a tube preamp, using a bridge with a CT transformer doesn't make any sense because it will give you a bipolar one and you don't need a negative supply here. The 0.9 factor is the DC equivalent of the RMS AC voltage before any filtering. Once you add filter caps, the DC voltage will be very near the peak AC voltage minus any diode drops, so ~1.4x the AC RMS voltage. Since the transformer is rated for 330VAC at 140mA, it's going to be higher than that if loaded less, which is a given here since 2-3mA is a typical 12AX7 current, so 465VDC is on the low side of what you'll see on the HT here. It will probably be more like 500V+. Not a deal breaker but you'll need significant power resistors to drop it to a usable level. I typically like ~350VDC for a HT supply in a 12A_7 based preamp. Typical bias currents for 12AX7s are less than 3mA so even if you did 5 stages at 3mA you're only looking at 15mA. Let's round up to 20mA for fun. With 20mA, in order to get 350V from 500, you'd need 7.5k of series resistance capable of dissipating 3W. There are lots of 5 or 10W power resistors around so that's not a big deal. Given that much series resistance you could easily do CRCRC filtering and not bother with the choke at all. So the first thing to figure out is what tubes you're going to use and how you want to bias them so you can get a bead on how much current you need from the supply. That's usually how you decide on a PT but there's no reason we can't work with what you have. That said there are lots of pretty cheap options out there from places like Edcor or Triad Magnetics. @beans-on-toast put me on to this one last year:

Invalid Link Removed

It's a basic 120V primary, 230V secondary toroid at 110mA, which gives you about 325V at 110mA but more like 370V at the 25mA I'm drawing out of it. Doesn't have a filament winding but you could buy a Hammond 166 for that. You'd still be looking at under $50. Edcor has a bunch of tube PTs that are around $50 as well.

Another note on rectification: Diode drops aren't going to be significant in the HT but they will be in the heater supplies. I would normally default to regulated DC heaters in any low level signal application but you might not have the margin on those heater windings to accommodate a diode drop and the regulator drop. You might get away with it using the 3.5A heater winding because with 3x12A_7 tubes (right?) you'll only draw about 0.9A so the winding will be lightly loaded and will be more than 6.3VAC under that condition. You'd want a low dropout regulator at any rate. I usually use LM317s for this but they drop around 2.5V at 1A. There are LDOs that only drop a volt or so from in to out (like the LT1086 for instance) so you could use one of those. That said, I'm using a Hammond 166 filament transformer rated for 1A in one of my projects with a single tube (300mA) and I was able to use an LM317 with 1N5822 Schottky diodes to get 6.3VDC for my heaters. You could also get away with a bit less than 6.3V and in fact lots of folks recommend doing that. The only penalty is that the tubes will take a little longer to conduct.

Don't bother with a standby switch for this application. It's totally unnecessary.

 

[Mark Reccord]

Also, you could drop a good bit of that excess voltage from the PT by using a tube rectifier...

Fast recovery diodes make less switching noise than regular ones. There is potential for that high frequency noise to make it through the power supply riding on the DC and then intermodulate with the audio signal, especially if you're using gain stages with no power supply rejection like these. I think it's mostly a non-issue given good PSU filtering but UF4007s aren't significantly more expensive than 1N4007s so I tend to go with the fast recovery ones.


EDIT: A "goo" bit?

 

[beans-on-toast]

As was mentioned, voltage multipliers are an option to consider. Attached is a voltage multiplier design guide.

The amount can vary but when you double the voltage, the current is reduced. Typically a doubler reduces the current in half. Unless you want to run it off an external switching supply for regulatory reasons, I think that it makes more sense to use the right transformer for the job. I've seen multipliers work well in amps with a single tube.

I sometimes use toroidal transformers for preamp power supply prototypes. An advantage is that you can easily add or unwind turns to adjust the output voltage as needed.

A power supply needs to be able to respond well to the demands of the amp and to be able to recover quickly. The lower the note, the more energy it requires to be reproduced. Some preamps loose headroom and distort when the lowest bass note is played. Some can't deal with a five string with a low B. Some can't keep up with quick passages and distortion or compression sets in. A proper balance in the power supply's time constant has to be found to make it work well.

 

[beans-on-toast]

Here is a low cost DIY regulated heater power supply. DIY PCB - Tube Amp - LV Tube Heater DC Power Supply - 6.3VDC from 6.3VAC

With low dropout components, you can use a 6.3VAC transformer to generated a regulated 6.3VDC. This is neat.

Of course, the current capacity has to meet the requirements of the amp, so it depends on the tube compliment. I haven't tried this out but it looks interesting.

 

[rufus.K]

So much good inforation here. I'm going to gut my chassis this weekend and take a bit of inventory, both of hardware on hand and of what was just said, before proceeding on power supply.
With all of that going on, maybe we can start talking about the first stage of the audio path in parallel to the PSU discussion.

 

[beans-on-toast]

I think it's mostly a non-issue given good PSU filtering but UF4007s aren't significantly more expensive than 1N4007s so I tend to go with the fast recovery ones.

I use ultrafast as well. Can you hear a difference in this type of application? Not sure, but I can see a difference on a scope. When I look at the baseline noise of a 1N4007, you can see a lot of fine hairy spikes, those aren't there with the UF diodes.

 

[static0verdrive]

sub

 

[agedhorse]

For a heater regulator, a simple Zener diode with pass transistor is a simple, effective compromise.

 

[CarpeBasso]

Subbed.

 

[negativefx]

Once we get the preamp built and explained, there is no reason we couldn't carry on and build a phase inverter and output section. I'd love to talk about PI theory when we get there.

What do you think of the SS vs Tube rectifier question?
What do you think of the heater question (AC or dc)?

SS rectifier is probably best for now. Tube recs are finicky.

 

[rufus.K]

For a heater regulator, a simple Zener diode with pass transistor is a simple, effective compromise.

What is a pass transistor?

 

[rufus.K]

Invalid Link Removed

For those unfamiliar, a Zener diode acts like a diode in that it passes voltage in on direction (clipping off one half of AC if present), but with one main difference. That is, it has a value that, if exceeded, the diode will allow it to flow the opposite direction.