the 1954 Limiter!!

[davemascera]

Most of the old compressors were designed for radio transmission, so the timings were fixed  (they were playing mastered recordings after all).  For recording purposes, you need a selection of attack and release times.  The normal way to do this is to keep the cap value constant but switch in various series resistors to feed it for attack, then a similar switch in parallel for release.  The advantage of switches is that you can make a record of the settings for the track, not really possibly with just a pot.

0.1uF can allow a slight ripple which on some feedback compressors can turn into motorboating,  1uF is used on the Fairchild and others, but many other comps get by with 0.25 or 0.5uF.  Resistors that add with each position are mainly used to avoid switching noises, changing caps will probably cause some  sudden charging noise.

DaveP

Hmmmm interesting... so if I'm understanding this, throwing in a 300k pot will probably make my attack around (if fully engaged) 33 ms, as opposed to a fastest-case (and apparently not plausible) 3ms. That's if the first time constant cap remains at .1 uF, which you're saying may create motor-boating, or rather does in some circuits.

It might be a good idea to omit the second time constant altogether. I get the feeling that this was also created with television advertisement broadcasts in mind.

So if I then threw a 2.5 Meg pot in front of an 800kohm resistor, that would supposedly range from about 80 ms to 330 ms on the release...

And all of this is ideal and likely won't really perfectly reflect the reality of the time constants.

Lots of experimenting to be done I suppose. I don't care for switches (though I'm already using one of those Chinese dual 24-step attenuators on the input). This will likely be used on vocals and I compress before it hits the tape.

 

[DaveP]

Hmmmm interesting... so if I'm understanding this, throwing in a 300k pot will probably make my attack around (if fully engaged) 33 ms, as opposed to a fastest-case (and apparently not plausible) 3ms. That's if the first time constant cap remains at .1 uF, which you're saying may create motor-boating, or rather does in some circuits.

It might be a good idea to omit the second time constant altogether. I get the feeling that this was also created with television advertisement broadcasts in mind.

So if I then threw a 2.5 Meg pot in front of an 800kohm resistor, that would supposedly range from about 80 ms to 330 ms on the release...

And all of this is ideal and likely won't really perfectly reflect the reality of the time constants.

Lots of experimenting to be done I suppose. I don't care for switches (though I'm already using one of those Chinese dual 24-step attenuators on the input). This will likely be used on vocals and I compress before it hits the tape.

You may actually get 3mS.  The fastest attack speed relies on the source resistance of the driving circuit.  A Fairchild for example has an output resistance of 100 ohms before its diodes, which enables its speed to get down to uS.  The 12AT7 is run fairly cool so it is probably around 25k, then there are a couple of 10k resistors as well.  I would try losing the 10k's and putting in two cathode followers from another tube to drive the 6AL5, this would take the source resistance down to around 250 ohms.

Have fun

DaveP

 

[emrr]

It might be a good idea to omit the second time constant altogether. I get the feeling that this was also created with television advertisement broadcasts in mind.

Mmmm, no, nothing to do with advertising.  The dual time constant can be found in many things, BA-6A's, Stalevels, etc. The early NBC vari-mu AGC's were meant to be driven at an average gain reduction of 15dB, and the slower dual time constant sets a baseline for the faster to work off of.  It reduces the amount of pumping and also extends the useful range of compression.  It sort of makes for a virtual threshold when you are into heavy amounts of compression.  You can also put it on a switch, like a Stalevel. 

 

[davemascera]

So far so good... build is basically done... only issue thus far is that beyond about 25 ms of additional attack delay, the circuit does indeed begin motorboating, but I also haven't 'balanced' the unit yet. Fortunately, that's a fine parameter for me anyway. The sucky part is getting a new pot.

B+ is a tad cold on the 6SK7's and a tad hot on the 6SN7... not horribly worried about it.. but might **** around with the dropping resistors... gonna check plate dissipation soon.

Gonna check for hum next shot I get (used all DC heaters and about 45000 uf of filtering so hopefully that bit pays off despite the low hum risk of this circuit anyway... I had a filament transformer sitting around and a lot of room on the chassis), then freq response and give it a good listening. If anyone cares I'll post a picture and some sound samples when its totally finished. Then if someone cares EVEN more (something makes me doubt they will) I'll post a schematic.

Sadly I got one of those 'rugged' OD3's that isn't quite as gaudy as the rest  :-[. Project was built into an old Magnavox Radio chassis and power xformer.

 

[DaveP]

Well done

DaveP

 

[scott2000]

The sucky part is getting a new pot.

??? Why...... Are you having a hard time locating one??? Or just in general because of not having it on hand? What one??

 

[davemascera]

??? Why...... Are you having a hard time locating one??? Or just in general because of not having it on hand? What one??

https://i.ebayimg.com/images/g/JwoAAOSwz2lXCB7S/s-l1600.jpg

Gaudier

No, there just aren't many electronics stores around me, so it's a matter of ordering from the internet and paying for shipping... but I got the motorboating to stop.... the B+ was actually way more off than I thought it was, and the regulator tube was getting REAL hot and the foam bit around it began to melt (or the glue holding it did at least). Realized this baby wants 250V unloaded coming in, not for the node to READ 250V. After that and balancing, motorboating ceased and this thing seems like it's functioning as intended.

So I haven't gotten to plug this in at the studio yet and feed it with a real input and hear what it really sounds like, but it seemingly works. Provides more than enough output gain. Meter strapped to the cathode shows 'limiting' for sure, but it's a 1 ma meter and only drops about .2 ma at maximum gain reduction, so that's a bit ineffectual.. I tried 10 ohms initially. Brought the needle up to nothing. 100 ohms on the pot lets the needle hit about .8 ma.. no signs of melting or anything terrible... considering getting a smaller meter, but stinks that I already have the chassis cut for this one. Maybe having a full, balanced signal will make me change my mind.

If I remember correctly, a signal coming in at about 600mv with no attenuation outputs at about 2.4V... if I throw on the gain reduction, I can cut that to around or below the signal level.

But all of this led me to check what the idling plate current is on the 6sk7's... ~1 ma... so something feels horribly wrong on that front. Either that or I don't get the circuit as well as I thought. Probably the latter, especially (now that I think of it) considering that any normal amount of current coming through those plate resistors would require a rather large B+. Did try out the 'balance' feature using the cathode pot and the screen pot... pretty cool feature and a bit of a lifesaver with that whole motorboating thing.

 

[scott2000]

Scary about the regulator.... I was thinking about how those ss octal replacements I linked are.......I guess not cool looking though compared to the gas...........

What do you mean by it wants 250v unloaded???  B+ should be 250??


Also, The original 1954 circuit is a bit vague on the metering and I haven't grasped how it is supposed to be done...

Do you have a drawing of yours yet???

Did you implement the one balancing circuit using the 6.3vac as well???


Excited for your progress...... I thought the original poster of this thread loved the circuit but I came across a thread in the black market , posted some time after, where he actually ended up selling it because he couldn't get it to work???? So I'm confused if anyone ever actually tried it......

 

[emrr]

I thought the original poster of this thread loved the circuit but I came across a thread in the black market , posted some time after, where he actually ended up selling it because he couldn't get it to work???? So I'm confused if anyone ever actually tried it......

The OP got out of audio altogether, I recall he had them working fine. 

 

[emrr]

What do you mean by it wants 250v unloaded???  B+ should be 250??

Also, The original 1954 circuit is a bit vague on the metering and I haven't grasped how it is supposed to be done...

It should work with 250V B+ measured at the node when connected, that is what the drawing means.  Layout could cause motor boating, making the coupling caps too big could cause it, you can always throw more filter capacitance at those two points and see if it clears it up at the higher voltage.  How far down is B+ for it to work?

1mA idle is totally believable given the resistances. 

Look at the BA-6A metering, which is the type PRR referred to back on page one.  Same as many other limiters.  Look at SA-39.  Notice all the resistances here are scaled way up from the BA-6A, so current will be lower, will need a more sensitive meter. 

 

[scott2000]

Look at the BA-6A metering, which is the type PRR referred to back on page one.  Same as many other limiters.  Look at SA-39.  Notice all the resistances here are scaled way up from the BA-6A, so current will be lower, will need a more sensitive meter. 

Thanks! I'll look over it again.  Both the op and davemascera seem to have had issues with metering and it would be nice to "see" what worked best.....at least in the 1954 build.........Lassoharp never really mentioned any issues in his design so, maybe it works in this???

I recall he had them working fine. 

I thought so too and have been interested in this build since.......I was just confused when he sold it...??

https://groupdiy.com/index.php?topic=59153.0

 

[davemascera]

Scary about the regulator.... I was thinking about how those ss octal replacements I linked are.......I guess not cool looking though compared to the gas...........

What do you mean by it wants 250v unloaded???  B+ should be 250??


Also, The original 1954 circuit is a bit vague on the metering and I haven't grasped how it is supposed to be done...

Do you have a drawing of yours yet???

Did you implement the one balancing circuit using the 6.3vac as well???


Excited for your progress...... I thought the original poster of this thread loved the circuit but I came across a thread in the black market , posted some time after, where he actually ended up selling it because he couldn't get it to work???? So I'm confused if anyone ever actually tried it......

Well, I turned off the regulator before actual damage was done, but the glue had turned to liquid for sure atop the foam. The weird thing is that it wasn't drawing anywhere near maximum current according to the datasheet. It was at about 33 ma. So it's probably just a matter of aging parts. I should probably send a message to the ebay dealer if emrr is right and get my money back.

The only drawing I have I did with pen on a piece of computer paper and it's really just for me to understand... a lot of the PS values have changed, etc. I just put a 1 ma meter beyond the 22 kohm resistor to ground and put a 100Ω pot across it. It definitely functions correctly, but really doesn't dip enough for my liking. I have a switch for it in case it causes distortion. Also made a switch for the second time constant, and of course, created adjustable attack and release on the first one. Weird part is that adjusting the attack should also affect the release. Not a problem for me though.

I used DC heaters like I said from a seperate filament transformer. I hit a bridge rectifier, 45000 uf with .33 in between 15000 and 30000... got about 19mv of ripple. Thing sucks in about 3.2 amps of AC from the transformer, which is well below rating. So in order to implement the balance circuit, I took the 6.3 ac line off of the power transformer itself, let it hit a pilot light on the front panel (which has since burned out as it was likely about 80 years old) and let that hit the balance circuit.

I'm unsure cannikin ever got his working. I wouldn't be surprised if he did not considering that he didn't post audio. But maybe he did and we can no longer see it here.

It should work with 250V B+ measured at the node when connected, that is what the drawing means.  Layout could cause motor boating, making the coupling caps too big could cause it, you can always throw more filter capacitance at those two points and see if it clears it up at the higher voltage.  How far down is B+ for it to work?

1mA idle is totally believable given the resistances. 

Look at the BA-6A metering, which is the type PRR referred to back on page one.  Same as many other limiters.  Look at SA-39.  Notice all the resistances here are scaled way up from the BA-6A, so current will be lower, will need a more sensitive meter. 

B+ is about 196V now at that node. Regulator tube is warm, but drawing only 23 ma. I have definitely not overloaded on filtering (choke-20-30-30). This project was a lot of using crap I already had (aside from the interstage) so I didn't want to create too much sag  cause I was using a 5Y3.

However, I discovered a grounding problem in the sidechain.  Fixed it, might go back to the way it was for some more tests. Still, shouldn't result in the regulator smelling like that. That's what I get for picking the rugged one instead of the bare tube, I guess.

And yeah, you're right.  I'm not positive what kind of meter I'm going to need if I want to see a proper resolution. Will have to do some tests with a rather beefy signal.

 

[davemascera]

So I plugged it into my technics stereo... still haven't taken it down to the studio...

I figured out that my 'motorboating' is not 'motorboating.' It's a static-y sound that gets made when the sidechain hits my dual ganged pot that is supposed to delay the attack. Very strange. When the sidechain is as intended, the circuit works. The release adjustment seems to work just fine though, and so does turning off the second time constant. Also, why did it go away when I took the B+ down? Maybe just cause the 12AT7 wasn't getting the voltage it wanted.

The big problem I have with the circuit is that it's light on the compression. If I had to guesstimate I'd say that if I put in a signal as loud as it can go without getting a large amount of clipping from  presumably the 6SN7, I can attenuate it maybe 4 decibels maximum using the threshold knob. Once it's attenuated it doesn't clip though. Basically a full volt of signal coming in will almost clip it if I turn the limiter off or relieve the threshold a certain amount.

I know emrr left some instructions for making the threshold a bit more extreme. May try and implement.

So the last three mysteries I have are can I increase the plate voltage on the 6SN7 to get myself more headroom, why does my dual ganged attack pot create a heavily distorted 'static-ish' sound whenever attenuation occurs, and how can I make the compression effect more extreme. Maybe the A-18 is handing the 6SN7 a hotter signal than it wants.

As an aside, the only way I could get the OD3 to have less than 40 mA on the cathode was to hit the first stage with about 230 volts. Weird. And I tried two different tubes, and beefed up my resistors considerably. Maybe therein lies the clue to the subtlety of the effect.

Here are some pics of this grungy thing. Gonna make a plate to cover that gap up front. It's not too terrible with hum.

 

[scott2000]

Cool!

I like the repurpose thing....

Hard to follow along as easy for me without a drawing.... You have an interstage too?

What are you using for input and output transformers?

On that note, what is the transformer output called in this 1954 ? The B+ is confusing to me???

 

[davemascera]

Cool!

I like the repurpose thing....

Hard to follow along as easy for me without a drawing.... You have an interstage too?

What are you using for input and output transformers?

On that note, what is the transformer output called in this 1954 ? The B+ is confusing to me???

I mean trust me my drawing wouldn't help you. It's basically mostly scribble. I was gonna make something totally new if this works out.

Basically it's the 1954 schematic, the only things changed on THAT section are I took the single 3.3 MΩ resistor (not the one paired in series with the capacitor... that pairing is on a switch so I can turn it on and off) and turned it into a 1.3MΩ resistor and a 2 MΩ potentiometer (to change release time).  I put in a dual ganged 300kΩ pot on each leg of the cathodes of the 6AL5. Also, instead of 250 volts hitting the circuit, I have 230, because the OD3 was outputting 42 ma with 246V, which is above its rated maximum.

I put in a meter between the 24 kΩ resistor that hits the cathode resistors of the 6SK7's and ground and put a 100 ohm pot across its terminals to allow current to escape and not melt it. This is also on a switch so I can bypass it.

Instead of an output transformer, I have this hitting an interstage (A-18). The primary center tap is just grounded like in the 1954 limiter scheme. On the other side of the interstage is a 6SN7 with the output stage and biasing resistor values from the Federal AM-864 implemented. I have about 200 volts hitting it, where on the federal, 185 volts is called for. Not really a large difference. The B+ to this tube goes through the actual output transformer just like the Federal.

The power supply is a 5Y3 hitting an 11 henry choke, hits 20 mfd to ground, 1500 ohms in series, connection to B+ (230V) of 1954 limiter stage and 30 mfd to ground, 1500 ohms in series, 30 mfd to ground and about 200V to the 6SN7.

Input and output transformers are edcor WSM 600/15000's I had from another project I didn't care about.

 

[davemascera]

https://www.youtube.com/watch?v=cFr0QON3x8s

Here's a ****** video in case anyone wants to hear. It really doesn't get loud enough to be useful until about 40 seconds in. The other thing is that I show the balance circuit at the end and the frequencies probably didn't get picked up by the phone mic. I'm a failure.

 

[emrr]

Nice drum choice in the video.

Cannikin's story seems to have changed between OP and sale.  Who knows. 

OK, if I'm reading this right, the 12AT7 feed is still coming from the same place.  Take it from the output of the 6SN7,  a place with more signal level.  It may need more B+, from lower value plate resistors, like down to 33K, which'll draw a lot more current and give greater drive power, consider the cathode resistor again.  It may need voltage divider attenuation on it's input if the other places are too hot, could be a separate threshold pot. 

What orientation is the A-18?  I'd run it with the higher Z side as primary.  Consider the original circuit calls for a A-27 with 100K primary.  The load affects compression range. 

Attack control, not sure what you're describing, doesn't sound right.  It should be series resistance between the 6AL5 and the other time constant parts.  Look at the Altec 436C.    You're wiring a rheostat, so connect wiper to one side for securities sake. 

Where is 6SN7 B+ tapped from?  Should be right off the output of the PSU, straight to the transformer primary CT, to the 6SN7. 

The series resistor into the 0D3 can be larger to lower current.    The implementation strikes me as odd.  Many times a regulator is used to control absolute threshold reference, and/or GR stage plate voltage, here it appears to be primarily controlling screen voltage, though admittedly with a 2K series resistor it's gonna mostly control everything.  Screen as 'most controlled' looks odd to me.  I'm not even sure why this needs a regulator, it's not a broadcast limiter meant for a transmitter, and plenty of comps like this don't have one at all. 

October 1954 Radio-Electronics, page 77

https://www.americanradiohistory.com/Archive-Radio-Electronics/50s/1954/Radio-Electronics-1954-10.pdf

Hilton Remley as author is a straight up ringer for RCA, this is their circuit. 

 

[PRR]

On a pentode, the effects of G1 are proportional to Vg2, hardly at all to plate voltage. So it makes sense to regulate only Vg2 (and Vk).

It still seems odd and lame. The idle current in the 6SK7 must be very small, maybe less than 1mA. The current in limiting must be even smaller, so the limited output can not be large. The graph shows -17dBm, 110mV in 600 Ohms.

The stated purpose is to trim "several dB" in commercials. By that goal, it is not intended to limit uncontrolled music.

FWIW: despite his assertion, *everybody* had limiter amplifiers by 1954. RCA was pushing them onto stations just as hard as GE and Gates. A good limiter would double your audience area and could nearly double your ad revenue. The FCC was attentive to repeat overmodulation. (The deal with commercials is that they are mixed hot and often pre-compressed to be SURE they cut through.)

 

[emrr]

FWIW: despite his assertion, *everybody* had limiter amplifiers by 1954. RCA was pushing them onto stations just as hard as GE and Gates. A good limiter would double your audience area and could nearly double your ad revenue. The FCC was attentive to repeat overmodulation. (The deal with commercials is that they are mixed hot and often pre-compressed to be SURE they cut through.)

Oh yeah, everyone had them by 1940.  It was suicide NOT to have one by then.  WE, RCA, Collins, Gates, lots of home made. 

 

[davemascera]

Nice drum choice in the video.

Cannikin's story seems to have changed between OP and sale.  Who knows. 

OK, if I'm reading this right, the 12AT7 feed is still coming from the same place.  Take it from the output of the 6SN7,  a place with more signal level.  It may need more B+, from lower value plate resistors, like down to 33K, which'll draw a lot more current and give greater drive power, consider the cathode resistor again.  It may need voltage divider attenuation on it's input if the other places are too hot, could be a separate threshold pot. 

What orientation is the A-18?  I'd run it with the higher Z side as primary.  Consider the original circuit calls for a A-27 with 100K primary.  The load affects compression range. 

Attack control, not sure what you're describing, doesn't sound right.  It should be series resistance between the 6AL5 and the other time constant parts.  Look at the Altec 436C.    You're wiring a rheostat, so connect wiper to one side for securities sake. 

Where is 6SN7 B+ tapped from?  Should be right off the output of the PSU, straight to the transformer primary CT, to the 6SN7. 

The series resistor into the 0D3 can be larger to lower current.    The implementation strikes me as odd.  Many times a regulator is used to control absolute threshold reference, and/or GR stage plate voltage, here it appears to be primarily controlling screen voltage, though admittedly with a 2K series resistor it's gonna mostly control everything.  Screen as 'most controlled' looks odd to me.  I'm not even sure why this needs a regulator, it's not a broadcast limiter meant for a transmitter, and plenty of comps like this don't have one at all. 

October 1954 Radio-Electronics, page 77

https://www.americanradiohistory.com/Archive-Radio-Electronics/50s/1954/Radio-Electronics-1954-10.pdf

Hilton Remley as author is a straight up ringer for RCA, this is their circuit.

So I just said f**k it and upped the voltage back to 246... hitting the 6SN7 with 230 now too. I don't really care if the OD3 is a bit off. It isn't getting that hot with the non-rugged one. Cathode is ~45 mA now once I let it settle down. Whatever. And there was a loose solder point on the attack pot.

I don't know if this makes sense, but the gain reduction and clipping has been functioning much better since I did that (as if following the circuit as written might be a good thing...).  So I'm gonna do some tests on some actual tracks and see what happens before I touch the circuit further.

I am using the A-18 'backwards' meaning the primary on the circuit is 80k and the secondary is 10k. The B+ for the 6SN7 is indeed coming through the Edcor output.

The attack pot had a loose solder point. It works now, but also reduces gain reduction in total. It's not like the average modern compressor in that sense. I'll try the way you described and is on the Altec and see what happens.

I've been tempted to try the cathode follower since I accidentally drilled an extra hole in the chassis, but I'm a bit burnt out at this point. If I am dissatisfied with the level I'm getting I'll give this idea a shot.

I have 120 Hz hum in this build. Likely because the output wire passes directly over both rectifiers. Too lazy to change it cause it's far too quiet to hear over any reasonable signal level.