Fender FR-10/FR-12 noise reduction mod

[Lysander]

Came up with a relatively easy mod to kill pretty much all hum on the FR-12! And, given it uses the exact same preamp board, this should work on the FR-10 just as well. Read on for details

To preface things, this IS a noisy amp. The culprit is not so much the Class D power section (which is in fact very good quality!), but the preamp - the active EQ section, in particular. The preamp is surprisingly old-school, both in its through-hole PCB construction, and design, which is basically a long chain of opamp gain stages.

The problem is that the EQ section alone uses 12 opamp stages, all built around TL084 chips. This is a low-cost part which is also technically "low noise", but its noise performance is pretty poor by modern standards. So, when you stack 12 gain stages worth of these in a row, each slightly amplifying the noise from the previous one, it creates very audible hiss at the output. The EQ controls and volume will do little to tame it, and the high cut control can help a bit - but won't do wonders.

As a result, this amp has a noise floor of ~34 dBA without input, which is exceedingly high. The measurements below are for the amp off, and then on, with volume at 4, EQ flat, and high cut at 0, in a very quiet room, and ~1" away from the grill. Note that my meter cannot measure levels below 30 dBA.

WKSmith at the Fractal Audio Forums started looking into this issue pretty much at the same time i did, and came to the same conclusions. His approach to a fix was to add a switch to bypass the EQ section entirely, and look into redesigning the EQ section from scratch; i wanted something that would tame noise while retaining EQ, which, noise aside, i find really useful on the FR-12.

So, i got to work.

 

[Mongillo19]

Came up with a relatively easy mod to kill pretty much all hum on the FR-12! And, given the fact they share the exact same preamp board, this should work on the FR-10 just as well. Read on for details

To preface things, this IS a noisy amp. The culprit is not so much the Class D power section (which is in fact very good quality!), but the preamp - and the active EQ section, in particular. The preamp is surprisingly old-school, both in its through-hole PCB construction, and its design, which is basically a long chain of opamp gain stages.

The problem is that the EQ section alone uses 12 opamp stages, all built around TL084 chips. This is a low-cost part which is also technically "low noise", but its noise performance is pretty poor by modern standards. So, when you stack 12 gain stages worth of these in a row, each slightly amplifying the noise from the previous one, it creates very audible hiss at the output. The EQ controls and volume will do little to tame it, and the high cut control can help a bit - but won't do wonders.

View attachment 12980

As a result, this amp has a noise floor of be no less than 4.3 dBA without input, which is exceedingly high. The measurements below are for the amp off, and then on, with volume at 4, EQ flat, and high cut at 0, in a very quiet room, and ~1" away from the grill. Note that my meter cannot measure levels below 30 dBA.

View attachment 12981View attachment 12982

WKSmith at the Fractal Audio Forums started looking into this issue pretty much at the same time i did, and came to the same conclusions. His approach to a fix was to add a switch to bypass the EQ section entirely, and look into redesigning the EQ section from scratch; i wanted something that would tame noise while retaining EQ, which, noise aside, i find really useful on the FR-12.

So, i got to work.

More coming soon...

 

[Lysander]

My idea was simply to replace opamp ICs with modern, high quality low-noise versions.

• On the EQ section, i chose to replace all TL084 ICs with OPA1604. This is an ultra low noise part - about 7x lower than the TL0xx line.
• On the input stage and output crossover, both using RC4558, i replaced the ICs with OPA1656.

Unfortunately, neither part is available in DIP packaging (through-hole), so i had to solder adapters for them, which took most of the time for this mod.

Note that, while recommended, replacing the RC4558s is not really necessary to tame hiss. I tried changing these ICs alone first, and noise levels were unaffected. WKSmith used a pair of OPA2134s, which is a similar to OPA1656, and reported tonal improvements; i didn't notice any on my amp.

The total cost was about $50 USD ordered from Mouser, with most of the budget going into the OPA1604s and OPA1656s. Both are considered "premium" performance ICs, and since i had to order new parts anyway, i searched for the lowest noise, audio-oriented alternative i could find. I reckon any low-noise opamp pin-compatible with the TL084 (and RC4558, if you care) will do just fine here; these are extremely cheap general use ICs, easily to source anywhere in the world - which is likely why Fender chose them in the first place.

Replacing the ICs on the preamp board was painless. Given its though-hole construction and relatively low component count, the board is a pleasure to work with. I had all five replacements installed in about 1h - including a quick test for the smaller chips getting replaced first.

And... voilà. The result is STAGGERING: hiss has became basically inaudible. Tone and EQ behavior remains the same, but the amp is so silent now that i cannot measure hiss over ambient noise, even if i place the sound meter right next to the grill. Take just a couple steps back, and the amp is dead silent.

So, that's it! If you have experience with a solder iron, replacing three chips is all that it takes to tame the FR-12's hiss down to normal, reasonable levels. I really liked this "FRFR" amp before, but i just love it now.

I really hope Fender acknowledges this as a flaw, sits down and redesigns the EQ section in the future. It's the sole weak link on what's otherwise a very well built, great sounding amp solution for modelers.

2023-11-03 EDIT: @norpe confirms that OPA4227, available in DIP packaging as OPA4227PA, works perfectly as an alternative replacement
2023-11-06 EDIT: @aflynt reports that OPA4228 (a faster version of OPA4227) also works, but only if a subset of the ICs are replaced, as this opamp is not unity-gain stable. OPA4227 is still the recommended DIP replacement.
2024-01-11 EDIT: Fender is now sneakily applying this mod on production amps: FR-10, FR-12 There's no official acknowledgement of the issue, nor any way to identify a "factory mod" amp vs. an original other than opening it up.

 

[Alex Kenivel]

 

[Stone]

Thats Fantastic , Hope Fender takes note of this

Great Job

 

[Piing]

Great job! I think that this is what I will do when I get my FR-12, in the case that it is noisy

 

[nolesj]

My idea was simply to replace opamp ICs with modern, high quality low-noise versions.

• On the EQ section, i chose to replace all TL084s ICs with OPA1604s. This is an ultra low noise part - about 7x lower than the TL0xx line.
• On the input stage and output crossover, both using RC4558s, i replaced ICs with OPA1656s. Note this is not really necessary; i tested the mod changing just these two chips first, and noise levels were unaffected. It's a much better quality part, though. WKSmith used a OPA2134 here, which is a very similar part, and reported tonal improvements. I didn't notice much of a difference on my amp though.

Unfortunately, neither part is available in DIP packaging (through-hole), so i had to solder adapters for them, which took most of the time for this mod.

View attachment 12985

The total cost was about $50 USD ordered from Mouser, with most of the budget going into the OPA1604s and OPA1656s. Both ICs are considered "premium" performance parts, and since i had to order new ICs anyway, i searched for the lowest noise, audio-oriented alternative i could find. I reckon any low-noise opamp pin-compatible with the TL084 (and RC4558, if you care) will do just fine here; these are extremely cheap general use ICs, easily to source anywhere in the word - which is likely why Fender chose them in the first place.

Replacing the ICs on the preamp board was painless. Given its though-hole construction and relatively low component count, the board is a pleasure to work with. I had all five replacements installed in about 1h - including a quick test for the smaller chips getting replaced first.

View attachment 12986

And... voila. The result is STAGGERING: hiss has became basically inaudible. Tone and EQ behavior remains the same, but the amp is so silent now that i cannot measure hiss over ambient noise, even if i place the sound meter right next to the grill.

View attachment 12987

So, that's it! If you experience with a solder iron, replacing three chips is all that it takes to tame the FR-12's hiss down to normal, reasonable levels. I really liked this "FRFR" amp before, but i just love it now.

I really hope Fender will sit and redesign the EQ section in the future. It's the sole weak link on what's otherwise a very well built, great sounding amp solution for modelers.

Great work! One question though- the wires connected at the board labeled “to pwr amp” and “3 pin plug”- how do you disconnect those? They don’t appear to be actual connectors.

 

[Lysander]

Great work! One question though- the wires connected at the board labeled “to pwr amp” and “3 pin plug”- how do you disconnect those? They don’t appear to be actual connectors.

Those are soldered in. That plastic base looks like a connector, but it appears to be just a protection so the solder point isn't stressed by the cables flexing. The actual connectors are at the other end, going into the power amp sensing / passthrough board.

I never ever considered desoldering those, tbh. The board can be easily worked on even with a few cables sticking out of it.

 

[JCW308]

Dang dude, Fender needs to PAY you for this! Great job!

 

[Lysander]

Dang dude, Fender needs to PAY you for this! Great job!

I'll take my pay in freshly minted Stomps, thank you!

Honestly, Fender doesn't even need to switch to expensive opamp ICs to fix this issue. Just redesign the EQ section so it uses half the gain stages (which is completely doable!), and even the lowly, cheap TL084 will perform fine here.

 

[aflynt]

That’s awesome! Did you socket the chips? Any sonic differences? Wondering if it’d be worth splurging on something like 3 $10 OPA4227PA or OPA4228PA chips to avoid having to do the SMD stuff…

-Aaron

 

[Lysander]

That’s awesome! Did you socket the chips?

I was planning to, but it turns out the SOIC adapters i use wouldn't fit the sockets, so i ended up just soldering them in

Any sonic differences?

So far i couldn't detect any, other than hiss being basically gone. Unnecessary complexity aside, this is a relatively straightforward circuit, so opamp selection should really not affect tone much - if at all.

Wondering if it’d be worth splurging on something like 3 $10 OPA4227PA chips to avoid having to do the SMD stuff…

Yeah, i think that part should work just fine - specs look compatible with the TL084, and it is very low noise too (3 nV/√Hz). Besides price, the main reason i went with SOIC parts is that i wanted a good quality JFET opamp for the input as well, and had a really hard time finding a suitable part in DIP packaging. There's simply way more options available if you're willing to solder behind a magnifying glass.

 

[aflynt]

I was planning to, but it turns out the SOIC adapters i use wouldn't fit the sockets, so i ended up just soldering them in



So far i couldn't detect any, other than hiss being basically gone. Unnecessary complexity aside, this is a relatively straightforward circuit, so opamp selection should really not affect tone much - if at all.



Yeah, i think that part should work just fine - specs look compatible with the TL084, and it is very low noise too (3 nV/√Hz). Besides price, the main reason i went with SOIC parts is that i wanted a good quality JFET opamp for the input as well, and had a really hard time finding a suitable part in DIP packaging. There's simply way more options available if you're willing to solder behind a magnifying glass.

I’m thinking I might get 3 of the OPA4228PA since they have a higher slew rate. Not sure if that matters in this circuit though. From your findings it seems like it might not be worth it from a hiss perspective to replace the 4558’s right?

-Aaron

 

[Lysander]

I’m thinking I might get 3 of the OPA4228PA since they have a higher slew rate. Not sure if that matters in this circuit though.

Some people swear by fast opamps in audio, but . More will never hurt, but the 10 V/us of the OPA4227PA should be more than enough on the EQ section. For reference, the RC4558, which has a (sonically) way more critical role in the differential input for the preamp, can "only" do 1.7 V/us.

From your findings it seems like it might not be worth it from a hiss perspective to replace the 4558’s right?

Nah, it won't do much for hiss. 4558s are used only for the XLR combo input + volume, and at the very output to crossover the signal to the main power and tweeter amps.

There're much better ICs for these roles than the RC4558, so it's not a bad idea to replace them as well if you're opening the amp up. If you want to do this, use a JFET-based opamp. OPA2134 is a very good option in DIP package, though it's becoming harder to source these days.

 

[aflynt]

Some people swear by fast opamps in audio, but . More will never hurt, but the 10 V/us of the OPA4227PA should be more than enough on the EQ section. For reference, the RC4558, which has a (sonically) way more critical role in the differential input for the preamp, can "only" do 1.7 V/us.



Nah, it won't do much for hiss. 4558s are used only for the XLR combo input + volume, and at the very output to crossover the signal to the main power and tweeter amps.

There're much better ICs for these roles than the RC4558, so it's not a bad idea to replace them as well if you're opening the amp up. If you want to do this, use a JFET-based opamp. OPA2134 is a very good option in DIP package, though it's becoming harder to source these days.

Thanks!

I ordered 3 OPA4228PA chips and sockets from Mouser. The data sheet has the slew rate at 2.3V/us for the OPAx227 and 11V/us for the OPAx228. I should have time to swap them in next weekend.

-Aaron

 

[SwirlyMaple]

My idea was simply to replace opamp ICs with modern, high quality low-noise versions.

• On the EQ section, i chose to replace all TL084s ICs with OPA1604s. This is an ultra low noise part - about 7x lower than the TL0xx line.
• On the input stage and output crossover, both using RC4558s, i replaced ICs with OPA1656s. Note this is not really necessary; i tested the mod changing just these two chips first, and noise levels were unaffected. It's a much better quality part, though. WKSmith used a OPA2134 here, which is a very similar part, and reported tonal improvements. I didn't notice much of a difference on my amp though.

Unfortunately, neither part is available in DIP packaging (through-hole), so i had to solder adapters for them, which took most of the time for this mod.

View attachment 13002

The total cost was about $50 USD ordered from Mouser, with most of the budget going into the OPA1604s and OPA1656s. Both are considered "premium" performance ICs, and since i had to order new parts anyway, i searched for the lowest noise, audio-oriented alternative i could find. I reckon any low-noise opamp pin-compatible with the TL084 (and RC4558, if you care) will do just fine here; these are extremely cheap general use ICs, easily to source anywhere in the world - which is likely why Fender chose them in the first place.

Replacing the ICs on the preamp board was painless. Given its though-hole construction and relatively low component count, the board is a pleasure to work with. I had all five replacements installed in about 1h - including a quick test for the smaller chips getting replaced first.

View attachment 13003

And... voila. The result is STAGGERING: hiss has became basically inaudible. Tone and EQ behavior remains the same, but the amp is so silent now that i cannot measure hiss over ambient noise, even if i place the sound meter right next to the grill. Take just a couple steps back, and the amp is dead silent.

View attachment 13004

So, that's it! If you have experience with a solder iron, replacing three chips is all that it takes to tame the FR-12's hiss down to normal, reasonable levels. I really liked this "FRFR" amp before, but i just love it now.

I really hope Fender acknowledges this as a flaw, sits down and redesigns the EQ section in the future. It's the sole weak link on what's otherwise a very well built, great sounding amp solution for modelers.

Well done! This is the kind of thing that happens when a company splits hairs over every last cent of parts cost. Some of my buddies used to work as EEs in a similar large-corp world, and many of their stories of job frustration were about things like this. They spent more time reducing the cost of their designs than making the designs in the first place.

This makes me wonder what corners were cut in the TMP…

 

[the-trooper]

So, that's it! If you have experience with a solder iron, replacing three chips is all that it takes to tame the FR-12's hiss down to normal, reasonable levels

Or just, you know, don't support companies selling shit products by not buying them.

 

[quadgear]

Great outcome here for the DIY crowd.

Wonder how this design made it through the process? Perhaps the mindset of building an amp (versus an "FRFR"/monitor) biased the team into thinking it was good enough (especially for a $500 price point)?

Will be interesting to see how Fender responds. Official line today is "it's normal and as designed".

Will there be a design/parts change for the production line?

Will there be a retrofit board for units already shipped? Or a warranty recall/replacement?

 

[quadgear]

Would variations in the fab of TL084's be a likely reason why the hiss is louder for some people and quieter for others?

I assume the SNR will vary across different batches of chips, and some people got lucky?

 

[Lysander]

Wonder how this design made it through the process? Perhaps the mindset of building an amp (versus an "FRFR"/monitor) biased the team into thinking it was good enough (especially for a $500 price point)?

I think the design was deemed good enough for the intended price point, which is normally the way these things go...

Will be interesting to see how Fender responds. Official line today is "it's normal and as designed". Will there be a design/parts change for the production line?

Eh, i wouldn't hold my breath on it. We might see a preamp board MKII down the line, but i doubt Fender will ever go back and provide a "fix" for existing users.

Would variations in the fab of TL084's be a likely reason why the hiss is louder for some people and quieter for others? I assume the SNR will vary across different batches of chips, and some people got lucky?

Someone asked the same on the Fractal forums earlier today.

While it is possible to find minimal variances (within tolerances) across chips, the TL084 is an extremely common part, manufactured by many companies all over the globe; it's capabilities, and build tolerances, are very well known by now. And, while the cascaded design of the EQ means that small opamp noise floor differences can significantly impact the output... i just don't think that's the case at all. All chips on my FR-12 were from Texas Instruments, which is an excellent brand.

It's just an inherently noisy preamp design, coupled with cheap ICs with average noise performance As noted, Fender could easily tackle this issue by removing half a dozen gain stages in the EQ section while keeping the existing chips in.

I believe that inconsistent noise reports are just people's different tolerances to noise, or differences in how controls (high cut, in particular) are dialed.