compressor, EQ, phase rotator, - what order?

[Opcom]

Well time is coming to start hooking things up for testing. I could use some advice on which order, from any radio station hardware hackers out there.

Here's what I would,like to use:

 -MIC (RCA-by-Turner desk mike from VHF FM base station - do not laugh, it actually sounds very clear on voice)
 -MIC preamp/mixer (Altec 4-channel, serves just as the mic preamp, and has a VU meter.)
 -Phase rotator (symmetra-peak, passive)
 -EQ (an old metal-cased Pioneer home stereo unit)
 -Compressor (RCA BA-6A)

I think the right order is:
MIC -> MIC amp -> phase rotator -> EQ -> Compressor -> modulator.

The brochure for the phase rotator says to put it before any AGC amp.

My confusion is as to where the EQ goes. On one hand, it would be useful to set frequency limits for the passband before the compressor. On the other hand, it could also help remove objectionable artifacts from the compressor's output, if any. Another mystery is that the EQ might be sloppy on phase, and what might that do WRT the efforts of the phase rotator? This question might be un-answerable, except to test.

The schematic of an EQ that uses the same design is attached, in case someone can comment on what happens to phase in that kind of EQ. I beleive it is the "Baxandall" design of EQ. All I know about the EQ is that it still works well and was part of my stereo back in the '80's.

The compressor is straightforward, RCA BA-6A, acting on both the positive and negative sides of the signal. It looks like it is not so much a clipper/limiter but a real compressor/AGC amp. (attachment).

I knew this would happen. Too many thingmabobs, which order to put them in?

[K1JJ]

In my opinion, your line-up looks OK as you have it.   As for the phase rotator, I've never used one, so no experience with it. Let's see what others have to say about that.

Yes, put the EQ before the compressor.  I find that very useful for compressing some added low end boost that wud normally come out muddy when the EQ is after the compressor.  Plus, some form of limiting is good just before the transmitter for best peaks control.  My Behringer 9024 has a 6-band compressor  AND a limiter on the output. I set that limiter on both AM and ssb and find it impossible to overmodulate or overdrive the linear, as a result.

T

[KB3DKS]

 Well, it's early in this thread and you will get swamped but here goes a couple points in my opinion.
  The BA-6 is fine where it is since it was commonly used as the driver out of the Studio to the transmitter site. Lots of possible gain and output level along with nice sounding iron. Very smooth even with serious compression.
  The home stereo EQ has one major shortcoming... it is a -1o level, high impedance
operating device and will not interface properly in 600 ohm signal chains. Matching transformers would help but I don't advise it. Especially driving the BA-6.
  The passive ?? phase rotator might need to see a proper matching impedance to work properly so that is a consideration.
  The Altec mixer/preamps are fine but it is real easy to build up a better sounding simple Mic pre and eliminate the unneeded circuitry. You should be able to find a passive VU meter that can be bridged anywhere in a 600 ohm line for level indication.
  I won't get into microphones as there are just too many good sounding, and bad sounding ones out there.
  What I did to confirm the quality of my audio chain is to simply plug a set of 600 ohm studio phones at the output of the final stage and listen and scope out the results. If your modulator is reasonably flat then what goes into it will come out of it.
  That is another story itself.

My chain is an old RCA SK-46 ribbon, a Marti 20-H program amplifier for about 50db gain,
a JBL Urei 7110 limiter, a Urei 546 Dual Parametric EQ, and a CBS 4000A Volumax that drives a reverse connected line output transformer to the grids of the 6CA7 modulators of my VKII.
Everything is transformer coupled with the output shield floated. No hum or ground loops.

Next........

Bill, KB3DKS in 1 Land

[KC2UBF]

Personally, I think the best sounding way to do it is to split the signal before the compressor, have one split go straight into the compressor inputs, and the other split go to the eq inputs.  then take the output of the eq and directly feed the compressor's detector with that signal.  On most compressors the detector input is called the sidechain or 'key' input.  In my opinion that's the best way to do pre- compression EQ, as it provides you with the most control over the compression and leaves the original signal purest. (the EQ is just adding control to the compressor, and not really acting as a normal EQ anymore.)  If you do this, and still want an eq in the chain to be used in a conventional sense, i'd get a second one and put it after the compressor.

this is a method that many mastering facilities use... if it doesn't work for the radio i'm sorry.  I'm new into radio, but definately not new into audio engineering.

hope it works out

[K1JJ]

Personally, I think the best sounding way to do it is to split the signal before the compressor, have one split go straight into the compressor inputs, and the other split go to the eq inputs.  then take the output of the eq and directly feed the compressor's detector with that signal.  On most compressors the detector input is called the sidechain or 'key' input.  In my opinion that's the best way to do pre- compression EQ, as it provides you with the most control over the compression and leaves the original signal purest. (the EQ is just adding control to the compressor, and not really acting as a normal EQ anymore.)  If you do this, and still want an eq in the chain to be used in a conventional sense, i'd get a second one and put it after the compressor.

this is a method that many mastering facilities use... if it doesn't work for the radio i'm sorry.  I'm new into radio, but definately not new into audio engineering.

hope it works out

I'd like to try that myself, but I don't have a 'key' or 'sidechain'  input on my 9024 compressor.  In this situation, do you recommend the compressor be placed before the EQ?

ie, I now have EV RE-20 mic > mic preamp > 31 band EQ >  9024 6-band compresser w/ limiting > into transmitter.

Do you feel I should swap the EQ and compressor positions?

T

[kc2ifr]

Tom,
The compressor/limiter should ALWAYS be the last device before the transmitter.
Bill 

[KA1ZGC]

Tom,

With what UBF was describing, the EQ would actually be in parallel with the compressor.

The compressor normally splits the audio into two paths internally, one is used to derive the controlling voltages for the other. With sidechaining, you're doing that external to the unit, and seperately shaping the audio that is used to derive those voltages. You feed both units with identical audio, and the output of the EQ feeds the AGC input seperately.

It's one way to shape the response of a monoband compressor. Since you've already got six bands, it might not be worth the effort.

[K1JJ]

Bill and Thom,

Tnx for the info. Looks like I'll keep things the way there are.  Always looking to try something new...

[KB3DKS]

My chain is an old RCA SK-46 ribbon, a Marti 20-H program amplifier for about 50db gain,
a JBL Urei 7110 limiter, a Urei 546 Dual Parametric EQ, and a CBS 4000A Volumax that drives a reverse connected line output transformer to the grids of the 6CA7 modulators of my VKII.
Everything is transformer coupled with the output shield floated. No hum or ground loops.

I have the 7110 limiter ahead of the EQ to make up a little more gain for the ribbon mic and just use  a 2:1 ratio with only about 6 db limiting on peaks. Otherwise would put it after. With a normal mic or proper mic pre I would place it after the EQ to utilize the adjustable average/peak function control. Would also insert the other half of the 546 parametric in the sidechain. The 7110 is a very desirable limiter for Studio use and in my opinion from longtime experience sounds much better than the DBX and Symetrix products of the same period. Well worth picking up if it can be found reasonable.
I also mainly use the 546 for it's variable hi/low pass filters with only a small amount of presence boost EQ engaged. I keep a reasonable bandwidth to make the best use of the PW 100 watts I have to work with. Hopefully this all comes out ok from your receivers.

Bill, KB3DKS in 1 Land

[WBear2GCR]

well, this is a trick question, imho...

There is a difference between a compressor and a limiter.

There are also two types of "compressors" - if you look in the Handbook section there is a nice discussion of them...

The limiter keeps the max level from hitting "O vu" (whatever value you set as your max).

The compressor tries to keep your average loud to soft "even" - or more even than nothing, depending on the amount of compression you dial in.

So, if you back OFF the mic, it dials up the gain, and if you move in it dials it back.

Whereas the EQ knows nothing about all this, it merely adjusts the gain of different frequency ranges. It does alter the phase within the passband, but we can ignore that mostly.

Here's why it is a trick question... IF you put the EQ before the compressor, then the compressor tries to level the amplitude of the signal - if it is NOT a multiband compressor, then it finds the loudest part of whatever you've sent (it does this anyhow) and based on it's time constants and "detection" method (fast/slow, weighted, etc.) and tries to set the output as a constant level. If it is multiband then it sets each band according to the band's detected signal levels. How that all comes out depends a great deal on the EQ applied and the innards of any given compressor.

But, if you put the EQ after the compressor, then your max level is going to be boosted by any amount that the EQ adds boost in a given band. Otoh, if you only CUT instead of boost to get your curve, then you don't need to worry at all about "too hot" a signal at the end of your chain...

I think the phase rotator belongs in the penultimate position - last or next to last. After you do all the dorking to the signal, then you want it properly phased, if you phase rotate it then compress it and EQ it the phase information is doubtless altered. Plus you also have to keep track of the absolute phase of each piece in the chain otherwise.

The last position, I'd want to be a soft knee limiter, to keep from going "over" and ever clipping your modulator... ( so I go with IFR here, but imho you want the limiter not a compressor...)

              _-_-bear

[k4kyv]

I think the phase rotator belongs in the penultimate position - last or next to last. After you do all the dorking to the signal, then you want it properly phased, if you phase rotate it then compress it and EQ it the phase information is doubtless altered. Plus you also have to keep track of the absolute phase of each piece in the chain otherwise.

The last position, I'd want to be a soft knee limiter, to keep from going "over" and ever clipping your modulator... ( so I go with IFR here, but imho you want the limiter not a compressor...)

I agree.  I would put the phase rotator just before the peak limiter, which is what should feed the audio to the transmitter.

My limiter has two stages: one is called RMS limiting, and the other, Peak limiting.  I think the RMS limiting is basically a form of compression with slower attack and release times, designed to bring up the average signal level, and the peak limiter has extremely fast attack time and maybe a slightly slower release time, designed to keep the modulation from exceeding a certain threshold, which on my unit, is separately adjustable for the positive and negative directions.

I assume the phase rotator adjusts the audio for symmetrical peaks, which would be highly desirable with a transmitter that has little or no headroom beyond 100% positive modulation.  If the transmitter has the undistorted headroom, a reasonable amount of asymmetry (up to 125-150%) actually allows for a slight increase in the average modulation percentage without breaking across the 100% negative modulation barrier.  Those high positive peaks are normally occasional and of short duration.  One drawback to the extra headroom is that this makes the modulator stage run at a lower average efficiency, so the tubes will run a bit hotter.

You say the phase rotator is a passive unit.  I wonder exactly how it works.

[KB3DKS]

Don and all,
Phase Rotator info ++

http://sharon.esrac.ele.tue.nl/mirrors/w3am/8poleapf.html
http://masterdigital.com/24bit/radioprocess.htm
http://www.icycolors.com/nu9n/images/Broadcast-Chain-Tutorial.pdf

Just Google Phase Rotator for more.

Bill,
KB3DKS in 1 Land

[Opcom]

wow it really is a trick question. makes more questions and more to think about. I do have Hi-Z to 600 Ohm matching transformers, but maybe it would be better to go to the music store and find a used EQ with balanced I/O. They cost about $200. Maybe later for that..

On the rotator - it's a symmetra-peak, and is described on W3AM's website http://www.w3am.com/
Pics of mine attached. - dug them out of an old AM station along with a BTA-250.

I got some time to work with the modulator yesterday. One nagging thing is that if you have the modulator up first, you need to know what it will do, so as to determine the parameters for the modulated stage. This is bass-ackwards from what most people do. Most build the RF stage first, then apply a modulator to it. In my case, I am "stuck with" certain modulator hardware and don't really know what it will do into a load. One can attach all kinds of resistive loads onto a modulator, many will work, some wont, but no matter what, the modulator needs to see a certain load to work best.

I came up with a simple way to test the modulator into various resistive loads. This involved some large resistors, etc. but it worked FB. The only issue was the tedious calculations to determine what the power output was with each load and at different plate voltages. So I made a spreadsheet that lets me put in the load resistance, a small tap at the 'cold' end, and convert the RMS voltages measured there to peak-to-peak voltages and RMS wattage. Another sheet in the same book lets one enter the p-p modulation transformer secondary voltage at 100% modulation and the expected PA current under carrier conditions, and provides a whole bunch of figures for the RF stage behavior.

I still do not know how much my modulator can do, because the temporary driving amplifier is a bit short on power and I had to short the modulator bias supply's output to GND since I have gone to 3-500Z's instead of the 304TH's (the old arrangement was AB1 and the supply has too high an impedance for 3-500Z's). The MOD is supposed to be able to be driven with <25W, but the temporary driver amp isn't doing it that well. Also, it blew an electrolytic and I had to repair it.
This was hair raising as it happened during the first test, about 1 second after I turned on the modulator B+ - 3KV. I didn't think I could turn off so many switches that fast. Anyway the MOD had no problems making 500W, but that is as far as it would go with the 30W hi-fi amp. I may have an impedance matching issue to work out. I did note that the frequency response of the modulator tapers off fairly linearly to about -6dB@5KC. The transformer is thought to be from an RCA KW rig that used audio feedback, so will consider that later.

Fun on Saturday.

[Opcom]

Here is the spreadsheet. It is disguised as a DOC file since zip files cannot be uploaded.

[KA1ZGC]

There's a little voice in my head saying you should go with a low-mu device. It never tells me why. Got any 304TLs? You could easily modulate a 4-1000 or similar with that arrangement.

[Opcom]

Driving the modulator: (everything in brown is what I have calculated)

I made an error - -the 304's used were TL type with amplification factor of 12 rather than the 304TH amplification factor is 20. So the OM got it right. 3-500Z have an amplification factor of 160.

I'm going to stay with the 3-500Z's at this point since that modification has already been made (sub-chassis, blower, etc.). I changed because I did not have a good pair of 304's, and they are now rare and too costly and I kept getting cheated by the COBs, who sold me "display versions"

The mod driver inside the XMTR is a 40 watt UTC CG-512. (+/-1.5dB 40C to 10KC). The data was sent to me by someone on the board, quite some time ago.
Secondary  G-F-F-G 
500 Ohm Input  Connect to 1 & 5  Join 3 & 4 = 1:2.5 (secondary is 3125 Ohms. Does the ratio indicate the full secondary G-F-F-G or just half of it?)
200 Ohm Input  Connect to 2 & 5  Join 3 & 4 = 1:3.8
50 Ohm Input  Connect to 4 & 5  Join 1 & 4 , 2 & 5  =1:7.5   

The difference is some grid current, and also the peak grid to grid voltage. For similar conditions, the three types at 3000V plate voltage call for:

304TL (class AB1) (bias -290V) (peak grid volts per tube 290) (drive 0 W/ Ig 0mA) (plate (2 tubes): 130/444mA, 12000 ohms CT, 730W out)
304TL (class AB2) (bias -290V) (peak grid volts per tube 390) (drive 55W/ Ig 199mA/ 1385 Ohms CT) (plate (2 tubes): 130/800mA, 9100 ohms CT, 1800W out)

304TH (class B) (bias -150) (peak grid volts per tube 420) (drive 6W/ 20mA/ 14.8KOhms) (plate (2 tubes): 134/667mA, 10200 ohms CT, 1400W out)

3-500Z (class B) (bias 0) (peak grid volts per tube 100) (Grid drive 25W/ Ig 244mA/ 102 Ohms (408 Ohms CT)) (plate (2 tubes): 300/770mA, 8600 ohms CT, 1420W out)

There may be some dispute towards the Eimac datasheet claiming the 3-500Z 300mA resting current to be class B. I considered that AB2, but Eimac seems to be going by grid conduction angle, not plate conduction angle; one could argue that Eimac is always right.

The end of this tube drive numbers mess is that there now exits a very different drive requirement. 1/4 the originl voltage - very different. I am not sure what this means in relation to the original driver transformer except that it would not need to step up as much. It was never designed to drive 3-500Z's. I used a multi-tap 30W transformer to match the driver amplifier's 16 Ohm output to the modulator. This discrepancy between the apparent drive voltage and current requirements of the 3-500Z's vs. the origintal types (304TL class AB2) might also explain why I got more satisfactory drive from the same speech amp when putting the amplifier's 200 Ohm output into the 500 Ohm modulator input (making the mod driver look like 1250 Ohms instead of 3125 Ohms). More experimentation is needed here.

Some searching on easy drive for 3-500Z modulator was found here, although it will pain me to remove the 40W UTC driver iron to substitute something less classical:
http://www.hamelectronics.com/k1deu/pages/ham/transmitters/am/pages/line_to_voice_2-32.htm

I should add that I beleive this is worth studying because I used a 70V transformer to drive a pair of 6146's in AB2 directly from a car stereo 4 Ohm output and it worked FB, having a 60W/CH dynamotor-powered tube amp in the old van. No more van, and no room for it in the pickup truck.

IIRC the CT on the "70V" side would be the tap that provides 1/4 the wattage (1/2 the voltage). So on a 32W transformer, the 8W tap would be CT and you would have 32-8-0 for the winding. However, the aforementioned transformer is too small to have the fidelity I am looking for so I'll look for others, and at the same time, spend some time getting the 100W speech amp going. I've learned that using a 4x or larger amp allows a fair mismatch without increasing distortion because it can supply more voltage, or more current, than necessary at reduced ratings. This claim was also 'proven' in practice by the use of a 175W amplifier to cathode modulate a 250TH. The match was not very important as long as there was enough voltage.

Any comments on this insanity are also welcome.


silly but unrelated:
http://www.geocities.com/matt2kw/BLACKCAT.html
http://grumpy.proboards.com/index.cgi?board=AMPS&action=display&thread=4293

[KA1ZGC]

There may be some dispute towards the Eimac datasheet claiming the 3-500Z 300mA resting current to be class B. I considered that AB2, but Eimac seems to be going by grid conduction angle, not plate conduction angle; one could argue that Eimac is always right.

Or that they're talking about the class the tube is being driven in, which means you go by grid conduction angle.

[Opcom]

I would agree with that. Another old Eimac datasheet with the same specs for the 3-500Z calls it AB2. Might have been changed one way or another for marketing purposes.

In the meantime this little figure for so-called line matching transformers came to mind.

[KA2QFX]

Wow, lot's to work with here.  I agree with Bill IFR; a MONOband compressor is best left until the last thing in the chain.  The order you have now looks good to me. 

A note on your EQ however.  Very interesting circuit. The filters act like "tuned grounds" for the individual passband frequencies, providing a loss in feedback signal in the otherwise flat summing amplifiers. They are also lower order filters than usually seen in EQs. Therefore, where phase is most greatly shifted, at the passband edges of the filters, these filters offer less and less response at the summing amps.  I would expect this design to introduce VERY little phase shift in the overall response of the EQ.

Mark

[Opcom]

That's interesting about the circuit and phase shift. The only other kind of EQ I have seen makes use of an LC circuit in each band which likely messes up the phase quite a bit. Those were found in old car stereo equalizers too. Maybe I can try the EQ and see how it does. There is enough room inside the metal case to mount XLR jacks in the rear panel and a couple of those little inexpensive matching transformers inside. It's mostly empty space, large only so as to have big impressive hi-fi EQ sliders on the front. 8-)

[Ian VK3KRI]

Although I have used Graphic Equalisers in the past, and still have a 1/3 octave
unit in the rack here, I'd say that for tailoring vocal audio, your probably better off
with something that provides simple 6 , 12 or 18 db roll off and or  boost.

Its not like your trying to notch out the freq that the PA takes off at in some
acoustic nightmare of a venue (but then again...).

Also with analog eqs boosting or cutting a frequency band can have significant
effect on the adjacent bands . Trying to generate a known freq response by
sending noise thru a graphic eq and then looking at the output with a spectrum
analyser  is what made me move to simpler circuits.

Of course, the eq you have works better than  one you don't have so go ahead
and play with it until you find an eqp-1 to go with the ba-6a. 

BTW the opamps in that eq circuit are configured as 'gyrators'
which simulate an inductance  in series with the caps.

[Opcom]

Curses!! foiled again!! and after going though all these gyrations! I admit to having a few commerical-potted-transformer-like low pass filters designed to 10K, 50K or 600 ohms in and out. One's at 3500Hz, a few slightly above and below. I been collecting. You are right, but I have no audio spectrum analyzer, so I should sweep it and see what the response looks like, then judge.

I didn't see a schematic for the EQP-1 on the WWW. it would be nice to look at.