New amplifier and shaker table

For the pressure sensor labs, I’ve built a homemade shaker table using a function generator to drive a Dayton Audio TT25-16 PUCK tactile transducer mounted on a 9″ circle of ½” plywood sitting on large rubber feet.  I finally got a chance to try hooking up the 20W LP-2020A+ Lepai Tripath 20W amplifier that took a couple of weeks to arrive from China.

Perspective view of the shaker table, showing the clip intended to hold the water reservoir.

Top view of the shaker table. The two holes on the top right were used for mounting an ADXL335 accelerometer breakout board. Unfortunately, the shaker table does not move vigorously enough for that to produce a strong signal.
The extra cutout for the wires is too large—I was in a hurry and did a sloppy job.

Bottom of the shaker table, showing the Puck screwed to the plywood with 6 6×½” wood screws.

Before hooking up the amplifier to the shaker table, I decided to look at the output waveform with just the oscilloscope as a load. With the tone controls switched out, the amplifier looks fairly linear (on an X-Y plot on the oscilloscope) up to about 1kHz, but at higher frequencies I see a lot of ripple, and with a 7.5 kHz input there is a huge ripple for a signal at about 65kHz.  Once the 65kHz oscillation starts, it remains present even if the gain is turned down to 0.  Turning the gain back up can reduce the oscillation, until the gain is turned up too high.  The harmonic distortion is truly terrible around 12.5kHz, with the 5th harmonic being about half the amplitude of the fundamental.  At 10.35kHz, the 6th harmonic is large, at 9.53kHz, there is a ripple at 6.5 times the frequency of the fundamental, at 6.22kHz the ripple is at 10 times the frequency, and at 12.8kHz the ripple is 5 times the frequency.  So this ripple seems to be around 62–64kHz, with the frequency pretty much independent of the input.  The magnitude of the ripple depends on the input and is much larger when a harmonic of the input excites the instability.  When the gain is set too high, the ripple becomes quite large, even before clipping occurs.

The amplifier is advertised as a “class T” amplifier, which seems to be a slight variant of the “class D” digital amplifier design.  The 64kHz is probably the frequency of the underlying oscillator for determining when to switch the output transistor on and off, and the ripple is inherent to the digital amplifier design.  It seems much larger than I would have expected, though.  The ripple is not an artifact of the high-impedance load, as it is still very strong even with the 16Ω Puck as a load.

To avoid clipping with the 16Ω Puck as a load, the output voltage has to be kept down to about 6.5V RMS, which is about 8dB more gain than I got from the op-amp-and-transistor amplifier I used in Characterizing tactile transducer again.  I get a strong resonance from the amplifier around 64.5kHz and a mechanical resonance from the Puck around 23Hz.  I may have to use some softer feet, as the shaker table dances around a lot at the mechanical resonance.

I think that we may be better off using the op-amp-and-transistor amplifier, despite the lower power, because the 64.5kHz ringing of the LP-2020A+ amplifier may be too confusing to the students and may throw off RMS voltage measurements.  We could up the power of the op-amp-and-transistor amplifier. With a slightly higher power transistor, like the 2N4401, the op-amp-and-transistor amplifier would be limited mainly by the 6V power-supply limit for the MCP6002 op amp I used.  I don’t know whether this amplifier application is enough justification to change which op amp we plan to use.

For the driver amplifier, do we have the students design their own amplifier for the shaker table, do we give them a design and have them build it, or do we build it for them? This comes down to a decision about whether the shaker table is to be thought of as a pre-existing piece of equipment (like the oscilloscopes or multimeters) or as something that they are making (though we will do the mechanical construction for them, as they do not have access to a wood shop nor the time to make the shaker tables).

It might be useful to have them do the design, but then we’d have to talk about bipolar transistors as current amplifiers, and I was thinking that we would remove a lot of the usual device-level modeling from the course in order to have room for the op-amp design stuff.

 

 

About these ads