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

March 28, 2024, 02:09:34 PM

Login with username, password and session length

Recent Posts

 

Popular electronics 22 watt amp

Started by cbg Rick, April 30, 2015, 06:57:48 PM

Previous topic - Next topic

Loudthud

The output transistors 2N4921 and 2N4918 are (were) silicon. The bad news is they are only rated at 30W and 3A collector current (if you get Motorola parts). A  different version, MJE4921 and MJE4918 are a little beefier at 40W. But no matter, these parts haven't been around for probably 30 years. Note that those parts have a reverse pinout compared to most TO-220s. Like many parts of that era, the beta falls off substantially as maximum collector current is approached, less than 20 at 3A.

The two RCA diodes that mount on the heatsink were kind of cool in their day because you didn't have to insulate the package from the heat sink. They are long gone too I suspect. I just use a TO-92 VBE multiplier face down with an improvised clamp.

SWTP amps weren't known for their reliability. They would usually burn up if subjected to any serious use. Today we have much better parts and know how to beef up the weak points. But this amp does not need a Zobel, it doesn't have enough gain or bandwidth to need one.



cbg Rick

I'm going into this project knowing that there are better ways to build an amp. I realize the transistors are probably no longer available and substitutions will have to be made. Its a learning exercise.
The points you guys are making about this circuits short comings are not falling on deaf ears, I understand it is a far from perfect amp. I don't think any children or small animals will be harmed if I build it, so I am going to forge ahead. :cheesy:
In case I didn't mention it before, it has been 30 years since I did any real electronics work so I have forgotten a bunch of stuff. It is slowly coming back as I study the replies and google the stuff that I don't understand. I need to look into this VBE multiplier that you mention..... Would it also work to use the BE junction of a TO-220 transistor to replace the diodes and clamp that transistor/diode to the same heatsink? 

Loudthud

The VBE multiplier is the circuit around the 2N3904 transistor in the schematic I posted in reply #7. For simplicity, replace the 68 Ohm resistor with a short. The circuit multiplys the Base to Emitter Voltage of the transistor so that it acts like several diodes in series. The pot adjusts the amount of multiplication so that the power transistors turn on just enough to eliminate crossover distortion. As the power transistors heat up, they need less Voltage to conduct the same current and the VBE multiplier (hopefully) reduces the Voltage applied to the Bases.

A power transistor could be used to make it easier to mount to the heat sink. You will need an insulator. This circuit works better with a high beta transistor, the beta interacts with the resistor values to some extent. It takes a little experimentation to get the thermal compensation just right. But this is a guitar amplifier, it doesn't need to be perfect.

cbg Rick

I vaguely remember needing to add a small resistance to help clean up crossover distortion, I just don't remember it being called a VBE multiplier. The posts here are helping a bunch and I appreciate you guys taking the time to add your thoughts and sharing your knowledge. I've downloaded LTSPICE and am working on figuring out how to use it. Seems like a great tool.

Roly

Quote from: cbg RickI'm sorry but, you keep referring to this amp as having Ge transistors, I checked the transistors in the parts list and I'm not seeing any Ge transistors listed. Am I missing something?

No, you may well be right, but as teemuk points out above, this is the least of your redesign problems.  This is still a case of "jack up the hubcaps and put a new car in-between".


The VBE multiplier is only one way of getting a temperature sensitive bias, such as a string of diodes, but only the transistor circuit operates as a "base-emitter voltage multiplier" (thanks to the gain of the transistor).

Quote from: cbg RickWould it also work to use the BE junction of a TO-220 transistor to replace the diodes and clamp that transistor/diode to the same heatsink? 

It not only works, it's my personal favorite.  BD139/140's for example have a very handy mounting hole through the middle.


Quote from: cbg Rickso I am going to forge ahead. :cheesy:

Popcorn?  Check.
Hard hat?  Check.
Smoke mask?  Check.

;)
If you say theory and practice don't agree you haven't applied enough theory.

cbg Rick

#20
No smoke yet, but it was close. I followed the original schematic and breadboarded the circuit. I used 2n3904 and 3906 transistors for Q1,2 & 3 for the output transistors I substituted NTE 152 & 153. With a 35V supply I adjusted R5 for half the supply voltage at the emitters of the output transistors as per the instructions, after about 5-10 seconds the output transistors were to hot to touch. Not wanting to let the smoke but wanting to hear it work I put a 50 ohm transistor between the collector and supply, collector and ground of the output transistors, readjusted for half the supply at the emitters. Everything was cool  8) and the amp does amplify.
So my question is what is the proper way to get the current under control in the output transistors? Would lowering the values of R9 and R10 put less bias voltage across the BE junction of the output transistors and in turn lower the CE current?

Roly

{here starteth the redesign...  8| }


Quote from: cbg Rickwhat is the proper way to get the current under control in the output transistors?

Okay, assuming that your output transistor are mounted on a sufficient heatsink (and by "sufficient" I really mean "excessive" - such as a couple of redundant CPU coolers o.n.o.), you are discovering why almost all amps are fitted with emitter resistors in the output pair.  These could be just about any value from 0.1 ohms to a couple of ohms, but typically 0.22/5W.

{the bigger the heatsinks the greater their thermal inertia and the more time you have to hit the "off" switch before your output transistors melt.}

What you are seeing is thermal runaway, what these resistors are there to prevent.

The reason for thermal runaway is that the VBE for each output transistor depends on temperature, falling with increasing temperature.  If the bias voltage is fixed then as the transistor gets warm and its VBE drops, it turns on more, and gets hotter.  This is a positive feedback loop that quickly leads (as you found) to thermal runaway, and if not quickly caught the destruction of the output pair.

With an emitter resistor between each emitter and the half-rail the result of more current is more voltage drop across the resistor, and this is in opposition to the drop of VBE.  If the resistor is large enough, and it doesn't need to be very large, the voltage across the emitter resistor will rise faster than VBE falls, reducing the thermal loop gain below unity and making the output stage thermally stable.

So the first thing to do is add a couple of emitter resistors.

Another thermal feedback loop exists between the transistor cases and the three bias diodes and ideally at least two of these should be tightly thermally couple to the output transistor cases (to increase the loop frequency response/reduce loop delay by greatly reducing the delay between the transistors getting hot and the diodes following).

Purely for physical reasons transistors in cases such as TO-220 are a whole lot easier to mount in intimate contact on the (TO-3) output transistor mounting bolts (thermally coupled but electrically isolated).  My personal favorite is to mount a couple of BD139's (or BD140's, doesn't matter, just get the polarity right) on or very near the output devices, and use the E-B junctions as two of the bias diodes.

In some stereo amps you will find a 3-way tagstrip mounted on one of the transistor bolts, carrying a normal diode bent over so it rests on the top of the transistor case and with a glob of thermal paste to improve coupling - a bit agricultural, but effective.


Quote from: cbg Ricklowering the values of R9 and R10

It might, but at the expense of increasing the dissipation in the drivers, and it doesn't address the basic problem of currently having a thermal loop gain above unity, and runaway - the output transistors get hotter faster than the thermal compensation diodes in the bias chain can pull back the base current injection.

A way to reduce the bias is to reduce the value of R6/100r, but that alone won't reduce the thermal loop gain, just start it off at a lower value.  This might make the amp thermally stable at idle but won't stop runaway once it is driven.

It may seem strange but this thermal loop between output transistors and their compensation diodes is subject to the same stability criteria as an electrical feedback loop.

If the loop gain remains well below unity, say x0.7, the output stage will finally settle to a steady state, but the closer it is to unity the longer it will take until it settles, and may significantly over- and under-shoot before it does.

As it comes to unity (e.g. x0.99) the whole amp may go into a slow thermal oscillation, continuously ramping up and down but never settling.  {in a marginal case you can even get an oscillation on an oscillation, an idle current wobble that dies away only to keep reappearing.  This can be over periods of tens of seconds to minutes, and can depend on the ambient temperature.}

As it comes above unity the output stage will be thermally unstable, what you have now, and may well self-destruct if not caught in time.

The acid test for thermal stability is a half hour run into a dummy load at about 2/3rds full power at a high ambient temperature (heat lamp), and you are looking for the heatsink temperature to settle under drive, then for the output pair current to ramp down to idle as soon as drive is removed.

HTH
If you say theory and practice don't agree you haven't applied enough theory.