kahn Symmmetra-Peak

[k4kyv]

I have located a Kahn Symmetra-Peak at a midwest broadcast station.  The engineer asked me to make an offer. No guarantee that it works due to its age, sold as-is.  Anyone have any idea what would be a reasonable price to pay for one of those things?  I don't have a  clue what they sold for originally, nor what one would be worth to-day. I get the impression the station hasn't used it for a while and that it is just sitting in a studio rack or in a store room.

[Opcom]

brochure attached. There are two models with different frequency bandwidths, check which one it is.

I sold one with all the paint "missing" from the front panel but electrically super nice inside for $80. I think it may have included shipping but do not recall. That was a few years ago, shipping was probably cheaper than today.

For what it is, I'd offer $50-100 depending on condition because it's the real thing and they don't make them any more.

The patents have virtually the schematic, so if the internal box of potted entrails ever dies, it can be rebuilt in the same case.

It's very easy to test a Symmmetra-Peak with a sweep generator and a 2-channel or XY scope.

[W3GMS]

There is a modern op amp version available which is touted to do the same thing.  I forget where I saw it but it was an internet item.   

So based on the description, this would eliminate being able to run extended positive peak modulation and produce a symmetrically modulated signal.  This should help with the typical diode envelope detector distortion when extended positive peak modulation is being detected. 

I agree with the mentioned dollar value.
 
Joe, W3GMS   

[W3GMS]

Here is the op amp version along with some information about it.

http://www.w3am.com/8poleapf.html

Joe, W3GMS

[W4NEQ]

I'm thinking those things had the components potted up, and if the caps have dried out, could be involved to fix.  The op-amp versions are really simple.  You could easily ad one to your modulimiter.

Or the ebay version:

http://www.ebay.com/itm/Audio-All-Pass-Filter-Phase-Rotator-Module-Board-/150665396265?pt=UK_Sound_Vision_Other&hash=item23145b8029

[W8IXY]

I purchased one of those phase rotators as seen on ebay, installed it in the Symetrix 528 and use it for my broadcast radio and TV voice work.  I deal with broadcast audio processing as part of my job, and am very familiar with the "sound" of those devices.  For $10.00 it was a bargain.  It does remove most asymmetry from voice.

In broadcast AM, that is a good thing because you can then use a modern audio processor to add a controlled amount of asymmetry to the audio to maximize the "loudness" of an AM broadcast.   The FCC allows up to 125% positive peaks in AM broadcast, and that can add about 2db extra "loudness" compared to audio limited to 100% positive and negative.  If the transmitter/modulator can stay linear at that audio level, you can gain quite a bit of "loudness" with well controlled audio.

I designed all the audio factory presets for the Omnia ONE broadcast audio processor, and its quite effective in keeping the average modulation level as high as possible to overcome the inherent noise, which is getting worse every day, when receiving AM signals.

See you in Dayton this weekend.

73
Ted  W8IXY

[ke7trp]

I have the ebay version along with a chassis and parts sitting here on the bench. I need to finish it.  Maybe this week.  Should take an hour.

C

[k4kyv]

I'm thinking those things had the components potted up, and if the caps have dried out, could be involved to fix.  The op-amp versions are really simple.  You could easily ad one to your modulimiter.

Or the ebay version:

http://www.ebay.com/itm/Audio-All-Pass-Filter-Phase-Rotator-Module-Board-/150665396265?pt=UK_Sound_Vision_Other&hash=item23145b8029

I looked at the e-bay version, already assembled on a neat circuit board (if I built one from scratch mine would undoubtedly be dead-bug style construction).  The price is certainly right, but it's been a while since I did any buying or selling via ePay.  A few months ago when I purchased a small-order buy-it-now item, I was able to simply use a Visa card, just as with any other online purchase, but that was from a domestic US seller.  Now they want me to log in with a password (I think I have an ePay account with a password, but it is long forgotten), and apparently "via Pay Pal" means you also have to have a Pay-Pal account too.

Am I overlooking something simple and obvious, a way to purchase an e-bay item these days, using a simple credit card, without all the hassle of accounts and passwords?

[ke7trp]

Ask the Seller to send you an "invoice".   He can send one using his paypal.  With that, You can simply pay it using your card.

I had to send these invoice for years for people that did not have a paypal account. Recently, We added online ordering to my website to get around those extra steps.

P.S  Dont forget you need a -15 and +15 volt DC supply to run that board.  If you can build one great,  But that guy has a small PS board complete with filter/regulator that simply hooks up and powers the board. This will drive the cost up but overall, Its pretty cheap.

C

[W3RSW]

Reinstate a new ebay password. Several pay pal transactions I've made recently give you the option of entering your credit card as " guest" or using your PayPal ID and password.
I guess all receive their cut regardless.

[Opcom]

I like the old fashioned way. That's why I have kept the original unit. No power needed, no ICs to blow up.

[K5UJ]

I guess there's something here that I am missing.  I always had this idea that for AM, asymmetry was good and, provided the phase is correct, a desirable thing.  Now I'm getting the impression that symmetry is important. 

By "asymmetry"  I don't mean wildly high positive peaks (150% positive or more), but more like 120% at most.  Maybe that's the idea with the Kahn?  If a voice is naturally very asymmetric, to control it?

[k4kyv]

Thanks for all the input on the subject.

The unit would be useful with transmitters with limited headroom.  For example, my 8005/805 rig flat-tops beyond about 110% positive.  I suspect measurable distortion begins to set in closer to 100% positive. Limiting unprocessed positive peaks to just beyond 100% means that the negative peaks are limited to roughly 75%, depending on the voice characteristics of the operator. If you have a transmitter capable of 150% positive, you can exploit that to achieve more "talk power" with close to 100% negative peaks and positive peaks that go where they may. But with a transmitter of limited headroom, a more symmetrical modulation waveform means more sideband power for a given carrier level. Interestingly, the literature on the Kahn unit stresses that it is most effective with voice modulation, while it achieves limited effect with music, which tends to be naturally symmetrical from the outset.

If I were to eventually incorporate one of these units in my station, I would make it switch-able, along with an attenuator pad to keep the level the same in or out. It would make a good A/B comparison, kinda like the 160m inverted vee vs shaded dipole test. 

I like the Kahn idea of a completely passive unit, but the active circuit seems simple and cheap enough, so if I am able to acquire both versions, I might run tests to compare their effectiveness and reliability.

[K5UJ]

Okay I see.  Which brings up another interesting topic but probably for another thread:  What can be done to increase clean audio positive limits on a rig limited to 100% up and down?  I assume a rebuild of the modulator section is pretty much needed--separate modulator p.s., more robust mod iron...

Instead of a Kahn, maybe another method albeit more expensive, would be to employ a fast peak limiter driven by a compressor to get enough gain reduction to limit the positive peaks but allow the negatives to get up around 95%.

[W4NEQ]

I guess there's something here that I am missing.  I always had this idea that for AM, asymmetry was good and, provided the phase is correct, a desirable thing.  Now I'm getting the impression that symmetry is important.  By "asymmetry"  I don't mean wildly high positive peaks (150% positive or more), but more like 120% at most.  Maybe that's the idea with the Kahn?  If a voice is naturally very asymmetric, to control it?  

This is a good question, and it depends upon the rules of the game, and the capabilities of one's transmitter.

For minimum distortion, Bob Orban argues that symmetrical AM modulation ( + -  99% ) is best.  On the other hand, if clipping is used just on negative peaks after dynamic compression, there is an argument for less distortion letting the positive peaks reach the transmitter's capability.

BUT the primary reason that AM broadcasters use asymmetrical modulation is to achieve more loudness (translation: peak power) with higher positive peaks.  [EDIT: Actually more headroom to achieve more average sideband power.]

The RULES of their game restrict steady state carrier power, but allow positive peaks to 125%.

The RULES of our game restrict PEAK power to 1500 watts.   If you really want to operate according to the rules, and your carrier power is 375 watts, you cannot legally modulate more than 100% positive.

If you run 250 watts carrier, then you have more legal headroom for positive modulation - again with the same 1500 watts PEAK power restriction.

If you run less carrier, and more positive modulation, you'll have more loudness due to a higher proportion of sideband energy, and you're heading in the direction of double sideband, suppressed carrier.  And you lose some "quieting effect" of carrier.  It's assumed you are limiting negative modulation to less than 100% to prevent splatter.

BUT, standard envelope detectors start to distort past 90-100% so sync detection is required to receive it without increased harmonic and IM distortion.   Which undermines intelligibility.

On the other hand, modulating + - 99% with maximum carrier, results in the most compatible modulation - more receivers will demodulate it undistorted (+ - 90% would actually be better in this regard)  With less loudness.

BACK to the legality:  The unfortunate restriction of 1500 watts peak power is undisputed. The rules are mute however on the subject of what exactly defines a peak; Is it a pulse lasting 1 microsecond, or a millisecond, or a second ?  This matters a bunch because many modulation schemes produce significant short-duration overshoots, particularly without hard clipping.  In order not to overshoot in excess of the peak power limit, overall modulation must be reduced.  Complex modulation monitors have been developed for broadcasting which allow selectable integration time constants for the peak flasher.  In our case, I think most hams use a CRO to visually see peaks.

SO, it's a very mixed bag.

[k4kyv]

I see extended positive peaks more as sort of a safety valve to alleviate the distortion and splatter that would otherwise be caused by positive peak clipping, whether intentional via hard limiting/clipping, or unintentional, as a result of driving the transmitter beyond its positive peak capability.

Very little sideband or "talk" power resides in these short-duration positive peaks that exceed 100%, but with the naturally asymmetrical human voice and no audio processing (aka pre-distortion), limiting the positive peaks to 100% or less means the negative peaks are limited to somewhere roughly between 50% and 75% with most voices.  And those occasional peaks are of very short duration, meaning the main, energy-carrying peaks are limited to far less than 100%.  Therefore the average sideband power to carrier power ratio is very low.  By allowing the negative peaks to reach 100% and the positive peaks to go where they may, full modulation and maximum sideband power is achieved.  This  requires that the transmitter have the modulation capability to accommodate those peak without flat-topping, both at the modulator stage and the class-C final stage.

Most hams (and broadcasters) get away with flat-topping at the positive peaks because of the infrequency and short duration of the peaks, which somewhat limits the severity of the distortion.  This partly explains why many apparently  clean-sounding AM signals draw complaints from adjacent-channel slopbucketeers who claim the signal is "wide". (Of course the probability that those same slopbucketeers are generating precisely the same kind of distortion even though their signal may sound "clean", is somewhere near 99%).

Allowing the full natural positive peaks to be transmitted without distortion reduces this more subtle form of splatter, and makes the signal sound cleaner at the received end.  I would venture to guess that audible positive peak clipping distortion is more objectionable at the receiver than audible distortion caused by the weakness of the envelope detector in handling positive peaks beyond 100%.  I haven't seen any real test data on this, nor run any tests myself.  I would be interested in reading any technical articles on this subject if indeed any such articles exist, whether in a periodical or an engineering textbook.

A similar objective was proposed by O. G. Villard, Jr., W6QYT in the DSB reduced carrier system, using the "upside down tube" circuit, which was a partially un-balanced balanced modulator, as described in his 1948 article in QST. This topic eventually led to the re-introduction of SSB to amateur radio, something that had generated little interest in the amateur community when first proposed in the 1933/34 series of articles in R/9 magazine. Villard's original upside-down tube article, IIRC, was entitled "Overmodulation Without Splatter".  The intent was not to significantly increase sideband-to-carrier ratio, but to accommodate the kind of occasional short-duration negative overmodulation peaks that generate splatter, but severely limit the average percentage of modulation if the audio level is dropped back far enough to avoid.

Negative overmodulation and positive peak flat-topping generate exactly the same kind of splatter and distortion.  Villard's system greatly reduced this splatter by increasing both the positive and negative modulation capability of the transmitter beyond 100%.  Unfortunately, negative-peak modulation beyond 100% of the carrier with the DSBRC system produces a phase reversal in the envelope of the signal accompanied by a sharp "bump" at the negative peak, which generate severe distortion in the envelope detector.  Villard's objective was to reduce spurious distortion type splatter, on theory that some increase in distortion was preferable to overmodulation splatter.

Positive peak expansion and asymmetrical modulation achieve the same purpose as the upside-down tube circuit, but only in the positive direction.  This system takes advantage of the fact that most human voices are naturally asymmetrical in waveform, and if the transmitter has adequate positive peak modulation capability, one only needs to phase the audio line feeding the modulator in the proper direction.  To preserve this asymmetrical waveform, phase and amplitude distortion throughout the frequency range of the modulating signal must be kept to a minimum.  The Kahn Symmetra-Peak reduces or eliminates the asymmetry of the voice waveform by deliberately introducing phase distortion to the signal.  This allows near 100% negative peaks without the necessity of increasing the positive peak capability of the transmitter far beyond 100%, which is always an expensive proposition, and actually reduces the efficiency of the class-B modulator or the final stage of the AM linear or grid modulated amplifier.

As for the legality in amateur radio, this is another grey area in the ill-conceived p.e.p. rule, since this rule includes the qualifier "under normal conditions".  It can be argued whether or not these short duration positive peaks that exceed 100%, but which add very little to the average sideband power, comprise "normal operation", or if they are to be considered incidental parasites to the lower amplitude peaks of longer duration, which generate most of the sideband power and carry nearly all the intelligence-bearing information contained in voice modulation.

[k4kyv]

He sent me some photos of the Kahn units he wants to sell.  He says they are Symmetra-Peak 2000.  I did a Google search and found this on a BC e-mail forum.

"Around 2000 or so, Leonard Kahn started making these again without the rack case. He sent me one to try out and it was just the potted part with some wires coming out."

So, apparently these are newer and more compact than the old rack-mount units, but he didn't give any specs.  Doubt he still has the manuals.

 

[Opcom]

Wow. Modern units - you ought to get them. Or one. I didn't know Mr. Kahn had made any more. I think he could supply a manual if you contact him via his website.

The Symmetra-Peak works in an odd way where by the frequency dependent phase shift it moves the high frequency and low frequency components of waveform peaks relative to each other so they are not stacked on top of each other as is the usual case with the human voice. The S-P was not really intended for music, but for voice.

Okay I see.  Which brings up another interesting topic but probably for another thread:  What can be done to increase clean audio positive limits on a rig limited to 100% up and down?  I assume a rebuild of the modulator section is pretty much needed--separate modulator p.s., more robust mod iron...

Why not make the modulator better able to faithfully reproduce whatever has been fed into it. Ham gear modulators are not very good in this department. It might not really be necessary to replace iron but it might be important to increase the plate swing or B+ to the drivers.

Examine the old broadcast transmitter manuals for 100-1000W rigs and look at the way that feedback is taken from each modulator plate and fed in a negative manner back to the cold end of the input transformer. Many of them used the same scheme. Attached is the BTA-250 modulator schematic with parts values marked but every modulator is different. Those old designs were done that way because it worked and those transmitters apparently sounded good or no one would have bought them.

[k4kyv]

From what I have been able to find out about them, the older units used a transformer or two (input and output?), UTC LS series, I believe.  The newer units either work directly at 500/600 ohms in and out, or otherwise used miniature transformers potted inside the modules.  They are supposed to dig them out at the station, so maybe I'll know more about them after the first of next week.

I'm leaving to-morrow morning for Dayton.

[K5UJ]

This is a good question, and it depends upon the rules of the game, and the capabilities of one's transmitter.
....[etc.]

That's a nice lucid statement of some obvious facts (the power limit part has been argued elsewhere  ) but it's not an answer to my question, probably because I didn't make it clear enough for which I apologize.   
I gather the point of the Kahn circuit for hams is a higher average without putting the ham rig mod iron and other components at risk. 

[K5UJ]

I did not know about the modern 2000 version either.  I wonder if it is also potted.   

[Steve - K4HX]

A higher average with potentially less distortion and/or processing required (no need to tamp down big positive peaks).

I gather the point of the Kahn circuit for hams is a higher average without putting the ham rig mod iron and other components at risk. 

[W2XR]

From what I have been able to find out about them, the older units used a transformer or two (input and output?), UTC LS series, I believe.  The newer units either work directly at 500/600 ohms in and out, or otherwise used miniature transformers potted inside the modules.  They are supposed to dig them out at the station, so maybe I'll know more about them after the first of next week.

I'm leaving to-morrow morning for Dayton.

Don & K5UJ,

I have a Kahn Symetra-Peak. The I/O xfmrs in that thing are not the UTC LS-series; I have no idea whom the mfr. of the xfmrs was, but you can be sure they are not the LS-series.

I have no idea how that misinformation got started about the transformers in the Symetra-Peak, and I have seen this numerous times before; probably some seller of the item (who used it for recording purposes, etc.) on eBay, as an attempt to enhance the value of the item.

73,

Bruce

[Steve - K4HX]

It's not like you would need LS series performance for amateur radio service any way.

[Bacon, WA3WDR]

It takes a lot of peak power capability to support high positive-peak modulation.  Pushed to its optimum, the positive peaks might reach 200% positive, requiring four times as much modulator peak power, compared to 100% modulation, and 2.25 times as much peak power capability in the modulated RF stage, compared to 100% modulation.

This means about four times as much transmit audio, of course, so asymmetrical supermodulation has been a popular technique in amateur and even commercial AM.  However, it requires remarkable transmit equipment for any given carrier power level.

What I have found in recent years, argues in favor of 'phase randomization', for the purpose of reducing modulation asymmetry.

1) It seems that 'phase randomization' increases 'modulation density', allowing less remarkable transmit equipment to be louder, without distortion.

2) More and more receivers today apply RF-peak AGC, even in AM mode.  This causes a perfectly clean, asymmetrically supermodulated AM signal to sound squashed, as though it had poor quality modulation compression.  The inappropriate compression actually comes from the receiver AGC attacking sounds having the most asymmetrical waveforms much more than sounds having less asymmetrical waveforms.

3) Amateur power regulations add peak power constraints that we did not face in earlier years.  Lower peak to average power in the RF envelope is more desirable today as a result.

4) Quality audio compression, and better sounding speech clipping (if used), is more easily accomplished when 'phase randomization' is used.

It looks to me that the 8-pole phase randomizer would do a fine job.  For best quality, the capacitors in the phase shifters should be mylar or preferably polystyrene or mica, and in particular not ceramic, tantalum or electrolytic - because significant audio voltage appears across them in normal operation, and they are not really 'clean' capacitors.  The values of these capacitors makes this an easy choice.  Also, the op-amps need to be of a type that has low common-mode distortion, because significant common-mode audio exists on their inverting and non-inverting inputs.  Fortunately, many such op-amps are available today.