Even More Stupid Questions about MOSFET Sidechain Driver

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Swedish Chef

Well-known member
Joined
Jun 7, 2004
Messages
351
Location
London
Ok after the idiotic hiccup earlier on, and consultation with a couple of producers and engineers I have decided that it's be useful to have a grunty amp to drive the sidechain on my Vari Mu comp to get the min attack time down towards 660 ballpark. I am running a 175/6 type of topology with a 6sn7gt as cathode follower on the o/p, and a 6bc8 as the Vari element.
Paul PRR I think you spanked me in the face with a MOSFET amp idea last time I asked about attack times (ignorance is painful... :grin: ) so I have a head guard on, but does anybody have any pearls of wisedom about where to start? :thumb:

The door is on the latch...

chef
 
> I am running a 175/6 type of topology with a 6sn7gt as cathode follower on the o/p,

175 does not have a cathode follower output. What are you really doing?
 
Ok 176. Sorry I thought they were similar but have been using the 176 schem for a basis. I have a 6bc8 interstage tx, 12ax7 and 6sn7 as I said. I also have the 6AL5 but and as I have the threshold controlled with a DC ofset on said valve I have not changed that to sandstate diodes yet!

I have no difinitive schem yet i'm afraid... :?

chef
 
> I thought they were similar

175 176 177 are all similar. None of them have cathode follower outputs. So I don't know what to suggest when I don't have a clue what you are really doing.
 
STEVE ...
whake up boy'O ... what are you doing ??
:green:
You haven't been putting your tongue on the HT again ... have you ??
 
ok I think this is a terminology issue. I am a dumbass, I am desperately trying to absorb as much info as possible from as many places as possible and I'm afraid I may call a certain layout something in my head which may not be quite correct, or worse totally wrong!!!!!, but it suffices for my internal conversations...I know what I'm talking about. Then when I come to trying to explain what I'm doing and all of the stupid mistakes I have made in my head come out, I have to get everything straight in a public manner, feel like a fool, look like one, generally am one, but I'm getting over my embarassment pretty swiftly as I want to learn.
There's that famous proverb about the man whom everyone thinks wise until he opens his mouth, well I've resolved that I'd rather have everyone think I'm an idiot but increase my knowledge, as opposed to having people think i know more than I do and remain ignorant.

Well now we all know how little I understand... :green:

I am running the 6sn7 in the same way as the 12bh7 in the 176, with almost the same res values in fact. Looking at it again I suppose that this isn't a cathode follower. What is it called then? :? Looking at it now, I can see it is not. Ok seeing as I am a chancer I am going to have a stab in the direction of a White cathode follower as the cathode of the 12ax7 connects to the Anode of the 6sn7/ 12bh7 it looks, at least to me, a bit similar :sad: . Is that right or am I totally missing the point?

I'll try to get a schem drawn up today. :grin:

yours penetently

chaff
 
> this isn't a cathode follower. What is it called then?

The output stage of the 175/176/177 is a plain old tube stage. Technically a common-cathode stage, plate-loaded, transformer coupled, yada yada....

The reason I asked is: you were asking for a faster rectifier. The 175+ design gets rectifier power from the output stage. If you use something other than the plain old 175+ design, then that makes a difference.

I have seen a private sketch. Right off I don't see a Release resistor. As-drawn, the first time it goes into GR, it will stay there all day (well, for a minute or so). Just to get it working, put 1Meg from where the time-constant caps connect to the input grids. That will be a usable value, exact value to be determined after much musical abuse.

Ah.... you have deviated from the UA time-constant network. Looks a lot like a Fairchild. OK. The Fairchild costs more ($5 of switched caps instead of $1 pots) but I suppose that's unimportant in a box that will be worth $500-$20,000.

You can't just use the Fairchild time-network because the Fairchild's rectifier really needs to be driven from 600Ω, and the 175+'s 12SN7 output is more like 6K per side. Taking from the 600Ω secondary won't give you the voltage. The Fairchild time-net really needs a 10 Watt rectifier driver.

SO: use the Fairchild network BUT raise all the impedances 10 times higher. Bigger resistors, smaller caps. A 2Meg resistor becomes a 22Meg resistor; a 1uFd cap becomes a 0.1uFd cap. Now the output stage and rectifier and time-constant network are all working at good impedance.

Problem now is that the output of the time-constant is a very high impedance, higher than the rated grid resistance of most tubes. It may be OK with selected 6BC8, if you burn-in for a week to getter the gas and keep them hot. But you also have to select balanced 6BC8 to avoid thump (the balance pots won't fix a badly balanced tube). SO you might end up with an awful lot of 6BC8 that can't be used.

FIX: put a unity-gain DC buffer between the time-constant output and the input grids. I think the lame old TL071 is perfect: very high input impedance and plenty fast for sub-milliSecond attack times. However you will need a -30VDC supply for it. (Tip: a pair of 15V 1W Zeners in the center-tap leg of the power transformer.)

Also in private viewing:

You raised the 12AX7 grid resistors to 25 times higher than UA's values. Why? UA used low-value to swamp unbalance in their not-centertapped interstage transformer. Even with a better tranny, I think you need to swamp too. I can't read the tranny ratio but remember the primary sees 10K, so the secondary can probably be loaded with 2*20K (what UA did).

You have two balance pots in the 12AX7 cathode, they both do the same thing: screw-up the gain-balance! Maybe there is a "sound" reason you deviated so far from UA's feedback network, but I just see setup difficulty. You also seem to be set up for lower output stage gain than UA had: maybe this is a change in interstage tranny ratio, maybe you want lower noise at the cost of higher 6BC8 distortion.
 
[quote author="PRR"]> Lots of stufff[/quote]

:shock: :grin:
Blimey!
that is fantastic. sorry there is a 2M res after the cap network. that was a careless scribo/ typo. thanks for pointing it out though! :thumb:
I will spend all w/e going over that stuff and will have to get back to you next week with some half intelligent responses.

I will just say that the interstage ratio is 2.25+2.25: 2+2 and the reason that those grid values are so high is because I didn't have the inter tx in there from the begining and had based the grid resistor on a previous schem with a different Vari Mu element. this thing has been rolling for ages; build, dismantle, re-build etc. etc.

Digesting and breadboarding for me then!!!! :thumb:

Is it legal to send alcohol through the mail???? :sam: :guinness: :wink:

chuffed
 
[quote author="Swedish Chef"]Blimey! ... that is fantastic. [/quote]
yes indeed
excellent as always
:thumb:

... sorry about the back-hander Steve but I just couldn't resist. :green:
 
Kev I will never talk to you again :evil:


Oh ok then... :green:

We both know why you did it :grin:

Lovely DIY :thumb:

Chorf
 
> I may call a certain layout something in my head which may not be quite correct

> I suppose that this isn't a cathode follower. What is it called then?

Input and output are 2-terminal concepts. Tubes have three working terminals. Therefore one tube pin has to be common to input and output. From that you can define these six possible connections:

in out common - Use
G__P___K ------ Plate loaded, normal
G__K___P ------ Cathode follower, unity voltage gain
K__P___G ------ Grounded Grid, unity current gain
K__G___P ------ abnormal, no known use
P__G___K ------ Inverted, high-voltage sensing
P__K___G ------ abnormal, but used in GE limiter

> have a stab in the direction of a White cathode follower as the cathode of the 12ax7 connects to the Anode of the 6sn7/ 12bh7

Well, no. Take the output-to-input feedback off before you study the basic stage. Plate-loaded 12AX7 is resistance-coupled to 6SN7 grid, 6SN7 is plate-loaded with a transformer.

White Cathode Follower: 2-tube, cathode follower plus plate follower all in series across the power supply. Both tubes are similar (nominally equal), not like an AX7 and SN7. The WCF is a semi-push-pull affair. It is wasteful of supply voltage but gives the best ratio of damping to supply current of any transformerless tube stage. A WCF typically will not give the large voltage swing you need for a vari-Mu limiter's rectifier. The 175's synergy comes with the transformer: you have 10K push-pull to whack voltage into the rectifier, and also 600Ω floating to drive a low-Z line.
 
:shock:
Oh my.
the more I learn...the less I know... :?

I'll keep on plodding though. Thanks for your patience and idiot level explanations :grin:

Chepkf
 
right after having been stuck under our irritating Vx for most of the last 2 days, I have at last had a chance to get to the mods.
All good.
I have done it in stages and made sure it was still kind of working in between stages. Nothing really changed until I revamped the f/b stuff around the 6sn7. I put in too small caps to begin with and ended up with a square wave osc for some reason. Anyway the largest ones I have at high enough V rating are 0.33uF and they seemed to get things back to normal again. Apart from the fact that the 6bc8 started to get slammed with about -20V and doing that thumpy very LF osc thing.
I have changed the grid res on the 12ax7 to the 20k you suggested, took the trimmers out and reverted to pretty much the same as the 175. Same with the cathode bias res on the 6sn7. Raised all of the res after the ac-coupling caps from the sn7 by a factor of 10, and reduced the caps in the time net by a factor of 10. I didn't change the res values in the little circuit which creates the DC bias for the 12al5, is that ok, or do I need to rehash that too? :?
LF442 is rockin' off a bench supply and I am getting compression but I have noticed the release is very slow, even though I have changed the release res back to 2M.
I'll have another stab after some food and see if i can get some sort of idea for the attack time :thumb:

Loosey goosey baby :grin:

Chez moi
 
:? :shock:
OK scratch that.
She has woken up a new woman. A feisty one. I'm now getting high frequency junk when I connect the bench supply ground to the star gnd point, but with out it I have humm... :?
Also the release is now far too short. Back to the 20M.
:shock:
Here we go again... :green:
 
Confuse-a-cat?
This could confuse a whole cattery...! :grin:
Right.
When the 12al5 is biased much lower than 65-70V (minimum comp) it all turns into a bumble bee with a frequency of about 1.5KHz. At this point (biased at 72V), with Sine input of about 1V (measured with s/c disconnected) between each grid of the 6bc8 and ground , the s/c is still whacking out around -10V but keeps producing a positive transient of around +10V every second or so.
When running a drum track through it, on the transient of each hit the CV blasts up to like +15V momentarily before going negative and reducing the gain.
Do I need to bias the diode even more +ve.
My back is hurting and my head is hurting :sad:

chump

EDIT: The +ve transient thing was because i didn't have the bench supply properly grounded to the chassis of the compressor...
 
right here's a teaser! :sad:
The LF442 is happily running on just about -30V on the -V terminal and ground on the +V terminal.
The problem is that there's an oscillation at about 1.5KHz. It goes away when I disconnect the o/p of the s/c from the i/p tx, but then there are no suprises there are there? :grin: It DOESN'T go away though, when I stop powering the opamp...? :shock: If I knew more I may not find this so strange I suppose...
It is doing this with no audio passing through the unit, and I can improve it by sticking a 0.1uF cap between the point where the CV goes into the distribution resistors and ground, but there is still what looks like 50Hz junk on there... :mad:
Oh, and as soon as you start putting audio through, the osc comes back when the CV drops below about 0.5V in amplitude, ramping down from HF and coming to rest at our favourite 1k5Hz

Any bright ideas :?:

(Deep intake of breath)....Siiiiiigh...
 
This is what happens when you send 50ms bursts of 2K (500ms cycle time) into the comp at the moment.


You can see the nasty spikes on the odd LF rubbish before the burst happens. Its attack is around 500us as it (I think) shuts the valve down, compresses far too much and decimates the waveform, then you can see it release. After the release you can see it begin to self oscillate at a HF which ramps down then the next burst comes. It seems to take a few cycles to calm down too...

This is a closer view:


Before I changed the resistors around the opamp to 33Mohms it was worse...anyone got any 10Gohm resistors... :shock:

:mad:

EDIT:1Gohm seem to be better than 33M
 
Back
Top