12AT7/46 Direct Reactance Drive Amp

[Steve Cresswell]

Steve,

As shown, the schematic has 20.85mA going through the 46 cathode resistor and 9mA through the C3g making 30mA in total being drawn through the 46.

At 250Vp-k @30mA for the 46, Vg-k should be closer to -28V than the 31V shown on the schema.

Ah Well,

On the surface that might appear to be the case.
However if you look carefully at the schematic, you'll see that what is
actually happening is that 20mA is being drawn through the 46, then 9mA of that 20ma is being stolen, by the c3g, from the top of the cathode resistor to provide it's working conditions.

Kirchoff's current law states that if approximtely 20mA is entering the system (the 20mA or so having been set by the 9K4 cathode resistor) then 20mA or so has to leave the system at the bottom. Therefore if 9mA branches off via the c3g then the the remaining current to flows to earth via the 46s cathode, where it once again meets the 9mA going through the c3g and we are back again to about 20mA.

That's how these DRD/Free Lunch/Monkey amps work. Jeremy Epstein has a good explanation of this in his notes for his "Free Lunch" 2A3 amp. Google "jeremy epstein free lunch and his site"appears at the top of the results list.

Epstein argues that the cathode resistor calculation needs to take into account the stolen current and subtract it from the total current going through the system, which means that in theory my cathode resistor ought to be a bit bigger than 9K4. However the DRD is a slightly different setup to the Free Lunch.

I have built three DRD amps and have not found any need to worry about increasing the cathode resistor value. The initial total current requirement at the top of the the power valve, in conjunction with the usual loadline stuff, always seems to give me the right voltages at the driver plate, so I dont bother.

The voltages seen on the schematic are the actual voltages measured in the working amp. These agree spot-on with the calculations I have described.

Steve

 

[mach1]

Kirchoff's current law states that if approximtely 20mA is entering the system (the 20mA or so having been set by the 9K4 cathode resistor) then 20mA or so has to leave the system at the bottom.

Steve, I totally agree: only problem is you have 30mA coming out at the bottom. I am very familiar with Jeremy Epstein's writings and the argument that the cathode resistor calculation needs to take into account the stolen current and subtract it from the total That is PRECISELY what I am proposing here. I don't see how the DRD circuit is immune from ohm's law: Jack Elliano places the additional resistance needed to attain the correct Vg-k of the output tube in the cathode of the output tube, whereas Jeremy places it in the branch that supplies the choke and plate of driver tube. Fortuitously, your amp uses a choke with the correct resistance to achieve the correct Vg-k of the output tube, and therefore the circuit does not require any added resistance. Is it therefore a lunch, drd or haemaphrodite?

 

[mach1]

This 46 drd circuit

http://fullrangedriver.com/singledriver/46amp.html

Calcs out perfectly. 38mA is drawn through the 46. Of that 29mA is routed through the 7k cathode resistor (202.5/7000), while the remaining 9mA is routed through the 6AN4 cathode resistor (1.275/135). The subtraction of the driver current is clearly illustrated in this example.

 

[Steve Cresswell]

Thanks for that.

Looks like I have the theory right, which is why everything appars to work out. It's the application that is not quite right, due to me not quite getting what Epstein meant in his notes.

Looks then like I'm overrunning the 46s by almost 50% at about 7W dissipation. As in the design you linked to, they don't appear distressed in any way and the amp sounds great. It is arguably the best amp I have yet built. But all the same I don't like the idea of overrunning these things. Working life must be being reduced. I just hope it doesn't end up eating a pair of tubes every six months.

Problem is, it sounds so good I'm loath to start messing with it.
It is the opposite of James amp you linked to. It's blindingly fast and dynamic with lovely treble and very deep lows.

I'll get a couple of spare 46s in. If it starts eating them I'll lower the currents, but the plate structure looks very hefty to me so you never know I might be OK.

Steve

 

[Steve Cresswell]

Here's the schematic again, hopefully right this time

An externally hosted image should be here but it was not working when we last tested it.

 

[Sheldon]

Looks then like I'm overrunning the 46s by almost 50% at about 7W dissipation.

As I read the data sheets, the max average dissipation is specified at 10W per tube in class B operation. Average is the same as constant for the purpose of calculating safe operating conditions. Now, that is with G1 and G2 connected together. It's not clear that the conditions shown for class A1 operation, other than voltage, are the maximum permitted values. As long as G2 isn't getting too hot (and you can play with the G2 resistor here) I think you should be fine.

Sheldon

 

[Steve Cresswell]

Hi Sheldon,

What you've said sounds logical.

I feel a bit more reassured now.

Thanks.

 

[Steve Cresswell]

Anyway

Here is the finished product

An externally hosted image should be here but it was not working when we last tested it.

I changed the rectifier from a Mullard GZ37 to a Mazda 4V UU5 IHR on a British 4 pin base
This put the icing on the cake as far as the sound was concerned.
it is fast, dynamic, detailed and with a lovely treble sweetness that seems to have been imparted by the UU5.

Nice.

 

[mach1]

Steve,

I wouldn't worry the least bit about running the 46's at your operating point. I have always found them to sound good when 'juiced up' a bit, and they do have a relatively large plate structure for the stated max dissipation. The main thing is that the amp sounds good !

Love your work btw: excellent aesthetic sense and craftsmanship. Your Jetson's 6AS7/6080 pp

http://wduk.worldomain.net/forum/showthread.php?t=2933&highlight=steve&page=13

is a classic that belongs in a design museum.

 

[Steve Cresswell]

Steve,

Love your work btw: excellent aesthetic sense and craftsmanship. Your Jetson's 6AS7/6080 pp

http://wduk.worldomain.net/forum/showthread.php?t=2933&highlight=steve&page=13

is a classic that belongs in a design museum.

Thanks

That 6080 was a nice amp.

You could say I've now gone minimalist in comparison.

Steve

 

[Richard]

modified C3g 46 amp

Hi Guys,

After spending some time following this thread, reading up on direct coupling and direct reactance amps, I've managed to put together a modified version of Steve’s amp. Thanks to Steve for permission in allowing me to modify his circuit and other advice he’s offered. This is my fist amplifier design and will be my first amp build. I’ve done a couple of tube dac outputs and want to make the leap.

I was considering building the direct reactance version but wanted to try regulating the driver’s voltage and leave the option open for LED or battery bias etc. I’m using a using a 175mA transformer so I'm not worried about the “free lunch” aspect, as there is current to spare.

I’m wondering if you guys would be kind enough to comment on this circuit.

My main query is, if the filaments to the 46 are heated before B+ is applied, would it matter if there is a short delay in applying the grid voltage (driver plate voltage) due to the action of the VR tubes?

Cheers,

Rich

 

[serph1nia]

What a great amp it is you have there.

Have you had a chance to compare it to your px25 drd amplifier?
or the 2a3 and thee 45 for that matter.
http://wduk.worldomain.net/forum/showthread.php?p=62277

In the px25 amp it seems you like the ultrapath circuit.
Have you tried this in this amp too?

Also, have you tried a 'normal' choke loaded, cap coupled design with better quality parts on the 46 cathode.

Sorry for just selfishly throwing questions at you, but since I might built an amplifier like this i'd really like to know

Thank you.

 

[Sheldon]

Re: modified C3g 46 amp

My main query is, if the filaments to the 46 are heated before B+ is applied, would it matter if there is a short delay in applying the grid voltage (driver plate voltage) due to the action of the VR tubes?

I don't see much problem with plate dissipation. The thing you have going in your favor is a pretty stout cathode resistor on the output tube, which will limit the current. Also, as the current increases, the plate voltage will decrease, further limiting the plate dissipation. You might check the potential grid current, though. Do a few calculations of failure scenarios. If it's a concern, you could use a zener string in parallel with your VR tubes, but about 50V higher (or enough to light off the tubes).

Sheldon

 

[Steve Cresswell]

Hi Rich

Looks like a nice design. Look forward to hearing how it sounds.

Serph1nia

Compared to the PX25 DRD, the 46 version sounds a bit warmer across the midband and more beefy at the bottom end. It doesn't quite have the treble air of the PX25 amp, but to me it communicates the music a bit more effectively.

This could be due to the fact that the PX25s are new production TJ mesh plates.
I have heard it with old Globe PXs and with these, it is a stunningly
musical device. However I simply cannot afford the prices being asked for globes.

I have not tried the ultrapath connection with the 46 amp as I am too busy enjoying the sounds it produces Maybe its something I ought to try. However it would mean a complete rebuild as I simply don't have the room on the chassis for a pair of big oil-filled ultrapath caps.

I have not tried a normal cap-coupled amp. Whether I would still have the speed with cap coupling is debatable.

As it is, I am really pleased with the way this amp has turned out. It is definitely worth building.

 

[Sheldon]

Re: modified C3g 46 amp

I’m wondering if you guys would be kind enough to comment on this circuit.

Another option to consider - a Loftin White version.

Connect the cathode bypass for the 46 to B+. Something like a 10-20uF film cap is plenty.

Omit the C3g bypass cap.

Cap couple a resistor from the 46 cathode to the C3g cathode. Value of the resistor should be the C3g cathode resistor, times u of the C3g. In this case 360X40 = 14.4k. You'll want this adjustable so you can trim for minimum PS hum. Cap should be about 1-2uF film (bigger OK).

It will be a very quiet amp, even with a fair amount of PS ripple.

The only limitation for the design is the reactance of the choke at low frequencies, which will lower the gain, due to degeneration at the cathode resistor. With the kind of speakers you'd drive with this kind of amp, it shouldn't be an issue.

For reference: http://www.diyaudio.com/forums/showthread.php?s=&threadid=125488&highlight=

Even if you don't try the LF, I'd try it without the bypass cap on the C3g. Also, use a film cap on the 46 (10uf is plenty).

Sheldon

 

[Richard]

Hi Sheldon,

Thank you for all the excellent advice and links, it gives me a lot of options to try. Love the winding jig

You might check the potential grid current, though. Do a few calculations of failure scenarios.

Forgive my ignorance but how would I do this?

B+ for the regulator tubes seems high enough to light them, I’ll wire up the power supply with load resistors and see if they’re ok.

The ultrapath cap connection seems to have a strong following. Have you tried the tubecad option which uses both a cathode bypass cap and the ultrapath cap?
http://www.tubecad.com/2008/08/blog0147.htm

Steve,
I’ll post my findings here but I can’t build and tweak as fast as you can, I also had a read of your Direct-Coupled PX25 Amp adventure. Have you considered any similar mods to your 46 amp?

Rich

 

[Sheldon]

Forgive my ignorance but how would I do this?

This (meaning calculating grid current): When I do an amp, I like to imagine the possible failure modes (got in the habit of doing this by reading the Tube Cad site). In your case, you might look at failure of the VR string, or failure of the input tube. In either one, you'd have the full B+ on the grid of the output tube. In this case I'm not sure how much grid current you'd have, or if that 13k supply resistor would limit it enough. The tube specs don't shed much light on how much the first grid will tolerate.

You could do a test without the input tube, and regulators. Use something like a 50k resistor from B+ to grid, and see what you get. Remember, the cathode resistor will limit your total dissipation, because as the current increases, the plate voltage will drop. You actually reach a maximum plate dissipation at around 25mA. Higher than that, and the dissipation actually drops. However, your cathode resistor now takes the heat, so to speak. So make sure it can take it. If the tube basically acts close to a short with that high a resistor in the grid circuit, then you don't have much to worry about.

If you don't want to play with it, you have at least 3 options:

1. don't worry, be happy

2. use a parallel zener string, as mentioned before

3. fuse the B+ at something like 30-35mA.

The ultrapath cap connection seems to have a strong following. Have you tried the tubecad option which uses both a cathode bypass cap and the ultrapath cap?
http://www.tubecad.com/2008/08/blog0147.htm

I haven't tried that, but in some ways it's similar. For instance, you'll note if you look at the entire schematic, that the cathode is tied to ground through the final PS cap. However, this is in series with the cathode to B+ cap, which is different from a regular cathode bypass in subtle but important ways. I found Darius's adaptation of the Loftin White approach the most elegant. He posted on this forum a while back. Seems he was a bit too rigid to enjoy the more freewheeling aspects of life here, but I learned a lot from his postings and his site. You really have to study it a bit. He was right about one thing; Loftin and White's work is very much under appreciated and poorly understood. That's obvious if you look up Loftin White on the net and see that most adaptations miss the whole point. It took me a while to grok, but Loftin and White were very clever fellows. Ask Kenpeter.

As for results, look at my amp and the cap sizes. I use this thing as a headphone amp, it's that quiet. Remember, these guys were designing for a time when capacitance was expensive and so were rectifiers. One example they show even uses half wave rectification.

Sheldon

 

[Richard]

Hi Sheldon,

All you help is invaluable for a beginner like me, so thank you once again. Its always nice to have options, I may take the easy way out and use a fuse in the B+. I’m going to build Steve’s version first to get it going, then tweak it by adding the ultrapass cap. Then the vr tubes and change the choke.

This weekend I managed to start building the power supply on a breadboard. For those following its:
385 - 0 - 385 >5X4G > 8uF (oil) > 12H 70 ohm choke >50uF. Split into:
Right 300 ohm > 50uF
Left 300 ohm > 50uF

The 50uF caps are 2 25uF MKP motor run 450V AC

Cheers,

Rich

 

[Sheldon]

I may take the easy way out and use a fuse in the B+. I’m going to build Steve’s version first to get it going, then tweak it by adding the ultrapass cap. Then the vr tubes and change the choke.

Good plan. Leave yourself plenty of room to work. One of the best things about DIY is the ability to experiment and try for yourself, different variations. I have a habit of making my designs compact, which makes it harder when I get the urge to experiment.

Sheldon

 

[Sheldon]

I should add, that a parallel zener string will be the most sonically transparent of the added safety features. It does nothing until there is a failure in the regulator tube or input tube. Doesn't necessarily mean that you would hear a difference between that and fuses in the B+.

Sheldon