Plate Reverb Circuit Feedback

[Sam0311]

Hi all,

When I first became interested in DIY electronics, I really wanted to try to make a Plate Reverb. I naively thought it would be a fun little project to keep me busy for a couple of weekends, the concept behind how it works is quite simple but it wasn't long before I realised this was way above my skillset and had to put this idea on the shelf.

A year has passed, my interest in electronics has been sparked again and I'm feeling more confident to try and tackle this task again. I spent the weekend putting together a schematic for this project. I'm feeling pretty confident that it could work how it is intended but due to my lack of experience and the fact that certain parts of the schematic were done by guess work, I'm putting this out there in the hope that some kind soul might tell me where I'm going wrong. Fingers crossed I've not messed up my calculations.

To summarise the intention behind the circuit, I want to be able to drive 2x4ohm transducers (likely between 10-20W) from a line level input to excite a metal sheet thats going to be around 500x300mm (closer to an EMT 240 than 140). The power amp section should have a good amount of headroom but I'm not overly concerned with clipping and noise since I'm trying to excite a metal plate not build a HiFi system. I want to have control over low and high frequencies both pre and post reverb. I will be using piezo discs to pick up the sound from the plate that I assume will need a preamp to bring to line level and suitable output impedance. I have added in feedback network as a means of controlling the reverb time.

Key Points:
- The inputs and output of the circuit should be able to function as mono or stereo and will be balanced line level.
- The nominal input voltage will be +4dbu or 1.228 vrms, by my calculations, once the input signals have been unbalanced and summed the voltage will be 4.912v.
- At a supply voltage of +/- 15v, the LM1875 should be able to output roughly 10W, the circuit has been designed so the nominal input level will leave around 15db of headroom (not including the feedback network).
- The LM1875 has a minimum gain of 10. I will need to reduce the input voltage by about 30db to give me my desired headroom.
- The idea behind R8 and RV3 was to give me the required attenuation then to have control over the input level to the LM1875, I'm not sure if this will work but I think it will?
- The transducer impedance will be 8 ohms (or two 4 ohm transducers in series)
- I want a 2 pole Low and High pass filter with variable cutoff points on both the input and output signals, how I have gone about this is partly guesswork.
- The values from the TL072 and onwards have either not yet been chosen or are subject to change, the plan is to first measure the voltage that the piezo discs output then work out how much gain will be required
- I wasn't sure how I should approach the feedback network but what I have put in the schematic seemed like my best bet

Questions:
- Does the secondary impedance of the input transformers work as if they were in series with the resistors going to the input of the summing amplifier, changing the values needed for R3 and R13 or is what I have done here fine?
- The components around the LM1875 were mostly copied from the data sheet. I have not included a capacitor in the feedback network for the other op amps, should I be doing this to avoid DC offset?
- I believe the NE5532 needs a load impedance of no less than 2k. In this circuit is the load impedance the same as the output transformers primary in parallel to RV8/RV9 or just the primary itself?
- How should I calculate the values for R28/29 to give me unity gain, I think I’ve done it wrong here but should it be all of the resistors before the NE5532 in parallel? - on second thoughts if this is the case things would be quite unstable because of RV2, RV5 and RV7 acting as variable resistors. Maybe I should learn how to implement a Sallen-key filter instead?
- I now think the voltage divider in the feedback loop is unnecessary if the signals going into the summing amp are at line level, I was forgetting about the 6db gain I was getting from balancing the input signals.
- Maybe I should put a buffer after the pot in the feeadback network?

I appreciate this is a bit of a chunky post so no expectations for anyone to read it all the way through but would be very grateful to anyone that can chime in on this.

Cheers,
Sam

 

[ccaudle]

The first thing that jumps out is U1B is going to saturate right away and clamp at one rail because it only has positive feedback at DC. Is that just a mistake and was intended to be a unity gain buffer?

The next is that you have pots feeding output transformers. The impedance of a transformer is not fully resistive, and the distortion will increase with increasing source impedance. That means that you will likely get higher distortion as you turn down RV8 and RV9. Better to have the pot in front of a buffer amp, and let the buffer amp drive the transformer directly.

Input impedance is only 3.3K, is whatever you are using to drive that circuit OK with a load in that range? Most pro gear would be, so shouldn't be a problem unless you have a more consumer style output circuit driving that.

It looks like your input jack labeled "L/Mono" will disconnect from the other side when a jack is inserted, so I think the R input jack should be labled R/Mono, and will drive both inputs unless you insert a plug into the L jack. Note that the driving circuit would then be driving a 1.6K load, that is getting a lot harder even for pro gear.

The idea behind R8 and RV3 was to give me the required attenuation then to have control over the input level to the LM1875

Seems like it should work, that will give close to 30dB attenuation as you have it drawn.

Does the secondary impedance of the input transformers work as if they were in series with the resistors

No, generally the impedance on each side is reflected across, so for a 1:1 transformer the impedance seen by the driving circuit is just the impedance of the circuitry on the secondary side. At very high and very low frequencies that breaks down, but should work in the middle frequency range.
So for the secondary side the driving impedance seen by the circuit should be relatively low if the output stage driving is low impedance, but at high frequencies the inductance of the transformer would increase the impedance, and at low frequencies the impedance would decrease (but limited by the DC winding resistance).
Don't necessarily take my analysis as fully correct, I rarely work with transformers, so I would defer to those who use transformers more regularly.

I have not included a capacitor in the feedback network

The input offset voltage could be from 1mV to 15mV, so at the output that would turn into 165mV at room temperature, increasing at increased die temperature. That is only about 3mW dissipated in 8 Ohms, so I would not worry about it very much, assuming your driver coils are built to handle 20W+.

is the load impedance the same as the output transformers primary

The load impedance is the secondary load reflected through the winding ratio. Since you have an output transformer labeled as 16:1 impedance ratio, that implies a 4:1 winding ratio. The impedance seen at the primary side will be 16x higher, but the voltage at the secondary side will be only 1/4 the voltage at the primary side. That is a 12dB attenuation of the output level, is that really what you want?

How should I calculate the values for R28/29 to give me unity gain

You can directly connect the output back to the inverting input. The recommendation you may have heard about using a value equivalent to the source resistance at the non-inverting terminal is to try to minimize input DC offset. Since you have a lot of coupling caps in the design that isn't really a concern. Note that some op-amps have a minimum recommended feedback resistor (usually current feedback topology) so always check the datasheet of the device you are using.

 

[Sam0311]

Great feedback here, really appreciate you taking the time to write this.

The first thing that jumps out is U1B is going to saturate right away and clamp at one rail because it only has positive feedback at DC. Is that just a mistake and was intended to be a unity gain buffer?

Ahh yes, I had intended for this to be a buffer.

When I was looking back over the schematic I had not realised I had had initially put the op amp the wrong way around and then ended up confusing myself trying to work out how to get the amp to be at unity gain with the resistance on the input, thanks for pointing this out.

The next is that you have pots feeding output transformers. The impedance of a transformer is not fully resistive, and the distortion will increase with increasing source impedance. […] Better to have the pot in front of a buffer amp, and let the buffer amp drive the transformer directly.

Nice, I’ve been unsure for a while the best way to implement an output level control if there’s a transformer on the output.

It looks like your input jack labeled "L/Mono" will disconnect from the other side when a jack is inserted, so I think the R input jack should be labled R/Mono, and will drive both inputs unless you insert a plug into the L jack. Note that the driving circuit would then be driving a 1.6K load, that is getting a lot harder even for pro gear.

Haha, I have a really hard time visualising things like this. I need to be able to see the thing in person and mess around with it to get a grasp on what’s happening. I have a bit of a habit of wiring my jacks up wrong in schematics. Good point out the output impedance, I guess there’s no real need for Mono/Stereo outputs, if I want a mono reverb I can just plug into one of the outputs.

The load impedance is the secondary load reflected through the winding ratio. Since you have an output transformer labeled as 16:1 impedance ratio, that implies a 4:1 winding ratio. The impedance seen at the primary side will be 16x higher, but the voltage at the secondary side will be only 1/4 the voltage at the primary side. That is a 12dB attenuation of the output level, is that really what you want?

Transformers are one of those things that are seemingly really simple to understand but the more I learn about them the more I realise I have no idea what I’m doing what you say about impedance being reflected from one side to the other and affected by the turns ratio makes sense and is something I could get my head around.

What I don’t understand is what is the difference between a 1k:1k and 10k:10k transomer? Are the values stated the nominal impedances on either side of the transformer for it to function as intended?

In terms of attention, I’m not super worried about it at the moment since I still need to find out how much gain I’ll need to add to the signal coming from the piezo discs but if I find that I can’t easily get the overall gain I need them I’ll need to reassess.

You can directly connect the output back to the inverting input. The recommendation you may have heard about using a value equivalent to the source resistance at the non-inverting terminal is to try to minimize input DC offset. Since you have a lot of coupling caps in the design that isn't really a concern. Note that some op-amps have a minimum recommended feedback resistor (usually current feedback topology) so always check the datasheet of the device you are using.

Ah okay great. Took me a while to get my head around what you said here but makes sense now thanks.

Cheers,
Sam

 

[ccaudle]

It was intended to be a unity gain buffer.

Then you just got the connections backwards in the schematic. Unity gain buffer should be input to + input, feedback from output to - input.

Would I be fine to keep the filters as is and remove the resistor from the feedback loop?

You always need a DC feedback (i.e. output to inverting input) connection around an op-amp. There are compound circuits where the feedback path is around multiple components back to the inverting input, but that doesn't really apply here, you need a connection from op-amp output to op-amp inverting input.

Transformers are one of those things that are seemingly really simple to understand

The basics are easy to understand, but then you realize real world devices aren't so basic, and there are several layers of secondary effects to understand.
I won't pretend I understand them well, because I essentially never use audio transformers. This is a pretty good online resource which explains some of the considerations:
Audio Transformers chapter from Handbook for Sound Engineers (provided by Jensen Transformers)

what is the difference between a 1k:1k and 10k:10k transfomer? Are the values stated the nominal impedances on either side of the transformer for it to function as intended?

I think that would typically be the typical load impedance (the second value), and the reflected load impedance that the primary side driver would see (the first value). The primary side impedance would be the reflected load impedance, not something you would want to try to create as the source impedance of the driver (i.e. you would not want to artificially increase the driver impedance by inserting a series resistor).
As to why one might be labeled as a 1K:1K and another labeled as 10k:10k since both would be 1:1 turns ratio, I'm not sure. It might imply that the second one uses finer wire, and therefore has higher winding resistance, so you would have higher losses if you try to drive a lower load impedance. Admittedly that is a bit of a guess on my part, check the manufacturer's data sheet for recommended use conditions. I would defer to anyone actually using transformers on a regular basis if someone else has a more practical explanation of why transformers may be labeled with a particular impedance.

 

[Sam0311]

Thank you so much, unless I’ve completely misunderstood what you’ve said I feel like I’m having a bit of a breakthrough.

A transformer doesn’t have an inherit impedance, it’s reliant on the individual circuit. For output transformers, the intended load impedance will be known, so the impedances given will reflect that. The secondary impedance will be the intended load impedance and the primary will be whatever the load impedance is reflected on to the primary. This will be reversed for input transformers.

 

[dmp]

I have a ecoplate that I am in the process of rebuilding the electronics - look at the schematics here for the ecoplate by Jim Cunningham.
The driver has a sallen-key HPF followed by a bass and treble control. The pdf shows the frequency plot. This also takes values directly from the lm1875 datasheet and has some electrolytic values that are oversized (10uF and 2.2uF) since they are following the sallen-key filter. I am making these smaller in order to use film caps instead of needing BP caps. I don't really like the lack of voltage bias in this design across the electrolytics.
I am making the filter bypassable, since I usually use my plate with a MUCH less drastic EQ, or none. And since these electronics sit at the plate, I would rather use a EQ by my monitors instead of running into the next room with the plate. In mine I also added a L & R input (with ina137 receivers) followed by a ina134 summing stage (from the datasheet). Somewhat unnecessary however, since it is easy to sum in the daw, before the send
The receiver is a JFET opamp with some fixed gain , a level control, and then a line driver. I would not bother with any EQ in the preamp. You'll probably prefer to pre-EQ a plate. And the preamp will need to be near the plate / piezos and why would you want to be running over to your plate to EQ it?
The line receivers / drivers may be a better choice than transformers, since it doesn't seem you have a reason for designing to use transformers.

 

[Sam0311]

Oh awesome, thanks for this. I’ve yet to look at this in any great detail but it seems to be exactly what I need. Maybe I’ll need to alter it slightly to suit my needs but I imagine that will be mostly gain staging.

I’m not necessarily going to be using this with a daw which is why I wanted L/R ins.

Fair point about the EQ, I was uncertain which I would want so put it in pre and post. I probably won’t be running over to the plate though, it’ll likely be within reaching distance.

I went for transformers because (although I don’t think this anymore) they seemed a bit less daunting to try and understand, and let’s be honest, transformers are cool. But you’re right, maybe op amps would be better. THAT Corp’s ICs have been recommended to me in the past.

 

[Sam0311]

As well actually re the transformers, I was leaning towards getting some nice transformers for the output thinking I might prefer the sonic characteristics of them. I also have a couple of OEP 10k:10k transformers I was able to get for free that I thought I may as well use.

 

[dmp]

I probably won’t be running over to the plate though, it’ll likely be within reaching distance.

If you are listening to the plate through monitors and the plate is in the room (without sound isolation) it will feedback. The sound in the room will drive the plate. But if you build a box around it to isolate it can be in a control room.

maybe build on a protoboard to figure out what you want.
For building a plate, the electronics is a small part of the project. The metal is the hard part and where the magic is.

 

[analag]

https://groupdiy.com/threads/can-i-do-it.81407/
I built a plate reverb the other day. I kept the electronics flat meaning no EQ curve was built in. There is no need for that in todays world. I created the curve in my DAW. I kept all inputs and outputs transformer balanced. I also bought a 20W power amp kit on ebay for about $30 and I got parts to make a stereo 20Watt amps. I cut the cone out a 4 inch speaker and if you look at the link I provided you will see how I got it to work. I used low noise FET opamps with 1:10 input transformers to amplify the transceivers. I chose not to use piezo pickups. I am very happy with how it turned out.

 

[Sam0311]

If you are listening to the plate through monitors and the plate is in the room (without sound isolation) it will feedback. The sound in the room will drive the plate. But if you build a box around it to isolate it can be in a control room.

I very rarely actually use monitors but it’s a good point to be aware of, thank you.

 

[Sam0311]

https://groupdiy.com/threads/can-i-do-it.81407/
I built a plate reverb the other day.

Nice thread man thanks for sharing. Funny you took the cone off some speakers for your drivers, the thing that gave me the idea to pick this idea back is a opened a pair of headphones that had stopped working and thought it would probably be quite easy to repurpose the speakers inside and use a cheap headphone amp to drive to drive them.

I can’t tell from your thread what you’re using as pickups? Is it the same idea as the drivers but wired in reverse?

 

[analag]

I can’t tell from your thread what you’re using as pickups? Is it the same idea as the drivers but wired in reverse?

In the first pic. The four coneless tiny speakers are used as microphones. (I bought 4 but I'm only using 2) I glued the voice coil to the plate and attached the body of the speakers to the frame with L shaped aluminum (non magnetic material is important here) brackets. I set the whole thing up so the plate can vibrate freely. The pickup pre amp is a pair JFet opamp transformer coupled mic pre. I'm using LT1113 low noise opamps in two stages. The first stage is variable gain and the second stage is fixed gain and that drives an LT1010 buffer opamp which drives a 1:1 transformer.

 

[analag]

In the feeding of the plate I use a dynamic EQ plugin to control the resonant peak of the transducer/ plate combination. It's also nice to insert a limiter or better yet a clipper which helps to get and maintain the signal in the sweet spot. The plate return is also resonant suppressed to get rid of any metallic artefacts. In the originals the EQ curve is hardwired in the circuit, but with plugins I can surpass that easier and much more creatively.

 

[dmp]

I built a plate reverb the other day.

Real DIY being posted in the brewery?? WTF

Very interested to hear what a 4'x3' plate can sound like. Can you post a dry & wet sample? Mine is 8'x4' and it is tough to have the space for it.

 

[Sam0311]

Very clever, thanks for sharing!

 

[analag]

Real DIY being posted in the brewery?? WTF

Very interested to hear what a 4'x3' plate can sound like. Can you post a dry & wet sample? Mine is 8'x4' and it is tough to have the space for it.

Mine is 5'x4'. I suppose I can post a sample when I find the time.

 

[Sam0311]

Quick update on this. I've started ordering parts so I can start to get the non-electronics part of the build going. I've been hacking away at the circuit today. Trying to tidy up some of the things that we're mentioned earlier in this thread and improving on the EQ controls.

Might look into changing the output transformers at some point.

 

[Philbo King]

Following this thread.
A couple years ago I hung steel roofing under my deck so I could sit out there and play guitar while it rains. Stomping on the top of the deck creates an awesome sound like shaking a spring reverb, so I've been thinking about putting transducers on it to see if it could be used for reverb.