Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 28, 2024, 06:54:20 AM

Login with username, password and session length

Recent Posts

 

Biasing opamps in a Marshall Lead 12

Started by dazz, July 05, 2018, 05:10:57 PM

Previous topic - Next topic

dazz

#15
Quote from: phatt on July 10, 2018, 09:17:00 AM
Err??
It may pass signal but it still not right,
Pin 5 has no bias  :o
R9 needs to go to pin5


Damn it, I put that coupling cap just to screw the bias. Noobs will be noobs.

Quote from: phatt on July 10, 2018, 09:17:00 AM
There are other problems as well which are a bit hard to explain.
Some circuits are hard to convert to single supply and you will have to alter more than R9.
I'll see if I can redraw it for you in the next day or two.

Phil.

I'm not sure I feel comfortable with having you do all the work for me, Phil. You know I appreciate your help immensely, but I think I need to figure it out myself as a part of my learning process. Maybe if you can simply name what those other problems are I can then research them and find solutions I can later post for you to evaluate?

Quote
Opamp Rule of thumb; The positive input has to have a DC path to a reference voltage for single supply or ground if it's split supply.
Phil.

That's something I sort of inferred from the way you biased the Casino 12, and that's the way it was supposed to be, but I messed it up with the wrong placement of C7

This should do the trick, right?

phatt

Was a late night after a gig,,, but now that you moved R9 it looks like that may work.

Other options;
Delete C5, connect wiper of 22k pot and C6 to "Vref" instead of Ground then you can delete R9 & C7.
The output of IC1A is already at Vref which then deletes the need to isolate IC1B.
I'm fairly sure that will work.
And less parts ;)

I always breadboard circuits before committing to a pcb as there is always some little bug. :grr
Phil.

dazz

#17
Quote from: phatt on July 11, 2018, 02:00:34 AM
Was a late night after a gig,,, but now that you moved R9 it looks like that may work.

Other options;
Delete C5, connect wiper of 22k pot and C6 to "Vref" instead of Ground then you can delete R9 & C7.
The output of IC1A is already at Vref which then deletes the need to isolate IC1B.
I'm fairly sure that will work.
And less parts ;)

I always breadboard circuits before committing to a pcb as there is always some little bug. :grr
Phil.

Awesome! thanks Phil.
You know what's weird? That's exactly what I had initially in mind! (see first post attachments) and sure enough, the simulation seems to confirm it works.

So the only problem I had was the reversed opamp inputs in LTspice. unfriggingbelievable.. that crap send got me trying alternatives until I figured it out, then it didn't occur to me the other tweaks might not be necessary. I'm not sure if I should get a sense of accomplishment or feel more stupid. LOL

Re: breadboarding, I tried that for the first stage of the preamp, and it worked, but I wasn't sure how it was supposed to sound so decided to rely on the simulation in what seemed a more "scientific" way of making sure the freq response was equivalent to the original circuit. I've found breadboarding to be impractical for circuits containing more than a few components (I also need to get better in that department) . In fact this project is nothing I plan on building, it was just something that I'm using as a sandbox project to learn LTSpice & Eagle enough to build a proper PCB for your Casino 12, as the veroboard I used ended up being a complete mess with all the modifications, trial and errors, desoldering and resoldering stuff on top of the board... it's fugly as hell and I want to do justice to it, haha

But anyway, there's still something unclear to me. It's the part where you say

QuoteThe output of IC1A is already at Vref which then deletes the need to isolate IC1B.

Seems to me the circuit I posted on July 10, 2018, 04:12:10 PM also has it's output at Vref... yet it seemed to need that isolating cap... I'll see what I'm missing now because there must be something different in those circuits cause it now works fine in LTspice without that cap.

EDIT: no! it only needed C1 to isolate IC1A's input, hence having it's output at Vref through R4. I think I get it now

One thing I really need to do is to figure out how to simulate my circuits from Eagle, because copying it to LTSpice and back will certainly introduce errors in the PCB

phatt

I would just keep bread boarding because you will learn so much faster as you will clearly see (make that hear) where and how noise issues, psu filtering and ground paths can make a big difference to the build.

I use sims as well as real working circuits before I commit to a pcb.
Note the picture of my very messy bread board.  :)
This is a compressor on the front and a cab sim circuit on the back. The small switches are just to by pass so I can hear the difference.
I built this last month for a mate as he was amazed at the presence of my guitar rig.
He noted that I hardly touch the strings but still have lots of punch where he has to belt his to get any volume, The classic muddy sound you get without good EQ.
I explained that a lot of guitar rigs have way too much bandwidth and although this makes a clean strum of open chords sound bigger it will trash a full bore distorted guitar tone.
Distortion only works well if you limit the B'width.

Distortion is the easy part,, getting the right tone shape is where the magic comes from.
Take Santana as example 90% of his tone is all under 1kHz.
But he is using HB pu's so with SC pu's you need a higher roll off.
If you are into HMetal tones you will want a different tone shape but the distortion mechanisms are often much the same.

Re sims and eagle;
I don't know much about transferring Schematic's from Eagle to LT spice but I believe some platforms can transfer spice files.
Too much stuffing around for me I just print out the schematic from Circuit-maker then read off that into KiCad.
This also forces you to actually read and understand the circuit.

Re the DC thing;
Your sims should have a probe setting to read DC at any part of the schematic. If you DC probe the output of U1A it will be very close to Vref. Then switch to AC and you will see that the Zero crossing of the AC signal is centered on the Vref.

Keep going as it does take a while to find your way and if you think you are silly spare a thought for me,, that last build I made,,I messed up in a similar way,,
I linked a resistor to Ground instead of Vref on the pcb which made the volume pot very scratchy. Lucky it was a simple fix but you do feel like banging your head against the wall. :duh
Phil.

dazz

#19
Quote from: phatt on July 11, 2018, 10:18:17 PM
I would just keep bread boarding because you will learn so much faster as you will clearly see (make that hear) where and how noise issues, psu filtering and ground paths can make a big difference to the build.

Understood. I'm just about to order a larger breadboard, that should help a bit

Quote from: phatt on July 11, 2018, 10:18:17 PM
I use sims as well as real working circuits before I commit to a pcb.
Note the picture of my very messy bread board.  :)
This is a compressor on the front and a cab sim circuit on the back. The small switches are just to by pass so I can hear the difference.
I built this last month for a mate as he was amazed at the presence of my guitar rig.
He noted that I hardly touch the strings but still have lots of punch where he has to belt his to get any volume, The classic muddy sound you get without good EQ.
I explained that a lot of guitar rigs have way too much bandwidth and although this makes a clean strum of open chords sound bigger it will trash a full bore distorted guitar tone.
Distortion only works well if you limit the B'width.

Distortion is the easy part,, getting the right tone shape is where the magic comes from.
Take Santana as example 90% of his tone is all under 1kHz.
But he is using HB pu's so with SC pu's you need a higher roll off.

I think I know where you're coming from. You mentioned the importance of limiting the bandwidth in the Casino 12 thread too (filter lows to avoid muddy bass with weak power supplies, and of course highs, that I guess also helps with picking up interference) and that Lead 12 preamp is for the most part unfiltered for all I can tell (although it's missing the tonestack). I will definitely keep all that in mind for my future projects though. THis one, I don't care, someone else asked me to prototype it and I simply took the opportunity to cut my teeth on the software, and learn something about biasing opamps. So yeah, my next reading will be on tone shaping networks, filtering techniques, etc. As always, thanks for the info

Quote from: phatt on July 11, 2018, 10:18:17 PM
Re sims and eagle;
I don't know much about transferring Schematic's from Eagle to LT spice but I believe some platforms can transfer spice files.
Too much stuffing around for me I just print out the schematic from Circuit-maker then read off that into KiCad.
This also forces you to actually read and understand the circuit.

Well, it's not so much about transferring schematics, Eagle Autodesk is supposed to have an LTSpice pluggin to simulate the schematic used to build the pcb. It just seems convenient to me to have it all integrated in a single software package. I'm not sure I want to use a third piece of software to transfer schematics from app A to app B which I'll have to review anyway and adds another potential point of failure. So if I can't get the LTSpice integration to work properly in Eagle, I'll do as you do and manually transfer de schematic myself.

Quote from: phatt on July 11, 2018, 10:18:17 PM
Your sims should have a probe setting to read DC at any part of the schematic. If you DC probe the output of U1A it will be very close to Vref. Then switch to AC and you will see that the Zero crossing of the AC signal is centered on the Vref

Yeah, works exactly like that :cheesy:

Quote from: phatt on July 11, 2018, 10:18:17 PM
Keep going as it does take a while to find your way

I will, thanks for your input and encouragement, Phil  :dbtu:

dazz

#20
OK, after a few months of having this project in stand-by I'm finally getting back to it, hopefully to complete it once and for all.
I noticed there's a huge problem with my approach here to feed this preamp 9V since it's originally designed to run at +16V/-16V. Obviously reducing the supply to just 9V affects the clean headroom available to it (what a stupid overlook on my part) and since the gain pot seems to form a low pass filter with C7, if it starts to clip earlier than the original it will do it at a different frequency response.

So someone suggested a charge pump chip to produce the symmetric supply (an LT1054 or ICL7660(S)), but I decided against it because for all I know, the result is not really symmetric, being the negative rail weaker (my LTSpice simulations show that with a 10mA load it's more like +17V/-16V) and I fear that might induce asymmetrical clipping. Are my concerns here justified or am I simply talking nonsense?

So i designed the charge pump as a voltage multiplier with two 7660's in series instead, so I get a single 20 to 24V supply to use with the biased op amp circuit. It's not the 32V swing of the original design but it's much better than 9V, and I believe I can compensate that by increasing R8 from 10K to 15K or 20K to attenuate the signal and achieve pretty much the same thing as the original design.

Maybe I'm just talking myself into using this single supply solution for no good reason, but I've been doing some reading about this stuff and I think it could have an advantage over the symmetric supply: no ripple induced in the negative supply V- pin of the op amp since it's now tied to ground. The PSRR of the positive rail of the TL082 I plan on using is much better than the negative one, hence it should help with ripple (as opposed to having two +/- rails with ripple on both). Does that make any sense?

I learned there's a 7660S with a boost pin 1 to increase the switching frequency from 10KHz to 30KHz outside the hearing range, although I ordered some cheap ones off of ebay that might be fake 7660S (rebranded 7660 without the boost feature). I had no idea this could happen, ugh! But I'm wondering, even if I end up with a 10KHz chip, what would be an acceptable level of ripple to have? The PSRR of the TL082 V+ pin at 10KHz is 80dB, so even if I have 100mV ripple it would be attenuated by a factor of 10^(80/20)=10000. Considering that the preamp's max voltage gain is 62dB, I take it that worst case scenario I would get a 80dB-62dB=18dB ripple attenuation at the output, or 8 times less. Are those numbers correct and

I know I can just go get a proper ICL7660SCPAZ or a LT1054, and I probably will, but I want to take this opportunity to learn about these things, so I'll be running some tests with whatever I receive in the end, different filter caps, etc... I plan on using a large 100uF 50V filter cap in parallel with a 1uF 50V ceramic to maximize capacity and minimize ESR which I heard is critical in these charge pumps.

Hope that wasn't too much of a mess

ETA: forgot to attach pictures of the charge pump simulations. The first one using the boost pin, the second without it. Using tantalum 100uF filter caps in the sim that I hope I can replace with those electrolytic + ceramic in parallel. I get "moderate" amounts of ripple there that I don't think will be an issue considering the PSRR of the TL082, even at 10KHz, is that right?

Ripple without boost: 3.5mV, 10KHz
Ripple with boost: 1mV, 30KHz

ETA2: I should have used the preamp's voltage gain before the tonestack, which is obviously higher than at the output. At 10KHz it's 66dB, but the 7660 switching freq gets lower as temperature goes up- The preamp gain at 6Khz is 70dB, so considering 80dB PSRR that's only a 10dB attenuation of supply ripple.. hmmm...

edvard

Quote from: dazz on November 25, 2018, 01:44:14 PM
...

So someone suggested a charge pump chip to produce the symmetric supply (an LT1054 or ICL7660(S)), but I decided against it because for all I know, the result is not really symmetric, being the negative rail weaker (my LTSpice simulations show that with a 10mA load it's more like +17V/-16V) and I fear that might induce asymmetrical clipping. Are my concerns here justified or am I simply talking nonsense?
...

Many, many, many distortion circuits utilize asymmetrical clipping as a desired feature, so I think you're OK there.  As far as clean headroom, I think you're still OK as long your preamp isn't overloading the front end of the power stage.  Even the venerable TDA2050 had a maximum input voltage of ±15v, and for most real-world situations, it was probably much less than that.

phatt

hi Dazz,

Wow,, hey Don't over think it,,
Here is how you run that circuit on single supply.
Even with higher headroom it will still distort early ,, that is the way it's designed.
I have not simulated the tone section but it looks like the treble is insane high,, I doubt I'd like the sound but it could be tweaked to back off the extreme treble. :-\

Sorry can't help on all that split supply thing you are talkin about,, I'd just find another transformer,, likely cheaper in the end.
But up to you. :-X
Phil.

dazz

Quote from: edvard on November 25, 2018, 04:01:41 PM
Quote from: dazz on November 25, 2018, 01:44:14 PM
...

So someone suggested a charge pump chip to produce the symmetric supply (an LT1054 or ICL7660(S)), but I decided against it because for all I know, the result is not really symmetric, being the negative rail weaker (my LTSpice simulations show that with a 10mA load it's more like +17V/-16V) and I fear that might induce asymmetrical clipping. Are my concerns here justified or am I simply talking nonsense?
...

Many, many, many distortion circuits utilize asymmetrical clipping as a desired feature, so I think you're OK there.  As far as clean headroom, I think you're still OK as long your preamp isn't overloading the front end of the power stage.  Even the venerable TDA2050 had a maximum input voltage of ±15v, and for most real-world situations, it was probably much less than that.

True, and as someone else pointed out to be elsewhere, opamps that aren't RRIO will clip asymmetrically anyway. It's just that I thought unbalanced rails could alter the way the opamp sound as they clip by adding even more asymmetry.

dazz

Quote from: phatt on November 26, 2018, 08:03:02 AM
hi Dazz,

Wow,, hey Don't over think it,,
Here is how you run that circuit on single supply.
Even with higher headroom it will still distort early ,, that is the way it's designed.
I have not simulated the tone section but it looks like the treble is insane high,, I doubt I'd like the sound but it could be tweaked to back off the extreme treble. :-\

Sorry can't help on all that split supply thing you are talkin about,, I'd just find another transformer,, likely cheaper in the end.
But up to you. :-X
Phil.

Hi Phil! You're not the first one to tell me I'm overthinking this hahaha.

Thanks for the schematics, I thought I already had that OK. I had only R2 and VR1's mid lug going to Vref in my circuit, the rest going to ground as they are not DC coupled to GND, except for R3, I added a cap there to hook it to GND (I can see now that wasn't necessary, could simply go straight from R3 to Vref)
Yeah, it distorts early  and a lot with a +16/-16V supply, but with a 9V one I think it will distort at almost any gain setting.
True, it's very trebly, not for me either. This is a project I got into by some else's suggestion. I'm building a RoG Azabache for myself for more fender like tones. I'm sure you've heard about the Azabache. But anyway, it's just that i saw the opportunity to learn how to bias opamps, last time (with the Casino 12 build) you had to do it for me, no more!  :grr :lmao:

This is a different beast though.

BTW, I think I should have used the spare opamp in the Casino 12 design to buffer the voltage divider.. it works great as is so I guess it wouldn't have made too much of a difference. Let me thank you again for all your help in that thread, The amp now sports a 25W class D chipamp and it's loud AF  :dbtu: