Suggestions for a Medium Gain Guitar Amp Topology...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Fed by a bench power supply capable of 1.7 amps at 650 volts continuously, the PC board got traces burned off, the resistor vanished, and the OPT is now a magnet.

Ouch! I had similar, if less clamitous troubles the other week when, whilst fixing my Leslie cabinet, I mistook the schematic symbol for a thyrister for a diode.
On the up side, now I know what a thyrister is :).

Also, the smell of a burning, 40 year old, carbon comp resistor is also strangely pleasing, or is that just me?

Matt.
 
Has 'cheap as chips' not made it across the atlantic?
Matt.
I've lived in Asia, Africa, the land of Trump, and Canada, and grew up reading a lot of books by British authors. So at this point in my life, my head is full of a jumbled polyglot of different spellings, different words, different popular phrases, and terminology from many decades.

So is it "valves" or "tubes"? Tyres or tires? Checks or cheques? Metres, or meters, or feet? Should we measure velocity in metres/second as scientists do, in kilometers per hour as most lay people do, or in fathoms per fortnight (as some Americans would probably prefer, simply to be as different from the rest of th world as possible)?

I'm thoroughly confused, in other words! :D

-Gnobuddy
 
Furlongs were nominally OK in the UK until the Weights and Measures Act of 1985.

Is that so. These days, the furlong still crops up as a measure in horse racing. To the best of my knowledge, the Weights and Measures Act dictates the size/quantity of a measure of ale/spirits. In which case, a furlong of ale for me...:).

Matt.

Ps. it occurs to me I may be confusing the weights and measures act with something else. Regardless, I still want a Furlong of Ale...:)
 
I've lived in Asia, Africa, the land of Trump, and Canada, and grew up reading a lot of books by British authors. So at this point in my life, my head is full of a jumbled polyglot of different spellings, different words, different popular phrases, and terminology from many decades.

-Gnobuddy

I embrace wholeheartedly the differences between the 'Queens' English and that of the English language spoken elsewhere in the world.

Every variation represents a possible new entry into the OED, and even where one fall short of official recognition it still represents a new word to add to our shared vocabulary.

A shared language leads to a shared culture, I'm all for that, although by no means do I wish to suggest the English language is the only means by which to share a culture.

I'm on the borderlines of politics here so I'll say no more, except that as an eleven year old boy - in a Disneyland gift shop, on holiday in Florida - I'd never have asked the price of the Mickey Mouse 'Rubbers', had I been aware of the language barrier...;)

Matt.
 

PRR

Member
Joined 2003
Paid Member
...the furlong still crops up as a measure in horse racing. To the best of my knowledge, the Weights and Measures Act dictates the size/quantity of a measure of ale/spirits....Ps. it occurs to me I may ...

Weights and Measures Act 1985

It starts by defining Yard and Pound in french units (I thot this was old news??) Then it specifically says "shall not remove the pint", which would affect your drinking. (Gee, even in the US we drink by the liter.)

In Myanmar, highways are marked in mixed miles/furlongs. (A remarkable precision-- here if the exit is 284-1/4 miles down the road, we round-out the 1/4 mile. Do they have furlong odometers?)

Yes, horse-racing changes slowly. Australia went over in 1972 but UK US Canada and Ireland cling to the old lengths. There are multiple definitions of furlong, mostly too small to matter to anybody avoiding the metric system. Toronto was laid-out on a 10 furlong grid, which now appears as a near consistent 2km between highway exits.

Authority for furlongs per fortnight.
 

Attachments

  • Miles-Furlongs.gif
    Miles-Furlongs.gif
    324.5 KB · Views: 91
Last edited:

PRR

Member
Joined 2003
Paid Member
How about a firkin of ale instead? That used to be a legitimate unit of beer at one time...

A lot to drink at once. Basically "a keg". 11 Gallons/44L, give/take a brewer's exactness.

The subject came up here. Beer is often advertised by the pint. Someone thought his pint was short. This is indeed, on the face of it, a matter for the Bureau Of Weights And Measures. Who got involved. For a while it looked like all the pubware would have to be scrapped, and new calibrated glass with a line on the side bought. Which works out to much more money than the alleged shortness. Eventually the idea sank.

I can see it for gasoline pumped right into your tank. You never see it, and you can't be sure you got every drop charged. But a pint of beer, you see it, generally before your money is taken. You can speak up then. Or you can drink it, and decide it was not worth the price, go down the street to another pub with more generous drafts. Or to the beer/bait shop, buy BUD which is always the exact amount marked on the can.

And yeast-water is cheap. You say you like the taste but the real "product" is the alcohol. Do these pubs and small craft brewers have to proof-test to be sure their product is 5.5% as claimed? (There was a converse controversy: some craft beers were brewing significantly stronger than their nominal %.)
 
And yeast-water is cheap. You say you like the taste but the real "product" is the alcohol. Do these pubs and small craft brewers have to proof-test to be sure their product is 5.5% as claimed? (There was a converse controversy: some craft beers were brewing significantly stronger than their nominal %.)
It's interesting to me: I can find good tasting ales in the UK that are +/- 4% but I struggle to find them elsewhere in the world (often beer with the same label ends up at 5.5% when packaged and exported).

Germany also seems to be able to make a decent tasting medium strength beer: e.g. a Düsseldorfer Altbier is inevitably shy of 5% and yet full flavoured and complex.

Finding a near-beer (sub 1%) with taste definitely requires a trip to Germany.

The current fad for 7%+ dry-hopped IPAs just does my head in. In every sense!

Meanwhile, back on topic, has anyone an opinion on Tristan's Merlin High Gain Preamp mated to a SE beam tetrode?
Is that cutting of one's nose or is it creamy-tone city?
 
Last edited:
Layout question.

Afternoon all,

A quick question regarding layout:

From the perspective of keeping noise out of the circuit, do I need to locate the anode (load) resistors as close to the anode as possible, or am I ok to mount these resistors in a sub-assembly elsewhere on the chassis?
Obviously, the wire(s) which then connected the sub-assembly to the respective anode(s) would need to be sympathetically routed.

I think I'm right in saying that most of the noise in a valve is generated by the grid stopper, and that the first pre-amp valve is also largely responsible for the overall noise level.
The grid stoppers will be soldered directly to the valve sockets, using screened wire between the input and the first gain stage.
Board space is limited, and relocating some components to another part of the chassis would help the design flow. The anode resistors seem a likely candidate. Am I missing something?

Matt.
 
From the perspective of keeping noise out of the circuit, do I need to locate the anode (load) resistors as close to the anode as possible, or am I ok to mount these resistors in a sub-assembly elsewhere on the chassis?
Matt, there are two very different types of unwanted signals, in this case, both lumped under the term "noise".

When you say the grid stopper generates most of the noise in a valve amp, you're talking about thermal noise. This is an unavoidable physical consequence of the fact that we're not at zero degrees absolute temperature, and there is nothing you can do about it. Running longer or shorter wires to the resistor in question makes absolutely no difference to the amount of thermal noise.

The other type of "noise" is something entirely different - excessive wire lengths acting as antennae, picking up stray electrostatic and radio signals from other electrical devices.

Relocating your anode resistors further from the valves will not change the thermal noise (hiss). But, as your post suggests, it does increase the possibility of picking up hum, radio signals, or other forms of interference. It also increases the chances of the amp becoming unstable and oscillating, because some part of the output signal coupled back into the long wires, causing unintended positive feedback.

Traditional forms of valve guitar amp construction (turret or eyelet boards with lots of wires hanging off them) have always struck me as having really poor layout (compare with inter-component track lengths on any modern PCB guitar amp.) The fact that they work in spite of the poor layout tells you that guitar amps are surprisingly forgiving of long wire runs - the limited frequency bandwidth is probably helping out here.

(Try the same with a modern integrated-circuit power amp, and you're very unlikely to get away with it. PCB design is critical for stability with these!)

So: if you must relocate an anode resistor by a few inches, you will probably get away with it. If necessary, consider using shielded cable to connect the remote anode resistor to the valve anode, with the screen grounded at one end.

If you're planning to relocate multiple anode resistors to a distance, and/or move them more than a few inches from the valves, things start to get more dicey. It isn't really a great idea, and if forced to do it, I would definitely use shielded cable for all those remote anode resistor connections.

Poorly designed commercial guitar amplifiers (like the Fender Blues Junior) have sometimes implemented bad ideas and got away with it. The Blues Jr. uses several inches of ribbon cable to connect the chassis-mounted output valves to the main PCB. The amp is also known for having marginal stability, and will often burst into oscillation if you simply move those ribbon cables a centimetre from the factory routing.

Back to your situation - is it an option to make a smaller daughter board, separate from your main board but mounted very close to it, to carry your anode resistors?

-Gnobuddy
 
Last edited:
...This is an unavoidable physical consequence of the fact that we're not at zero degrees absolute temperature, and there is nothing you can do about it...

-Gnobuddy

Hi,

The universe as a whole would be a much duller place were everything in it at 'absolute zero'. I'll accept a little thermal noise in my amp by way of compromise :).

Essentially, by increasing the lengths of wires, I'm increasing the risk of picking up noise from outside the amp, without decreasing the noise in the circuit - a loose/loose situation.

The amp chassis I'm using is something of an odd shape, and requires longer runs of wire than I'd like, in it's current configuration:

Osc (42).JPG Osc (33).JPG

I've actually abandoned the idea of locating the anode resistors remotely since I posted, so apologies for wasting your time there - although your advice was helpful nonetheless.

On the whole, I'll probably need similar lengths of wire whichever way I cut it, but moving the anode resistors created a lot more problems than it solved.

Back to the drawing board...

Matt.
 
The universe as a whole would be a much duller place were everything in it at 'absolute zero'. I'll accept a little thermal noise in my amp by way of compromise :).
Fair enough! :)

And those of us who prefer valves knowingly make thermal noise much worse, because our amplifying devices of choice are heated to around 1200 Kelvin instead of being at room temperature at around 300 Kelvin. Four times the thermal energy per degree of freedom, twice the thermal noise voltage in a given bandwidth!

The amp chassis I'm using is something of an odd shape, and requires longer runs of wire than I'd like, in it's current configuration:
Looking at those photos, it looks like you could have a double-decker board layout, with a second board on tall standoffs an inch (sorry, a couple of centimetres) above the first one. Maybe shorter than the lower board, so you can still get to that bias trim pot when needed.

Leonidas' wouldn't have done it, to speed up assembly, and save pennies in construction cost. But for a one-off project, why not?

Back to the drawing board...
...the way of all truly DIY home construction projects! :D

-Gnobuddy
 
Looking at those photos, it looks like you could have a double-decker board layout, with a second board on tall standoffs an inch (sorry, a couple of centimetres) above the first one. Maybe shorter than the lower board, so you can still get to that bias trim pot when needed.


-Gnobuddy

That's the amp as it stands, essentially Vox AC4 taking up a lot of space! I'll strip it back to the bare bones before I start this project, and replace the rather dubious on/off switch and mains lead while I'm at it.

A lot of the layout choices are defined by where the valve sockets are punched, and the siting of the transformers (although it's hard to see where else the trannys would logically go). The valve sockets are an especial nuisance as they're just a little closer together than I'd like, so fitting the associated parts adjacent to each valve is proving a headache.

In addition to the schematic posted above, there's also a reverb circuit to find room for and possibly a harmonic tremolo, so space is at a premium. For reasons which elude me, I never seem to make things easy for myself :).

Matt.
 
When you make decisions about where to locate components for each stage there are 3 things to think about (actually probably more but these are the most important).

1) The impedance of the circuit node: Low impedance nodes are much less susceptible to capacitive pick up of noise and other signals which may cause stability problems (stray wiring capacitance is what we are talking about here). The cathode circuit is the lowest impedance node on a gain stage so having the cathode bias resistor and bypass cap mouted remotely is the best choice rather than grid circuit components or anode circuit components which are higher impedance nodes.

2) The signal voltage level at the circuit node. The anode circuit will have the highest signal voltage swing and so stray capacitive coupling to other circuits is more critical. I would keep the anode circuit components such as the anode load resitor and the coupling cap to the next stage close to the socket.

If the anode signal needs to cross other signal lines then make those crosses at right angles to keep capacitive coupling to a minimum.

3) Any grid stop resistor neds to be as close as possible to the grid pin. When doing PCB layout I would not hesitate to use a surface mount resistor for the grid stop.

Hope this helps.
Cheers,
Ian
 
Last edited:
Hi Ian,

Thanks for the advice. I think I had most of that covered, but it's always nice to have someone else confirm your theories; plus there's some stuff there (particularly, capacitative coupling in point 2) I hadn't put together for myself, although like most things it seems obvious once it's spelled out for you :).

I've abandoned the idea of moving components. I've done some re-measuring and I think I've space to get all the components adjacent to the valve they're associated with.
I'll cut the turret boards from hardwood offcuts, so I have the advantage of being able to put things wherever I want - down to the 10th of a mm if needs be - although I'm not sure that level of accuracy will be needed here...:)

Matt.
 
Matt,
On the noise from grid stop issue:
For a high gain preamp the thermal noise from the grid stop resistor on the input (1st) stage is indeed the highest contributor to noise.
The noise is proportional to the resistance value - so don't use higher than is necessary.
The old "rule of thumb" from the golden age of tube design is that to be absolutely free of parasitical oscillation problems then the gridstop should be not less than 8/gm. For a 12AX7 triode with a gm of 0.0016mA/mV that means 5K minimum.

You may have noted that Merlin recommends 10K for the input stage.

That input stage gridstop also acts with the Miller Capacitance to form a low pass filter to block RF from entering the amp. If you want to maintain the roll off given by the "traditional" 68K gridstop then add 470pF from grid to 0V or grid to cathode in addition to dropping the gridstop to 10K.

Cheers,
Ian
 
Hi Ian,

That bit I had got down. I'm using a 10k stopper on the first gain stage with a 680pF cap from grid to ground, as Merlin recommends.
I've been reading his preamp book to guide help me through the design process. I was reasonably savvy with lead dress and guarding against hum etc, but I'd never previously encountered noises Johnson, excess, shot and flicker. I may have been obsessing over them slightly :D.

I'm determined to keep the noise floor as low as possible, but - as gnobuddy pointed out - RF noise etc is probably the bigger practical concern. There's no sense in having a super-low SNR if the rest of the amp is a giant antenna!

Matt.
 
I'm determined to keep the noise floor as low as possible
Have you considered using a JFET as the input device? If you pick one with a low specified flicker noise, they are much quieter than valves, and probably the quietest readily available device for source impedances in the range of a typical guitar.

There is a cascode JFET/MOSFET circuit that seems to have gained use as the input stage in some high-gain guitar amplifiers, both DIY and commercially manufactured. The JFET provides the good low-noise performance, while the MOSFET copes with the typical high-voltage B+.

As a bonus, JFETs and MOSFETs are very small, and you can get an extra gain stage while using very little physical space in your cramped chassis.

mjd_tech posted a version of this FET cascode in post #43 of another thread: http://www.diyaudio.com/forums/instruments-and-amps/314730-tube-emulation-eq-5.html#post5263157

The 2N5485 seems to have gone the way of the dodo, but 2N5457s are still available, as are J111, J112, and J113 JFETS.

-Gnobuddy
 

PRR

Member
Joined 2003
Paid Member
The classic Fender input-jack wiring makes 34K, not 68K, series.

Hiss of a 12AX7 is very-very roughly 2uV or like a 20K resistor, widely variable one tube to the next. Never a whole lot lower. Can be much higher. The 34K has the secondary "benefit" that all amps in the showroom are likely to hiss similarly (when set to same gain and tone) by setting resistive hiss a couple dB higher than the good/best tubes.

Note that you probably come from a *guitar*. The guitar's noise resistance may be 5K in the lows but over 100K in the highs. This makes 5K or 50K rather moot. You also have a knob on the guitar and some players actually move it from full-UP. This just throws-away signal/hiss performance. But today everybody has a PedalBoard and the noise figure is set in those boxes, typically not by the amplifier.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.