A TRUE Piezo Super Tweeter (piezo tweeter revisited)

PREFACE:
It is my opinion that the only legitimate use for piezo drivers in audio is as a Super/Ultra tweeter. Their full range output is quite nasty and not HiFi.
They are however very fast, hence their use in ultrasonics.
THE GOAL:
To augment speaker systems that drop-off around 14Khz, and/or a method of helping older people with a degree of high frequency hearing loss.
To achieve this with a minimal amount of 'overlap' with existing tweeters that would cause H.F. 'smear & comb filtering'.
PICK YOUR PIEZO:
All the drivers I have investigated for this project are non Motorola current generics. You can try any piezo, but I recommend the 95/98mm circular horn that is available on Ebay for about $10 a pair. I was quite surprised to find that my pair were excellently matched in both capacitance and output.
THE PARTS and CIRCUIT:
Excluding the piezo drivers, there are only 4 parts required per Super Tweeter. In Australia the cost per Super Tweeter is about $50. (the U.S. would be less)
For the sake of simplicity, all components are standard value parts.
The heart of the 'drive/filter' is a 5W audio line transformer, used in reverse.
Find the 8 ohm side of the transformer by measuring the wires (or terminals) that are less than 1 ohm DC. This side of the transformer connects to the amplifier via a 1.5uF (high quality) capacitor and 2.2 ohm 5W resistor in series with one of the wires. This completes the input to the transformer.
( facing both sides of the transformer, the right-hand connection is 'POSITIVE' )
Now, on the other side of the transformer find the two terminals next to each other that measure about 40 ohms DC. (all other terminals should be ignored)
These '40 ohm' terminals directly drive the piezo WITH a 1.5mH inductor in parallel. This inductor does not need to be an expensive heavy duty type.
That's it - circuit complete.
FINAL RESULTS & THOUGHTS:
For this project, 15Khz can de considered a reference point. It is where the 'filter/drive' has peak output and also where the impedance is 8 ohms.
Above 15Khz , 18Khz is -2dB, and 20Khz is -2.5dB.
Below 15Khz , 10Khz is -5dB, and 5Khz is -23dB.
This means we have achieved a true 18dB per octave/third order cross-over slope.
The achieved impedance is as follows:
85 ohms@1K , 28 ohms@3K , 17 ohms@5K , 6 ohms@10K , 8 ohms@15K and 5 ohms@20K.
Unless you have a Spectrum Analyzer with microphone, you will need to pick 'Phase Orientation' by ear. It is usually pretty obvious when you get it right.
If the output sounds too loud, simply insert a series .33 or .27uF capacitor in series with one of the piezo terminals. The lower the value the lower the output.
Rather than cutting holes in boxes, I built this project directly onto nice 'baby slabs' of hardwood with felt feet, to place on top of existing speaker boxes.
THE SOUND > Very revealing. Depending on your speakers and ears, this can be like lifting a veil off your sound system, especially with natural sounds.
PS. Don't let that 5 watt transformer rating worry/trick you. It is a 'nominal' value tested @ 400Hz, which does not apply here. Transformers are far more efficient at
high frequencies.
 
Interesting I was just playing around with some Piezo
Mainly interested in Impedance Data which is hard to find.

But did manage to get a trace off a old Motorola Datasheet
for KSN1041A
1684236627019.png

Accuracy of Frequency Response debatable of course.

But mainly interested in impedance to see how
typical filters work.

KSN1041A-SPL.jpg
KSN1041A-Impedance.jpg


Trying think of solution, finally came to me.
Just adding typically impedance compensation.
I could get normal filters to work
15 ohms seemed to work well. Filter Cutoff Q
more easily controlled adding 3.3u or 4.7u in series
with 15 ohm resistor.

Impedance then behave like typical 8 ohm drivers
and even creates small rise instead of falling.

KSN1041A- XO-Impedance.jpg

KSN1041A- XO-schema-1.jpg

Ironically was looking for a 3rd order around 15 to 18K
as well

KSN1041_Crossover.jpg
 
  • Like
Reactions: 1 users
yes indeed higher order filter would help

with the vintage conical waveguides.
sure the wave guide makes it very directional at high frequency.
Be interesting to find non waveguide/horn element with data

GRS uses larger piezo elements for better lower frequency performance.
for super tweeter a smaller element and 3rd order be more ideal
 
Last edited:
Not bad at all.
In the nineties, Audax had a very expensive piezo tweeter HD3P, iirc it was around $300 a piece. Never seen it used, never seen it even described as to what it's sound quality, nor filtering/integration with other units went.
Quite the mysterious unit.

Doing a search (much easier than in the nineties I guess;-)

https://www.hi-fiworld.co.uk/index....y/371-hd3p-gold-dome-replacement.html?start=1

Not an easy filter, hope it helps..
 
the late Dennis Kleitsch aka "DKA" gave some info on using CTS- Motorola KSN1016A piezo tweeters. I used a 1uF cap to Zoebel to stepup transformer and got ~103dB at 2.83v sensitivity and 1/4 of 1% distortion using 5KHz sine.
 

Attachments

  • DJK KSN1016A APPLICATION.jpg
    DJK KSN1016A APPLICATION.jpg
    229.9 KB · Views: 133
  • Like
Reactions: 1 users
Interesting circuits, showing how the higher impedance characteristics of piezo's do lend themself to
transformer use for higher output SPL. Some of that looks VERY LOUD - probably PA use.
I wonder if it was at all HiFi? It is important to remember that any series resistors with piezo drivers
decreases output at very high frequencies like 20Khz.
It is also interesting to see how many people have used Zobel Networks, to then use standard 8 ohm
filter topology.