Ok, I'm still trying to find a good preamp circuit for my power amp (maybe amps, depends how this first one goes!). I've built a small 25W stereo power amp that runs on a single +12V DC supply. So, that pretty much eliminates any bipolar designs (this time). I built a Professor Tweed ROG pedal for a friend and tried it in front of my power amp but found it MUCH too quiet. At anything less than say 70% the volume was way too low. Even at 100% the power amp was not pushing any decent output. I have verified the power amp is good with a mp3 player and hi-fi speakers so I know that's not the issue.
so, I'm at a loss now...
1) is it reasonable to contstuct a preamp running off +12V DC with sufficient gain to "properly" drive my power amp? I would like the most clean headroom as possible as clipping tripath chips is generally not advised. Of course, all volume and tone controls will be passive.
2) How about using a LM386 as the preamp stage? I've read some posts indicating success with a ruby circuit as a preamp. I'm concerned that the low output impedance of the 386 won't be a good match for the power amp though.
here is the datasheet for the power amp chip I'm using. http://www.datasheetcatalog.org/datasheet/Tripath/mXyzxwww.pdf
I would really appreciate any suggestions or advice you guys could give as I'm a bit in the dark here and don't want to waste a lot of time and money building circuits that won't work for my purposes.
FYI, I do have all the parts to build this: http://www.beavisaudio.com/Projects/ValveCaster/MatsuminValveCaster.gif which I will try at 12V, but I don't have high hopes, so I'm looking for back-up plans.
-Matt
anyone? 120 views an no opinions? :(
1) Yes, you can build a preamp that is powered from single 12V supply and that is capable of driving most power amps to their full output power level. If the ROG designs do not provide a sufficient output signal amplitude that could do this then just fit a suitable voltage amplifier stage to follow them.
By the way, just because you are stuck with a 12V single supply does not automatically mean that bipolar designs are out of the question. You can easily derive the "half supply" reference voltage needed for bipolar designs with a resistive divider. This reference voltage potential naturally won't be zero volts as it is in a bipolar circuit, it is 6V, but in reference to this so-called "virtual ground" potential the rail voltage potentials of 12V and 0V (ground) will behave similarly as the rails of a bipolar circuit behave in reference to zero volts. Just look at how most OpAmps are used in 9V battery-powered circuits to get the hang of the idea. It applies to any circuit - not just OpAmps - but with those it is likely easiest to realize. Since the new "zero" reference becomes 6V you naturally need to AC couple the stages to get rid of the DC.
2) LM386 can naturally work as a preamp gain stage. Its low output impedance is beneficial when you want to push a lot of current to a low impedance load (which is what LM386 was designed to do in the first place). The low output impedance makes it an ideal power amplifier - and it does not cause a drawback in coupling to higher impedances either; in fact, it's beneficial in that as well since for optimal voltage transfer you need to couple from low impedance to high. The only thing you loose by increasing the load impedance is the amount of current pushed to the load (because I=U/R) – which does not matter at all if you're using the chip as a preamplifier that mainly amplifies voltage.
Other issue is how practical the chip is for this kind of application in the first place. The chip's gain is stuck between the minimum of 20 and maximum of about 200, and it is difficult to adjust. If you are only building a voltage amplifier stage then an OpAmp will be a much better device choice for that application. It allows you to define any arbitrary value of gain – easily.
Thank you very much for your reply. When you say the gain on the LM386 is hard to control are you indicating that you don't believe the 1K pot (set up as a variable resistor) technique used by the Ruby design provides very good gain control? I have built a Ruby amp with adjustable gain so I will give that a try once I reassemble my power amp.
-Matt
That depends entirely on what you want. The potentiometer in Ruby controls gain between figures of 20 and 200. It's a simple question of what if you don't need that much gain? Or what if you need a stage with a specific voltage gain ratio? Something like that is not very easy to build using the LM386. That chip is optimised for certain applications and plain voltage amplification is truly not one of them. If you want flexibility in setting up a voltage amplifier stage with any arbitrary amount of gain then OpAmp-based circuits are a way better choice than LM386.
Could you suggest a good op-amp based circuit that I could take a look at? Most of what I have seen so far are distortion pedals. Ideally I would like to have a switchable clean/dirty channel on the amp (which is kind of why I was leaning toward the Ruby as a preamp since you can increase the gain to add distortion... not a switchable solution, but should work).
-Matt
It's questions like this that make sure you will either get a multitude of different answers or no answers at all.... How do you define a good circuit / preamp anyway? (Whether based on OpAmps, tubes FETs, BJTs or any other possible choice of device). No one can read your mind and know what YOU consider as a good circuit.
All that I can really say is this: Are there some solid-state amps / FX pedals etc. that have the tone, features and other stuff that you like? How about you start by checking out their schematics. The Internet is full of those; if you have a specific one in mind and you can't find it you can always post here (or some other forum) asking if anyone has it or perhaps try to contact the manufacturer. Also, what was wrong in the mentioned ROG circuit aside the fact that it had too low output signal? Tose kinds of issues can be easily compensated.
If you know a suitable clean circuit and a suitable distortion circuit (say, from a distortion pedal) why not just combine them? Often all that is needed is a simple switch that selects between two signal sources. If making these kinds of adaptations is the problem then you need to overcome it by studying the electronics more. Otherwise you stuck to following pre-made designs and in such cases you are pretty much always dealing with things that present a compromise between what you wanted and what you got.
You are right, my general electronics background is quite weak at the moment but I'm doing all I can to improve that (Hence my questions here). I have a pretty good understanding of simple circuits like voltage dividers, RC filters (to some extent) and simple volume controls. Basically, the stuff you learn in a simple college intro to electronics class. The various amplifier circuits are where I begin to struggle. What makes this more difficult is that I'm not actually a guitar player :) I have taken up this hobby because I have a few friends who do play guitar but don't have the time/desire to build amps/effects.
So, to answer your question, here's what I am after in a preamp:
1) Tone: Generally a nice "bluesy" sound... In this case that would be ranging from nearly transparent "clean" to a little dirty and breaking up. So, either a single preamp channel with variable gain or two channels (one clean with fixed gain and one dirty with variable gain to control the distortion)
2) Provides sufficient output to drive my power amp(s). I know this is a tricky one to quantify but the issue I had with the ROG circuit is a prime example. This preamp circuit (if built to spec) would not sufficiently drive my power amp to its full potential. I expected this was easy to resolve (as you indicate) but I'm not sure how to do this while still preserving the character of the preamp's tone.
3)Won't destroy my power amp... As with number 2, this is quite obvious but still a concern to me. One power amp I have should never exceed about 5.5V on the inputs or you risk destroying the chip. Obviously a simple voltage divider can ensure that never happens, but you have to attenuate the signal across the board just to guard against that one possible voltage spike. I would much rather retain the full output of the preamp and simply clip the output before it reaches a dangerous level for the power amp. In one of your other posts you point out that typically a preamp must be capable of much more gain than you would theoretically need to achieve a 1V RMS output. The tone and volume controls along with input/output impedances between the power amp and preamp can attenuate your signal to the point that you don't get enough total gain under normal circumstances.
Sorry for the vague questions but I'm trying to find a good solution to my problem without too much trial and error (which gets expensive ;) ) I would love to be able to just design a circuit that meets my needs but for now it seems my best bet is to try building other people's designs and study how they work.
-Matt
http://www.generalguitargadgets.com/richardo/distortion/ (http://www.generalguitargadgets.com/richardo/distortion/)
This helped me out a lot when I was trying to figure out the basics of gain stages and filters. You should be able to make a simple IC gain stage to use after the Pr. Tweed and before the power amp after you read the article. It will probably take a little trial and error to get it how you want.
Hey Armstrong... I just wanted to drop a note to say I'm basically at the same stage as you, both in my electronics education and in my amplifier build... Instead of the Prof. Tweed, however, I'm going to try the the Eighteen, as I've read that it has been used as a preamp with better results than the Prof tweed. Also, my poweramp is an LM3886 which I'm powering at +-18v (so it should be in the 40ish watt range)... I'm 90% done with the Eighteen (I need to add the pots and properly bias the transistors), and I have not yet assembled the LM3886, but I have the power supply finished (~80% of the work!), so I should be ready for testing in a week or two (with my busy schedule). When I get it built and tested, I'll let you know my results. Also, if you end up going with an IC voltage amp stage between the pre and power, let us know how it works, and what, exactly you did. I'd be very interested!
Cheers,
DDC
Hey, I just had another thought. I don't know about your Tripath chip, but the LM3886 has an easily adjustable gain (here's alink to a page describign basically how I'm going to build the LM3886 http://dogbreath.de/Chipamps/ThreeResAmp/ThreeResAmp.html (http://dogbreath.de/Chipamps/ThreeResAmp/ThreeResAmp.html)) You just adjust the value of two resistors (Rf and Ri in the schematic). If you are having low volumes with the preamp (prof tweed) you built, can you not just rais the gain of the poweramp chip? I know when I was designing my amp, I made sure that I would be matching the amount of gain of the poweramp stage to the output level of the specific preamp I am using (eighteen). Now of course I have not finished yet, so I'm not sure if this will work out. Does anyone see any issue with this? It would certainly be best if one did not have to build in yet another gain stage between the pre and power (simpler is better, right?)...
Cheers,
DDC
It will work out. The chip (LM3886) is basically just an OpAmp that can feed high current to low impedance loads. Pretty much all rules of OpAmp design can be applied sovereignly with this chip. There is no reason why one should replicate the component values of the datasheet circuit or the popular gain clone designs. Those values just happen to be in the ballpark of a basic application that has a gain of about 20, which is pretty desirable for amps that are coupled to generic domestic audio devices. It might not work that well with designs that are coupled to guitar preamps feeding a line level signal, or worse, FX pedals that have who knows what signal level.
Yet I wouldn't build the circuit shown under the link. It has no DC coupling so if the source has any DC offset it will be fed to the input of the chip, which also reflects it to the output and the speaker. To make matters worse, there is no capacitor in the feedback loop that would set a unity DC gain, which means any DC offset will also be amplified. If by chance the DC offset happens to be as high as 1 V (can happen some times), then a popular value of gain that is about 20 will mean the DC offset in the output will be 20 volts! That means you can say goodbye to your expensive speakers. Even half or quarter of that DC offset value is not much better. Note that there can be DC offset even due to power amp circuit itself, so you can have this issue even if the preamp is fine on that regard.
Second, the circuit lacks all measures to enhance stability. The input should have a low-pass filter to suppress RF interference; in the feedback path Rf should be bypassed with a low-value (10 – 47 pf) capacitor (sometimes in series with a resistor) to suppress any chance for oscillation. The output should include Zobel network (x uH inductor in parallel with a 10-ohm resistor), and a series 4 - 10-ohm resistor and a 47 - 100 nF capacitor shunt to ground. These will enhance the stability of the amplifier when it drives the complex load of the speaker cable and the speaker. The power supply is not shown very well but the Vcc and Vee rails of the chip should be decoupled to ground with 100nF caps, preferably straight at the corresponding pins of the chip. If the wiring length to main filter caps is considerably long there also should be 100 – 470 μF caps located physically closer to the power supply pins of the chip. All these additions are even more important if you cannot do as neat PTP job as shown in the link. Most of us use PCB for convenience, and the design is not always top notch so the power supply decoupling and enhancements of stability are issues to concern.
Overall, these minimalist designs are appealing to audiophiles who believe the blatant claim that fewer components equal better audio quality. In some cases it is true but I don't think its good engineering to make simple circuits that endanger stability and reliability. Even more, the mentioned additions won't have any audible effect whatsoever - but they will make the circuit a lot more reliable. Take a look at any modern, commercial chip amp and you will find out that it utilizes the mentioned features.
Okay, I'm having problems posting a schem... I get "Error 500 - Internal Server Error" every time I try to post it!!! Anyway, here is the text the post I wanted to make, and I guess I'll have to try to post up the schematic later on... sorry about that!
Original Post:
"Actually, now that I go back to read that page I see that it was only one of the pages I used when fully understanding how to design my circuit, and is actually not at all the same as the one I designed. I've attached a scan of my pencil drawn schematic, which, now that I've thought about it more, I actually drew more from the first "typical" application schematic in the datasheet than from any design I found on a website. You'll have to forgive me for the mix-up, I designed the circuit several months ago, but haven't gotten around to working on it again until now because I've been working oversea's for the last couple of months! Now I have learned the "hard" way not to post before thinking!"
Well, I'm still getting a "500" error everytime I try to post a schem, so I guess I'll have to switch tracks here.
I looked again at the preamp I started building before I left two months ago, and I realized/remembered that I was actually building a Professor Tweed, not an Eighteen! I remember now that I was going to build the eighteen until I read a couple of posts over at the stompbox forum (http://www.diystompboxes.com/smfforum/index.php?topic=52322.0 (http://www.diystompboxes.com/smfforum/index.php?topic=52322.0) and http://www.diystompboxes.com/smfforum/index.php?topic=21396.0 (http://www.diystompboxes.com/smfforum/index.php?topic=21396.0)) where people were having volume issues with the Prof Tweed, but solved them by changing the value of the capacitor in the feedback loop (the gain is set by an NF loop). The capacitor (C7 in the schem I have) in the original design over at ROG is supposed to be between 10uf and 1uf, but in reality it should be much lower than that to ensure a proper amount of gain. I just finished building the circuit, and I used a value of 0.01uf (I made it socketed, so I can also try other values), and I got PLENTY of volume. In fact I was so impressed with my trial today I made a YouTube vid of me playing through it while it was hooked up on my bench. Here's the link: http://www.youtube.com/watch?v=ErnjbJJwqoM (http://www.youtube.com/watch?v=ErnjbJJwqoM) In the vid, I plug the Prof Tweed staight into some crappy computer speakers (which are looking for about a 1v input signal, much the same as a poweramp chip would be), and I get fantastic volume. In fact, I had to dial down both my mic (I'm a harp player) and the volume pot on the computer speakers because I was getting too much feedback!
So, Armstrong, if you are still working on your amp, I suggest that you switch out the feedback cap before you try any of the other suggestions (the extra op amp stage, etc.)... It should work for you.
FYI, the only other mods I did to the original Prof Tweed circuit was to increase the decoup,ling caps from 22n to 0.1uf (better for harp), to add an addition decoupling cap tot he input (also 0.1uf), and I changed the value of the caps in the speaker sim section of the circuit from 2n2 to 5n8 following some advice that it would increase the bass response and make it "warmer" which is what I wanted...http://www.diystompboxes.com/smfforum/index.php?topic=63944.0 (http://www.diystompboxes.com/smfforum/index.php?topic=63944.0)
Sometimes I can't but wonder what the heck were the guys at runoffgroove.com thinking. First off, copying a tube circuit and just substituting tubes with FETs is never a good idea. Sometimes these kind of circuits work, sometimes they don't, but what is certain is that they do not behave like the original circuit with tubes.
Secondly, the designer obviously tried to replicate the negative feedback loop of Princeton but failed epically. See how in in Professor Tweed the signal passes through two common source gain stages. That's two inverting amps in series which in total makes the output signal NON-inverted. Then this signal is fed back to the source of the first circuit. This is positive feedback!!! How did they fail to see this?
The reason it works as negative feedback in Princeton is because one can invert the polarity of the output transformer's secondary so that the signal phase is proper one to provide negative feedback. Then there's those awful drain trimmers that assure that you can't build these circuits consistently.
hmmm... I took a look at the original princton schem (http://www.freeinfosociety.com/electronics/schemview.php?id=802 (http://www.freeinfosociety.com/electronics/schemview.php?id=802)) and the feedback loop does connect to the output of the output tranny. It also connects to the grid (? I think it's the grid, I'm not a tube guy) of the 6y6, which I think is what the ROG guys were copying by having the loop derive from the last FET's drain, although that would go against their logic for the tube to FET switch. Here's what the the ROG guy's say they did:
Quote"We took the 5F2-A schematic and copied it part for part using MPF102 JFETs in place of each tube stage. Each tube Grid was replaced by a JFETs Gate. The tube Plates were analogous to a JFET Drain. Finally, a tube Cathode was replaced with a JFETs Source. We used a 100k trimmer for the plate resistors on the schematic due the fickle nature of JFETs and the much lower power supply involved."
One interesting thing I see when I look at the original princton schem is that the NF loop is connected to
both halves of the 12ax7, and the Pof tweed has it connecting just to the second analagous half (ie, the second FET). Does anyone know a practical reason for changing this? Also, it seems that while the (positive) feedback loop of the Prof Tweed connects to the source of FET #2 (second analogous half of the 12ax7), the original schem connects the NF loop to the plates of the 12ax7 (which are supposed to be the analogical drains of the FET's)... Anyone know why that would be done?
All that being said, why then does the circuit work and even actually sound good? Not ever having heard a real fender princeton, I cannot say myself that the circuit sounds anything like it is "supposed" to sound, but I've read comments on the stompbox forum that indicate that it does actually sound similar. Is this all a happy accident?
The Princeton 5F2-A (which runoffgroove.com refers to) realizes the negative feedback loop with a 22K resistor connected from the speaker output to the cathode of the last preamp tube, which is, as said, one half of 12AX7 and feeds the power tube. That's the only feedback loop there. Don't mistake the power supply for feedback loop....
By the way, to get this over with...
Grid = analogous to transistor's "base" or FETs "gate"
Cathode = analogous to transistor's "emitter" or FETs "source"
Plate/anode = analogous to transistor's "collector" or FETs "drain"
Professor Tweed is theoretically very similar to 5F2-A, except the important lack of output transformer, which can perform phase inversion so that signal appearing at speaker output has an opposite phase in reference to signal appearing at the preamp tube's cathode where feedback is directed. All you need to do for that is to reverse the leads of the secondary. This makes signal flow inverting, inverting, inverting (resulting to inverting) in oppose to Prof. Tweed's inverting, inverting (which results to non-inverting).
Anyway, amps/circuits with positive feedback tend not to be very stable and their behaviour can be somewhat unpredictable. I don't say it's bad engineering to use positive feedback (as it has several useful applications, such as oscillators to name one) but if you don't know what you are doing the chances for "happy coincidences" where the circuit works impeccably are slight. I don't think the designer of Prof. Tweed really knew what he did with that circuit.
I don't know why it works as well as it does, though. Maybe because the amount of positive feedback is very small and the 22K resistor and 1uF cap in series make a pretty effective shunt to ground through the low 1K5 source resistance of Q2, which effectively attenuates the signal a lot. When you decrease the value of capacitance you decrease this effect, hence more gain. You could also experiment with the value of the series resistor and with a modification where feedback is inserted to the drain of Q2, instead of its source. That would be the conversion to negative feedback.
I hate to disagree with Teemuk , however there are high-voltage FET's that can directly replace certain tubes(not all). However , in Instrument Amplification positive feedback has been used, I would not recommend it for DIY's , but it is used often in production amps . Back to the circuit if your wanting a Fender Champ(tube champ) tone, you will not get it with a solid state circuit, sorry bro , but there 2 different animals, but have fun trying , I have built tons of amps , modified tons more , you can get close , but no cigar. Also the output transformer in a tube amp does 2 things 1st it takes the 2 split phases of the signal and combines them on the primary side (the side of the tubes) 2nd it matchs the impedance on the secondary(the side that hooks up to the speaks) the impedance from the tube is high(around 5k ohms) speaker impeadnce is low(4-16 ohms) transistors, and opamps (generally) have a low output impedance .
Rock On
Sometimes I can't but wonder what the heck were the guys at runoffgroove.com thinking. (2)
Dear Teemuk, I couldn't agree more with you.
ROG's ideas are the worst "design" technique I've ever seen, and I've seen quite a few (just as you). Original (tube) circuits component values are NOT "pulled out of the blue" , but carefully calculated to get a given optimal "working point", gain, bias, signal handling and, where appropriate, distortion. None of that is considered by ROG, who probably thinks that "original" resistor values are somewhat magical and shouldn't be messed with. But they mess with everything else!!! Just as an example: a 12AX7 with 820 ohms in its cathode and 100K on its anode has a gain around 50/60, can output around 60 clean RMS volts and stands around 1,5 Vac in its grid (and will stand without dying around 100Vac there, if so supplied by a previous tube). An MPF102 or similar will be grossly underbiased by an 820 ohm resistor (it would need from 2k2 to 10k) and the gain is a meager 3 to 6. Just Proto one and see!! To make matters worse,the so called "100k bias trimmer" is a "max"value 100 k Load Resistor and never used in that value, but typically from 1k5 to 10k at best. The rest of their assumptions are similarly way off the mark. Do they sound good? Well, some do, being minimalistic Fet preamps, sometimes (easily) overdriven. Do they sound like a Princeton/Plexi/Matchless/Hiwatt, etc.? Definitely not!!!! Can Fets be used to get a good guitar sound!! Yes, emphatically, but using datasheets ,measuring them, and using normal good design techniques.
These are stompbox circuits guys, not tube amp replacements--or even preamps originally in spirit, it's others that have found use in applying slightly higher voltage to drive power amps. I think the runoffgroove guys would be the first to admit that they're not trying to veritably replace tube amps (and anyway if I'm not mistaken it's Jack Orman and Doug Hammond that get credit for adapting these FET techniques to guitar circuits).
What were they thinking? They were experimenting and came up with what many feel are great sounding circuits. what's the big deal here? like it or not, decades of "bad" design principles, accidents, and cheap shortcuts have shaped present ideas of good guitar tone (who at the dawn of guitar amplification could have predicted extreme distortion characteristics would become desirable, or even a singular focus of amplifier design?). I'm honestly surprised you guys dont appreciate the spirit behind these articles, or are you "beyond" constructively adding to the discussion?
Anyway I would love to see or hear what you guys think are more well designed FET based circuits. I read support here for marshall valvestate amps, which I think are kind of a strange choice (and I love early florida death metal!).
QuoteI'm honestly surprised you guys dont appreciate the spirit behind these articles, or are you "beyond" constructively adding to the discussion?
We are not adding to the discussion constructively? Why not? Because we are critisising bad design practices or weird things the ROG circuits feature because their designer didn't seem to understand how the original circuit works? Did you even read the whole thread?
I like the spirit of that site's articles but I just fail to see why every FET gain stage should have a 100K drain trimmer when the world is filled with designs that work greatly without them. Even if someone wants to fine tune the drain resistor value what's wrong in calculating a trimmer value that's approximately in the correct ballbark to start with. They also fail to notice that when you fine tune the drain resistance you are also messing up with the amount of gain as well as with all the RC circuits formed in conjuction with the coupling capacitors. With those trimmers in there you have a hard time in building anything consistent.
If you want to convert tube preamps to FET preamps you actually should do more design work than just replacing all the plate loads with 100K trimmers and all the tubes with J201's.
FET's are tricky and even designers of commercial applications have troubles with them but that's where handpicking or techniques like constant current loading come in.
QuoteI would love to see or hear what you guys think are more well designed FET based circuits.
Well, how about Roland's Blues Cube series preamps, Gallien Krueger's G.I.V.E. variations, Randall FET-based overdrive circuits, etc.