New Schematic - PP UL 6V6-types, CCS-fed

[sorenj07]

Here it is. I'll call it the "6V35" because it uses Dynaco Z565 output transformers. The actual output tube shall go unmentioned Normally I'd just plow right ahead but this is only the second time I've ever implemented constant current sources, and the first time I've tried to design one into a new circuit.

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

The CCS in question is a 10M45S. I have four of the TO-220 package types. I guesstimated about 65 ohms to set each tube to 37.5mA using the graph at the bottom of this page:

http://www.tubecad.com/2006/11/blog0087.htm

(I'm also considering using a 50-ohm pot on top of a 47 ohm resistor.)

Will the output stage work as described? I'd heard that it's good to bypass CCS' with capacitors, so I have done that. I also guessed 100K for the grid resistor, is that OK?

I think I'm going to need at least +-21V to drive these to full output, working from the curves here: (I just picked the 300V/37.5mA point)

http://www.polstra.com/projects/spice-examples/loadlines/

11.25W is pretty conservative for 14W total plate+screen dissipation, so this amp ought to last a while. And I suppose I could always dial up the current.

As you can see, I have not decided on a driver stage. I was planning on rectifying and maybe doubling the 6.3VAC heaters to feed a CCS of some sort, but don't really have any ideas so far. Does anyone have any suggestions? I have a couple 12AT7's and a few 12AV7's that might work..

 

[Shoog]

If you don't bypass the CCS then you get no signal.
When I use CCS in the tails I usually take the node between the two bypass caps and reference it to ground via a 1meg resistor (which keeps them biased). The effective value halves but it forces the valves into differential mode and helps to clean up the sound.

The IXY chips are very good in this capacity (I am using therm as well), but you can expect quite a bit of variation between units - in the order of 1-3mA, so that trimmer might be a good idea.

Shoog

 

[sorenj07]

If you don't bypass the CCS then you get no signal.

I'm a bit confused... Serpentine (Tom) over at Geek Zone had this to say about bypassing the CCS' with caps:

...don't those 220uF caps across the CCS at the power tube cathode completely disable the AC-wise CCS function? IOW, what is the point of those CCS' except to set idle DC? And what do you think will happen when leaving class A operation?

(http://geek.scorpiorising.ca/GeeK_ZonE/index.php?topic=4152.msg46581#msg46581 )

Something doesn't quite add up.. This is the first time I'd feed an output tube with a CCS so I don't know the answer!

 

[Zen Mod]

CCS here is as very long tail in LTP

if you bypass it for AC ( with cap ) , you loose LTP tail , and that's then plain symmetrical amplifier , without inherent balancing feedback , where CCS is used just for biasing

 

[Shoog]

In order for that to work you need one CCS in a common tail (which is not what you have). If you try for one CCs in a common tail you will lose DC balancing which may or may not be a problem for you. My suggestion creates the LTP and the DC balance.

Shoog

 

[Zen Mod]

ooops - I didn't clicked on attached pic ( so I saw just a thumbnail) , so I didn't realized that there are two CCS-es .

Shoog is right ....... off course ;

 

[sorenj07]

My suggestion creates the LTP and the DC balance.

That would be ideal, but I'm having problems visualizing what you mean. You said:

When I use CCS in the tails I usually take the node between the two bypass caps and reference it to ground via a 1meg resistor (which keeps them biased). The effective value halves but it forces the valves into differential mode and helps to clean up the sound.

I thought that by "tails" you meant the LTP, and not the power tubes? But then you talk about bypass caps, which are under the power tubes. I'm confused though, wouldn't the "node" between the two caps just be ground?

 

[Shoog]

At the moment each bypass cap goes from the cathode to ground. What I am suggesting is taking the ground connections and connecting them together at a new "node". This new node still needs a trickle path to earth to keep them biased (ie the cathodes negative), inserting a 1meg cap from the new node to ground will do this nicely. Now the signal through the tubes can only travel through the low impedence path across the caps rather than the high impedence path to ground. This makes it functionally a LTP.

Hope that helps.

Shoog

 

[Zen Mod]

nice tip, Shoog .

 

[Shoog]

Hope you spotted the diliberate mistake - I mean't 1meg resistor not cap.

Its such a simple mod to try that you could rig up a bypass switch to short out the 1meg resistor and do a dynamic comparison.

Shoog

 

[sorenj07]

So let me get this straight: Since the sum total current of the two output tubes should stay constant (class A), the cathode bypass caps don't need to be connected to ground except by a 1M resistor, as long as they're connected to each other? It's beginning to make sense.

Also, is there any problem with using another 10M45S with a 300-330 ohm current set resistor to feed 9-10mA into a LTP of 12AV7's? I'm opting for simplicity here.

 

[Shoog]

It should work fine in the frontend, but you will probably need to generate a small negative rail. If you have a spare heater winding this would do.
However in this application I have successfully used LM317's and other basic voltage regulators. Because the impedance of the bypass is proportionally so low the relatively poor frequency response is no really to much of an issue on the outputs (people claim they are more important on the LTP. My only point is, you may want to save the IXY chips for an application where their voltage rating will be more useful.

i just noticed - how were you setting the bias on the LTP in the original schematic.

Shoog

 

[sorenj07]

Layout Pictures!

Almost all set on the layout. There will be a stiff bar of aluminum clamping the 10M45S' up to the chassis, insulated with mica of course

The one thing that's missing is the wood for the sides. I'm about to cut, sand and varnish that now.

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

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

That 5V4GA sure looks a little shrimpy doesn't it? I have several 5Y3GT's that have a much bigger plate structure. I hope it doesn't explode while I try and grab a decent 5AR4 from somewhere to plunk in.

 

[Shoog]

I would consider moving the IXy's to the side of the case and running flying leads to the tubes. My experience is that the area between the valves can get hot. Remember to use use gate stoppers on the IXY chips as they are high gain mosfets and tend to oscillate.

Looks good.

Shoog

 

[sorenj07]

I would consider moving the IXy's to the side of the case and running flying leads to the tubes.

Thanks! Yep, I have some 1K carbon comp resistors exactly for that purpose. Two questions: How much power are these suckers dissipating if I set them for around 40mA each (via 75 ohms to ground)? Bias for the tubes is right around 20V, if that helps. And should I worry about potential oscillation/feedback if I do run long leads? Would a ferrite bead help?

 

[Shoog]

They will generate one watt of heat each(20x0.040=0.8W). Just so long as the gate is stopped with the 1K resistor, then they should stay stable regardless of lead length. Watch out for hum pickup.

My experience, and Tubelabs, is that they are quite variable. Build them and test them using a 30-40V power supply with a 375R resistor in series. Check the drop across the resistor and calculate the current (voltage/375=current). Design them such that you can adjust the current set resistor by adding or taking away an ohm or two resistance. Aim for matching for a push pull pair of about 1mA.

Shoog

 

[athos56]

I like the way thats looking. Two questions, who made those transformers? And what did the final schematic end up looking like?

 

[sorenj07]

They will generate one watt of heat each(20x0.040=0.8W). Just so long as the gate is stopped with the 1K resistor, then they should stay stable regardless of lead length. Watch out for hum pickup.

OK, thanks. Sounds like I'll want to chassis-mount them to make super sure they never die.

Build them and test them using a 30-40V power supply with a 375R resistor in series. Check the drop across the resistor and calculate the current (voltage/375=current). Design them such that you can adjust the current set resistor by adding or taking away an ohm or two resistance. Aim for matching for a push pull pair of about 1mA.

Yeah, I've got a 68 ohm resistor followed by a 20 ohm, 20-turn pot. I ought to be able to match these suckers really close. No 10 ohm resistor but I'll just look for 2.652V across them for 39mA, 2.72V for 40mA, etc.

I like the way thats looking. Two questions, who made those transformers? And what did the final schematic end up looking like?

Thanks! Just wait till she's done. I'm going to spray the perf metal black to match the transformers. That's an Allied 6K7VG for power, a Hammond 193J for choke input, and two original Dynaco Z565's on the output. I took apart the Hammond and Allied transformers and used a razor/sandpaper to strip off all the varnish, and resprayed 2 coats of matte finish metal enamel. An hour or so in the toaster oven at 150 degrees and they were ready to be bolted back together.

I need to redraw and scan the final schematic, that'll be up soon.

 

[athos56]

Thanks! Just wait till she's done. I'm going to spray the perf metal black to match the transformers. That's an Allied 6K7VG for power, a Hammond 193J for choke input, and two original Dynaco Z565's on the output. I took apart the Hammond and Allied transformers and used a razor/sandpaper to strip off all the varnish, and resprayed 2 coats of matte finish metal enamel. An hour or so in the toaster oven at 150 degrees and they were ready to be bolted back together.

I need to redraw and scan the final schematic, that'll be up soon.

Ahh, that's it, a few coats of matte enamel makes anything look better

 

[sorenj07]

Here we go at last. Shots of the wiring as soon as I get bored of listening to music.

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

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

Slight hum in the left channel but you have to have your head right in front of the speaker to it.. Might have to do with issues outlined below.

I didn't have enough bypass capacitors so I used two 150uF capacitors in the left channel, and two 220uF caps in the right channel. These two caps go right to the 10M45 pins, meet at a floating point, and get connected to ground from there.

The amp has a weird phasey noise that starts up when I flip it on, swings from channel to channel, and quickly disappears. It squeals when I turn it off. Oh well, maybe the 5V4 rectifier doesn't like all its "new-school" solid-state neighbors