Another nice one from Scott Hamptone

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I hope I am not committing sacrilege by saying this but I am not terribly impressed with Scott's design.

The gain make up preamp connected to the tank output needs to be low noise but Scott chooses an inverting op amp stage with a 100K input resistor. This resistor is in series with the output impedance of the tank as far as noise is concerned so the noise from the 100K is added to the noise from the tank. This very easy ti fix,  Just change it to a non-inverting stage with 1ooK from the + input to ground. The 100K is now in parallel with the tank output impedance and does not increase noise. While you are at it you can change the feedback network from 1M and 100K to 10K and 1K which will reduce the noise generated by this network by 20dB.

The mixer G2 sums the wet and dry signals. Here he uses a non-inverting stage when a virtual earth mixer would have been better. The signals are line level so noise is not a problem so you can safely use the inverting stage a virtual earth mixer requires. He uses a 5.23K feed resistor for the wet and 47K for the dry. This appears to be to get 20dB gain for the wet and 0dB gain for the dry but it also causes serious interaction between the wet and dry controls. I am not sure if this is deliberate but it is certainly odd. A VE mixer would do the job without the interaction

Lastly, a minoe point, he uses an inverting satge for the tank drive circuit. Nothing wrong with this except he adds EQ to the input arm which means the input impedance varies with frequency, The data sheet for the part recommend a non-inverting set up and again it is easy to convert it and avoid a variable input impedance.

As I said, I am not sure if this is sacrilege as Scott is well respected.

Cheers

ian
 
I´d say noone should be beyond reasonable criticism  ;)


You have good points ruffrecords, for me the interesting thing is drive with power amplifier for lower recovery gain (and noise etc.).

I think the focus in the Hamptone DIY designs is on SIMPLE.
Same with the FET preamp, I guess people are reading esoteric stuff into it...


I have good experience with the ETI reverb (I built it on stripboard with IC make-up, and I won´t do it again :eek: ):
http://www.diystompboxes.com/smfforum/index.php?topic=111671.msg1028607#msg1028607
It´s just much more complex :p
 
ruffrecords said:
X chooses an inverting op amp stage with a 100K input resistor. This resistor is in series with the output impedance of the tank as far as noise is concerned so the noise from the 100K is added to the noise from the tank. This very easy ti fix,  Just change it to a non-inverting stage with 1ooK from the + input to ground. The 100K is now in parallel with the tank output impedance and does not increase noise. While you are at it you can change the feedback network from 1M and 100K to 10K and 1K which will reduce the noise generated by this network by 20dB.

Is this in reference to the johnson noise that is produced by the resistors?

 
buildafriend said:
Is this in reference to the johnson noise that is produced by the resistors?

  That and there's also the opamp current noise multiplied by the resistor, depending on the opamp and the chosen level one might be  higher than the other.

  Notice than the opamp also has a voltage noise that remains the same despite the resistors around it, if using very low value resistors and the voltage noise of the opamp becomes higher than the other two factors there's not reason to make them any lower, is not like you put a short circuit in the feedback and you get a zero noise voltage follower.

JS
 
buildafriend said:
Is this in reference to the johnson noise that is produced by the resistors?

Mostly although as other have pointed out there are other sources of noise. The noise in a 100K resistor is just below -100dBu, The preamp has a gain of 20dB which raises it to just below -80dBu and there is up to another 20dB gain in the mix amp which raises it again to just below -60dBu and that is before you add in any noise due to the first op amp itself.

Edit: I just look up some tank specs and the highest DCR figure I can fine is 800 ohms. So, if you change to a non-inverting topology, the 100K is now in parallel with 800 ohms so the noise resistance is near enough 800 ohms. The noise in 800 ohms is below -120dBu, in other words 20dB less than in the original.

Cheers

Ian
 
ruffrecords said:
.......This very easy ti fix,  Just change it to a non-inverting stage with 1ooK from the + input to ground. The 100K is now in parallel with the tank output impedance and does not increase noise. While you are at it you can change the feedback network from 1M and 100K to 10K and 1K which will reduce the noise generated by this network by 20dB.

This isn't easy fix because the signal should be inverted for inphase mixing in the next stage (or wires inside the tank should be reversed if it is possible).

The mixer G2 sums the wet and dry signals. Here he uses a non-inverting stage when a virtual earth mixer would have been better. The signals are line level so noise is not a problem so you can safely use the inverting stage a virtual earth mixer requires. He uses a 5.23K feed resistor for the wet and 47K for the dry. This appears to be to get 20dB gain for the wet and 0dB gain for the dry but it also causes serious interaction between the wet and dry controls. I am not sure if this is deliberate but it is certainly odd. A VE mixer would do the job without the interaction

He obviously likes to not flip the phase when the circuit is inserted in the chain.

Lastly, a minoe point, he uses an inverting satge for the tank drive circuit. Nothing wrong with this except he adds EQ to the input arm which means the input impedance varies with frequency, The data sheet for the part recommend a non-inverting set up and again it is easy to convert it and avoid a variable input impedance.

He noticed that in the article.  I like his simple design of EQ-ed voltage (power) amp. Anyway, this circuit is a part of musical instrument, not a module in AP. Nice article in overall (IMO), with the accent on  most problematic parts for DIY.
 
Couldn't one flip the phase pretty easily somewhere else to make up for the inverted signal while not ending up with higher "noise?"

There is a great chart in Douglas Self's small signal design book that charts out Johnson noise in resistors by R value. It looks as Ian explained, the higher the R value the more Johnson noise. This is a fairly serious thing to look at when working with low THD+N specs like that of modern op amps. Maybe not as much in a tube amp..? But overall it's noteworthy theory.

 
buildafriend said:
Couldn't one flip the phase pretty easily somewhere else to make up for the inverted signal while not ending up with higher "noise?"
Yes you can. He has an inverting preamp followed by a non-inverting mixer so overall on phase flip. I suggested a non-inverting stage followed by an (inverting) virtual earth mixer so also overall on phase flip.
There is a great chart in Douglas Self's small signal design book that charts out Johnson noise in resistors by R value. It looks as Ian explained, the higher the R value the more Johnson noise. This is a fairly serious thing to look at when working with low THD+N specs like that of modern op amps. Maybe not as much in a tube amp..? But overall it's noteworthy theory.

Exactly. We tend to use 10K bus resistors for mixing but we are working at line level. Using a 100K resistor increases the noise by 10dB, which might not be too bad at line levels, but in the tank preamp we are talking levels more like mic levels so we cannot afford any unnecessary noise sources.

Cheers

Ian
 
My gut reaction is the inherent noise floor of a spring tank system will swamp any contribution from the amps. 
 
ruffrecords said:
Yes you can. He has an inverting preamp followed by a non-inverting mixer so overall on phase flip. I suggested a non-inverting stage followed by an (inverting) virtual earth mixer so also overall on phase flip.

As I said, this way direct signal will be mixed with out of phase reverb signal, no matter which type of summer is used.

My gut reaction is the inherent noise floor of a spring tank system will swamp any contribution from the amps.

For sure.
 
Almost every design ever done, could be tweaked a little better, but sometimes it does not make an audible improvement.

It is good mental exercise to understand best design practices, but that does not mean every design that doesn't optimize every single link is inferior.... 

If the design is well respected maybe it already sounds OK as is.

JR

PS: I am too lazy to look up the schematic to second guess it.  When doing scratch designs do everything as good as you can.  When copying an existing design be careful when second guessing the original designer's decisions.
 
In this case it's clearly described as a DIY simplicity reduction of a product he sells, which he's lived with and bench tested extensively. 
 
emrr said:
In this case it's clearly described as a DIY simplicity reduction of a product he sells, which he's lived with and bench tested extensively.
Or maybe he intentionally made it lower fidelity than his full price version...  :eek:

JR
 
moamps said:
As I said, this way direct signal will be mixed with out of phase reverb signal, no matter which type of summer is used.

For sure.

But phase is a non issue. The reverb signal is delayed which itself is equivalent to a huge phase shift.

Cheers

Ian
 
Reverb is inherently random phase.

The reverb tank is not so very hissy; it is indeed an 800 Ohm resistor. The usual reverb hiss is hasty implementation: insufficient drive, and high-hiss recovery amp or mix-stage.
 
PRR said:
Reverb is inherently random phase.
Then why make the make-up amp inverting, where high impedance/low noise  and no mixing is needed?
PRR said:
The reverb tank is not so very hissy; it is indeed an 800 Ohm resistor. The usual reverb hiss is hasty implementation: insufficient drive, and high-hiss recovery amp or mix-stage.
I think driving the tank with a 20watt amp is something I haven´t thought off (LM386=1watt max.). Curious about the benefits. Someone ( ;D) has to build it and report!
 
L´Andratté said:
Then why make the make-up amp inverting, where high impedance and no mixing is needed?I think driving the tank with a 20watt amp is something I haven´t thought off (LM386=1watt max.). Curious about the benefits. Someone ( ;D) has to build it and report!

Precisely. There is no need for it to be.

Cheers

Ian
 
Made me look.... Agreed, not optimal. Does it matter ???

Back when I ran my kit business, EE students would be required to change some circuit for class credit. Maybe he engineered in a number of circuit features to change?  ;D

JR
 
Back
Top