Trimax A30 Limiter , what do the tubes do in this circuit ?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

gary o

Well-known member
Joined
Dec 28, 2004
Messages
1,531
Location
uk
http://www.retrovox.com.au/trimaxA30.pdf

Im trying to understand what each stage of this limiter does, if anyone could shed some light...but be gentle with me i may not be a technical as most.

Can i begin by saying what i think each bit does & maybe someone or someones can correct me and fill in the gaps

seems like the T1 & 1620 on left of schematic is a little pre amp

C7 of input amp is feeding the grid of 6SN7 below which Im told is a phase splitter ?

the next 6SN7 Im told is cathode follower ?

6X5 is rectifier

So is this little circuit basically an amplifier fed via the input amp (1620) C7 to provide a control voltage coming from the daul diode 6X5 ??

Going back to the output of the the 1st amp(1620) OP via T2 thru time delay circuit T3

So now the side chain is making the control voltage before the actual signal arrives at the output

The 6SN7 after T3 is the limiting part of circuit 7 control voltage fed via the 6X5 rectifier...tho I dont understand how it controls the level here ?? ....Help!

Is R32 R33 make up gain level of the push pull make up amp 2X 6SJ7s plus one last 6SN7 T4 OP tran

Silly question here what is EM bottom right of schematic ?? meter ?? & R46 ? is that for zero

I am looking around for threasholds and ratios & time constants theres a little switch by the 6X5 wonder what that does do ? ah I think thats control voltage out for another unit..

Im supposing the side chain is taken from the input amp because thats before the time delay lattace thing.
There the output amp does have to be a push pull as in usual vari mu but I spose push pull is less distortion type amp

I notice the OD3 voltage reg feeds vari stage and side chain

Seems the control voltage comes from the cathode ? of 6X5 in usual vari mu comes from anodes ?? I dont understand much about how tubes actually work yet sadly

Does anyone know what these limiters sound like ?

Thanks for your patience would love to hear any thoughts on this one cheers.
 
The second tube looks like it is being used in the classic 'variable-mu' arrangement.

Take a look at the PRR vari-mu thread for a masterful description if how this principle works. -In addition, read the notes preceding the schematic diagram, in particular the bit about 'thump' feed-through... it ties in with PRR's explanation.

Does anyone know what these limiters sound like ?

From the specs... they don't claim much. 10kHz frequency response is noted. -It's described as suited AM or FM broadcast, AM doesn't do anything over 5kHz, FM is dead by 15kHz (19kHz and above is where the L-minus-R stereo information is located, so L-plus-R mono has to be silent by then, with a reasonable 'guard' margin)

Assuming you can get some transformers and feed/load them correctly, you can get the response up to whatever you want. (Can you really get those exact transformers these days, even if you wanted to?) but by the time you've changed the transformers that much, why not just build PRR's Vari-mu, learn the operational aspects and if you REALLY want to be clever or build a higher-spec unit, design and build your OWN variable-mu compressor...

Keith
 
Hi Kieth thanks Is the 6SN7 a remote cut off tube ?
I have a reasonable understanding of the usual vari mu circuits but this one doesnt look usual to me ?
 
Cool yeah didnt think it was the usual vari mu type...makes me very interested as to how this type sounds.

Thanks
 
The designer was grinding several axes. But in the end it is just a broadcast safety limiter, and was not widely copied.

There is always some attack time (otherwise it would just clip brutally) so he added a delay line.

Incidentally this demands a feed-forward topology.

He also dislikes the existing gain-cells, for reasons outlined but not fully explained in paragraph 2.2. There is an astonishingly complete bibliography on page 9.

But theory goes out the window when you find a happy simple answer.

> looks like it is being used in the classic 'variable-mu' arrangement.

Only backward.

The classic affair reduces current to reduce gain. As a limiter, this means minimum current at maximum output, which is just a bad idea. This is why the best of the breed run very high idle current. Even so they always run out of steam.

Note that the A30 takes a positive control voltage. The variable tube idles at low current. When big stuff happens, its current is increased. This seems to be more appropriate for the high level out of a limiter.

So as input rises, V2 grid voltage rises, V2 current increases.

> I dont understand much about how tubes actually work yet

You should probably grok tubes a LOT better before you get into tube limiters. Simple fixed-gain tube work is hard enough, and befuddles many designers; controlled variable gain is another level above that. Any mature tube man will grasp the basic explanation in paragraph 2.2. A tube is a resistance. We often use a Plate Resistance number for quick estimates. The cathode also has an impedance. It can be figured 1/Gm, or it can be figured as Rp/Mu. You need to reach a point on your path where you can see that these two expressions say the same thing!

6SN7 data page 5 shows that tube "constants" vary with current. You can derive the curve of Cathode Impedance from this data. For a very hasty plot, divide Rp by 20, the nominal Mu. Obviously Rk is infinite at zero current, high at low current, and goes toward 400 ohms at high current. The A30 feeds the cathode through 5K. So for V2 current of 5mA or 10mA we can get substantial attenuation, 5400/400 or 23dB. If V2 is flowing 5mA, then 5mA of signal will distort but 1mA of signal may be quite clean. Since this comes through 5K, we can put quite substantial voltages through T2 without distorting in V2.

What is the current in V2? Here we hit a happy trick. Once V2 flows enough to matter, current is largely fixed by R21 R27 5K resistors. So current is nearly proportional to input amplitude. This seems like a good way to make a feedforward limiter. In fact it is wrong because the impedance of V2 is not proportional to current. The happy accident is that he found a set of resistances and voltage levels for which multiple non-linear effects can be made to cancel over almost 10:1 of input range. It truly is a cancellation: if you get a bit off, the limiting level drops as input increases, which is usually wrong. The result is S-shape and the best you can do is line up the two inflections for a semi-flat portion. As input rises past the upper inflection (V2 socked to maximum current), output level and distortion rise VERY fast. This works when station operators respect the system and attempt to keep levels nominal: the limiter catches slip-ups to 16dB over maximum. In rock-tracking it will probably be a nasty BLATTT box.

> Is the 6SN7 a remote cut off tube?

No, and we do NOT want remote cutoff in this plan. Everything is backward. What we want is infinite saturation. Sadly, no tube can do that.

> the side chain is making the control voltage before the actual signal arrives at the output

This is a frill and a gimmick. Over-clever idea. Look-ahead limiting came into its own with digital audio. Implementing delay in analog is either way too costly or has no audible effect (except this plan appears to have massive phase-shift in the top octave?).

> FM is dead by 15kHz

This box predates Stereo FM. Mono FM can exceed 15KC. (So can AM.) The 10KC spec reflects the fact that AM was still king and most transformers had uncontrolled >10KC response. Also that the "delay" is aimed between AM and FM conventions.

> I am looking around for threasholds and ratios & time constants

Time constants are, obviously, all those Rs and Cs around V7 and V8. Two time constants and some back-bias. If you abuse it, there is another time constant as C13 C18 poop-out.

Threshold is built in, by 6SN7 Mu and by regulated supply voltage. You can move levels around with pads... there has to be some pad ahead of this box (probably sealed), R32 R33 set the output level to suit your transmitter.

There are no knobs because this is NOT a creative tool. It is a safety valve. Ideally it never does anything. When it is called upon, it should do the minimum possible harm. When it does get called to action, we do not know what sort of event will be going on, and it should not happen for long, so there is no notion of "tuning for best effect". Gross overload "shouldn't happen", indeed inputs 20dB above nominal max may be impossible (clipping in master board), so gross behavior in extreme abuse is acceptable.

> very interested as to how this type sounds.

If it has "a sound" in normal operation, flat or light limiting, that's just wrong. I bet it "sounds" a lot like a Volumax, neither crystal clear nor obviously flavored. In heavy abuse with pathological inputs (drum solos) it probably "does something", and different from all others.... but who wants to build all that and then find out it just sounds bad on almost everything?

The only tube-y part is V2. All else could be implemented in transistors. Not trivial: the input audio must float, the control voltage can be quite large and may not be easy with common chips, etc. I had whomped-up a plan with a bunch of transistor pumps, and one tube, but it failed to hold interest, and was probably lost in the last drive crash.
 
Thanks for yr knowledge PRR fantastic information...as always. I will read over & try to digest...as you say I need to learn more basics abouts tubes..Im a solder it & hope kind guy I learn that way & have ended up with audio boxes I could have only dreamt of couple years back....Iv also fried a few components here & there.

I did bread boarded gain control circuit (v2 I think not lookin at schem now ) I feed it control voltage from my DIY BA6a I didnt know the control voltage was positive in this circuit....prob just one of the reasons it didnt do much I think it made nasty noises if i remember correctly

I also asked lots a questions.. & new york dave gave me some great info on the spec of the inductors, I fiddled with Inductors and capacitors to try & make the delay circuit connected between my various pres & recorded results couldnt get any audioble delay

Like I say I enjoy solder & see, would still be interested in having another go at the gain control part & possibly the side chain...maybe i could use amps I already have to generate the voltages

Im currently testing some 6AB6s triode wired into the BA6A instead of its original 6SK7s croc clips all over the place ....sounds good but 6SK7s some better....designed for the rest of circuit I spose

Thanks.
 
The 6SN7 after T3 is the limiting part of circuit 7 control voltage fed via the 6X5 rectifier...tho I dont understand how it controls the level here ?? ....Help!

Think of the 6SN7 as a dual voltage-variable resistor.
Call the "lower" half of the tube R1000 and the "upper" half R1001.
The 6SN7s grid voltage (from 6X5), controls the resistance between the 6SN7s cathodes and plates.
The push-pull signal from T2 enters thru the right sides of R21
and R22 and exits at the junctions of R21/R1000 and R22/R1001.
The "right/left" and "upper/lower" above refers to the layout on the PDF.

The following diagrams are HIGHLY simplified!!!
SIGOUT+ drives V3 > V5a, SIGOUT- drives V4 > V5b in standard push-pull.
When 6SN7s grid voltage is low, the A-K resistance is high, giving a higher output voltage.
100K100K.GIF


When 6SN7s grid voltage is high, the A-K resistance is low, giving a lower output voltage.
100K10K.GIF


And it's a FEED-FORWARD compressor!!! The signal is detected BEFORE
any gain modification.

Thanks for the PDF
I O U A :guinness:
 
> The following diagrams are HIGHLY simplified!!!

Thanks.... I was too lazy to draw.

Here's more realistic values.

ery5pi.gif


Your R1000 R1001 is, as you say, the 6SN7 Cathodes. I would see the "cathode-grid" impedance, rather than "A-K" resistance. True, the current must come from plate. But action is relative to grid voltage.

The cathode impedance is 1/Gm. The GE sheet shows Gm varying from zero at zero current to around 2,500 uMho (2.5mS) at high current. I didn't ballpark the minimum A-30 current, but I am sure they idle the tube so Gm is way under 200uMho or way over 5K ohms, "infinity" for practical purpose. At the other extreme they can get 50V control, most of that across the 5K resistors, so up to 10mA, Gm=~~2,500uMho, Rk=~~400 ohms.
 
Yeah, I was trying to keep it ultra simple hence the arbitrary R values.
The plates are "near" AC ground via R57 (on the PDF) and the PSU filter caps and as the grid voltages go higher, the A-K resistance/impedance (because there's AC and DC) go lower, shunting more signal to "ground".

Either the CBS Volumax or Audimax did a similar type of shunting audio signal but with a solid-state diode bridge where the DC current thru the bridge was varied from the output of the detector. It didn't sound great and the gain reduction range was low and got really nasty if overdriven.
(Nasty in a bad, non-artistic way).

I would like to hear one of these things though.
Maybe Started/Incomplete Project #11.

:cool:
 
It was the CBS Volumax 400 and it wasn't a diode bridge.
Just two back-connected pairs of diodes where the current through them
from the detector modulated their dynamic resistance/impedance thus
changing the amount of signal loss.

schemo: http://www.freeinfosociety.com/electronics/schemview.php?id=313
 
> Volumax ... it wasn't a diode bridge.

It is a 4-arm bridge. Two D, two R. (Each "D" was physically two diodes to get the awkwardly low knee-voltage up.) It could have been implemented as an all-diode bridge. But diodes cost money! (Then.) And to inject control current they needed either a transistor current source (real money!) or large resistors. So let two arms just be resistors.

> The plates are "near" AC ground

So are the grids. Both by being tied to the large time-constant capacitor, and because they are tied to each other across a balanced source.

I like to see it as cathode-grid impedance 1/Gm. True, it is equally valid to see Rp/Mu against plate voltage. Since there's no big audio difference between plate grounding and grid grounding, it does not matter. (Except Mu is not dead-constant at low current, so Rp/Mu is more squinting and math-making than just 1/Gm.)

> Started/Incomplete Project #11.

Only eleven? You can do better than that. I lost count long ago.
 

Latest posts

Back
Top