12SN7 MHz Oscillation

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Jazzy_Pidjay

Well-known member
Joined
Jun 7, 2004
Messages
370
Location
Paris (FRANCE)
Hi,

After searching for days and days about Neutralization, Grid Stopper
and Oscillation all over this and other places, i'm still strugglin'
So i decided to ask the question, i hope it's not already debatted here, in this case excuse me...

I'm about the end of the design of a mic preamp, a variation of the Asymptot/SlowBlow.

The first stage which is a 12SN7 give me some oscillations at about 2MHz
I tried differents gridstopper from 1k up to 39k right at the pin, and yes with 39k
which give a corner freq, with miller effect, etc.. of 46kHz...
i can still see some oscillation :?. The oscillation is pretty low, the smaller scale on my portable scope (only one i got) i believe it's less than 1mV.

I really ain't know any other tricks, and even if the oscillation is very low, i would like to kill it, both for the preamp and for my knowledge/understanding of that.

Any help will be very appreciated,

thanks :thumb:
 
You got a 12sn7 to oscillate?
Congrats!

Usally the wide spacing on the octal sockets provides a better layout since you can route the wires easier,

we need not a schemo, but a jpg of the layout.

the freq of osc will change all over the place.
even with temperture.

dont feel bad, this is how the therimin was discovered.

start by moving grid wires while it is oscilatring.
use an insulated stick.
i use viagra, being careful around the pwr supply.
but then, i'm just a bear with furniture.
:razz:


what kind of iron?
what res freq?
loads the sec with 610 ohms?
 
thanks for all the answers

i draw a rough schematic and i come back with it soon.

i actually don't know if its oscillation or noise picked up by the grid.
 
Ok here is a 'schematic'

12SN7.jpg


I measure with the scope between A and ground,
it's where i can observe an oscillation at about 110kHz going up and down from
0V to 1mVpp at maybe 1Hz.

On this preamp, there's only one half of the tube used at a time.

There's a relay on the primary of the transformer and one at the point A
they are used simultaneously to switch between one half (one tone) and the other.

B+ = 275v

Here is the first half :

Cathode Voltage = 4,6v
Current = 11mA

Rp = 8k2
Rk = 470R
Rg = 39k
Trafo = LL1538 (1:5)
Grid stopper from 1k to 39k

Here is the second half :

Cathode Voltage = 2,8v
Current = 10mA

Rp = 15k
Rk = 330R
Rg = 82k
Trafo = Sowter 6469f (1:7)
Rk bypassed = 100µF


I know the first one, is a bit 'unusual' at least the Rp is not two times or more the Plate resistance of the tube,
but it almost impossible to overload it in the vocal use it is design for, and sounds damn good without needing a pad.

But the first one is oscillating a bit more too... :sad:

There's no 'wires'

The Gridstopper is right a the pin, and the LL1538 is right at the gridstopper resistor leg,
not even 8mm of leg. the same for Rk and Rp,
all the resistors are right at the pin, and there's not even ONE
added piece of wire, on both

The Sowter 6469f is connected directly to the gridstopper too,
with maybe about 2,5cm of wire for the secondary.

I don't know what to try, i haven't understand plate stopper,
and if it's right for this problem, and it seems like the gridstopper just
lowpass filtered the oscillation, but the oscillation seems to still birth and exist,
with or without in the limit of an audio bandwith.

The ground/0V is the most quiet i had done in my life, star grounded with attention, and many tests.
So i don't think there's a ground issue, maybe i'm wrong

Thanks guys anyway for your help, i hope i can find a solution.
 
so it´s two different stages in parallel, both fed from the same B+ via Rp´s with no further decoupling?
i think a full schematic would clear up a lot of things.
and a picture from the used pcb as well.

-max
 
Not exactly,

Theres no PCB at all, just resistors and tube socket.

The two tubes are not parralleled, There's on half on circuit OR the other
by the use of one relay connecting the input XLR to ONE of the transformer primary
and one relay connecting the 100k pot to ONE of the 1µF cap.

So at any time, the two half are drawing current, but only on is on the signal path, the other is just disconnect completely.

But even, yes, each half is decoupled, each got a dedicated 330ohms/47uF RC cell on the B+

I can take a picture of the inside, i don't think it will help, it just a point to point soldering, without wires, just components one after the others.
 
Sometimes a little wire can add some needed inductance, plate wires on some of the fenders are a mile long.

I woyuld get rid of the fancy relay switching scheme and try to get both sides working seperately, then you know you have a problem with this unorthodox, but cool, relay idea.

Them, when you are satified that you have an open ground, or a sheared grounfd that should be switched out, you can re construct a new sw scheme.
Like a DP/DT instead of just sw-ing on lead.
 
Yep, the oscillation will still be there,

this idea of relay, (in first place because i hate an unused half :) )
come only after my first prototype.

And during my first prototype with only one half, the oscillation was there.

Maybe adding an Inductor between the Plate resistor and the Plate ??

I look for methods of killing oscillations, but i can't find so much about it on this forum, and i had search a lot.... :sad:
 
ok, hold on, i gots a bundle on that.
sounds like two problems, motorboating and ultra sonic.

all osc problems are caused by positive feedback. period.
so thats what we are looking for , a packet of phase shited electrons is going to where it should be forbiddin.

grid stoppers, small caps, inductors, this is a band aid approach to a bad design.

if its a good layout, the thing should not oscilate.

audio is easy, rf is a bitch, so this is a real piece o cake, but it tales patience.

reverse phase on the tramsfomer.

rememvber the grid stoppers added to the la2 with the sowter iron?

sometimes the outside winding should be inside, etc.
 
OK :grin:

Actually there's an 4k7 on the grid of the 12SN7 of the Sowter Side
but that just the last one, left in.

My small scope Velleman PPS10 give me a different looking oscillation
on each half. but it is still hard to read, whatever the time resolution of the scope, the oscillation just move, and look like random in some resolution

The 'average looking' seems a bit better on the Sowter side, but not satisfactory too....

In fact maybe 1mVpp of oscillation is 'normal' i really don't know, but it doesn't seems like a good health.

I try swapping the phase.

PS : Plate Resistors are big green wirewound 10Watts
maybe it's important ? cathode resistors are 3Watts Metal film
 
hmmm, maybe the inductance from the wirewound, easy to check, just swap in a 12 k 2 watt if you have it, or make one from a bunch of resistors in series or parallel

hey, what ver of php is this?

I need to enable a file on this new platform so my pics link.

sorry for the ot.

1 millivolt, you say?
maybe we aren chasing nothing.

how does it sound?

if it sounds great, there is probably nothing wrong.
 
The wirewound is not guilty, a 15k metal film 3watts doesnt change anything.

After the 20dB 5687 output stage, i can see a 13mVpp oscillation, still 'random' on my display, and 2.5mVrms (from the scope)

The preamp sound amazing, i'll share it. both half sound very different, but headroom is quite different too.

I was not sure of what can do an oscillation even at small amplitude,
maybe it was a sign of bad health, and sound can be better, or maybe it was just a growing oscillation at a little stade with no consequence, anyway if it's not a problem to keep a stage oscillating, i can live with, as the sound is very good.

you talk about PHP of my website ? probably the 5, is use EasyPHP but don't know the version on the server, if i can help ask me what you need?
 
Do you live in an area with high rf?
Or you have never seen this before, while living at the same place?

Take a 500 pf cap , a pair of jumpers, ground one end, go around from front to back, shorting out the rf, hopefully, with the cap.

see if that turns up any new news.

yeah, the wirewound is a res circuit with the .001 cap, way out of rf osc range, but we would have to bput that resistor on an inductance bridge and do a root ccalc to find f-res.
 
No, i don't leave in an heavy RF area,
it's my first DIY project building with the 'help'
of a scope, so maybe my others projects was worst

Take a 500 pf cap , a pair of jumpers, ground one end, go around from front to back, shorting out the rf, hopefully, with the cap.

I don't understand what do you talk about ? :roll:
 
ok, make it a 100 pico farad cap, a small ceramic will look as a dead short to rf, but not your audio. so you are being selective as to which frequencies you want to short to ground.

obviously, you do not want a cap that will short out your signal And the rf, as you will not beable to play the preamp with no signal.

so the cap shorts out the rf only.
where is the rf coming from?
at those freqs, it tends to radiate throut the entire chassis, so it is hard to track down the source.

but the cap can help localize the problem.
place it on the grid first, then the plate, then the input trans pri, then the sec...
 
Ok i got it, my understanding of english is not perfect

i test that tomorrow, need to sleep now

Thanks a lot for the help, it's appreciated !
 
I may be reading and thinking about this incorrectly.

If one half is unused, but still powered, is it possible it could be oscillating and bleeding within the bottle to the other half? Is switching a 100K dummy load resistor (to ground) onto point A of the unused half useful in this regard, or is the B+ line and decoupling sufficient to be considered a non-oscillation prone output load?

1 mfd is an awfully large coupling cap for a 100K pot. I've seen change from 0.1 to 0.22 cause motorboating in some preamps.
 

Latest posts

Back
Top