Cgs synth

hi all,

i built a tube vcf this last week using 1j17b. this is heavily based
on ken's tube vca idea. it is a sallen-key using the tube for gain and
vactrols control the cutoff frequency. i still have opamps at the
input and output as in ken's circuit.

a couple of questions and comments: Has anyone noticed the tubes going
into cutoff with larger input signals? I built the amp as ken's
circuit twice with 3 different 1j17b's and got the same result. the
problem was that as the input increased, the voltage at the control
grid decreased until the tube cutoff. it decreased a few volts. i was
looking at average voltage of -18 or -19V at the grid with a 10v p-p
input. mine only accepted very small inputs without clipping most of
the signal. looking at my archived email, it seems one other person
has had this problem.

I ended up getting the output more centered with soft clipping on both
the high and low sides of the amplified signal by making a few changes
(i tried very many resistor combinations plus a few circuit
configurations, and i tested this on two 1j17bs in two separate
builds). note, i set this for a gain of about 3 when a full 15v is at
the screen input. also note: i knew nothing about tubes a week ago,
and now i know only a tiny amount.

-connect the 330k grid resistor to -15v as one option allows.
-set anode (plate) resistor to 27K intead of 8.2K. I did this for some
gain, but it is balanced well with my other settings. i didn't notice
any terrible distortion even with 47K. i never got much gain with the
8.2k resistor. I expect this was mostly due to the tube going into
cutoff.
-set the tube's input resistor to 330K. this was 2.2K. i used a 330K.
with that, the grid current will cause it to clip at the on the
positive side of the input signal which means the negative side of the
output is clipping. the clipping is fairly soft on that side though.
-set screen resistor to 27K. I found this helped even things out also.
It seemed that there was still a tendency to clip at the positive side
of the output before the negative side until i did this.

now, it is like this: the positive side of the input signal, and
positive side of the output signal are soft clipping when the input is
large (+-5V is very large). since its inverting, the output will be
roughly symmetrical. note, there is a fair bit of distortion except
for at the smallest input signals (+-1 or 2V). no matter the input
signal, the tube limits the output to about 15Vp-p i guess.

so, for the filter version. it sounds nice i think. the thing will
resonate fine but requires some input to get it started. I have been
testing and i'm pretty happy with it. I'll be preparing a PCB layout
this week, and i'll be trying with some 1j24b's soon. i'll post some
schematics after that. here is a sample of its sound. it starts off
high resonance, low input. then high input high resonance. then high
input lower resonance.
http://www.homebuilthardware.com/tmp/sk-tube-vcf.mp3

for anyone interested there is a really nice article here:
http://www.freewebs.com/valvewizard1/Common_Gain_Stage.pdf
which describes these two types of tube clipping very well. i highly
recommend reading it!

-Ryan

>Has anyone noticed the tubes going
>into cutoff with larger input signals?

Yes, absolutely. I might have to pull mine out a tweak it a bit based on your observations.

Tony


To: cgs_synth@yahoogroups.com
From: synthlist@...
Date: Mon, 7 Jul 2008 22:02:38 -0500

Subject: [cgs_synth] 1j17b tube vca becomes vcf




















hi all,



i built a tube vcf this last week using 1j17b. this is heavily based

on ken's tube vca idea. it is a sallen-key using the tube for gain and

vactrols control the cutoff frequency. i still have opamps at the

input and output as in ken's circuit.



a couple of questions and comments: Has anyone noticed the tubes going

into cutoff with larger input signals? I built the amp as ken's

circuit twice with 3 different 1j17b's and got the same result. the

problem was that as the input increased, the voltage at the control

grid decreased until the tube cutoff. it decreased a few volts. i was

looking at average voltage of -18 or -19V at the grid with a 10v p-p

input. mine only accepted very small inputs without clipping most of

the signal. looking at my archived email, it seems one other person

has had this problem.



I ended up getting the output more centered with soft clipping on both

the high and low sides of the amplified signal by making a few changes

(i tried very many resistor combinations plus a few circuit

configurations, and i tested this on two 1j17bs in two separate

builds). note, i set this for a gain of about 3 when a full 15v is at

the screen input. also note: i knew nothing about tubes a week ago,

and now i know only a tiny amount.



-connect the 330k grid resistor to -15v as one option allows.

-set anode (plate) resistor to 27K intead of 8.2K. I did this for some

gain, but it is balanced well with my other settings. i didn't notice

any terrible distortion even with 47K. i never got much gain with the

8.2k resistor. I expect this was mostly due to the tube going into

cutoff.

-set the tube's input resistor to 330K. this was 2.2K. i used a 330K.

with that, the grid current will cause it to clip at the on the

positive side of the input signal which means the negative side of the

output is clipping. the clipping is fairly soft on that side though.

-set screen resistor to 27K. I found this helped even things out also.

It seemed that there was still a tendency to clip at the positive side

of the output before the negative side until i did this.



now, it is like this: the positive side of the input signal, and

positive side of the output signal are soft clipping when the input is

large (+-5V is very large). since its inverting, the output will be

roughly symmetrical. note, there is a fair bit of distortion except

for at the smallest input signals (+-1 or 2V). no matter the input

signal, the tube limits the output to about 15Vp-p i guess.



so, for the filter version. it sounds nice i think. the thing will

resonate fine but requires some input to get it started. I have been

testing and i'm pretty happy with it. I'll be preparing a PCB layout

this week, and i'll be trying with some 1j24b's soon. i'll post some

schematics after that. here is a sample of its sound. it starts off

high resonance, low input. then high input high resonance. then high

input lower resonance.

http://www.homebuilthardware.com/tmp/sk-tube-vcf.mp3



for anyone interested there is a really nice article here:

http://www.freewebs.com/valvewizard1/Common_Gain_Stage.pdf

which describes these two types of tube clipping very well. i highly

recommend reading it!



-Ryan






















_________________________________________________________________
Need to know now? Get instant answers with Windows Live Messenger.
http://www.windowslive.com/messenger/connect_your_way.html?ocid=TXT_TAGLM_WL_messenger_072008

[Non-text portions of this message have been removed]

On 7/7/08, ryan williams put forth:

>a couple of questions and comments: Has anyone noticed the tubes going
>into cutoff with larger input signals? I built the amp as ken's
>circuit twice with 3 different 1j17b's and got the same result. the
>problem was that as the input increased, the voltage at the control
>grid decreased until the tube cutoff. it decreased a few volts. i was
>looking at average voltage of -18 or -19V at the grid with a 10v p-p
>input. mine only accepted very small inputs without clipping most of
>the signal. looking at my archived email, it seems one other person
>has had this problem.

As I built a modified version using a different tube, my circuit is
different, but I did not notice that. It seems odd that the signal
on the input grid would effect the voltage on the control grid. Even
though Ken's circuit has a rather large resistor on the output (which
I included in my version), the control grid shouldn't draw much
current, so unless you are getting capacitive coupling, its voltage
should be determined by the op-amp. Also, it seems odd that you are
getting a voltage higher than the rails. I measured around -14.38V
with the gain potted all the way up.

>so, for the filter version. it sounds nice i think. the thing will
>resonate fine but requires some input to get it started. I have been
>testing and i'm pretty happy with it. I'll be preparing a PCB layout
>this week, and i'll be trying with some 1j24b's soon. i'll post some
>schematics after that. here is a sample of its sound. it starts off
>high resonance, low input. then high input high resonance. then high
>input lower resonance.
>http://www.homebuilthardware.com/tmp/sk-tube-vcf.mp3

That sure sounds like a resonant filter. I would love to see the
schematics, or a description of your circuit.

>for anyone interested there is a really nice article here:
>http://www.freewebs.com/valvewizard1/Common_Gain_Stage.pdf
>which describes these two types of tube clipping very well. i highly
>recommend reading it!

Thanks :)

mark,

sorry i can't explain it. i don't have much knowledge of tubes. but i
built it thrice and spent many hours testing and measuring. i
basically used trial an error to choose my resistor values for the
circuit. i have no idea how the voltage at the input grid went down
below the rail. that part is what confuses me, but the behavior i was
seeing was just like what is described in that article i posted. the
voltage no longer drops so low with the resistors i changed.

i have it on a soldered breadboard currently. hopefully next week i
can do schematic capture and pcb layout. i'll post schematics within a
few weeks. for now here is a description:

taking ken's vca circuit. i insert a sallen-key filter node like this:
http://mypeoplepc.com/members/scottnoanh/sitebuildercontent/sitebuilderfiles/scotts_mutant_filter.pdf
but without the feedback path created. i keep the opamp to buffer in
between the RC and the tube input. that opamp might be removed but it
was easier to just leave it. then the tube amp follows set for gain of
3. following the tube amp is the opamp buffer again. connecting from
its output is a pot used to divide the output before inserting back
into the feedback capacitor of the sk-filter node. the 1Meg resistors
in the posted schematic are really vactrols, vtlc2/3. so it is not a
fully tube filter but a tube is at least used in the most important
place. maybe an all tube version could be made. it might be better to
make one using only tubes in the feedback path? another option i tried
which is actually used in that sound example was to put the pot
divider before the tube amp instead of after it.

i removed the amplitude control. that could be used for resonance but
i found a bit more even overdrive clipping when fixing it to +15v with
a 27k resistor. the output is limited this way to something like +-6
or 7V but with the fixed gain of 3 in the tube amp, it is easy to
overdrive the filter.

the vactrol current source was copied from scott stites mutant vactrol
filter (which was came from rene schmitz 4096 vco). it has exponential
response, but vactrols don't react linearly so cv is not precise.
http://mypeoplepc.com/members/scottnoanh/birthofasynth/id17.html

i put the schematic up on my web page. this is for reference to anyone
who wants to see how it works. i have tested this.

http://www.homebuilthardware.com/index.php/projects/sk-tubevcf/

-ryan
-