RCA/175 Style Limiter DIY

[emrr]

Be curious how you feel the Sowters perform, and how they measure. 

AX7 bias.  That's a bitch in my experience, if critical.  Try a single AX7 cathode follower circuit ala Fender (http://www.ampbooks.com/home/classic-circuits/fender-bassman-voltage-amp/) and swap through a pile of new to modern AX7's.  You'll see all sorts of results, mostly maddening. 

 

[alexc]

But .. boy is this thing quiet  8)  Silent like. No noise gate needed here!

Doug - I really like that fender config - I use it more than any other - not sure why, I just like it.

But not with AX7 which I genrally don't use. I have a bunch of early 70s rca ax7s as well as the more modern, mostly russians.

Don't like any of them. Almost always sub and bias for something else. I prefer AY7s, even the  modern ones.

My silly bias probably but I just prefer a colder grid voltage - nearly everything I have is just too hot for typical AX7 settings. If I need the gain of an ax7 I mostly use a pentode nowadays.

 

[alexc]

I am pretty speechless really.

The Sowters sound absolute class.

This thing is oiled glass with a mink fur coat.

But more .. elastic  8)

Sh*t man. I'm floored.

 

[alexc]

Still floored ...

 

[alexc]

And I'm now sure this what some of the historical monster guitarists used...

I'm thinking Jim McGuinn, Clarence White, Scotty Moore, Johnny Marr 

 

[alexc]

Yep - no doubt.    ... #1 diy limiter of all time, for me so far    ;D

#2 - diy la2a, #3 pm660  then the rest  .. diy 1176, diy vactrol, vca et al

Ross and the like great for unbalanced stuff.

Hey! Everyone has an opinion !

 

[ruairioflaherty]

Obviously you'll need an outside beta-tester.  Should I PM you my address?

 

[alexc]

;D

I should say my last diy limiter is the fester tg. We'll see.

 

[alexc]

Another thing regarding Drip's 175b pcb .. there are options aplenty.

Some of the more interesting are :

- the option for switched interstage trafo 

  -> at a couple of hundred $ (aust) I want to keep this guy in circuit ALL the time

- option for center tap  or resistive-network fed B+ on interstage traffo primary

  -> lots of discussion on this and the relative merit

as well as 'fine' adjust for the attenuators, grid stoppers and prolly a few other things that I missed

 

[emrr]

Very cool.  The transformers on a real 175 are a bit crap, I never liked any originals I ever used.  The Sowters should make for a very different beast.

Agreed on ax7.  I'm not a fan in most cases.

Congrats on this one, sounds like great listening.

 

[DaveP]

Nice work Alex,

That's another one off the to do list ;D

best
DaveP

 

[alexc]

While I do more detailed tests and tweak the bias of the ax7 closer to spec, I'm having a stab at the 'utility power' board I want to use.

This may well be a waste of time, and I may ditch the utilities if so, but here goes ....

First up I am doing the voltage multiplier to give V+ sufficient for a pair of relays and a tl071 based vu buffer. I need 11V or so minimum to click the relays and (17+17)=34mA to keep them there  plus whatever the tl071 uses.

As I have 3.15V-0-3.15Vac for the heater, I am using a full wave Cockcroft Walton voltage multiplier - that won't interfere with the existing heater circuit.

That amount of current is quite an ask, so breadboarding is a must!

I've done 1 stage with 1N4007 diodes (x4) and 1000uF, 35V electros (x3).

I see +7.3Vdc at the output wrt ground powered from my 175b

Happily, the noise floor appears unaffected so I'll go to 2 stages.

Now +15.3V no load, dropping to  +12.2V with 340ohm (1.2W) resistor.

So that seems to be supplying 36mA which is in the ballpark.

Seems steady, nothing blowing ! and noise floor is completely unaffected.

So, connecting up the relays and all good. They switch with a happy willingness.
Supplying 33.5mA at 12.4V.

So it seems that it works in principle. I could up the current capability a little by increasing the overall capacitance and even adding another cap across the load, but it all begins to get a little silly when things get too big!

So I'm just doing a quick veroboard layout to figure how large it will all come to before deciding whether to go on or change tack.

 

[alexc]

Well that all went well - completed the utility psu and utility wiring.

Now the relays are all in place, the light show is nice and showy and the noise floor is unchanged.

So time to wire in the vu buffer and GR wiring.

Here's a pic of the inside with the utilities and meter wired up - some tidying up to go and done.

 

[alexc]

Just have the final tweak'n'test to go ....

 

[alexc]

and lastly ...

Here's a funny - didn't realise my RTA was way out of cal. I had done a 'restore' of my pc to a saved image where TrueRTA wasn't cal.

So I went thru my procedure as I documented some time ago and sure enough, I get back to my historical 'loopback' noisefloor of -83.4dBu (spl style of measurement, 20-20KHz).

So that means my measurements are way too bad by around 6dB - it's that bal/unbal thing and how TrueRTA assumes everything is unbalanced, even though the Motu is balanced and all that.

So there it is : cal and recal and test and retest and take nothing for granted.

OR 'get a pro analyzer!!'


Anyway, I'll be putting up some results of the test shortly

 

[alexc]

Final build testing is now done.

Did some tube subbing - found my best 6BC8 (vintage Raytheon). Balance at idle rock solid with plate delta <10mV and cathodes delta <1mV.

The 6BC8 is idling too hot right now - my schemo says 3.27mA per side (my sheets say 3.75mA per side) and mine is measuring at  5.3mA per side (sheets say 3.6mA !) - so more checking there.

Found my best AX7 (modern Mullard). That fixed up the balance in the AX7. It's fine - there is an error on my schemo. Plates should be around 200V - the ax7 run at some 0.32mA, which mine do.

BH7 sub made no diff. Didn't bother subbing rectifiers and glow plug.

I could probably get the B+ a bit closer to spec too with more dropper resistor selection. I'm coming up at 296V instead of 310V with my current now at 60.7mA HV. No biggy but I may improve that once I stitch up the GR tube idle current.

Did my meter cals - had to add 8K in series with GR meter to calibrate - this is due to my 6BC8 idling quite a bit too high. It will need further cal when I measure the GRs and compare.

Good thing is the  meter is totally and completely stable, which is a nice touch.

VU meter - all good - nice movement off the nissei with the jlm buffer.

Did some sensitivity testing - the wiring is all pretty stable but there is quite a bit of microphonics. The tube subs improved it some but but does remain.

The 'tap test' quickly recovers to the steady state noise profile, however.

The noise profile is rock solid with no unwanted twerking about - very well mannered indeed.

No 'optimisation' of ground scheme needed - this is almost perfect as is. The Drip PCB is clearly outstanding in this respect. Thank the Lord above for that   .. and Drip, of course!

It does provide seperation of 'church' and 'state', meaning the psu ground and audio grounds are kept apart with a single jumper point only between the two.

No heroics necessary here as it is almost perfect with a simple interconnect of the two.

Apart from that, I did the usual things - XLR pin1 to chassis at the xlr, make sure good continuity around the case (not always a given, especially with my cheap chassis), star chassis connection for the heavy currents, including my 'utility psu board'. Ensure shielded audio cables where appropriate, such as pots. Where the shield is not explicity used, leave with one end grounded, one floating (as like as not!).

And use heavy gauge ground wire - no exceptions!

Finally, adding the lid made no difference to noise profile at all - a good sign of stability.

So - that is now pretty much buttoned up. 

 

[alexc]

------  noise tests with no GR

With my  motu 828mkII re-calibrated for the dBu convention, I have :

- loopback noise floor  at -84.35dBu      (20-20KHz 'spl style' of measurement) 
  50Hz < -120dBu, 100Hz -110dBu and rest lower than that

- 175b powered up and hard bypass same as loopback

- 175b in circuit, no limiting and unity gain (-10dBu input, 1KHz sine) noise floor at -84.28dBu
  50Hz  -108dBu, 100Hz -96dBu,  300Hz -109dBu, rest lower than that

- same as above but now with +5dB gain (with output atten) noise floor at -84.10dBu
  50Hz  -107dBu, 100Hz -97dBu, 300Hz -108dBu, rest lower than that

- same as above but now with +10dB gain (with output atten) noise floor at -82.50dBu
  50Hz  -103dBu, 100Hz -93dBu, 300Hz -105dBu, rest lower than that

- same as above but now with +15dB gain (with output atten) noise floor at -79.0dBu
  50Hz  -98dBu, 100Hz -88dBu, 300Hz -100dBu, rest lower than that

- same as above but now with +19.5dB gain (max makeup) noise floor at -75.5dBu
  50Hz  -94dBu, 100Hz -83dBu, 300Hz -83dBu, rest lower than that

ie. 10dB gain  adds 1.7dB noise, 15dB gain adds 5.1dB noise and 19.5dB gain adds 8.6dB

So that is completely stellar basic performance and ranks at the top of all builds I have done. 

I checked that the noise floor is same regardless of Hi/Low Gain switch setting.
(compensate for the gain on the out attenuator to give unity gain)

------------

Freq response looks like +1dB at >20KHz and -1dB at 20Hz at -10dBu sweep

At unity gain, 2nd harmonic is 57dB below fundamental, 3rd harmonic 47dB, 4th 80dB, 5th 76dB
At max  gain, 2nd harmonic is 51dB "                  "      , 3rd              47dB, 4th and high  75dB

--------------

Next up, some testing under various conditions of GR.

ie. how much reduction I get, how much distortion under GR and the time constant measurements.

I will be looking at 6BC8 operating points in detail and considering possible subs as well.

Also looking for 'thumps' (CV breakthru into the audio path) and getting a feel for transient behaviour of the 'slam'

THEN some more listening tests!

All good stuff  8) 8) 8)

 

[alexc]

Still sounding fine

Did a quick check with a 6n5p in the GR tube position (direct sub but with increased heater current)

Damn fine  8) 8)

This sub gives more gain overall, if somewhat 'grainier' or more distorted a result. But not so noticable when you reduce the gain a bit to compensate.

What stands out is increased GR available and most importantly, the release characteric is much smoother on the slower settings  ie not so 'lumpy' in the long 'let go'

That suggests there are less 'bumps' (or more maybe!) in the transconductance curve, I guess. 

I think that although the sound is a little less 'smooth' than a 6bc8 (this one at least!), the overall GR is more transparent because it recovers more 'gracefully'

I'm a guitar player but I wish I played the pedal steel - hence the obsession with compression.

In my mind the $$  are worth it when I sound more like Gram Parsons' steel guy  ;D

I really hate 'lumpy release syndrome' - my main gripe with (diy) 1176 and one of my great joys with the la2a and poorman 660.

I'm using an orange drop as the C5 0.1uF 400V release cap.

jus' sayin  ...  ???

 

[alexc]

Something else - this unit can get quite hairy with the input atten really contributing in a big way to harmonic distortion.

Fairly innocuous changes in the input atten make big changes to the both the balance of the harmonic orders' dominance as well as the overall THD.

Not sure if this is the input traffo doing it or the GR tube (was same regardless of GR tube type)

I'm using the Sowter 1009e which has a spec of :
- MAXIMUM INPUT LEVEL 0.5 % THD at 50 Hz (40 ohm source) +15 dBu

I'm not driving it all that hard -  prolly around -10dBu at the XLR and with the input atten around 10 o'clock. It shouldn't be causing any problem on the primary. But as a 1:9 turns ratio, doesn't take much to turn into a way-hot-pickle at the grids! So I guess it's at the grids.

I'll see some more when I scope the gain structure some.

----

This could be a good thing or make it more challenging to use, depending on your point of view.

I think it is a good thing - you can drive this pretty hard and get some rubbery grossness if you want.

Or be a lot more subtle and get some highly classy result.

It ain't rocket science, after all!

I think the Hi/Low Gain switch is an absolute necessity as well as a wonderful feature to have.

I'm pretty sure I'll now incorporate it on my other limiter builds which use the t-pad atten before the traffo.

 

[alexc]

OK - jumped back to 6BC8

For now, prefering somewhat 'lumpy class' over 'smooth brash'   

For now.