Plate Reverb pre EQ Issue

[deveng]

I designed a pcb for this circuit to provide pre-equalization to driver for my plate reverb.  Its the circuit from the Ecoplate.  The simulation was good and showed proper boost and cut levels.  However, when bench testing I found what appears to be a slight oscillation most visible on the 1kHz sine.  The problem becomes visible with oscilloscope at pin 7 and 5 of IC2B.    Boost is close to 20dB at 10kHz.      Here's the circuit schematic.

 

[deveng]

and here is my odd distortion riding on the 1kHz sine......

 

[deveng]

I've tried the usual,  feeback loop compensation, additional power supply filtering.  The only way to get rid of it so far is to reduce the 10k boost by adding an additional pole prior to IC2B but of course this sort of defeats the HF boost purpose of this circuit.  I can get about 12dB of  10kHz boost without this distortion, but that's it.    I have not tried soldering IC2B to the pcb.  Currently its in a gold pin socket. 

Could it just be layout (parasitics) or am I missing something here? 

 

[analogguru]

Put a 4558/1458 (instead of the 5532) into the socket and observe if this disappears.

 

[deveng]

Limiting open loop gain/ slew rate?

 

[Bo Deadly]

There's nothing obviously wrong with the circuit. At least the filter part is pedestrian. It could be a bad solder joint or some kind of grounding issue. Is this a PCB or protoboard or breadboard or what? If it's a breadboard, just break it down and redo. That's probably the path of least resistance. Breadboard jumper wires are notoriously sketchy. Or maybe a damaged pot. Something like that.

The circuit looks like it was designed for more of a high Z amp. With 150K on the input you'll get a little bit of an offset w/ 5532. That shouldn't be a problem with this particular circuit because there's no gain and it's AC coupled. But at high Z you're negating any noise benefits of a fancy bipolar op amp like 5532. It should probably be just a TL072. A plate is relatively noisy anyway.

Actually, it could be the LM1875. If you're running that on a breadboard, grounding will be really important. And you would need to test that with a proper load.

Post the schem for the "ecoplate". Does that use LM1875?

 

[deveng]

Analogguru,

Wired up a 4558 and the problem remains.

Squarewave,

Its all on a pcb.  In fact I built a second one and the same issue is there.  Its not the LM1875 as I designed my pcb without it.  So my output is at the 5k trimmer.    I believe the schematic is correct as well and it simulates perfectly.    My board uses TL071's for IC2A and IC2B and a THAT 1246 for the differential receiver.  The 1kHz signal is clean up to IC2B.    I've confirmed my circuit matches the Ecodriver wiring and values. 

 

[deveng]

Here's my schematic for reference.    PCB was fabbed directly from this.

 

[analogguru]

So you are using single opamps.... remove the second opamp and observe if the "visible problem on pin 5" still persists.

 

[ruffrecords]

What is the purpose of the 4.7uF capacitor C9? Why not short it out? Perhaps it is inductive enough at high frequencies to destabilise the loop.

Cheers

Ian

 

[deveng]

Analogguru,

Removed 2nd opamp and input 1kHz at pin 6.  Problem still there.  So the issue appears to be related to the 3rd stage.

Ian,
I'm not clear on the purpose of C9.  I shorted it and issue still there.

This one is baffling me!  When I see oscillation it usually doesn't appear like my oscilloscope picture from my 2nd post.  Usually the scope displays a fuzzy sine sometimes with the visible frequency riding on top of the entire waveform and it moves.  This one is odd as the distortion seen on the top of the my sine wave never moves, it stays constant.    Also,  I noticed the low end of the sine appears a bit flattened.  Very subtle.

 

[JohnRoberts]

Study the pcb layout for possible positive feedback...

If output current leaks into an earlier gain stage. 

JR

 

[Bo Deadly]

So you built two and both exhibit the same behavior? That rules out a few things. Check your probes and scope connections. I have some switchable 10x/1x and in the 10x position, small signals can be a little wonky. Does the distortion scale with signal level? It's not hard for a test rig to have problems for all sorts of reasons. If the waveform changes with signal level, that would suggest that there might be a problem with the test rig. Meaning your board could be just fine and it's your rig.

 

[deveng]

Right, that's the reason for the second.  Also easier to debug as it is not installed in the system.  I may have found what appears to cause the problem.  As you suggested (it could be the test rig) I tried a different signal generator and the distortion goes away.  Early on I had checked the signal on both channels of my scope and the 1kHz sine looked perfect.  But,  for some reason using that  bench generator with the scope and pcb,  the signal becomes distorted.  Using another signal generator the circuit is clean.  Never seen that before.   

Thanks for all your inputs.  Now I'm getting the response as attached (simulation results) and the distortion is gone, circuit is fine. 

Regards,
Jeff

 

[deveng]

....and the actual measured response......

 

[Bo Deadly]

Great!

Presumably your recovery amp has an inverse of that filter?

 

[deveng]

The pickup amp I designed has the 100Hz  boost of about 6-9dB  but no HF cut (see response curve attached).  That's pretty much how the EMT and Ecoplate response curves go.    I had a friend who got me some demo high end aerospace piezo transducers.  They have the internal amplifier so they are low impedance output.  They're flat from 20-20k and all that is required is a current source to power them.    The idea was that I don't have to deal with the high impedance issues and long cables.  They run into a simple buffer amp and the LF boost then to a pair of Jensen transformers.

Regards,
jeff

 

[jazzcrisis]

The pickup amp I designed has the 100Hz  boost of about 6-9dB  but no HF cut (see response curve attached).  That's pretty much how the EMT and Ecoplate response curves go.    I had a friend who got me some demo high end aerospace piezo transducers.  They have the internal amplifier so they are low impedance output.  They're flat from 20-20k and all that is required is a current source to power them.    The idea was that I don't have to deal with the high impedance issues and long cables.  They run into a simple buffer amp and the LF boost then to a pair of Jensen transformers.

Regards,
jeff

Do you have the model number for the aerospace piezos? I have a plate reverb I built and I feel like my ordinary piezo pickups are a weak link.

 

[deveng]

I have a pair of 7251 Isotrons.  They are expensive brand new but you can find them on Ebay  ranging from  $100-150 a piece used.  They require a 4ma current source and the proper coax cabling.  Self resonance is 45kHz, they spec at +/-.34dB from 20-20kHz with an full scale output level of +/-5v.    I'm still testing mine and have not finished my plate construction so I'm not certain how they'll sound.    I did an initial test (to verify the current source and receive amps) with a freshly mounted stainless plate and the results were not quite there yet.  At that point I had no pre equalization, my plate was not yet tensioned correctly and I was using a questionable driver, but the piezo's appeared to work great.  My work with the pre-eq pcb in this thread finishes up the full drive/receive electronics.  I'm cutting new thicker corner supports so I can increase plate tension.  Once that's done  I should be able to permanently mount the driver,  the drive/receive circuits and give a proper testing.    I've attached a picture of the build in its current state.

Regards
Jeff

 

[Bo Deadly]

Very professional build!

Is the weight of the accelerometer a factor wrt high frequency response? That one is apparently 10g which might not sound like much (about the weight of 2 US quarters) but normally it would be attached to a large mass and not a relatively lightweight piece of sheet metal. I would think there should be ones that are much lighter.

Update: Like this one for example:

  https://buy.endevco.com/contentstore/mktgcontent/endevco/datasheet/2222d_ds_090719.pdf

which is 1g, claims to be "the world’s most popular minature piezoelectric accelerometer for vibration measurement on mini-structures and small objects" and they're not terribly expensive used. It doesn't have an integrated amplifier but I don't see why that would matter much as long as you use a high quality coax cable.