RCA/175 Style Limiter DIY

[alexc]

Did some program music thru this guy


;D ;D ;D Man that's cool

A little lite on in the bottom end, in this case, yes  : BUT .. WOW  that's an awful lot of fun in the sound 8)

The compression is just insane - even just ticking over makes a real .. glue .. thing.

I've never really been one for full range material on vari-gm boxes BUT I think a full freq response one of these could well change my mind. (.. adding 3dB on the bottom fixes that ..)

Playing 'Big Star' song 'In The Street' - an alternative mix which is quite rough.

Until this

Still  laffing and playing on ...  put the guitar on top as well and the blending is just amazing.

Putting something  little more serious  - in this case a local Aust guy, 70s :
    - Richard Clapton - Girls On the Avenue

Quite .. different.

On just the program, it totally changes the relative balance of the mix elements.
I don't think I've ever heard that before. I've only about 5dB GR max but it's very ..  different.

Mixing would need to be judicious but the end result could well be out of the ordinary.  :-X


Seriously - if I had some more bux I'd do this box again, but with more respect.

It's  .. freakacious ... 'there is no emoticon to express what I'm feeling ..'

ps - no pumping, thumping or anything even remotely close!

---

OK - all kidding aside. That's a truly .. respectful .. treatment of the mix.

4dB GR, Release at 1pm, Attack 1pm, Gain back to unity.

Gram Parsons and Emmy Lou - 'Sin City'.  Wow.

--

To be real, I need my relay bypass. Nothing like a quick reality check

 

[alexc]

Whoever designed this thing .. Blumlein ..  Major Armstrong .. Western Electric .. RCA guy

Who knows?

But what a great job!

 

[alexc]

Every now and again, the twin gods of electronics and music grant you a perfect day.

Today was mine.

Sigh .....

 

[DaveP]

You always get full marks for persistence Alex! ;D
Excellent trouble shooting.
best
DaveP

 

[alexc]

Thanks Dave - as usual, I have drawn on your descriptions of these designs, and even though my 'production values' are very far from yours, I have taken on board your 'principles for successful construction' 

I'm just too much of a 'tinkering' engineer/musician type rather than a 'craftsman'.

And sometimes, but not always, the price to be paid is 'noise under applied gain'.

My wish would be to approach the kind of quality you build in from the start.

When I started these 'tube on top' builds they were mostly to use up spare parts and see what kind of performance one could achieve in 'organically assembled' (ie. hastily knocked together) builds.

Had I have known how great the results could be, I would have done a more respectful job of them!

-----

So - am poking about trying to improve the noise floor performance 'with gain applied'.

I have each channel about the same, at -72dBu or so at +15dB gain, which is a worsening of the 20-20KHz 'spl style' noise of 12dB.

I would like to improve that by 5dB or so.  :-X

Nothing is helping so far. Examined all the usual suspects.

- haven't found any wiring sensitivities
- haven't found any ground loops
- addition of B+ massive brute force additional pi-filtering makes virtually no difference
- addition of B1+, B2+ additional pi-filtering makes virtually no difference
- Adding the lid makes virtually no difference.
- Re-inforcing continuity around the chassis makes no difference.
- removing input XLR makes no difference

It is rock solid in that respect.

----

It's also not appearing to be noise from PSU - each channel is very similar  ie. the closest is same as furthest. No further gains in interstage traffo orientation or shielding.

Definately not audio cross talk between channels or PSU interactions between channels. I have very good seperation of PSU per channel and have checked for audio/noise cross interactions.

----

I can tune the noise somewhat with the balancing at the finals' cathodes - this alters the relative dominance of the even/odd harmonics in the low orders, and the total of the harmonics in the high orders. By favouring 2nd order distortion products, I can get a 2dB difference in the noise floor whilst still being near the 'balance point'.

Similar thing but to a lesser extent at the GR tubes' cathode and plate balance for <1dB.

Another thing is metal-can GR tubes must have the 'shield' pin connected to ground. It makes no difference to noise floor if not 'touched' and so on, but floating will pick up noise when 'touched'.
Best to ground that shield.

---

So - nada so far.

Changing tack - looking at the noise floor spectra at +15dB gain applied, the noise seems to be accumulating in the mid-freq  500Hz - 3K or so.

Sure,  the hum harmonics grow under gain, but smaller 'growth' in the mid and high freqs adds more to the 'spl style' of 20-20KHz 'headline' noise figure.

ie. the 'spl style' 20-20KHz measurement is a measure of rms  'spectral energy'. The higher frequencies contain more 'spectral energy' and so those peaks have a higher weighting in the rms 'spl' measure.

This is born out by the fact when I actually listen to noise floor under gain, I don't get the sense of hum.

So - I'm looking at 'hiss' rather than 'hum' for now - never done this before, so have no idea on how that might be achieved. 'resistor noise' in the grids maybe? That sort of thing.

---

I'll also do a bit more tube subs - it may make a small difference.

I will continue for today's session then toss in the towel and accept it for what it is :

PRETTY DAMN GOOD (although not totally great!)

 

[DaveP]

Don't sell yourself short Alex, you are doing some great work over there.

I would be quite happy with the noise levels you are getting.

It seems to be just hiss which is the problem, which is probably just down to tube and resistor noise and maybe operating points?

If you are using classic circuit values then there is not much more than can be done.

best
DaveP

 

[alexc]

Cheers Dave - the regard of one's peers is a good part of the process 

And I think I am seeing a fundamental limitation of my build - it would take the next level of quality to bring that performance measure also to the next level.

So - it's all done. Just doing some leisurely final measurements and testing at the bench before it goes to it's new home in a day or two.

And the vitalstatistix : weighs in at 12kg and cost around 1250aud of which around 300aud was parts shipping.


Here's the final pics, starting with the insides  ..

 

[alexc]

and front - the meter lighting is yellow and a bit dim in this light.

 

[alexc]

and the top

 

[alexc]

and the glamour shot 

 

[alexc]

So that concludes this guy, excepting any other interesting things I think of.

As always - thanks kindly to all those who helped motivate me to do these builds ..

prr, emrr and Dave in particular whose wonderful descriptions of his own work made me think it was all possible.


Cheers

 

[alexc]

Now this guy is in my music room and working out fine, I'm kind of sad really.

'Post-build letdown', as Lucy would say.

I would so like to do another one right now with premium traffos and take my time to do some .. crafting .. for the first time in my life.  But units like this really are a luxury build - I'm lucky to have made it this far!

I'll never really reach the lofty standard, that I know in my bones, these guys can embody :'(

Oh well. Such is life. Like parting with friends, it' sad and cathartic at once.
It's been a privilege to experience this build, and the 660 before it, even such as they are.

From here on in, just a couple of last units and that's it.
I'm feeling like I've done nearly everything that I set out to do and more is just repetition.

Anyway - I still have the listening for as long as I have left to hold them  

And the appreciation of the true aesthetic wonder that only 'tubes-on-top' design brings.

----

Here's a pic of my other garden friend - he took less time to grow than this limiter

 

[alexc]

After spending some more time with push-pull line amps that have cathode balancing pots, I think I have a lead on one of my issues with this build ..

------

The octal channel here had a low frequency -3dB point a little higher than the noval channel.
At the noval 55Hz -3dB point, the octal was around 1.5dB worse.

I noticed at the time the octals' 6V6 cathode balance pot couldn't quite adjust for an equal voltage waveform at the plates. At the time I thought it was within the error of 'stuff'.

Maybe 2V or so on the typical output traffo primary signal voltage of 200Vpp or so

The novals, on the other hand, allowed me to adjust and fine-tune the noise floor by around 2dB.

----

Working with my current build, which has a similar pp line amp, I had a similar but more severe problem with the range-of-adjust  on the finals cathode pot.

I saw that a couple of volts difference at the plates signal voltage can indeed make a couple of dB difference to the output at low freqs 20Hz-50Hz.

I replaced the pot with manually selected cathode resistors and that gave me for sure 2dB or so improvement at 30Hz.

---

SO - adding some extra resistance to the pot sides for this octal channel to give a bit more targeted adjustment range, so as to achieve a better finals' plate balance than I previously measured.

That should reduce the loss of inductance due to imbalance in dc currents 'just a bit' and *may* buy me the 1.5dB improvement at 55Hz which I am looking for.

;D

---

Apart from that, all is well with the world and team 'AlexC Audio'   

 

[alexc]

I did a lengthy session in my music room yesterday .. using the Drip 175b build which I have after a 'vintage channel strip (3)'

That is an api pre+gpultec tube+1176 rev_d  diy channel.

As I suspected, it was a vast improvement over the channel alone.
I mean, alone, the Vintage Channel 3 it is pretty good BUT the difference with the tube limiter after it was MAJOR.

Using the channel's 1176 as a fast attack/fast release and the 175b as a moderate attack/slow release.

The overall noise level at silence remained quite low, thanks to the Drip 175b not adding virtually any hum and the Vintage Channel 3 build long since shaken down for any excess noise.

Like 'my' other great 'discovery' of an 1176+LA2A combo, this 1176+175b combo is ..  um .. 'awesome'

I claim a 'partial credit' !