Using piezo tweeters wisely: a "how to"

[Paul C]

The only piezo tweeters worthy of high fidelity or pro audio use are the Motorola/CTS units, and of those, only the Powerline Series are suited for pro audio. The ones made by other companies are junk. I have seen them selling for as low as 75c each to dealers. If you pay more than $2 for one made by anyone but CTS, you are being ripped off.

Sometimes it is said that piezo tweeters are "harsh" or "bright". First, if they aren't a real Motorola/CTS unit, they are junk. But with the Motorola/CTS units I have come to see (and others agree) that they have higher sensitivity than the published 92-93 db/wt SPL ratings. They seem to be 3-5 db higher.

I have used the KSN1165a, rated at SPL = 92 db @ 1w @ 1m, with woofers in the 96-97 db range and they seem to have excellent balance. Using the KSN1165a with a 92 db SPL woofer, the tweeter does sound very bright. By reducing the tweeters output by 4 db it matches the woofer very well and makes for an excellent speaker. (more on this later)

Ignore everything you have read about resistors and piezos. Most all are quoting each other and MISQUOTING the manufacturer. Some will tell you to place a resistor in series with the piezo tweeter, some say parallel. And both are wrong.

Years ago with some of the Motorola line, not all, just some, and I have the paper from them about this, Motorola recommended a 20-50 ohm resistor in series with SOME of the tweeters. This gave protection to the amp from high freq (100khz or so) oscillations as well as some protection to some of their tweeters. They did not recommend using resistors for any of the Powerline series, which include the KSN1142a, KSN1165a, KSN1188a, and a few other in the series.

(The Motorola piezo speaker division was purchased by CTS some years ago)

The CTS/Motorola Powerlines were the very high quality ones for pro use, and contained built in protection circuits, quite sophisticated. This protection circuit uses a resistor and thermal breaker paralleled with a tiny light bulb. At about 100 watts the thermal breaker heats up, opens and now the power goes through the light bulb. As the bulb heats up it increases resistance. What this does is give "soft clipping", that is, a sort of compression of the highs. This is very mild and almost inaudible, and certainly not noticable. They burn out about 400 watts. These are once again available through Martin Sound.

http://www.martinsoundpro.com/item.asp?id=340

Anyway, don't use a resistor with the Powerline Series piezos for power protection purposes; it is simply not needed. If you do put a resistor in series (and even Martin Sound says for Lpad use a 20-30 ohm resistor) I say, go ahead, but it will NOT make a difference in sound. Do you know why? Because a piezo tweeter has an impedance of 1000 ohms or greater. If you put a 20 or 50 ohm resistor in series with a 1000 ohm speaker, well, it is a miniscule change in power to the speaker. Duh!

So, how DO you pad down a piezo?

The CTS/Motorola paper says you can put a small CAPACITOR in series with the piezo to reduce the level, and that can work. But if you put an 8 ohm resistor across the terminals (follow this whole discussion now...) now it looks like an 8 ohm load. The piezo will not know the difference, it works off voltage only, drawing almost no power.



Now, put an L-pad to this 8-ohm load. Make sure you use an 8 ohm L-pad. (or if you use a 16 ohm L-pad, use a 16 ohm resistor) An L-pad is actually two rheostats, one in series with load, one in parallel, and the taper of each is such that the load presented back to the crossover is constant. Now this piezo, plus 8 ohm resistor, plus 8 ohm L-pad looks like 8 ohms to the crossover.



What crossover? Well, now that it is 8 ohms, it needs a crossover to keep from putting unnecessary load on the amp below the frequency produced by the piezo tweeter, power that would be wasted in the L-pad and resistor. A simple cap calculated for an 8 ohm load (or 16 ohm load if 16 ohm L-pad and resistor are used) is all that is needed. For example, for the KSN1165a or KSN1142a, which roll in at 1800 hz, if these were to be crossed in at 2000 hz it works out to 9.9 uf, which is a non standard value. But 10 uf is standard and plenty close enough. If you wanted to use all of the 1800 hz-20 khz available from that piezo tweeter use 11 uf (10 uf + 1 uf in parallel). If you wanted to cross this at 3000 hz use 6.6 uf (5.6 uf + 1 uf in parallel). Etc. Just remember, you can't go lower than the piezo tweeter is capable of going, but you can go higher. Calculate the capacitor using an 8 ohm impedance.



So now with this setup you can not only control the level, but control the crossover point, too.

You may also use a "fixed L-Pad" like this:



-3 db R1 = 2 ohms, R2 = 6 ohms
-4 db R1 = 3 ohms, R2 = 5 ohms
-6 db R1 = 4 ohms, R2 = 4 ohms

These are approximate, as in, "close enough for jazz".

One caution--Do not try to use a steeper crossover, 12 db/oct, or higher. This is not needed for tweeter protection, and the piezo can react with the inductor in these and produce chirps on occasion.

[Celt]

Good info Paul. Thanks!

[ozmoid]

Thanks, Paul! Excellent info, as always!

[crackerkorean]

I will be putting this exact info to use soon hopefully. I cant wait.

Thanks again paul

[GordonW]

There's more than one situation here...

Many people using piezos, are using the KSN1005 type... the "fluted bullet" type, like the ones in the Dahlquist DQ10, basically. These CAN cause amp mis-behaviour... they go to as low as ONE OHM at extremely high frequencies (ie, 100KHz region), and can cause amp oscillation. Especially with vintage amps (which have less high frequency stability), it's a VERY GOOD idea to use a series resistor. I've personally heard MANY cases, where a harsh "gritty" sound was ELIMINATED, by the insertion of even 4 ohms in series... this damped and alleviated the "resonant circuit" created by the capacitive load on the amp output. If using a KSN1005-type piezo with the above circuit diagram, I'd place a 4 ohm resistor (the value isn't critical, anything from 4 to 10 ohms would be quite reasonable) just before the driver... after the resistor to ground...

Also, with Powerine type piezos... many people wind up using the 1" throat screw-on piezo drivers, with various horn flares. In these cases, the Powerline drivers oftentimes wind up having TOO MUCH output in their lower registers, due to the "horn gain" (horns have a tendency to cause a "tilted down" response as frequency increases, due to their pattern narrowing at low frequencies down to the horn cutoff, since more of the sound is projected on axis and less off-axis). The way to correct this, is to use a low-electrical-Q quasi-third-order highpass crossover... a cap, then a resistor to ground, then another cap, then the L-pad/resistor/piezo driver. This can be tuned to give a pronounced "tilt-up" to the drive of the peizo, which cancels out its falling response. As for cap values... I would say, start at the value that gives the frequency desired for a simple first-order crossover, for both caps... and then tune to taste. It is OK to use different values for the first and second caps, if that works out best...

Regards,
Gordon.

[macaltec]

[Paul C]

Yes, GordonW is correct about using a resistor in series with the KSN1005 (and its near twin, the rear mount version, KSN1001).

You will find the 1005's mostly in older PA gear.

[Paul C]

Hey, Gordon, have you used the KSN1165a?



This is a PowerLine model that produces 1800 hz - 30 khz. It is rated SPL = 92 db, but is actually more like 96 db. Has the horn built onto it, and has excellent sound.

[GordonW]

Originally Posted by macaltec

I went back and looked up one that I had done, for a Powerline on a typical 4x10 horn. I stole part of Paul's graphic, but changed the diagram layout to what I was talking about.

This was for a PA app, from a 10" woofer, so I went with a higher crossover point... IIRC, it was around 4000 Hz. You can scale the caps larger, to have lower crossover points. Changing the cap relative values so that the first one is larger and the second is smaller, will make more of a "gradual" rolloff... while making them "more the same" will result in a flat top-end, with a more pronounced "break-point" near the crossover point (ie, a sharper "knee" to the curve). This can be useful in tuning the horn/driver combo... if there's too much upper treble/lower mids, then make the second cap smaller, while leaving the first one the same (or maybe SLIGHTLY larger)...

[macaltec]

Thanks Gordon. Clear as mud now .

[GordonW]

Originally Posted by Paul C

Hey, Gordon, have you used the KSN1165a?



This is a PowerLine model that produces 1800 hz - 30 khz. It is rated SPL = 92 db, but is actually more like 96 db. Has the horn built onto it, and has excellent sound.

Actually, I did use one of those recently, to make a monitor cabinet, with a 15" Heppner woofer. Let the woofer run "open" on top, and used a resistor pad plus a 2.2 uf cap (IIRC, a 3 ohm resistor in series, and a 10 ohm in parallel with the piezo). Probably around 6KHz. Just to "extend" the top of the 15" (which sounded good by itself, just no treble).

For its app (PA monitor), it was VERY nice! Probably better than it needed to be!

Regards,
Gordon.

[Paul C]

Right now for my main speakers I have some some Speakerlabs (10" woofer, W1008b) with the KSN1165a's rigged up with L-pads as per one of my drawings.

Originally these were 3-ways crossing at 800 hz and 8000 hz, but the Polydax mids are no longer available. I reworked the crossover to roll the woofer off at 1800 hz, where the KSN1165a's roll in. With -4db cut on the L-pads, they sound great.

[GordonW]

Originally Posted by Paul C

Originally these were 3-ways crossing at 800 hz and 8000 hz, but the Polydax mids are no longer available. I reworked the crossover to roll the woofer off at 1800 hz, where the KSN1165a's roll in. With -4db cut on the L-pads, they sound great.

Get 'em while you can (since they're discontinued)... but the Vifa M10MD39-08 is a very good replacement for the Audax 12P25FSM (and given the Speakerlab parts, I assume that's what was in there?). Efficiency is very similar, as is impedence and frequency range...

Regards,
Gordon.

[Paul C]

No, I have long forgotten the Audax/Polydax number on the mid, the Speakerlab number was M408b. The mids I had were similar to, but not the same as the 12P25FSM. It was discontinued back about 1990 or so. I got the last two that Madisound had. Then the surrounds on one of those went south and I replaced the mid AND tweeter on each with one KSN1165a each.

The woofer and tweeter on the Speakerlabs were SPL = 92 db, that Vifa is only 88 db. No problem, the baffle has been modified and I am satisfied with the sound now. These speakers will soon be replaced with some Pi Speakers 4pi Theatre's, using Eminence Delta-15's and PSD2002/H290 horns. Mo' bigger is mo' better! I just need to wait for the rainy season to stop so I can get outside and cut some wood.

[crackerkorean]

Well they are hooked up and running on a Fisher TA-600 tube receiver. I have to say im impressed. This system is slowly coming together. The highs are detailed and coupled with the carlson cabs it sounds great.

I listened to a variety of things tonight and all was good.