20V valve amp - help please?

[markeebee]

I'm putting together a valve amp that will run from a 19V laptop power supply.  Front end is a standard valvecaster with ECC81 / 12AT7, "power" amp is an ECC86 / 6GM8 with both triodes strapped together.  Like this:



I'm just getting a bit of buzz and hum at the mo, but that can be sorted later.  Maybe.  If I'm way off course, maybe you could give me a nudge.

My immediate problem is the heater behaviour.  The heaters are are in series, and I've put R8 (it's actually 68R) in series with the heaters of the ECC81 to try to get the string to current share.  Is that right?

When I power up, the ECC81 glows like a mutha for about a second and then settles down a little.  I think this might kill it pretty quickly, is there anything I can do to limit this?

Thanks in advance for help and suggestions.

[markeebee]

Forgot to say.....

Resistor R8 gets REALLY hot.  And after a while there's that smell of hotness that usually means something's wrong.

[markeebee]

Also forgot to say.....

I stole the ECC86 idea from here:

http://philsvalveradiosite.co.uk/philsvalveradiosite/ecc86lowvoltageoutputstage.jpg

[CynicalMan]

The resistor is right. The ideal value for it around 84 ohms, but 68 should work. It's dissipating over 2W, so it should be a 5W resistor, and it will heat up.

My guess is that the problem with the ECC86 is that tube heater resistances are lower when cold. If the 12AT7s' heaters are starting off with a low resistance while the ECC86's resistance is higher, then the ECC86 will get more current and voltage until the 12AT7s get hot.

I'd probably just make a 6.3V heater supply with a LM7805. Just stick two silicon diodes from the IC negative supply to ground. No smelly power resistors needed.

[markeebee]

Thanks Alex.  Yeah, I was thinking about using a 7806 but I thought forcing the heaters to share would keep the heat and component count down.   Turns out the resistor is getting much hotter than a regulator probably will, and at least I could heatsink the regulator.

I think I'll go with your suggestion, and then post up all the other problems I'm gonna have 

[gmoon]

70 ohms is right on for getting to 330mA.

You could also try something like 210 ohms in parallel with each of the filament halves of the 12at7...

(edit) Ops, that's not right--that's figuring with both filament sides pulling 330mA, which isn't so...

More like 38 ohms.

[CynicalMan]

You'll probably want to put the 7805/6 in place of the resistor you have now, with the output connected to the junction of the heaters. That would give it a heat dissipation of around 2W. As long as the heater problems aren't huge, that will fix them, but you should measure the cold resistances of your heaters to make sure that they won't pull up the regulator's output past 6.3V. If you run the tubes directly off the regulator, the dissipation will be around 8W.

[davent]

What about leaving the tubes  in series and using a LM317 to regulate the current rather then regulating the voltage? Still need resistors...

[gritz]

I think that the problem here is that the ecc86 filaments want 330mA (according to the datasheet) and drop 6V3, but the ecc81's filaments are wired in series and want 150mA. Perhaps the simplest hack would be to wire the ecc81's filaments in parallel for 300mA operation at 6V3 and connect the two valves in series. Then you would need to "lose" 20-(6.3+6.3) = approx 7V4 at 300mA. So that's about 25ohms in series, dissipating about 2.25 watts. Remember that a linear voltage regulator would have to dissipate the same power. A resistor would also provide a bit of protection against that "cold current surge" - small signal valves generally tend to break at power-up, after all. A constant current source for the filaments would be elegant, but it would still get warm.

The maths above is approximate but a 27R 5watt resistor might be worth a go.

Regarding the noise thing: laptop switch mode power supplies can be pretty grisly so it may benefit from splicing in a choke (or dropper resistor) into the +ve line after the filaments but before the amp proper and following it with a reservoir capacitor to smooth things out. A low esr high temperature electrolytic designed for switch mode power supplies would be good here. Similarly you might need to decouple the supply to the preamp valve's anodes to further reduce noise and stop any instability that may come from the demands that the power amp section places on the power rail. This schematic is for a high voltage single-ended amp so you'll have to tweak the values a bit, but it shows the basic idea:



Hope this helps.

[PRR]

"Flash" at start-up is normal for some tubes.

Your series plus resistor is approximately optimum. Wiring for 6V and dropping 19V to 6V is MUCH more heat for no benefit.

Power output is a whopping 0.01 Watts.

Hum and hiss happen if the amp is not well boxed/shielded. Also you have WAY more gain than a guitar amp needs, and no gain control, which means hum/hiss is MAX volume all the time.

C1 could be 47 nFd, not uFd.

[markeebee]

Thanks gents.

Just to see if the cicuit worked, i used another wall wart to get 6v for the heaters. It works, but is very noisy. Maybe heterodyning because of the two supplies?

This evening I'll reinstate the PRR-approved series filament resistor and ditch the wallwart and try out gritz suggestions to reduce noise.  Then add gain, vol and tone.

I've got a really cool old fan heater enclosure that i want to renovate and build this into.  Here comes spring, there goes my summer.  I'm not a quick worker.

[markeebee]

C1 could be 47 nFd

Ah yes, it is.

Also R5 is 100k.

[amptramp]

The V1/2 stage has no cathode resistor and may be a source of distortion.  There will be a small amount of contact potential due to the grid resistor, but it is about a twentieth of the value that would give consistent biasing.  When the signal to the second grid goes positive, it will pull grid current and when it goes negative, it will cut off to the extent that the R4/C2 time constant of 7.2 Hz will set the amount of time the signal vanishes.  A cathode resistor will solve two problems: excessive gain and "gulp distortion" where the signal shuts off for a noticeable time.

[markeebee]

I put it all back together as in my diagram and, blimey, it works pretty well. 

Still quite a bit of hum, but I can work on that.  Still that disconcerting flash in the preamp tube on startup, but if Paul's happy with that then so am I.

Very old 6" speaker from a radio, 70V line transformer for the OT.  And do you know what?....*smug face*......it sounds pretty ok.  The volume is just about loud enough to sound like a proper amp, just about loud enough to drown out the clack of the pick on the strings. 

Despite the theory, there doesn't seem to be too much gain.  I was interested in hearing 'gulp distortion' - not really sure what I should be listening out for but there was nothing apparent. 

Overall, I'm quite chuffed that something so simple turned out so nicely.  Now I'm gonna pish it up by building it into an enclosure.

Quick and dirty cellphone vid, look out for the volume jump about 30 seconds in:

[PRR]

> The V1/2 stage has no cathode resistor and may be a source of distortion

That's good, no?

(And it is adopted from a distortion pedal.)

Anyway: V3 only needs about a Volt. V2 has gain of 30+. So V2 grid signal should be 30mV. The transition from 100+Megs to 1K grid impedance is typically 100mV wide. So a 30mV signal is "not much distorted". I think.

I do wonder if a V2 cathode resistor would change anything. But it isn't his humm, it sure isn't his heater flash, and he seems happy with the small sound (it sure scares the crap out of the cellphone mike!). "Proper Biasing" can wait for a post-build rainy-day.

[Perrow]

I just have to subscribe to this thread, looks really interesting.

[PRR]

> just have to subscribe to this thread

There's a button below the thread which will do that without posting a Reply.