• TalkBass has been independent since 1998. Add your voice.
    Create a free account to reply to discussions, view embedded media, and browse with fewer display ads.
    Join free Log in
    Want zero display ads or expanded classifieds tools? Compare plans.

DIY tweeter top box 2 - piezo units.

#22
Thanks to a very customer service minded Monacor I now have frequency response and impedance charts for a couple of their piezo tweeters. Now, before I try and design a filter/volume control circuit, am I right in thinking that frequency response curves are taken at a constant power setting? And therefore when trying to approximate a reasonably flat frequency response in my system I need to make allowances for the power the amp is delivering changing with frequency? If so does that imply that a piezo tweeter, where impedance diminishes with increasi/ng frequency, will tend to over deliver at the top of the range and it might be as well to have a shallow low pass filter in the circuit to even things out? Impedance is well in excess of 100 ohms at 5K, and around 30 ohms at 18K.
So that suggests I'd be looking at:-
high pass filter around 5K to take all the LF energy out of the circuit.
low pass filter around 15-20K to reduce excess HF response
voltage divider to match sensitivity / volume
This ought to be doable with a moderate cost RC circuit shouldn't it? A largish series capacitor to do the high pass, a series of resistors to do the voltage divider, and a small capacitor and resistor in parallel to the tweeter?
 
#23
Something like this? I reckon if I use 5w resistors it ought to be safe to a bit over 100w on the main input unless I'm completely confused...
154349131366025297441.jpg
 
#25
agedhorse said:
Is the frequency response based on applied voltage or power?
Click to expand...
Oops. yes, that's rather important isn't it! The question half occurred to me, but then I made the dubious assumption that these industry standard plots must be normalised against impedance variations.

The plots are all marked ATB, so presumably they were made with Kirchner elektronik ATB precision software/hardware, which Monacor appear to market. Invalid Link Removed


The following text from p12 seems relevant. You'll understand it at least 10 times more clearly than I, but I think this is probably saying constant voltage.


The output is balanced but not free floating. This means that the output voltage depends on whether the connection is balanced or unbalanced. Therefore, the connection is set in the measurement program to display the real output voltage. Caution: Because the balanced output is not free floating, the Out- may not be connected to ground when the connection is unbalanced.
Output data Output range
2dBV ... 16dBV balanced corresponding to 1.26V ... 6.31V

-4dBV ... 10dBV unbalanced corresponding to 0.63V ... 3.15V 2dBU ... 16dBU balanced corresponding to 0.89V ... 4.48V

-4dBU ... 10dBU unbalanced corresponding to 0.45V ... 2.24V in 1-dB-steps Frequency range DC ... 24kHz +0.1dB, DC ... 50kHz +0.5dB Output impedance 44Ω balanced, 22Ω unbalanced Mini. load impedance 600Ω balanced, 300Ω unbalanced

For speaker measurements a power amplifier is required. The balanced or unbalanced input of the amplifier is connected to the PC card output. The output power of the amplifier is adjusted with the volume of the amplifier. The program adjusts the output voltage to different measurements. Caution: For the impedance measurement the output voltage is set by the program to 6V balanced, 3V unbalanced. Before starting the measurement, turn the volume to zero and disconnect the speaker from the amplifier output.
Click to expand...
 
#26
So it appears that this is a constant voltage measurement not constant power (it ignores the changing impedance).
 
#31
As an update on this I built my boxes, and have used them on top of a pair of 1*10s in a small show, but I haven't got round to borrowing any equipment to do frequency response testing yet. They seemed to do an adequate job. This, FWIW (not too much I fear - not nearly enough science involved), is the circuit I used.

wiring.jpg


There's not a lot of science in this, but the conclusion I came to was that with the piezo tweeters it wouldn't be as critical as with a moving coil tweeter, so I could take a punt and experiment a little. My understanding of the advice I was given was that in a very basic setup like this I could get away without a proper crossover and not filter HF from the woofer.

I started with a capacitor to block LF (thus high energy) signal. I failed to find a reference I understood about how to calculate the frequency of an RC filter like this, what with having the capacitative piezo in the mix, but I came to the conclusion that 0.47uF ought to let through everything I wanted and block at least some of the lower frequencies. I don't think there's any way I can tell how good the choice is without doing a proper analysis of the circuit as the tweeter discards what it can't use.

Next is a 10ohm 15W resistor in series. AH suggested that a 10ohm resistor in series with the piezo tweeters would be good practice, and I am not going to argue.

The rest is a crude volume control in the form of a voltage divider using fixed 15W resistances and a rotary switch. This was way way cheaper than buying a reasonable wattage variable resistor. Again I didn't really find references I could understand to derive values for this. L-pads seem to me quite inappropriate with the piezos. So I decided that around 100 ohms total would probably be a reasonable first approximation, and then guessed some steps that seemed OK to me. With the series of 15W resistors (which are on an aluminium heat sink), and the capacitor to filter out LF I'm hoping the setup of a pair of these should be reasonably safe with my 250W rated amp, especially as I don't actually anticipate using this baby rig for anything that would need the amp to be turned up!

The tweeters are Monacor MPT-177, which are actually a twinned pair.

The result - well all I've actually used them for so far was with my PA 1*10s to play recorded music for a show in a small hall. It was a very significant improvement on the sound without them, and I reckoned that one notch down from max on the volume control sounded about right. I think so far the results are reasonably adequate for such a finger in the air exercise. Frequency response testing would probably be best done outside in the absence of a suitable facility, and can wait until summer!
 
  • Like
Reactions: agedhorse and BasturdBlaster
#32
IIRC, series connection of a Piezo tweeter raises its natural LF "crossover point" by about an octave. Because of the reactive nature of Piezo tweeters, they can be a real challenge (read as darn near impossible) to tune to an application. There are a lot of things working against a simple inexpensive solution when you start trying to do things that they are not inherently designed to do.

The reason for the difficult in calculating a crossover frequency is that you really can't... the impedance of the driver falls with rising frequency at roughly the same rate as the "crossover" cap does. In essence they cancel each other out and you are left with the raw LF rolloff response of the piezo.
 
  • Like
Reactions: JimChjones and Wasnex
#33
agedhorse said:
IIRC, series connection of a Piezo tweeter raises its natural LF "crossover point" by about an octave. Because of the reactive nature of Piezo tweeters, they can be a real challenge (read as darn near impossible) to tune to an application. There are a lot of things working against a simple inexpensive solution when you start trying to do things that they are not inherently designed to do.

The reason for the difficult in calculating a crossover frequency is that you really can't... the impedance of the driver falls with rising frequency at roughly the same rate as the "crossover" cap does. In essence they cancel each other out and you are left with the raw LF rolloff response of the piezo.
Click to expand...
If you build a crossover calculated for an 8 ohm load impedance and place an 8 ohm power resistor in parallel with the piezo driver, the resistor swamps the much higher impedance slope of the piezo. The crossover sees a more or less 8 ohm load across the frequency range.
 
  • Like
Reactions: agedhorse
#34
I can see that would make the crossover calculations straightforward, but in my ignorance of the fine detail of electrical theory it seems to me that the signal would mostly go through the power resistor and not the tweeter. Or is there something about the way the piezo works? I have trouble getting my head round how the piezo element behaves in a circuit in that sort of respect. I'm aware I have only a rather simplistic understanding of voltage and current behaviour.
 
#36
We're into something else I don't understand sufficiently in the TC tweetrr thread.

The pulsar paper talks about using a series capacitor to attenuate piezos. It gives values for a couple of specific applications. Now my rudimentary understanding of capacitors in an RC or RL network is that the capacitor value affects the frequency that passes. If that's the case why doesn't the capacitor value affect frequency as well as attenuation? Doesn't the presence of the woofer make it an RC circuit?

I could have used switched series capacitors rather than the voltage divider to attenuate I suppose, but instead of limited understanding of how it worked I'd have had none. Also with resistors the attenuated power comes out as heat, where does it go with a capacitor?
 
#37
It's a simplification, but if you think of a dynamic tweeter as a resistive load that can be attenuated with a resistor then you can look at a piezo as a capacitive load that can be attenuated with a capacitor. What you have to wrap your head around is that with a piezo you're not dealing with an RC or RL network.
 
  • Like
Reactions: agedhorse
#38
JimChjones said:
I can see that would make the crossover calculations straightforward, but in my ignorance of the fine detail of electrical theory it seems to me that the signal would mostly go through the power resistor and not the tweeter. Or is there something about the way the piezo works? I have trouble getting my head round how the piezo element behaves in a circuit in that sort of respect. I'm aware I have only a rather simplistic understanding of voltage and current behaviour.
Click to expand...
Actually, the crossover with a resistive load works very nicely for piezos. Because a piezo element is essentially a high impedance device, it is mostly voltage driven with low current. The load resistors sole purpose is to appropriately load the crossover network to pass the desired band of frequencies.

The voltage developed across the crossover load resistor is more than sufficient to drive a piezo. Of course, the load resistor current is wasted power, but because it's mainly high frequency, it doesn't really consume any more power than an 8 ohm driver would, in it's place.

Years ago I had some Norberg nearfield monitors that surprisingly, used piezos. If I recall correctly, they used such a crossover, and it worked very well.
 
#39
JimChjones said:
We're into something else I don't understand sufficiently in the TC tweetrr thread.

The pulsar paper talks about using a series capacitor to attenuate piezos. It gives values for a couple of specific applications. Now my rudimentary understanding of capacitors in an RC or RL network is that the capacitor value affects the frequency that passes. If that's the case why doesn't the capacitor value affect frequency as well as attenuation? Doesn't the presence of the woofer make it an RC circuit?

I could have used switched series capacitors rather than the voltage divider to attenuate I suppose, but instead of limited understanding of how it worked I'd have had none. Also with resistors the attenuated power comes out as heat, where does it go with a capacitor?
Click to expand...
Heat is caused by current flow and voltage drop across a device. The less current, the less heat created.
 
#40
Thanks folks, I'm not sure I grasp all the implications, but I'll see what I can figure out when I have a basic test setup together.

Just as a thought, can I measure frequency response really close up above 4K as I need to worry less about diff tone from diff bits of cone, and would that mean I could be less concerned about room effects? Or is it just as much an issue as the lower end?
 

Latest posts

Top Bottom
Back