Building a Solid State Amp. Really?

[cometazzi]

I don't have time to read through this right now, but I'll say I've had a lot of fun with the TDA2005 and other chipamps over the years. They can be great fun and can sound good too.


I'll come back probably tomorrow night and read the thread, but for now I want to state that I applaud this effort!

 

[TequilaCaster]

The Boss Katana amps use chipamps too...ST TDA7293
https://www.thegearpage.net/board/index.php?threads/boss-katana-amplifiers.1744135/post-25760197

It may be easier to do current feedback (as opposed to voltage feedback) to get more of a tube-ampish feel with a chipamp than a class-D amp, but I don't really know... just speculating.

 

[printer2]

The Boss Katana amps use chipamps too...ST TDA7293
https://www.thegearpage.net/board/index.php?threads/boss-katana-amplifiers.1744135/post-25760197

It may be easier to do current feedback (as opposed to voltage feedback) to get more of a tube-ampish feel with a chipamp than a class-D amp, but I don't really know... just speculating.

Shoot, just when I wanted to start stuffing parts. I guess I will have to read up about current feedback, was not going to bother. I wanted to keep this simple and if it turns out acceptable enough others could use the design. Class D is no different than an Class AB amp as long as you have a NFB path, I think.

 

[TequilaCaster]

Elliott Sound Products succinctly discusses SS guitar amp voltage vs. current and mixed mode feedback in this article... https://www.sound-au.com/articles/guitar-amps.htm

The Fender Frontman 25R used it...
https://guitarwork.ru/electronic/firm/Amplifier/Fender/Fender Frontman 25R.GIF

 

[cometazzi]

A lot of small-to medium guitar practice amps and small up to ginormous bass amps have been using chipamps since about the late 1980s. From a mass-production standpoint they are win-win, so why wouldn't they?

I love tube amps as much as all the rest of us here, but I've been saying for decades that it's time to drop the bad reputation that solid state guitar amps have. At least don't paint them all with a broad brush. Obviously, SS amps started off on the wrong foot because the early ones (late 60s?) were terrible, especially when it came to overdrive. They were designed by Engineers, not guitar players, so that's no surprise.

Today, that lives on a little bit in the small, cheap practice amps because they're designed to be cheap. Small speakers and basic drive circuits don't lend well to great tone. That said, there are some SS amps that sound fantastic. High gain is something SS amps do best, but I've had a Hughes and Kettner Blue 60 that had an incredible clean tone and was an amazing pedal platform. The gain channel was decent. Not awesome, but decent. I've currently (still) got a 1976 Peavey Pacer 100-SS that can do that clean-to-dirty transition as well as some tube amps. Nobody believes me when I say that until I show them! It also has a great clean, great reverb, and the distortion is tubey, if not a bit dry. Hitting it with a boost or Tube Screamer really livens it up. Full tilt it is a terror.

Back to DIY stuff, I actually started with SS chipamps before getting into pedals (unlike like most do). I started with chipamps, the LM386 being the first. Then bridged them, then moved up the LM-series and TDA-series ladders. I learned about power supplies by having to build them to feed these chips.

I've got a few Class-D chips but I haven't played with them yet. One of the last things I was doing before I transitioned to building tube gear was taking the tube-to-jfet designs from runoffgrove.com and pairing them up with a chipamp back-end. My digging around in the forums here over the last couple of years has turned up a few other examples of people doing the same.

It's great fun! This thread is timely, because I was recently curious about what would happen if I posted a SS amp build here.

I'm interest to see how this one turns out.

 

[printer2]

Elliott Sound Products succinctly discusses SS guitar amp voltage vs. current and mixed mode feedback in this article... https://www.sound-au.com/articles/guitar-amps.htm

The Fender Frontman 25R used it...
https://guitarwork.ru/electronic/firm/Amplifier/Fender/Fender Frontman 25R.GIF

But how do you approach it with a bridged amplifier?

 

[TequilaCaster]

But how do you approach it with a bridged amplifier?

https://xkcd.com/28/

 

[edvard]

But you are assume I have pedals. I also think a switch onboard that can convert from a Fender to a Marshall sound would be of use, wonder if I have room for one?

The Orange, the first 2:38 minutes sound like... ... crap. Ji-fet? No, not a mistake as he said it more than once. A video that is 97% high gain with enough reverb that it all sounds like mush. Clean also with reverb, just to show it is not a one trick pony. And that is the thing, easy to do clean or gainy stuff with SS. The world of almost clean to a mild dirt? Tubes still show how it is done.

You can get pretty good "almost clean" tones with inverter chips like the 4049. They clip softly, so transients get rounded off instead of clipped, and you can even fake an inverter stage with a PMOS and NMOS wired together like the 4049 datasheet shows; they automatically bias up even when they are mis-matched. The secret is to have a JFET or Op-Amp buffer up front, then use a smaller resistance in the inverter feedback loop instead of the 1Meg - 10Meg that the dirt boxes use, and cut out as much bass as you can stand (you can add it back in later). Much quieter too; inverter-based distortions are notoriously 'hissy', but that's due as much or more to the shot noise of using high-resistance carbon comp resistors in the feedback loop as it is the high gain that comes of it. Using metal-film resistors closer to the 100k - 220k range makes for a much quieter circuit.

I've taken to making circuits with the dirt coming from NAND and NOR gates; just short the two inputs and wire it up like an inverter; you get asymmetrical clipping (er, 'rounding') that way, and even with the gain jammed into multiple stages, rolling off the guitar volume cleans it up quite nicely. If I didn't have two cars to work on today, I'd post a clip of what I'm talking about. Maybe later...

 

[edvard]

It’s more than tubes vs transistors. A big part of the ‘vintage’ sound we all tend to get excited about involves transformers. Building a ss amp including transformers didn’t really catch on beyond a certain point in time - I’m guessing as a cost cutting measure. The warmth or vintage sound of old preamps in mixing consoles tend to have transformers. The way they saturate is a crucial part of that sound. The recent thread on the ss amp Brian May uses on occasion has them in the design.

The reason is definitely cost-cutting. When you can get most of the way there just running the speaker straight off the output with a yuge capacitor to block DC instead of a transformer, it adds up quickly. I'm looking for a design using MOSFETs running in class AB with an output transformer, but all the ones I've been able to find are quite complex; I don't believe it needs to be so, but I'm OK with being wrong.

 

[glenlivet]

Vox Pathfinder 15R Schematic

www.tdpri.com

This is the one I'd try to make. The schematic isn't *really* that hard. Signal path is pretty easy to follow, and it's a classic. To make it smaller and easier, you might be able to chop out the reverb and trem parts of the circuit.
Getting the specific IC's and some other parts might be an issue.....

 

[cometazzi]

Getting the specific IC's and some other parts might be an issue.....

Maybe, but it might not matter. I'm no Engineer, so this is all anecdotal: IME, unless you're distorting the OpAmp itself, it mostly doesn't matter what type you have in the signal path. Linear amplification/buffering at the frequencies guitar works at will be pretty consistent from modern opamp to modern opamp, unless you've got something really bizarre. Sometimes the impedance matters (JFET-based vs BJT) but I've often subbed TL0xx family for most anything. Most of the differences will be in gain factor (which is again, unimportant unless you're distorting it), slew rate and frequency range. Faster slew rates will make things a tad 'brighter' but usually not enough to notice. High frequency is going to be Radio frequency, so we don't care about that. That said, the NJM4558Ds were pretty available from places such as SmallBear or GGG last I looked, along with the usual Mouser and Digikey, but not Jameco. There is a lot of 'mojo' around the "4558D" but in reality they aren't that special in today's context(s).

The linear TDA series has been discontinued, though you can still find that TDA2030 on ebay from Soviet China, sometimes assembled onto a board with a heat sink, sometimes not- and the price is the same either way. I've heard some say they've gotten factory rejects or fakes, but all of mine seem fine. The linear series has been replaced by a Class-D series which is in current manufacture. They're not pin-for-pin compatible and wire up a little differently but if you're building your own board that doesn't matter.

What you *will* have a hard time finding are JFETs in the TO-92 package. Those all went SMT-only probably more than a decade ago. Places like AdaFruit sell breakout boxes so you can turn SMT JFETs into thru-hole package, but that'll take a hot-air rework station.

That said, you'd be surprised what you can find on ebay. Most takes the slow boat from overseas but it's there if you want it and are patient.

 

[printer2]

You can get pretty good "almost clean" tones with inverter chips like the 4049. They clip softly, so transients get rounded off instead of clipped, and you can even fake an inverter stage with a PMOS and NMOS wired together like the 4049 datasheet shows; they automatically bias up even when they are mis-matched. The secret is to have a JFET or Op-Amp buffer up front, then use a smaller resistance in the inverter feedback loop instead of the 1Meg - 10Meg that the dirt boxes use, and cut out as much bass as you can stand (you can add it back in later). Much quieter too; inverter-based distortions are notoriously 'hissy', but that's due as much or more to the shot noise of using high-resistance carbon comp resistors in the feedback loop as it is the high gain that comes of it. Using metal-film resistors closer to the 100k - 220k range makes for a much quieter circuit.

I've taken to making circuits with the dirt coming from NAND and NOR gates; just short the two inputs and wire it up like an inverter; you get asymmetrical clipping (er, 'rounding') that way, and even with the gain jammed into multiple stages, rolling off the guitar volume cleans it up quite nicely. If I didn't have two cars to work on today, I'd post a clip of what I'm talking about. Maybe later...

I have an inverter to try somewhere, picked it up for just this reason. My big problem is the stuff piled up on my bench. I took down all the shelves I had stuff on as I got cupboards to replace them with. That meant all my stuff is stuffed in boxes or piled on something else. I have other things happening in life also so this will not be finished as quick as I hoped.

 

[edvard]

Maybe, but it might not matter. I'm no Engineer, so this is all anecdotal: IME, unless you're distorting the OpAmp itself, it mostly doesn't matter what type you have in the signal path. Linear amplification/buffering at the frequencies guitar works at will be pretty consistent from modern opamp to modern opamp, unless you've got something really bizarre. Sometimes the impedance matters (JFET-based vs BJT) but I've often subbed TL0xx family for most anything. Most of the differences will be in gain factor (which is again, unimportant unless you're distorting it), slew rate and frequency range. Faster slew rates will make things a tad 'brighter' but usually not enough to notice. High frequency is going to be Radio frequency, so we don't care about that. That said, the NJM4558Ds were pretty available from places such as SmallBear or GGG last I looked, along with the usual Mouser and Digikey, but not Jameco. There is a lot of 'mojo' around the "4558D" but in reality they aren't that special in today's context(s).

The linear TDA series has been discontinued, though you can still find that TDA2030 on ebay from Soviet China, sometimes assembled onto a board with a heat sink, sometimes not- and the price is the same either way. I've heard some say they've gotten factory rejects or fakes, but all of mine seem fine. The linear series has been replaced by a Class-D series which is in current manufacture. They're not pin-for-pin compatible and wire up a little differently but if you're building your own board that doesn't matter.

What you *will* have a hard time finding are JFETs in the TO-92 package. Those all went SMT-only probably more than a decade ago. Places like AdaFruit sell breakout boxes so you can turn SMT JFETs into thru-hole package, but that'll take a hot-air rework station.

That said, you'd be surprised what you can find on ebay. Most takes the slow boat from overseas but it's there if you want it and are patient.

The NJM072D and NJM062D can be replaced with TL072 and TL062 respectively. They are both JFET input op-amps, one being low-noise (TL072) and the other being low-current (TL062). I see the NJM062D is being used as an active filter element.

The 4558D can be replaced by a 5532, as they are simply low-noise "Hi-Fi" op-amps. I know the 4558 is supposed to have 'mojo', but eh. I'm allergic to mojo, it gives me a rash of giggles.

The TDA2030 can be drop-in replaced by LM1875, which is still in production. They work quite nicely. The CA3080 I'm not sure about. The various transistors are anybody's guess, but the JFET I can spot is being used as a switch. You could probably use pretty much any N-channel JFET you could get your hands on and it'd work.

 

[archtop_fjk]

I’m watching this thread with interest and agree that the Vox Pathfinder would be an interesting circuit to clone. The preamp gain reminds me of the Marshall Bluesbreaker overdrive - maybe replace the hard clipping LEDs with soft clipping diodes in the second op amp feedback network?

I’m a fan of the old Roland blues cube solid state amps, and the BC-30 used a upc1188h chip for the power amp. What I recently discovered after perusing the schematic is that Roland used two zener diodes in the chip amp feedback network (see below). Soft clipping in the power amp? The BC-30 does sound tube-like so perhaps that’s part of its secret sauce.

 

[printer2]

I’m watching this thread with interest and agree that the Vox Pathfinder would be an interesting circuit to clone. The preamp gain reminds me of the Marshall Bluesbreaker overdrive - maybe replace the hard clipping LEDs with soft clipping diodes in the second op amp feedback network?

I’m a fan of the old Roland blues cube solid state amps, and the BC-30 used a upc1188h chip for the power amp. What I recently discovered after perusing the schematic is that Roland used two zener diodes in the chip amp feedback network (see below). Soft clipping in the power amp? The BC-30 does sound tube-like so perhaps that’s part of its secret sauce.

View attachment 946429

The preamp may have the special sauce from what I have read in the past. The zeners are 18-19V, the power supply at 24V. It is interesting that the chip data sheet lists supply voltage at max +/- 23V. Could it be soft clipping? Might have to try it.

 

[Gijs]

What you *will* have a hard time finding are JFETs in the TO-92 package.

The j113, for example, is in active production and probably will remain so for a while. From the Mouser site:
in stock 46.611 - Can Dispatch Immediately

 

[TequilaCaster]

The early Roland Cube amps used JFET preamps with chip amps and sounded real tubey.
https://medias.audiofanzine.com/images/normal/roland-cube-1167932.gif
Here we see the 3-JFET preamp again, predating those mentioned in post #22.

 

[TequilaCaster]

The j113, for example, is in active production and probably will remain so for a while. From the Mouser site:
in stock 46.611 - Can Dispatch Immediately

A guy (Jonny Reckless) over on diystompboxes.com posted his J113 preamp/amp design and a video demo of it.
schematic...
video...

 

[cometazzi]

The j113, for example, is in active production and probably will remain so for a while. From the Mouser site:
in stock 46.611 - Can Dispatch Immediately

True, but I seem to remember the J11x series being optimized for switching applications and don't make great amplifiers. I.e., they do work as an amplifier, but they don't have the same characteristics that make other JFETs great.

I could be misremembering though. It's been awhile.

 

[Gijs]

I seem to remember the J11x series being optimized for switching applications and don't make great amplifiers.

They are intended and marketed as mainly switches, sure, but that does not mean they have characteristics that make them unsuitable for audio. I have found an old FET databook here, you can find lots of info on different FETs and their manufacturing processes. It seems actually that some FETs marketed as 'general pupose audio' have been given that label because they are not good for anything else. The J11x series has a lower noise level and higher transconductance than many 'audio' FETs.
Of course they are not as good as audiphile devices like the 2sk170, but hey, they cost only 40 cents or so. For lo-fi purposes like guitar preamps they work really well i think.