Hi,
This is split from a thread in Instruments and Amplifiers. I was hoping to get more suggestions on a particular aspect of a guitar amp build. Delete if not allowed.
I've been designing and building a guitar amp over the past few months and I've already completed the power amp and am currently working on the preamp. The power amp is totally silent, no hiss/hum or anything like that. However, I'm getting a fair amount of hiss from the preamp. When the master volume is turned up full (as you may do if you want a very clean sound) there is an unacceptable level of HF hiss.
I have attached the schematic in case anyone has any suggestions on how to improve it, or if I've missed out something obvious.
The hiss seems to be generated by U3.1 as grounding the input to U4.1 removes the problem.
Any suggestions would be greatly appreciated.
Thanks,
James
This is split from a thread in Instruments and Amplifiers. I was hoping to get more suggestions on a particular aspect of a guitar amp build. Delete if not allowed.
I've been designing and building a guitar amp over the past few months and I've already completed the power amp and am currently working on the preamp. The power amp is totally silent, no hiss/hum or anything like that. However, I'm getting a fair amount of hiss from the preamp. When the master volume is turned up full (as you may do if you want a very clean sound) there is an unacceptable level of HF hiss.
I have attached the schematic in case anyone has any suggestions on how to improve it, or if I've missed out something obvious.
The hiss seems to be generated by U3.1 as grounding the input to U4.1 removes the problem.
Any suggestions would be greatly appreciated.
Thanks,
James
There's a lot more before U4 then just U3. The total gain is extremely high,
and set with very high resistor values. Not at all how to make a low noise amplifier.
and set with very high resistor values. Not at all how to make a low noise amplifier.
Last edited:
Are P1...P4 the inputs? What are J3 and J4 connected to?
R24...R27 are fairly large noise sources straight in series with the input and R33 injects a substantial noise current above a few kHz, but having zero experience with electric guitars and their reverb units, I can't tell if they are really an issue.
R24...R27 are fairly large noise sources straight in series with the input and R33 injects a substantial noise current above a few kHz, but having zero experience with electric guitars and their reverb units, I can't tell if they are really an issue.
Hi,
Thanks for the quick replies, these are exactly the suggestions I'm looking for!
I should have mentioned that the noise generated from U3 is present with its inputs grounded (or the channel volumes turned down). I expect I need to rethink the gain structuring somehow but I'm having trouble wrapping my head around how best to do it. It needs to have lots of gain available so that it can overdrive the clipping circuit on U3 but that means that, at low gain, the signal is attenuated a lot before U3, you have the turn the master right up and so you amplify a lot of hiss.
P1-4 are the inputs, they should connect to J3 and 4 where the wire is broken off. This is just because the jacks are going to be mounted on a separate board.
Thanks,
James
Thanks for the quick replies, these are exactly the suggestions I'm looking for!
I should have mentioned that the noise generated from U3 is present with its inputs grounded (or the channel volumes turned down). I expect I need to rethink the gain structuring somehow but I'm having trouble wrapping my head around how best to do it. It needs to have lots of gain available so that it can overdrive the clipping circuit on U3 but that means that, at low gain, the signal is attenuated a lot before U3, you have the turn the master right up and so you amplify a lot of hiss.
P1-4 are the inputs, they should connect to J3 and 4 where the wire is broken off. This is just because the jacks are going to be mounted on a separate board.
Thanks,
James
Ok, then scale down the impedances for U3 by a factor of ten (same gain, lower noise). Diodes remain the same.
56k --> 5.6k
560k --> 56k
22pF --> 220pF
56k --> 5.6k
560k --> 56k
22pF --> 220pF
Last edited:
As you have the problem with the volume potmeters turned way down, you could put more gain in front of and less after the volume potmeters.
And use low noise opamps for the frontend - TL072 isn't the quietest for low source impedances. If you pick a bipolar opamp for low voltage noise, make sure it has low current noise too (at least as good as the NE5534A). Inductive pickups have a lot of inductive reactance that will multiply with the current noise of the amp's input...
Series resistance is the enemy of low noise design, keep any series resistances at the front-end as small as possible (100 ohms sort of value, not 10k for instance). R3 is a big problem here - you need to lightly load the previous tone circuit, but a large resistance is not the way to achieve this if you want low noise, you need to buffer the tone circuit output, then the 100k/560k gain stage can use 1k/5k6 network and be quiet.
The same goes for feedback networks - that 200k/10k first feedback network can be 2k/100R for instance. Most opamps can drive 2k loads no problem, many can drive 600 ohms well (NE5534A for instance), so that is the lower limit for circuit impedances.
Series resistance is the enemy of low noise design, keep any series resistances at the front-end as small as possible (100 ohms sort of value, not 10k for instance). R3 is a big problem here - you need to lightly load the previous tone circuit, but a large resistance is not the way to achieve this if you want low noise, you need to buffer the tone circuit output, then the 100k/560k gain stage can use 1k/5k6 network and be quiet.
The same goes for feedback networks - that 200k/10k first feedback network can be 2k/100R for instance. Most opamps can drive 2k loads no problem, many can drive 600 ohms well (NE5534A for instance), so that is the lower limit for circuit impedances.
Last edited:
Hi,
I've had some time to make a few changes this morning and the hiss is now substantially reduced. I decreased the gain and lowered the feedback network impedances in U3. I noticed that I'd actually been using a 220p compensation cap, not 22p, and since I was happy with the sound, I scaled this to 2n2. I then increased the gains in the input stages by changing R5 from 220k to 680k and R11 from 150k to 470k. I also approximately doubled the gain in U2.1 by changing R3 from 100k to 56k. I added two extra LEDs in the feedback network to increase the clipping threshold at that stage. Since there is now more gain in the front end and less in U3.1, this had to be done to restore the clipping thresholds to where they should be.
This seems to have gone a long way to reducing the noise just by restructuring the gain but I will continue with your suggestions to attempt to improve it further. Mark, I just read your reply regarding impedances after I'd made these changes.
Do you suggest adding an op-amp buffer after VR1? Or an emitter follower? Would you also suggest this after VR4 and VR2 as both of these go into 56k source impedances.
The reason I've picked the TL072 for the front-end is because I need a 1M input impedance for guitar pickups. I don't think this would be achievable with an NE5532 or am I missing something? I could, of course, swap some of the other op-amps for NE5532 to see if this improves matters. I'll have to order some as I've run out.
Thanks,
James
I've had some time to make a few changes this morning and the hiss is now substantially reduced. I decreased the gain and lowered the feedback network impedances in U3. I noticed that I'd actually been using a 220p compensation cap, not 22p, and since I was happy with the sound, I scaled this to 2n2. I then increased the gains in the input stages by changing R5 from 220k to 680k and R11 from 150k to 470k. I also approximately doubled the gain in U2.1 by changing R3 from 100k to 56k. I added two extra LEDs in the feedback network to increase the clipping threshold at that stage. Since there is now more gain in the front end and less in U3.1, this had to be done to restore the clipping thresholds to where they should be.
This seems to have gone a long way to reducing the noise just by restructuring the gain but I will continue with your suggestions to attempt to improve it further. Mark, I just read your reply regarding impedances after I'd made these changes.
Do you suggest adding an op-amp buffer after VR1? Or an emitter follower? Would you also suggest this after VR4 and VR2 as both of these go into 56k source impedances.
The reason I've picked the TL072 for the front-end is because I need a 1M input impedance for guitar pickups. I don't think this would be achievable with an NE5532 or am I missing something? I could, of course, swap some of the other op-amps for NE5532 to see if this improves matters. I'll have to order some as I've run out.
Thanks,
James
I've used a differential configuration there so that the two inputs are in phase. One input has an extra inverting gain stage and the two inputs may be jumped together.
What does the impedance of a guitar pick-up look like over the audio frequency band?
You can make a 1 Mohm input impedance with an NE5532, but what you can't do (at least not without tricks that require additional active parts) is to reduce the noise current below the 0.7 pA/sqrt(Hz) of the op-amp. When the source impedance is greater than (18 nV/sqrt(Hz))/(0.7 pA/sqrt(Hz)) ~= 25.7 kohm, the 0.7 pA/sqrt(Hz) noise current of an NE5532 will do more harm than the 18 nV/sqrt(Hz) noise voltage of a TL072.
Modern FET op-amps have far less noise voltage than a TL072 and still have negligible noise current. For example, 8 nV/sqrt(Hz) for an OPA2134 or 4 nV/sqrt(Hz) for an OPA2156.
Reading around, this trade off between noise voltage and noise current, considering the high guitar pickup impedance, seems to come up a lot. Most seem to recommend that the FET op amp ends up being quieter.
I wonder if it would be worth using a discrete FET stage on the input, going into an NE5532.
I wonder if it would be worth using a discrete FET stage on the input, going into an NE5532.
That can work very well when you use a discrete FET with a fairly high transconductance and not too much 1/f noise.
Hi,
I've now made several adaptations, based on suggestions, and the hiss is drastically reduced. I also prefer the sound of the amp now (ready to be told it's just placebo effect!)
I made the following changes:
-Replace U1 and U2 with OPA2134
-Scale down feedback impedances on U1 and U3 by a factor of 100
-Replace U3 and U4 with NE5532 and U5 with NE5534
-Add resistors from non-inverting input to ground on inverting bipolar stages
-Increase value of compensation cap on U5 to 470p
-Increase R37 to 47k
-Add 1u capacitor between U4.1 output and master volume pot
I'm fairly confident that the design is finalised now but any further suggestions are of course welcome!
Thanks,
James
I've now made several adaptations, based on suggestions, and the hiss is drastically reduced. I also prefer the sound of the amp now (ready to be told it's just placebo effect!)
I made the following changes:
-Replace U1 and U2 with OPA2134
-Scale down feedback impedances on U1 and U3 by a factor of 100
-Replace U3 and U4 with NE5532 and U5 with NE5534
-Add resistors from non-inverting input to ground on inverting bipolar stages
-Increase value of compensation cap on U5 to 470p
-Increase R37 to 47k
-Add 1u capacitor between U4.1 output and master volume pot
I'm fairly confident that the design is finalised now but any further suggestions are of course welcome!
Thanks,
James
You might want a bleeder resistor on the inputs (10M or so) to reduce cracking on plugging in? Something like 100 ohm series on the input to protect the opamp inputs might be nice to have, help reduce RF getting direct to the inputs too, although that's less problematic with FET front end.
Hi,
The bleeder resistor presumably goes from before C1 to ground. Is that correct?
Whereabouts would you put the series resistor on the input?
Thanks,
James
The bleeder resistor presumably goes from before C1 to ground. Is that correct?
Whereabouts would you put the series resistor on the input?
Thanks,
James