Tube comp/limiter under development

[Larrchild]

Gus, I believe, mentioned a base-emitter junction from a bipolar as a choice.
Shottky's don't have the v-rating i think.

 

[SSLtech]

Yeah, I thought that the PIV would be the killer for Schottkys...

It occured to me that the third 'steering' diode also causes the compressor to behave a little 'independantly' since any limiters on the link buss will still release and attack at their own individual rates... though this is perhaps better than what happens when you link 1176's together and all the controls interact.

Anyhoo...

Dave, what happens if you "nail" that back-bias voltage down tight using a low-source-impedance supply (even a battery or two for testing porpoises) -Does the nonlinear behaviour diminish?

Keef

 

[NewYorkDave]

[quote author="SSLtech"]
It occured to me that the third 'steering' diode also causes the compressor to behave a little 'independantly' since any limiters on the link buss will still release and attack at their own individual rates... though this is perhaps better than what happens when you link 1176's together and all the controls interact.[/quote]

That was exactly my intention in adding that diode.

Dave, what happens if you "nail" that back-bias voltage down tight using a low-source-impedance supply (even a battery or two for testing porpoises) -Does the nonlinear behaviour diminish?

I don't know, but I intend to try it.

 

[analag]

If say the diodes were driven by a direct coupled sidechain amp, I bet the frequency related compression curve would look very different. Maybe we're talking about different things here. It is obvious that the impedance of the source driving the rectifier affects the speed of the unit also the biasing scheme which in the 660/670 is optimized for faster turn-on than the other things of a vari-mu nature. Lower source impedance/ flatter TC network behaviour, not perfect but...
Higher drive current/ faster attack but longer release time.

analag

 

[analag]

Any thoughts on the 6BJ6.

analag

 

[rafafredd]

These are the recommended ones by retro instruments:

6AU6, 6AK6, 6CG8, 6AH6, 6BA6, 6BH6, 6BJ6, 6CF6, 6265, 6265, 6136, 5749, 5725

Of all these, they say the best on is 6BJ6.

 

[analag]

I bought (50) 6BJ6, I plan on using (16) of them in this thing.

analag

 

[NewYorkDave]

Followup: sourcing the reverse bias directly from a regulated bench supply made no difference; the forward conduction at higher frequencies (relative to that at low frequencies) still pinches off at back-bias greater than 15V or so. It occurs to me that it probably has to do with pulse width in relation to the charging time constant in the sidechain (which is essentially a sort of PWM circuit, when you think about it), but I'm not inclined to sit down and math it out right now.

 

[NewYorkDave]

More tweaks, more measurements.

The bumpiness at the bottom of the curve might be from the GR kicking in at the start of the sweep. That reminds me, I really need to add a "hold" function to the sidechain.

 

[NewYorkDave]

(Image removed. See next page for details).

 

[SSLtech]

-Can we work on getting that a little more linear? :?

I mean ...come...ON!!! :twisted:

Keef

 

[analag]

Dave's been tweakin up the compressor. Nice.

analag

 

[clintrubber]

[quote author="SSLtech"]-Can we work on getting that a little more linear? :?

I mean ...come...ON!!! :twisted:

Keef[/quote]
I got fooled as well at first, but did you see that X-axis ? :shock:

He's been stretching & compressing & interval-fumbling that axis to get a straight line :thumb: :wink:

 

[NewYorkDave]

C'mon, Peter, I'm not that dishonest. I didn't manipulate it, that's just the way Excel laid it out. I messed with it for a half hour trying to figure out how to get a linear axis, then I gave up. If any of you know how to do it, I'm all ears. Or I can post the raw numbers if one of yuz wants to do the work for me :twisted:

Anyway, the point of that graph wasn't to show a straight-line Volt-per-dB characteristic but rather to illustrate the large range of GR that is achieved with a relatively small swing in control voltage.

 

[CJ]

1N5819 will take 40 PIV.
You could series 2 or three, but then FVD goes up.
Plus, not that linear. FVD goes up with current/voltage.

Interesting to note UTC A-26 is an output being used as an input.

Why?

#1 to get the right ratio
#2 to get the input level up

But, the A 26 is a single coil job.
No humbucking. What to do? Thrown on an A-33 ext shield.

 

[PRR]

> that's just the way Excel laid it out.

eXceL is an idiot.

What it did is actually "logical". Just not how we want to see it.

 

[AnalogPackrat]

If you don't tell XSmell to plot your two colums of data as x-y coordinates, it assumes one column is data labels or something "logical" like that. It gets especially confused if one of the colums contains text (you have to format the column as numbers if you imported it in a way that confused it's parser). Select your two columns and choose one of the several x-y graph types with a line plotted between data points (or not if you want one like PRR's). In many ways, I really miss good old 1-2-3...

A P

 

[clintrubber]

[quote author="NewYorkDave"]C'mon, Peter, I'm not that dishonest. I didn't manipulate it, that's just the way Excel laid it out. I messed with it for a half hour trying to figure out how to get a linear axis, then I gave up. If any of you know how to do it, I'm all ears. Or I can post the raw numbers if one of yuz wants to do the work for me :twisted:[/quote]
Hi Dave,

I was just kidding of course and dared so since I couldn't imagine you'd actually done a thing like that :thumb:

I myself do know of such a possibility for AudioPrecision-data though but hasten to add that I've never considered using it.

Impressive GR-range indeed!

Regards,

Peter

 

[NewYorkDave]

Thanks, guys. Choosing X-Y plot did the trick. As you can see, the result looks like PRR's graph.

 

[rafafredd]

still pretty linear.