RCA/175 Style Limiter DIY

[alexc]

Hi All

Here's the next of the couple of projects I have in the works at the moment.

It's a dual limiter, done 'poor' style but with some interesting bits and pieces.

Based very loosely on the rca style of limiter, and with bits and pieces from other  of the similar configured units like the urei 175 and so on ... and drawing a lot from the forum members discussions over the years .. and also with some truly inspired builds that some members have shared along the way.

One channel uses big octal tubes,  one channel uses small tubes and are the same apart from that.

So - classic style of limiter with utc input transformers and Edcor output and Power supply transformers with some classic old chokes.

Pentode GR tubes with interstage transformers feeding a moderately seriously beefy line amp. Sidechain is simple high impedance style with tubes for regulator and rectifier.

The rest is 'put' together using parts that I have already and my usual very generic 2RU case and home 'wreaked' panels with safmat labelling. All in all, fairly quickly done-as-I-go production values, fairly typical of my approach.

Meters are mci/triplett with vu buffer assembly - coolio. After using these there is no going back.
Absolutely exqusite sensitivity to music and ballistics that is just wonderful.

They are going to be switched between GR and VU which will be a little hack of those nice meter buffer cards

--

I end up with a very servicable 'prototype' kind of unit rather than a meticulously made build or commercial one.
This is probably the last of these kinds of builds for me - I have way too much 'prototype' style of stuff!

After these builds I am discribing in these preamp and compressor threads, that's pretty much it for building.
Anyway, there are one or two things left to complete along the way 

So there it is - a dual classic style of tube limiter done in a way that interests me.

Here's the basic chassis at the first interesting stage of assembly

 

 

[alexc]

And apart from that, I have a pic at the start of wiring up

 

[alexc]

and the first of the insides 

 

[alexc]

To be more detailed about the config.

As I said, it's a 'found parts' kind of thing, so the choices I made reflect a budget and the desire to explore some interesting parts, old and new. If I had more funds, I'd make different choices. This is what I had a chance to throw together, and doubtless getting a stable, 'nice' unit together is going to be a challenge.

signal path is balanced

utc a series 600:10K input  traffo -> input atten dual rotary switch > GR pentodes -> utc 10K:5K interstage traffo

then  makeup amp triodes -> makeup amp finals -> Edcore 8K:600 transformer -> t-pad attenuator

I'm using some octal can gss7 for GR on one channel and a 7 pin pentode for GR on the other,  a 6BJ6.

The finals are 6V6 on the octal side, and 6S4A style on the other.

Driver is 6SN7 on one side and a au7 on the other.

The PSU is an Edcor traffo, the same as used in the Poorman 660 builds. It is 250Vac non-CT at 200mA and 9Vac at 6A.

There is a 30Vac CT @ 1A for the utilities like lamps, VU meter buffer amp, leds and relays, of which around 120mA per 9V rail will be used.

I've used this transformer a few times. Certainly all the Edcor power transformers I have used have been a fine choice and very flexible in lots of projects.

The HV is with a diode rectifier feeding a pair of tube rectifier+cap/choke pi filter paths. The tube(s) is (are)  2x EZ40(s) which can use the same 6.3V dc heater as the other tubes. They provide up to 90mA each, which would be the max each channel work could work at.

The tube rectifiers are something of a fancy affectation but then so is the whole thing! I like a challenge and to do things a  little different each time.

I made the psu as seperate as I could for a couple of reasons, mainly to hopefully avoid inter-channel weirdeties in bias and so on that can happen with vari-mu configs of 'varying degrees of capability'

Each psu hv path can do 90mA of dc max - about right for this kind of thing. They are to max be around around 50mA in the line amp and probably around 25mA in the GR tubes. The remaining 15mA or so for the driver.

There will be some russian voltage regulator tubes as well as some 6AL5 diode tubes.

Heaters are all dc filtered and unregulated. I expect around 5.7A or so of 6.3V dc, and  I have 50K of filter caps there so far.
As always, a dropper power resistor gets me to the right heater voltage under load.

 

[alexc]

The side chain timing constants I'll be doing are pot variable attack and release - a very simple affair, based on a urei 175.

I'll maybe add an additional 'threshold' switch on the top deck. Maybe high/low sensitivity type of thing.

This unit is intended to be after a tube preamp I did some time ago - a dual fender 'brown face' era bassman/vibro preamp with traffo based line amp stages.

So it will see fairly high levels. I will probably end up adding a top deck 'input atten' switch to cater for that.

The great thing about the top deck is you can add stuff later and without messing on the front panel

This one parts would prolly run to around 1Kus with maybe 200 of that in shipping of some of the bigger parts.

 

[alexc]

There's a few things that makes these guys interesting to me and different from a close version of say a 175b, like the Drip pcb, which funnily enough I am also currently about half way thru.

I started that way back in the days when I had no real idea of all tube limiters.
I have a lot invested there with the sowter traffos.
But's it's been a long time sitting.

Later I learned on various others leading up to this build.
Anyway, I think the learning will be great to compare the differences between them.

Firstly, there is the extra-beefy line amp with transformer followed by a decently power capable T atten.
6V6 and similar are pretty strong which is interesting so need something there.

Then there is the use of pentodes  in pentode mode ! rather than triodes.
There's a lot of things going on with that there screen and for sure a world of interest to play with.

Using some interesting tubes for reasons of price and because they are different!
Some nice cheap us nos off the beaten track types as well as an whole raft of slightly different russian types.

That also means you have to work out all the operational points of the various stages and so on, which is really where the fun is. That is to say, to learn 'what is' 

And of course, the haut-couture of utc transformers and triplett meters.  8) which the Drip doesn't have 
It uses the more common ones  a littlw cheaper, easier to work with  and still not too bad a ballistic in the sub 1/2 kg category!

 

[alexc]

And of course, the back of it!

That monster Edcor PSU traffo is only part monster compared to the Hammond one as specified by Drip.

I bought that one, and Hammond are surely very nice.

But when I appreciated the sheer *scale* of it, I started what is my LAST project I ever imagined.
It's a tube guitar amp soonly.
But that's another story.

The Edcor only *borders* on nuts in a 3RU short depth rack.

 

[alexc]

Started doing the wiring on this dual rca style/175 style unit.

Now this is using the same Edcor psu traffo and Edcor 8K:600 voutput transformers as the pmSix61 limiter build I'm doing as well.

During testing for that I found a lot of hum - wasn't sure if it was psu radiating into output traffo or just ground noise. It turns out to be ground noise.

So I thought I would do some noise testing 'as I go' for this build, especially because this build has output ChB a lot closer to PSU traffo than ChA - that's everything in the circuit, including output traffo.

So, I've got the output traffos wired to the XLRs, the HV and heater PSU wired up but with no loads and the bipolar utility is just open circuit. The ground is a start more or less and is in place, such as it is.

Nothing else. I have terminated the output transformers in 8K at the primary and 600R at the secondary, connected to a lexicon balanced soundcard to my RTA.

I am checking the effect of grounding the CT of each output traffo primary as well.

Note the output traffos are around 30cm for ChnA and 18cm for ChnB from PSU traffo at centers.

My noise floors measures are not calibrated to absolute voltages - in this case 'loop back' is -78.4dBu or so 'below some arbitrary level'. It's pretty close to my other measurement rigs tho which are at -83dBu calibrated  or so.

Disclaimer - I'm not making any claims as to the validity of these tests - I'm simply trying to devise a standard and sensible imho test to watch for noise build up as my build progresses, rather than tackling a big problem all at the end.

This is probably the biggest and most complex pp wiring job I'll attempt as I wind up my journey into prototype builds.

----------

Power off DC ground  < 1mVpp,  Power on DC ground (no psu loads) < 1mVpp
XLR output grounded hot, cold  -78.4 dBu

Power Off
-----------

Chn A (furthest) : SPL -78.4dBu        Chn B (closest)  : -78.4dBu

50Hz  : -116dBu                                50Hz  : -116dBu
100Hz : -118dBu                                100Hz : -118dBu
150Hz : -120dBu                                150Hz : -118dBu

CT to ground : no difference

Power On
-----------

Chn A (furthest) : SPL -77.8dBu          Chn B (closest)  : -78.2dBu

50Hz  : - 90dBu                                50Hz  :  -101dBu
100Hz : -115dBu                                100Hz :  -108dBu
150Hz : - 93dBu                                150Hz :  -100dBu

CT to ground : -77.9dBu                    CT to ground : -78.3dBu   

So this first test shows no real noise induced in the no-loaded-psu case either by ground or direct radiating. In fact, the closer is quieter!

The thing to note is that things are very different, firstly when the 600R termination resistor is removed and if the primary 8K termination is increased.

With 600R termination removed, about 7dB is added to ChnA and 9db to ChnB - thats with CT not grounded. With CT grounded it is about 5dB and 7dB respectively added.

With 8K termination increased to 20K then noise starts to rapidly build up both at low harmonics of mains freq and also at harmionics thru to 2KHz or so.

----------------
So I'll be watching the noise as I build up this circuit.  So far it is good, in terms of radiated noise, providing my 600R termination is valid and the 8K termination is valid.

Next stop to check at : completing the HV section with rectifier tubes, chokes, caps and pi-filters.
Cheers

 

[alexc]

Now with the HV and dc heater rails completed and with around 1.2A of heater load and no HV load as yet.
The output traffo CT is connected to HV, primary terminated in 8K and secondary in 600R.

No significant change in the noise levels listed above. 

Next is to wire up the finals in both channels for cathode bias and have the grid signal grounded.

I'm using 6V6 for the big tubes channel and 6S4A for the little tubes channel.

 

[alexc]

I have now the finals in each channel fully wired in and working. HV current is some 60mA for the 6V6s and 50mA for the 6AS4s - so a total of 110mA HV at present.

The grids are ac coupled and the signal is grounded for now at the 6SN7 and 12AU7 respectively which are also only wired up for heater.

Heater current is at 4.2A for now.

I have a 100R 3W balance pot in each of the cathode bias circuits, unbypassed and am able to set the cathode balance voltage to 0.01V in each case.

So - the noise figures are pretty reasonable so far.

I haven't yet measured the signal gain on the finals.

Chn A (furthest) : SPL -72.5dBu                Chn B (closest)  : SPL -75.8dBu

50Hz  : - 87dBu                                      50Hz  :  - 88dBu
100Hz :  -80dBu                                      100Hz :  - 90dBu
150Hz : - 88dBu                                      150Hz :  - 92dBu

Added noise  : 5.9dBu (over power off)      Added noise  : 2.6dBu (over power off)
                      5.4dBu (over power on)                            2.5dBu  (over power on)

Est gain : +19dB(plate)                            Est gain : +18dB(plate)
              -11dB(traffo)                                            -11dB(traffo)
                +8dB (net)                                              +7dB (net)
 
So if the gain estimates are in the ballpark, it would look like ChnA is a little noiser and ChnB is very quiet. Even despite the fact the output traffo is nearly half the distance from the psu, which is more than half loaded up in terms of power delivered.

It may be that ChnA has more gain (very likely - could be some 25dB or so) and/or is a higher plate resistance - the 6S4A is very low on both counts.

All appears  pretty stable - alt grounds, wire jerking makes no diff, temp steel shields make no diff, lid nada.

Next stop - wiring in the 6SN7, 12AU7 and the interstage transformers to complete the makeup+line amps.

 

[alexc]

With the 6SN7 and 12AU7 wired in, as well as the interstage transformer  (10K:5K) with the primaries shorted to ground, the results are pretty good so far.

The traffo is just an interesting part I had - no fore-thought or science in that choice! Just interested in how it goes.  I may even add one of them fancy bypass switches there at the top deck.

HV current is 9mA for the 6SN7 and 12mA for the AU7, total now up to 130mA and heater at 4.2A.

Chn A (furthest) : SPL -75.6dBu                Chn B (closest)  : SPL -74.5dBu

50Hz  : - 90dBu                                      50Hz  :  - 92dBu
100Hz :  -85dBu                                      100Hz :  - 85dBu
150Hz : - 100dBu                                    150Hz :  - 83dBu

Added noise  : 2.8dBu (over power off)      Added noise  : 3.9dBu (over power off)
                    -3.1dBu (over previous)                              1.4dBu  (over previous)

Est gain : +19dB+23dB(plates)                  Est gain : +18dB+18dB(plates)
              -11dB(traffo)                                            -11dB(traffo)
              +31dB (net)                                              +25dB (net)


So, again, assuming my gain estimates are reasonable, the figures are really good.
ChnA actually improved a lot and ChnB degraded only very little.

ChnA badly needed supply decoupling from the B+. It had added some 17dB of 100Hz noise so I added a pi-filter stage for the 6SN7 and the big problem gone. ChnA didn't need any, but I'll pobably add to keep this similar across both channels.

I have a pretty decent imbalance in the cathode currents in both the 6SN7 and AU7, of the order of 0.5mA in each case. I don't have a balancing pot in this stage, so I will try and do some matching on the plate resistors to try and clean that up.

The GR stage will have both plate balance and cathode balance pots.

Next stage is to wire up the GR tubes : a pair of 6SS7 metal cans  for ChnA and a pair of 6BJ6s for ChnB.
I'll be running them idle pretty hot, at around -1.5V bias and current of about 11mA each 6SS7 and 10mA  each 6BJ6. That will be HV totals of 90mA for ChnA and 80mA for Chn B.

This will be an interesting stage - I haven't yet worked out the bias properly. The screens will be sourced from some russian OA2 type regulators with some 150V. The plates will come from input transformer CT fed B+ and with some balancing resistors. Not yet sure of whether I will run additional plate loads for the pentodes.

Anyway, at idle, the gains of the GR stage can be pretty big, so it will be interesting where the noise floor goes from here 

So far everything is going very well.  ;D ;D

 

[alexc]

Added the 6AL5 rectifier and SG1P (0A2 equivalent) voltage regulator tubes and also wired in the variable threshold circuit. Heater is now up to 4.8A.

No added noise   I also included the supply decoupling on ChnB same as A. It's noise floor improved quite a bit.

The figures are (with primaries of interstage traffo all still grounded) :

Chn A (furthest) : SPL -75.6dBu                Chn B (closest)  : SPL -75.9dBu

Those glow plugs sure are perdy

Next is wiring in the GR tubes. Thanks to DaveP's BA-6A build thread and reworked schematics, I have a reasonable idea of the biasing. I'm just working some last issues then I can get on and get those done.

I will be running plate resistors as per the rca style. I have the 175b build following which will be without them, for some variety. But these will include them so as to have an op-point around 1/2 B+  ie. 135V from a B+ of around 260V

So far so good.

ps I'm seeing again the buildup of a dc offset on the secondaries of the edcor when floating.

 

[alexc]

Being doing some calculations of bias and balancing schemes for the GR stages, and decided on some simplifications over the rca style.

Firstly, because I have not much extra HV current to play with, I'm going for a simplified cathode balance which doesn't reference the screen balance network.

Makes it more like a 175 or modded 436. Pretty simple just a balance pot in the cathode with some additional resistance thru to the meter and ground.

The screen balance I'm going to do as a simple balance pot with additional series resistors.

Now having gone thru the bias points, I'm changing from 6BJ6 to 6BD6 - mainly because I need a colder idle bias to accomodate the meter in the cathode circuit. The 6BJ6 needs too hot of a bias for the current I want, leaving me without sufficient cathode voltage to run the meter. I like to use 0dB GR at max deflection, not 0VU, so I need a few more volts there.

Also, the 6BD6 is closer to the 6SS7  - I want to be more similar across the channels.

Other improvement is I've ordered some UTC A33 shields - a reasonable deal on a bunch of them, so I can do this unit and my 660 as well. i figure I'll have the same noise issue here as I found there so every little bit helps. They spec '20dB reduction' of magnetic field, so hopefully that will translate into a few dB better noise at the grids.

Next stop is to wire in the GR tubes and await these parts to come in. This guy is starting to look the part - at some 15kg, it surely is the most impressive looking beast I've bashed together  

 

[alexc]

I've just done the preliminary wiring for the GR amps and am about to do some bias checks and then signal tests.

Here's a progress pic of the top. Thats 14 tubes and 9 irons.

----

Early checks are pretty good - the bias voltages and currents mostly where expected. Just going to balance things up now and get the meter voltages working. 

I am running 5.7A heater and around 85mA+85mA of HV.

Noise is still pretty reasonable - don't know the gains of the GR stage but I expect around 20dB or so each. So the overall gain from the GR tubes to the output traffo is high. -- Turned out to be +5.7dB and +8.2dB respectively. The plates themselves were in the >20dB gain but after the plate resistors and balancing termination to the interstage primary, overall gain dropped quite a bit.

Seems like I have around -72dB noise floor right now with 50:-94dB, 100: -78dB, 150: 95dB. Nope. Not even close! (yet)

If the gains are around what I think, that is stellar not stellar! - this is the idle condition of max gain, no GR and no feedback yet applied.
-- gains are very high overall and noise is also quite high. Have yet to rotate them UTCs and then shield them. Followed by the input attenuator and then negative feedback, all of which should reduce the noise to acceptable

I have the XLR wired to the UTC 500:10K then with 25K/25K balancing network on the secondary and to the grids. No attenuation right now. So this is the inputs unconnected or floating at the xlr.

The meters need around +0.9V dc for FS deflection, so I've figured that into my cathode circuit along with an adjustment pot of 100R linear. No probs so far.

Now to do more detailed checks and balances.

----

Everything balances, at this one 'idle current', pretty much equal, so overall there is very little to correct in terms of balances for now.

I'm running idle of 4.76V on the 6SS7 with 99V on the screens and 137V on the plates.
Thats a plate current of 5.8mA, screen current of 2.6mA totalling 8.4mA each phase, 16.8mA all up.

I'm running idle of 3.65V on the 6BJ6 with 113V on the screens and 170V on the plates.
Thats a plate current of 4.6mA, screen current of 1.6mA totalling 6.2mA each phase, 12.4mA all up.

So that is very, very close to my calcs. As I said, I will change to 6BD6 when they arrive.

Now it is all biased up, I'm doing some checks on noise - wire jagging and such.

-----

 

[alexc]

I moved this project to my main bench - I was building at a second space without cro etc.

I've  scoped and measured up the gains and such and everything is working properly. Which is good.

My noise measurements, which I said were all relative, bear no resemblence to anything valid on my calibrated RTA however  :-[

My gain estimates were in the ballpark which of course means huge, something like 45dB net gain. With around 35dB noise added.

Anyway, the measuring did help me avoid ground loops and psu coupled loops so it was helpful.

---

So now to analyse the gain structure, figure out the feedback amount that I should use while I make up the attenuators and such.
I'm guessing around 15dB of feedback. I'll try the standard  values and go from there.

That and up to 10-20dB atten at the output should bring me into the 20dB overall gain which is my initial target.
I'll do some more reading and see if I can figure any better.

So far, so good.

 

[etheory]

Don't feel alone with no-one else posting, this stuff is awesome, well done!

 

[ruairioflaherty]

Don't feel alone with no-one else posting, this stuff is awesome, well done!

Absolutely!  Tube design is above my pay grade but I'm watching and learning.

 

[alexc]

Thanks kindly for the interest. 

Just waiting for the weather to warm up at my bench to continue. Don't have the CRO and all that at my fireside diy spot.
My big table is still too cold!

These few builds of tube limiters and such have been a real learn for me - after these are done, and with a year or so of usage and twerks, I think I'll be ready to do them 'properly' 

The other big limiter is in usage - it also makes a valuable contribution to keeping my music room warmer!

This dual RCA/175 style and also the mongrel pre I'm doing are my main projects for the end of year DIY season!

 

[alexc]

OK!

Winter is now gone, temperatures are up, and I'm baaack in business.

First order is to finish the drip 175b pcb based build of the venerable UA 175b limiter I started.

Here's a pic of the build pretty much completed, except for some 'utility' things I'm yet to do.

Because the utility stuff can sometimes degrade performance due to added complexity, I'm testing up before adding them. These things are relay hard bypass, vu meter buffering, vu meter lights and some front panel led indication.

Particularly in this case, because I have to source the utility supply from the 6.3vac heater, as I don't have a dedicated winding on the psu traffo for it. That means a utility circuit with a rectifier/smoothing section as well as an additional voltage doubler to give my relay and buffer V+.
.
Not sure if it will work out without adding too much noise, so I'm testing the basic unit first .

Here's a pic of the insides without them .... 

(there's some pics of the outsides earlier in this thread)

---

Now for the basic testing while I do the first burn in ...


*** dc operation **

The dc conditions worked out mostly as expected.

I'm using the Edcor xpwr131 which gives the 320-0-320Vac HV as well as centre tapped 6.3V and 5V for heaters.

I wind up with an unloaded B+ of 428Vdc which drops to around 410V at 58mA of HV load.
(which is just tickling this beast!)

The filtered B1+ comes to 306V with my 1.8K dropper. Good enough.

Using ac heaters.

The  OB2 regulator tube comes in at 107.5Vdc at around 20mA supply.

Biasing for the 6BC8 is quite close to spec and with some adjustment of the plate and cathode balance pots, comes up extremely close to equal at idle .. like 1mV or so.

The AX7 is a problem bias - have to figure out why it has 2V on cathodes instead of 1.4. But these are biased so cold and with such high gain, even a bee fart makes big differences in conducted current. Probably a question of subbing till I get closer.

BH7 is pretty much spot on, with the plate balance a little out (no adjust). Something like 2V on 300Vdc or so. Close enough, but could be inproved with a balance pot on the cathodes, I think.

The 6AL5 rectifier is nearly spot on and the CV achieved is correct. Attack and release appear to be fine. I'll do time constant checking a bit later on.

*** front panel controls

No problem with the attenuators - I'm using CAPI units. Work as expected with around 68dB at the input and 85dB at the output of attenuation.

The gain hi/low switch works fine, with around 10dB of atten.

Limit switch does the thing - selects either open circuit or side chain CV to apply to grids.

Meter zero pot I brought out to front panel and seems OK.

**** noise floor and freq response

Quick testing here shows a very low unity gain noise floor - my Motu 828MKII interface calibrated for absolute voltages on the dBu convention to the best of my ability gives a 20-20KHz 'spl' style of measured noise floor gives a loopback of -76.8dBu.

The 175b with unity gain (-10dB u input, 1KHz sine) gives a noise floor of -76.4dB which is very, very good. 50Hz is at -97dBu, 100Hz at -88dBu and higher harmonics of psu less than those.

No sign of any 'twitchyness' - rock solid stable in the psu mains harmonics.

Considering the 'generous' size of the edcor  psu traffo and choke, and relatively close proximity of the Sowter output traffo to them,  I am very, very, VERY happy with this.

Adding 10dB of gain adds 5dB of 'spl style measurement' noise.
Adding 15dB of gain adds 9dB  "      "          "            "      "

So that's OK but not great. My PM660 style build (with similar psu traffos and Edcor OP traffo) was quite a deal better then that (some 20dB of gain only added around 7dB noise).

However, this is the first testing session, so not too bad really.

At least there is no sign of ground loops, ground voltage differences or psu induced noise

;D ;D ;D ;D

Frequency response at -10dBu sweep is flat 20Hz to 20KHz - no problem there.

Harmonic distortion seems a little high at first glance, but OK. More on that to come later.

*** Gain Reduction

No problem with reaching 20dB or so GR.

At 9dB GR harmonic distortion is reasonable. As GR increases, so does THD as expected.

Hopefully I can tweak/select this to improve.


-----

OK, so now after a few hours burn in - no smoke, dc conditions stable and basic performance is as expected. Hopefully I can improve the noise floor under gain and GR some, going forward.

Now to the first listening tests - the most exciting part of all! This is my first 175 style of limiting
ie. single GR tube with two triodes, (hi quality) interstage traffo, and moderate power (hi quality) push-pull output stage.

The sowter traffos cost a bomb, so I hope to hear something good ... better than the modest priced large frame edcor output traffos I am accustomed to.

----

Generally speaking this was a pretty simple build - the drip pcb works well and is quiet!

There was the error regarding orientation of the Sowter interstage traffo, which one absolutely must fix, but apart from that - all good. Kind of overkill really, but good.

----

Next up - the LISTEN    8) 8)

Thence to be test/adjusting  to a more detailed level.

Finally,  the utility functions and button up.

Cheers

 

[alexc]

Oohhhhhh yeesss  ;D

That sounds  ... sublime ....

So very classy - like a girl with some fur coat.

So very fine 

'Venus In Furs'