(note for moderators: I have crossposted this thread to other forums on the web where it seemed it might be relevant. I did not find anything in the rules against this, but if this is against the rules, I apologize)
Greetings. Yes, the title is exactly what I intend to build, as silly as it might seem.
I want to build a moderate power guitar amplifier using solid state actives. Why obsolete and salvage parts?
There's a few reasons:
- I have access to bunch of obsolete and salvage parts through my own collection of stuff, and membership at a hackerspace. At least some of my BoM will be cheap or free this way.
- Being constrained by what I can get for extremely cheap or free will make this somewhat of a challenge and force me to get creative with the circuit design, and might even be fun.
- And finally, these older and salvaged parts are probably not going to run as clean and linear as modern parts would. I see that as a potentially useful quality in a guitar amplifier. Exactly how and why the poor electrical performance of old tube circuits creates the je ne sais quoi sought by guitarists has been discussed extensively, and IMO, still hasn't been fully worked out. But in either case, if bad electrical behavior can be pressed into a musically useful role, that bad behavior is possibly desirable to design for.
Here's what I have on hand so far:
1 120V:24V 40VA power transformer from a furnace
1 120V:16V 15VA doorbell transformer.
1 40W 10" speaker left over from a long-dead practice amp.
1 reverb tank, left over from the same long-dead amp.
A few hundred different transistors, in varying quantities
I am willing to buy some parts, if for no other reason than I know I'll have to. I'm going to need various resistors and capacitors for filters, voltage dividers, setting operating points for the transistors, etc. I'm even willing to buy a few complementary transistors, as I saw almost no complementary pairs in the parts drawers. I may have to purchase some voltage regulators as well, depending on what power supply circuits I am able to come up with. Beyond that however, I really would prefer to limit myself to the transistors on hand for the design, because that's the essence of this game: How good of a sound can I make with this old stuff, even if I have to force it into behavior it was never designed for?
Design OutlineHere's the outline of the design that's been forming in my head over the past week:
- Class AB push-pull output section based on a complementary pair using some of the higher power transistors selected from my hackerspace's parts drawers, running at high voltage, into the primary of the doorbell transformer, stepping the voltage down and the current up to drive the speaker. Yes, I know this is unnecessary, appropriate BJTs or MOSFETs should be able to push enough current to drive the speaker directly. However, I believe the interactions between the output devices, transformer, and speaker are an important part of the vibe of the old circuits that have caught my ear. Thus, I think it is worth at least trying to design the output stage to use a transformer.
- A preamp section using a number of the small signal and not-so-high power transistors from the parts drawers, running at medium and lower voltages. Exactly how many gain stages, I'm not sure yet. I would like to span from nice clean tones, through edge-of-breakup, to a moderate crunch. Very high gain distortion is a maybe; we'll see what sounds I'm able to get out of the parts on hand.
- A power supply section running the furnace transformer, with a voltage doubler or quadrupler to get up to the voltages needed to make use of the doorbell transformer as an output transformer. Other portions of the power supply section can offer lower voltages appropriate for the transistors in the preamp section.
- Maybe use the reverb tank for a reverb circuit.
- All of this into a single combo chassis to keep the amp readily portable.
Timeline / StepsHere's how I see this going. I welcome any input and discussion as the project moves along through these phases. If this order of design is wrongheaded, I also welcome input on what might work better.
- Transistor Selection
I need to work out which transistors are going to appear in the output and preamp sections. This will determine the voltage and current requirements from the power supply. I am leaning towards using the few JFETs and MOSFETs in the pile. This is mainly because I understand those are more readily coerced into similar misbehavior as the high-mu triodes and power pentodes that came to define the voicing of many tube amp circuits. However if some of the BJTs, SCRs, UJTs, or other miscellaneous parts available in my hackerspace's parts drawers might also be useful, I am interested. - Power Supply Design
I need to design a power supply to meet voltage and current needs of the transistors selected for preamp and power amp roles in the previous phase. I don't have any super firm ideas here yet. I like the designs I've seen from Walt Jung, but I don't know if those can be realized at both the high voltage I want in the output section, and the obsolete parts I'm limiting myself to for this project. - Power Amp Design
Next, I will need to come back to the Power amp and work out its design in greater detail than the current vague idea of "high-voltage push-pull complementary pair across a doorbell transformer, because lol". This is where I expect a lot of the discussion on exactly how and why tube circuits misbehave and which of those misbehaviors are musically useful and worth attempting to replicate. - Preamp Design
Moving on to the preamp section, I expect even more time will be spent on how to get the transistors in this section to misbehave in fun ways, and how that misbehavior can be leveraged towards producing the kinds of voicing and character I want out of the amp. - Testing and tweaking
There should be some testing and tweaking of each of the above bits as they get built, but I expect there's going to be a lot more once start I connecting them together. I should anticipate lots of time playing riffs into the amp, critical listening, staring at bizarre oscilloscope traces with a "WTF?" look on my face, and coming back here with my observations to get help with the Why and How behind those WTFs. - Build the cabinet/enclosure
Honestly, this one gets into woodworking and cabinetry, which feels out-of-scope for this forum. If I remember to take pictures, I'll be happy to share them, though.
That's all I have for now. I am still working on transcribing the photos of my hackerspace's parts drawers. I figured I'd get this initial post up so I could get input on the basic design outline and the intended design process/sequence. I will have parts lists in my next post. They will be long. I am tempted to put them into tables for easier readability.[/list]
I have finished transcribing the pictures of my hackerspace's electronics parts drawers. This is not the full set of components. I have deliberately omitted a good bulk of the collection for being 7400 and 4500 series logic chips -- which I don't see as likely to be useful unless I start getting into some multi-voice channel switching fanciness.
I could post tables of the transistors on hand now, but they would be just tables of name/type/polarity, and nothing of their specs At 230+ devices that's a lot of device model numbers that are probably obscure, and you'd need to look up their specs to know if they're useful or not.
Would you guys like to see basic specs for each transistor type like max voltage, max current, max wattage, amplification factor, etc, or just a list of the devices on hand? The latter I can post right away. If I am to post the former, I will be spending several days hunting down datasheets for over 230 transistor types.
Hello Koreth,
I admire your conviction but are you aware this would Definitely takes years of R&D to perfect and in the end may not be any better than what is already available.
It would take years of bench testing to perfect,, I should know as I've worn out 6 breadboards (and counting) and sent many perfectly healthy transistors to smoke heaven in the last 30 plus years.
Regards the Tx output using discrete components is likely no better that using current Feedback which is very commonly used in a lot of Guitar amps now.
If you are chasing Tx output design ideas then find some of those Factory pa systems as a lot of those use Tx drive for line drive.
I have and old Inkel Pa with Tx output giving 4 or 8 Ohm as well as 70V and 100V out puts.
I used it a lot years back as it was 120W output driving a Quad box, gave me a massive sound.
Phil.
Is the 24v 40VA transformer centre tapped, or is it just one output of 24v?
If you have 12v + 12v it would make it easier I think.
Do you have any Op-Amps or will it be all transistors?
There are lots of old transistor amp schematics around that you could borrow sections from, maybe a proven pre-amp that sounds good already.
It all depends on what transistors you have, I would like to see just the part numbers, I don't need to see the specs for each one.
Cheers
Mick
Quote from: phatt on February 27, 2023, 05:01:48 PMHello Koreth,
I admire your conviction but are you aware this would Definitely takes years of R&D to perfect and in the end may not be any better than what is already available.
It would take years of bench testing to perfect,, I should know as I've worn out 6 breadboards (and counting) and sent many perfectly healthy transistors to smoke heaven in the last 30 plus years.
I'm willing to risk this ultimately being a waste of time. There is no doubt in my mind that it will at least be a fun and educational waste of time.
Quote from: phatt on February 27, 2023, 05:01:48 PMRegards the Tx output using discrete components is likely no better that using current Feedback which is very commonly used in a lot of Guitar amps now.
I saw this mentioned in a large thread on diyaudio.com. I will restate my understanding from what I read, and welcome any corrections
To my understanding, driving a speaker from tube plates (whether triode or pentode), through a transformer causes the combined circuit to act more like a voltage source, which is not the ideal way to drive a speaker. This is because you end up with a poor output impedance, and the speaker's impedance curve gets reflected back at the tube plates, causing frequency dependent gain. In essence, the non-linearities of the speaker heavily color the final output sound as a result. Most guitar amps
do use some negative voltage feedback to improve the output impedance of the output stage and cause it to behave more like a current source, thereby damping the speaker's non-linearities, but not very much. Thus, the the choice of speaker & cabinet to have a still large influence on the final sound, regardless of what the preamp is doing, especially on designs without NFB (e.g my Dual Recto employs no NFB, and sounds noticeably different driving an open back 1x12 vs a closed-back 4x12).
With transistors you have the opposite. Transistors are typically *much* better at acting as current sources when wired in a common emitter/source config, removing the need for an output transformer. The various solid-state output stages I've seen in schematics have typically had a complementary pair wired in a common emitter/source config, driving the speaker load directly. In this setup the final transistors act as current sources, providing the current gain needed to drive the speaker, once previous stages have provided the voltage gain needed for the +/- voltage swing required for the target output watts.
What I read was that providing current feedback in such a setup has the opposite effect of voltage feedback in a common-cathode->trafo->speaker setup: It
increases the output impedance, causing the output stage to behave less like a current source and more like a voltage source. This in turn reduces the dampening a low output impedance would offer and allows the guitar speaker to more heavily color the output tone, much like it does in tube & transformer designs.
Do I have that correct? If so, I am willing to consider that idea should an output transformer prove unfeasible with the parts to hand or what I'm able to get for cheap/free.
I do still want to pursue the output transformer idea, as I understand the poor output impedance and resultant speaker coloration is just one part of the sounds of the tube amps that have caught my ear. As I understand, there's other bits to that characteristic tone, including, but not limited to:
- the transformer's main and parasitic impedances interacting with the those of the output tubes and speakern forming some semi-complex filters
- magnetic saturation during hard drive doing...I'm not sure honestly, some kind of high-pass effect, I think? I understand Fender deliberately undersized the transformers for their power output in early models deliberately to try to protect the speakers from being blown out by excessive bass.
- inductive flyback voltages during crossover when the output actives are driven into clipping.
I could see the first couple items possibly replicable with judicious placement of reactive components in the output stage, though I might worry about phase margin when using negative feedback. Does the phase shift that comes with the various deliberate and parasitic filters inside the feedback loop give cause for concern about oscillations and stability with current feedback and transistors like it does with voltage feedback with tubes and a transformer?
I don't see how get or control the inductive flyback at crossover during clipping unless that's a characteristic of the speaker's inductance, not the transformer's (or FML if it's both).
Quote from: phatt on February 27, 2023, 05:01:48 PMIf you are chasing Tx output design ideas then find some of those Factory pa systems as a lot of those use Tx drive for line drive.
I have and old Inkel Pa with Tx output giving 4 or 8 Ohm as well as 70V and 100V out puts.
I used it a lot years back as it was 120W output driving a Quad box, gave me a massive sound.
I'll keep an eye out for such. Any other brands to look for besides Inkel and Factory?
Quote from: Tassieviking on February 27, 2023, 08:18:02 PMIs the 24v 40VA transformer centre tapped, or is it just one output of 24v?
If you have 12v + 12v it would make it easier I think.
The 24V 40VA transformer does not have a center tap on its secondary, unfortunately. I haven't seen a single 24V transformer intended for furnace or doorbell use that does. Rewinding is unlikely, the E and I laminations were welded together at the factory. In theory I could cut, re-wind and re-weld, but I am disinclined to try. Yes, I have access to a welder, but my welds make baby Jesus cry, and I don't know if the filler wires I have on hand are appropriate for magnetic applications anyway.
Looking at things, it think my best bet will be to make use of some voltage doublers, and stick the ground reference in the in the middle to get a bipolar power supply. It has already been pointed out to me that the votlage swings available from 24VAC are going to limit how much power I can practically get out of any design that uses the 120:16VAC doorbell transformer as an ouput transformer. I'd have to do my own center tap refernced to ground to get +/- 15VDC. A doubler in theory could get me roughly +/- 30VDC, a quadrupler roughly +/- 60. The latter would get me about 15W into the 8 Ohm load of the speaker when using the doorbell transformer as an output step-down, if my math is right.
Quote from: Tassieviking on February 27, 2023, 08:18:02 PMDo you have any Op-Amps or will it be all transistors?
I actually do, but I don't think they're fit for audio use. No 5532s, or 4558s, rather U741s and similar. I'll have them in the parts lists in my next post.
Quote from: Tassieviking on February 27, 2023, 08:18:02 PMThere are lots of old transistor amp schematics around that you could borrow sections from, maybe a proven pre-amp that sounds good already.
It all depends on what transistors you have, I would like to see just the part numbers, I don't need to see the specs for each one.
Cheers
Mick
That's a relief. Hunting down datasheets for 227 different transistor types, some of them obsolete before I was born, is no small task. I'll have the full list in my next post.
Sorted into tables for easier browsing. A note, these descriptions come mainly from label on the drawer or a quick web search. I might have some of these wrong.
BipolarGermaniumName | Polarity |
2N513 | PNP |
2N539 | PNP |
2N540 | PNP |
2N575 | PNP |
2N43A | PNP |
2N174 | PNP |
2N777 | PNP |
2N778 | PNP |
2N326 | NPN |
2N369 | PNP |
2N376 | PNP |
2N378 | PNP |
2N388 | NPN |
2N441 | PNP |
2N462 | PNP |
2N1012 | NPN |
2N1396 | PNP |
2N1412 | PNP |
2N1412A | PNP |
2N1556 | PNP |
2N1754 | PNP |
2N1760 | PNP |
2N2082 | PNP |
2N2138 | PNP |
SiliconName | Polarity |
2N1613 | NPN |
2N1711 | NPN |
2N1564 | NPN |
2N930 | NPN |
2N327 | PNP |
2N333 | NPN |
2N339 | NPN |
2N340 | NPN |
2N706 | NPN |
2N730 | NPN |
2N2151 | NPN |
2N2193 | NPN |
2N2197 | NPN |
2N2218 | NPN |
2N2219 | NPN |
2N2221 | NPN |
2N2222 | NPN |
2N2369 | NPN |
2N2405 | NPN |
2N2904A | PNP |
2N2905 | PNP |
2N2906A | PNP |
2N2907A | PNP |
2N2946A | PNP |
2N3016 | NPN |
2N3019 | NPN |
2N3053 | NPN |
2N3054 | NPN |
2N3117 | NPN |
2N3209 | PNP |
2N3250 | PNP |
2N2351 | PNP |
2N3300 | NPN |
2N3301 | NPN |
2N3302 | NPN |
2N3468 | PNP |
2N3567 | NPN |
2N3569 | NPN |
2N3585 | NPN |
2N3632 | NPN |
2N3642 | NPN |
2N3643 | NPN |
2N3646 | NPN |
2N3715 | NPN |
2N3724 | NPN |
2N3725 | NPN |
2N3740 | PNP |
2N3767 | NPN |
2N3772 | NPN |
2N3790 | PNP |
2N3902 | NPN |
2N3903 | NPN |
2N3905 | PNP |
2N3906 | PNP |
2N3947 | NPN |
2N3962 | PNP |
2N3964 | PNP |
2N4001 | NPN |
2N4013 | NPN |
2N4035 | PNP |
2N4036 | PNP |
2N4037 | PNP |
2N4062 | PNP |
2N4209 | PNP |
2N4234 | PNP |
2N4237 | NPN |
2N4238 | NPN |
2N4356 | PNP |
2N4398 | PNP |
2N4399 | PNP |
2N4405 | PNP |
2n4900 | PNP |
2N4901 | PNP |
2N4910 | NPN |
2N4914 | NPN |
2N4916 | PNP |
2N4918 | PNP |
2n4921 | NPN |
2N4922 | PNP |
2N4923 | NPN |
2N4957 | PNP |
2N5086 | PNP |
2N5087 | PNP |
2N5089 | PNP |
2N5137 | NPN |
2N5301A | NPN |
2N5032 | NPN |
2N5306 | NPN |
2N5038 | NPN |
2N5320 | NPN |
2N5321 | NPN |
2N5322 | PNP |
2N5323 | PNP |
2N5334 | NPN |
2N5415 | PNP |
2N5416 | PNP |
2N5447 | PNP |
2N5449 | NPN |
2N5629 | NPN |
2N5680 | PNP |
2N5681 | NPN |
2N5682 | NPN |
2N5880 | PNP |
2N5882 | NPN |
2N5883 | PNP |
2N5885 | NPN |
2N5886 | NPN |
2N5986 | PNP |
2N6051 | PNP |
2N6052 | PNP |
2N6053 | PNP |
2N6058 | NPN |
2N6059 | NPN |
2N6107 | PNP |
2N6109 | PNP |
2N6121 | NPN |
2N6213 | PNP |
2N6261 | NPN |
2N6282 | NPN |
2N2685 | PNP |
2N6287 | PNP |
2N6290 | NPN |
2N6295 | NPN |
2N6299 | PNP |
2N6316 | NPN |
2N6385 | NPN |
BD327 | NPN |
BU507 | NPN |
D41D2 | PNP |
D41D5 | PNP |
D42C2 | NPN |
D44C5 | PNP |
D44H2 | NPN |
D45C5 | PNP |
D45H2 | PNP |
D45H5 | PNP |
MJ10003 | NPN Darlington |
MJ10004 | NPN Darlington |
MJ10007 | NPN Darlington |
MJ10008 | NPN Darlington |
MJ10009 | NPN Darlington |
MJ10022 | NPN Darlington |
MJ10023 | NPN Darlington |
MJ11011 | NPN Darlington |
MJ11012 | NPN Darlington |
MJ11017 | PNP Darlington |
MJ11018 | PNP Darlington |
MJ11028 | PNP Darlington |
MJE105 | PNP |
MJE180 | NPN |
MJE200 | NPN |
MJE210 | PNP |
MJE253 | PNP |
MJE340 | NPN |
MJE371 | PNP |
MJ410 | NPN Darlington |
MJ490 | PNP Darlington |
MJ4033 | NPN Darlington |
MJE3055 | NPN |
MJE13002 | NPN |
MPS918 | NPN |
MPS3563 | NPN |
MPS3704 | NPN |
MPS4355 | PNP |
MPS6507 | NPN |
MPS6512 | NPN |
MPS6514 | NPN |
MPS6516 | PNP |
MPS6518 | PNP |
MPS6523 | PNP |
MPS6531 | NPN |
MPS8098 | NPN |
MPS8099 | NPN |
MPSA13 | NPN Darlington |
MPSA42 | NPN |
MPSA43 | NPN |
MPSU01 | NPN |
MPSU02 | NPN |
SPS2336 | NPN |
TIP121 | PNP |
TP2N45 | NPN |
Field Effect TransistorsJFETsName | Polarity |
2N4856 | N-channel |
2N4858 | N-Channel |
2N5114 | P-channel |
2N5115 | P-channel |
2N5462 | P-Channel |
MOSFETsName | Polarity |
BUZ11 | N-Channel |
BUZ54A | N-Channel |
IRCZ24 | N-Channel |
IRF250 | N-channel |
IRF130 | N-channel |
IRF234 | N-channel |
IRF123 | N-channel |
MPF6661 | N-channel TMOS |
MTH50N05 | N-channel TMOS |
MTP10N25 | N-channel TMOS |
MTP25N05 | N-channel TMOS |
MTP25N10 | N-channel (TMOS?) |
MTP30P06 | P-channel TMOS |
A lot of older amps have all transistor power amps in them, like the Marshall 12watt amps, Marshall 20watt and 30watt amps. They used Darlington power transistors in some of them.
Randal RG100 has all transistor pre-amps I think.
Yamaha G50 is all transistor apart from the reverb tank.
Roland Blues Cubes used to be all transistors except for one IC for the reverb tank.
Lets not forget the fantastic Albert Kreuzer 2002 BASS JFET Preamp, it might be re-voiced for guitar but might be too clean sounding.
Hopefully someone more experienced can make some suggestions of all transistor pre and power amps.
Just search around for some old schematics and borrow parts of them and get the breadboard out and experiment.
Also if you have some 4069 hex inverters or similar they can be used as Op-Amps.
Sunn Beta used the 4069 chip, and I know there are others that used them as well.
https://www.drtube.com/marshall-jmp-solid-state/
https://www.drtube.com/marshall-jcm800-solid-state/
https://music-electronics-forum.com/forum/amplification/solid-state-hybrid/923640-why-does-cmos-get-such-a-bad-rap/page2
Good luck, I will be interested to see what you do.
Cheers
Mick
Hi Koreth, Note; I'm not a qualified teck,
I just got sick of music shops exploiting my lack of knowledge so I started learning how audio gear works by reading lotsa books (long before computers and the net)
----
Now it sounds like you have a more in depth learning than myself.
Well I went deep trying to learn a lot of tecky stuff which sent me quite mad for a while.
But over time I started to come to understand a lot better by simplification and basic observation.
We all Ask why do Valve Amps have the magic mojo?
I came to understand a Valve power stage for *What it DOES*
rather than *How it Works*.
**It's a COMPRESSOR that distorts badly with big signals**
Most of the favored Valve amps of a bygone era were of very basic design.
My no 1 rule of thumb (Valve or SS);
If you want a great rock guitar amp, just build a crappy basic circuit.
Old OTx,s were cheap and suffered from all the quirks you likely already know about. Leo built to a budget so interleaved OT were out of the question.
(at least in the early days)
As the signal goes big the bandwidth gets truncated as they can't pass low freq well at high volume hence the classic flabby bass when cranked up.
The highs also get rolled off as well. This of course is exactly what you need for
Overdriven guitar Sound/Tone. Leaving plenty of bandwidth to cover the rather limited range of guitar. Some Tx spec sheets actually show this as a chart.
So by design cheap Tx's can't pass all the high freq hash which magically filter off all the nasty top end fizz. A simulation would likely show a lot of hifreq crud on the top edge of the signal at the plates of the power valve but vanishes on the secondary. Bingo a filter altering as signal goes big.
While in a SS power stage this does not happen hence they often got a bad wrap as being harsh. (a lot of those issues can be filtered out with smart design)
You have to spend a lot of money to get an OT to do a clean 20/20 hifi bandwidth.
Don't fool yourself into thinking Valve amps can't do hifi clean, they can be on par with even the best SS hifi rig. But OT will cost big money.
The compression in those classic guitar amps happens simply because of the often
**Overlooked soggy power supplies that were used**. It has a profound effect on the
outcome. plus Add in all of the other limitations and quirks you already know about.
A SS PSU is rigid while the old valve amps used high voltage supplies which were very soggy. So the moment you belted a big cranked power chord the HT drops like a brick and hence the compression effect.
Also don't forget the voltage drop from Screen grid to PI section. Just altering those voltage nodes along the supply paths can result in a very different performance.
I've tweaked a couple of later model Marshall's which use a 4k7 drop from screen to PI. Simply changing back to 18~22k makes them respond/sound much more like the older models. This thickens up and tends to smooth out the crunch effect.
the PI stage then distorts before the power valves.
Unlike most compressor circuits whether they be builtin or in guitar pedal format may well compress but seldom do they distort an if they do it's often crappy.
One way I found to get a simple SS discrete power stage to do this magic trick;
Years back I scored a little 20 watt combo with a 10 inch speaker.
While trying to tweak it and draw out the circuit I had the mains running through a light bulb limiter. It finally sounded really good but the moment I removed the limiter bulb the magic was lost.
So I permanently installed a 40Watt light bulb into the chassis with a switch to bypass it. A 20watt bulb was too low and the sound just cracked up badly.
I have fond memories of playing along with some Clapton songs and impressed that my tone sounded very much like Clapton's tones.
Darn stupid of me to swap it for bigger amp, thinking I could do the same trick and get a bit of extra volume but alas it had a Chip power stage and the idea did not work.
Those power chips are too perfect for that to work, obviously it only seems to work with a discrete power stage.
Now I'm sure far greater minds than mine have come up with even better ways but That's what I have come to understand after years of sending perfectly working transistors to smoke heaven.
I hope that helps, Phil.