the 1954 Limiter!!

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

cannikin

Well-known member
Joined
Feb 8, 2005
Messages
1,009
Location
Seattle Washington
Finally here is the fruit of my labors for the past 2 years (mostly searching for parts... especially the XFRMRs). I got the circuit from a 1954 RadioCraft magazine, hence the name "1954 Limiter".

I wanted this project/unit to be grand in size -- big chassis, big meters, big knobs -- just like the old days before everything was optimized to be small and compact -- With each unit in a seven space rack -- I think I achieved that goal.

Bear with me as I try to describe what's going on in the circuit.. still learning:
The limiter uses UTC A-12 (600:80K) input transformer and 2 6SK7s. The thing I liked is that it implements 250K pot to adjust the screen voltage on only one of the 6sk7 to get the 2 6sk7s screen voltage in balance. The bias is controlled by a 500K pot (input attenuator) and the grids even have a 500 ohm pot. The 12AT7 amplifier is used ahead of the 6AL5 rectifier to get a rapid attack - the 6AL5's cathodes are biased positive with respect to their plates and will not conduct until the output signal is fed thru the 12AT7 to have a limit level. Control of both bias if the 12AT7 and the screens of 6sk7s is obtained with the OD3 voltage control tube. Adjusting the 25K pot on the OD3 changes the limit level. the attack and release time is fast (for tube limiter). I really love the sound of it, it uses a UTC A-27 (100K:600) for the output -- shit with these transformers how could it sound bad? The switch is a 3 position switch, limiter "on" "off" and "Balance"-- balance makes is the unit a linear amplifier with no limiting.

1.jpg

2.jpg

5.jpg

6.jpg


The 2 6SK7s use DC heater power (I got the DC heater circuit from a Marshall amp that uses DC heaters on the input tubes), the other heaters are AC and have a hum balance pot -- just in case I'm getting some 60Hz in them. (everything the orange cap and couple of resistors to the right on the picture is part of the limiter on/off/balance switch)

heater.jpg


I used a bus ground scheme and the only ground point is near the input... seemed to have served me well.

turret.jpg


The meter circuit just toggles between input and output, I tried to implement a compression meter circuit (like the la2a or federal) but the result was a disaster - I burned 2 nice meters in which the needles on the meter melted and now are bent and always point down ;-) oh well I learned not to do that........

The Dual B+ power supply is 250V (for each channel) which I put in its own 2 space rack chassis to hopefully have ZERO AC hum issues (which worked!! in addition to the ac heater balance pot and dc heaters on the input tubes. 60Hz/120Hz hum has been my nemesis since my first project). The power connects to the units via a barrier strip =B+/GRND/Heater1/Heater2. I used a 6AX5 rectifier and since it eats 1.2amps, I had to add a triad filament transformer to power everything else. The first two filter caps are grounded to the center tap of the power transformer and chassis, the last Filter cap is grounded at the buss bar of the limiter circuit. This gave me a good ground point (zero volts).

3.jpg

4.jpg

7.jpg


The unit doesn't have a ton of gain like a federal limiter but the limiter starts at -30dbm on the input (input attenuator at max) when the output reaches -20dbm. I might have to build a line amp to put on the backside of the unit but I will test it the studio before I start another project....

8.jpg
 
That looks great! 9U for 2 limiters. Excellent, I like your style. :)

Is this schematic readily available? I'd like to see what it looks like.

I look forward to hearing how it sounds in action.
 
Let me be the first to congratulate you - they look fantastic. I admire the effort that goes into pulling a project like this together.

I haven't even got the time to assemble kits at the moment so clocking in at the end of the day is as close as I'll get to DIY for now. It's 1.22 am and I'm just finishing work - I started at 9am :?

Would love to hear a sound sample. Usually I have no interest in random audio samples but I'd love to get a sense of the character of these monsters. A before and after vocal or drum kit?

Stay well,
Ruairi
 
I type too slowly :grin: Perhaps it's the glasses of Vino Nobile di Montepulciano that accompanied the last hour of my work day... :thumb:

Cheers,
Ruairi
 
finally! Another DIY he-man equipment, destined to break toes and rodents when dropped. Plenty of room for additions and mods too! Looks great!
 
that is sick
those screw term strips: is that for filament, B+ etc.? and is there a cover that protects that?
 
[quote author="shabtek"]that is sick
those screw term strips: is that for filament, B+ etc.? and is there a cover that protects that?[/quote]

yup it is I have some plexi-glass cut to put over top of it.
 
> balance makes is the unit a linear amplifier with no limiting.

The original "Balance" position puts 60Hz on both 6SK7, which "should" cancel to the output.

The cathode trim will have most effect at high gain, your static trim. The screen trim will work over the full range, will trim the residual heavy limiting thump.

> the output reaches -20dbm. I might have to build a line amp

Indeed. The max clean output near 20dB GR is around 0.5V in 600 ohms. You don't get "real line level" from a couple low-current tubes. You either go big, or lots, or add an output stage. Narma liked massively simple. Most others felt that an output stage was the best path.

It is a fine low-level limiter. It isn't going to drive 10 miles of phone line to a transmitter. May be plenty ample for many studio jobs, if they are set-up for this unit.

The intended GR monitoring is a HIGH impedance voltmeter at the 6AL5 plates. As a $10 10-second hack, put your DVM on 19.9VDC. But DVMs don't dance properly. You could use a VTVM plan, without all the range-switch junk, but that really wants a dedicated floating 100V supply, or at least a good negative supply.

You could try a current meter in the 6SK7 cathodes returns, but it will be 4mA to 6mA, NOT 50uA. Use a 1mA meter with 10 ohms across its terminals, that will read low at idle, raise to 15 or 22 ohms. You could do this with a uA meter (or VU meter) but as you know, any screw-up is costly.

Attack time is ~~3mS. First release time is 330mS. A second time constant is around 1 Second.
 
HI PRR

I was so hoping you would give your expertise on this circuit. you are brilliant truly brilliant! "Attack time is ~~3mS. First release time is 330mS. A second time constant is around 1 Second." I can't believe you figured that out just by looking at the schematic.... wow! thats what it said on the article of RadioCraft.

Quote "By use of a Dual R-C circuit, attack time is rapid, while release time is prolonged. The parallel R1-C1 combination of 3M3 and 0.1uf charges very quickly; in the order of 0.00006 second, and discharges in about 1/3 second. By adding R2-C2 of 3M3 and .3uf in series across the R1-C1 combination, R2-C2 will charge slowly; in a little less than 1 second, and discharge even more slowly, in something like 2 seconds. Thus on the very short duration peaks only R1-C1 will be effective, since R2-C2 does not have time to charge, and recovery time is short. On Continued peaks, however, R2-C2 charges, and the recovery time is lengthened. This Prevents the "thumping" and "breathing" that occurs on some limiters and yet does not "dig a hole" in the program as would a quick attack long release type limiter" End Quote

this was designed as a limiter for the consumer Television set to knock down the volume of TV commercials in 1954

I will give your meter suggestion a try....

Thanks
 
an electric point

as i saw on the PSU photo, you have direct access to the high voltage socket, to me, it seems a little bit dangerous.

did you plan to cover the HV socket?
 
Almost EVERY piece of similar manufactured vintage gear has similar exposed electrical connections. Entire broadcast facilities were wired up in this fashion. It requires one to not be an idiot and keep ones digits in place, tongue in mouth, etc. It is quite a challenge to grab a terminal block like that and short anything; it can be done but the dividers are tall enough to prevent most accidental shorting.
 
I got that idea from looking at a lot of older gear, its going to be in my studio in a closed rack, I know not to reach behind there when the units are powered on, and they will hardwired to my patchbay so I'm not really worried about it. I do have plexiglass plates that I intend to put over the top of the strips away. Its fine.
 
Wow david that looks great & thanks for showing the schematic too looks like a cut down BA6A the attack release daul constenant set up seems the same as BA6 from memory.... I bet she sounds sweet ...who cares about exposed wires love it :grin:

What resistors do u have ...I love the old carbons not sure about the mojo ...know about the noise, but I love the look of them

Cheers
 
I see ( I think) makes sense to take output from first stage & next tube for side chain amp to feed the 6al5 for control voltage...the actual audio is just 12sk7s then...cool...no feedback for nice tubey sound & short signal path & output tran doesnt need to take HT......sorry if Im just repeating whats been said its just Im starting to understand these vari mu s more now...what a nice circuit
 
> I can't believe you figured that out just by looking at the schematic....

It is basic.

cantikin-2.gif


C1 charges-up from 12AT7 10K and 6AL5. I guess I estimated 12AT7+6AL5 plate resistance around 20K, so 30K total. 30K*0.1uFd is 0.003 seconds.

"The parallel R1-C1 combination of 3M3 and 0.1uf charges very quickly; in the order of 0.00006 second"

I believe they are wrong. The 3M3 has negligible effect. To get 63% control voltage in 0.00006 seconds, the charging source must be 600 ohms. Yet the 6AL5 must be several hundred Kohms, there is an explicit 10K resistor in the path, and the impedance at 12AT7 plates is dozens of ohms. 0.00006 seconds in their dreams.... they copied that number from a better sidechain, did not stop to think.

> knock down the volume of TV commercials

And 0.00006 seconds, 0.06mS, 60uS, is too short for this purpose. System overload is a non-issue, the only judge is the listener's ear. It will take many mS for the ear to cry "loud!". We don't want it ducking on every ignition pulse, only on "HELLo there, LET me Tell YOU aBOUT the NEW....."  Seveal mS response is good. 50mS would be fine.

There is another detail which makes it slower. In addition to ~~30K resistance, there is a 0.03uFd cap in the charging path. When the 12AT7 plate rises, charge divides between the 0.03uFd and the 0.1uFd. So it won't come close to filling on the first half cycle. Actual response time will be many mS.

The main decay is simple. 0.1uFd holds the control voltage, 3.3Meg bleeds it away. 63% will be gone in 0.1uFd*3M3= 0.33 seconds or 330mS. 63% control voltage is not the audible "release" point. Depends on your definition of "release" and on the 6K7's gain/voltage curve. But in that ballpark.

There is another R and C, the C is 3 times bigger, so something else happens 3 times slower, around 1 Second.

This dual-stage plan was floating around, but I suspect someone saw a Fairchild. Which does have 50uS attack, but from a much beefier sidechain than one 12AT7.

The interfacing is odd for consumer TV. It is approximately the level between volume control and first audio stage, but the ratio detector and volume control will not drive the 600 ohm input well.
 
Back
Top