Transistor equivalent

[spec]

I will buy that components in the next year's summer holiday, because 2 reasons:
- I will use my parents' account to pay for that components under they permission
- This year I have to pass two very important exams
I also want you post circuits:
- 2N3055-2955 amp. with 40w into 4ohm and 8ohm (because I don't know what types of subwoofer I will buy)
- other discrete transistors amps (that has good sound qualiy) with 20w into 4ohm and 8ohm (that amps will drive satellite speakers)
- I also want to know power supply requirements of above circuit
And finish, thank a lot peoples that help me!

Hi Nilolai,

I'm doing a circuit for an Amp and PSU to suit your requirements- will post soon.

 

[spec]

But dearly miss my Bogan tube stereo amp and large single stage 12" Phillips speaker with tuned ducts in 4cm walls of compressed particle board for the acuity of imaging of all sources of sound in stereo or quad. It wasn't the best but it sounded like live music.

Hi Tony,

Yes, the sound of live music that is the nirvana. The closest I came to it was at a hifi show at London UK: pair of electrostatics and a massive bass horn, fed from its own valve amp. The source was a transcription turntable and very expensive arm, probably SME, and the best MC cartridge. It was in a large room which had anechoic treatment- never heard anything as natural effortless and open as that since. Close was a pair of electrostatic headphone, again with valve amps, transcription deck and decent MC cartridge.

Like you, I really like the open sound which you tend to get with the minimalist speakers: in the UK Lowther and to a lesser extent, Epos and B&W. The Epos ESP14 is a twin speaker arrangement with just a single capacitor feeding the tweeter. The bass/midrange driver is tailored to roll off at the appropriate frequency.

When the house refurbishment is complete, I will be looking at a pair of B&W 685-S2s which use the same philosophy- not tremendously low bass but OK, and not too big.
https://www.bowers-wilkins.co.uk/Speakers/Home_Audio/600_Series/685.html

The old cinemas have massive horns with a single loudspeaker driver and valve amp, made by RCA I think- only about 20W. That was a nice expansive sound, and of course the room acoustics helped a lot. Loud too.

In the days when hifi was way beyond the average person's financial means, I used to build amps and speakers part-time. It made me laugh. You would set the system up in their lounge and play a record (disk). You could tell they weren't happy. Normally, they would imply that it was a touch tinny. They were very happy when I turned the treble down and the bass up- I soon learned! Then there are the reggae and heavy metal devotees; the more bass resonance the better- I too like that kind of music.

 

[spec]

In the student era, time was abundant, but cash was zero so we used to go to all sorts of antics to get a good sound with no cost. At the time, you could savage as many 5 to 8 inch speaker chassis from TV's radiograms and similar, as you liked. The first move was to put them in a decent box. The next was to use the biggest driver you could get and couple that to the smallest driver you could get via a capacitor. This was followed by doping the HF speaker cone with yacht varnish. The next, cut away the paper surround of the bass driver and replace it with rubber, either plain of foam. Finally dope the centre 75% of the bass unit, also with yacht varnish. Sounds crazy but, do you know, the final speaker didn't sound bad- certainly a vast improvement on the original.

A friend built a guitar combo with a valve amp made from TV bits and 10 salvaged speakers. I never heard it, but he said that also sounded pretty good- not compared to a Fender, Marshall, or Vox though.

 

[Nikolai Petrenko]

spec, can you post 20w transistor amp circuit?

 

[spec]

Sure. If you are going for 20W why not consider class A: very nice sound and simple to build! 2N3055s would do fine.

 

[Nikolai Petrenko]

class A, yes, sound very nice.
for subwoofer 40w too!

 

[spec]

40W class A could be done, but you would need 4x 2N3055s and would be more difficult. I will post a class A/B circuit and a class A circuit for you to consider.

 

[KeepItSimpleStupid]

Have you ever heard Stax earspeakers? Incredible.

 

[Nikolai Petrenko]

4x2n3055 not a problem. I have made circuit and pcb for 3 years only with permanent marker and paint, I made pcb for microcontrollers circuit (for diy soldering station) without any softwares, not difficult but very simple and save money.
But 2n3055 not good at high frequencies

 

[spec]

Have you ever heard Stax earspeakers? Incredible.

They sure are and heap big expensive too. I think the phones in my post were Stax. I once approached domestic management about buying a pair + matching amp- didn't go down too well!

 

[spec]

4x2n3055 not a problem. I have made circuit and pcb for 3 years only with permanent marker and paint, I made pcb for microcontrollers circuit (for diy soldering station) without any softwares, not difficult but very simple and save money.

OK Nikolai, I will investigate 20W and 40W class A amplifiers using 2N3055s.

The most critical area with class A semiconductor amplifiers is dissipating the power in the output transistors, the 40W version in particular. With the 20W version, you could build a stereo pair plus power supply in one case, but for the 40W version a separate case for the power supply and each amp would be best.

Here are some points:

(1) Constraints

(1.1) The junction temperature of the output power transistors (2N3055) must not exceed 200 deg C.

(1.2) Every output power transistors must be kept within its data sheet Safe Operating Area (SOA)

(1.3) The connections from the driver circuit to the output power transistors must be short. Ideally, the transistor leads should solder straight on to the PCB. This may seem at odds with the other requirements covered here but it is not, and can be done: more on this further down the line

(2) Implementation

(2.1) Use an insulator between the transistor case and heatsink with the lowest possible thermal resistance. Without going to esoteric and expensive insulators, mica is the best. Also ensure that that each power transistor is bolted down firmly with the biggest bolts that will fit the holes in the transistor case. A very thin layer of the best thermal grease will also be required. The heatsink under the insulating washer must be perfectly flat. As a side note, a commercial UK audio amp had problems with output transistors blowing. It was discoverd that the heatsink machining stage had been missed out!

(2.2) Use a heatsink with the lowest possible thermal resistance, heatsink to air. This normally means big and expensive. But you can make a heatsink at minimal cost.

(2.3) Use a separate heatsink for each power transistor.

(2.4) Use the mono-block approach where each amplifier is in its own case.

(3) Making a heatsink

(3.1) Get some scrap copper or aluminium and fabricate a heatsink. The metal will have to be thick in the area where the transistor bolts down, say 10mm min. Scrap yards, in my area anyway, are full of suitable aluminium scrap. Also garages throw out aluminium castings and the like that would be ideal.

(3.2) Use fans to cool the heatsinks. These can be salvaged from scrap PCs. Fan cooling a heatsink can radically lower the thermal resistance, heatsink to air.

(3.3) Use liquid cooling, as is used on some high-power PCs to cool the processor and graphics chips. Liquid cooling may sound crazy but it isn't that hard to do and would provide the ultimate in thermal dissipation.

(4) Power Supply

The power supply will also be big, but should not be too critical to design. You will need some very high value reservoir capacitors; many in parallel would be fine. I imagine at least 100,000 uF will be needed. 4 x 20A upwards rectifiers, or bridge will also be needed. You have mentioned the mains transformer before, and plan to use one from a microwave. That should be fine. Almost certainly you will have to replace the secondary winding to suit the amplifier. The type of core etc is not relevant to the electrical performance or the quality of the audio amp, especially in the case of class A. When considering the last knocking in amplifier design the type of transformer may be significant, but it is a moot point. As long as the transformer provides enough current at the right volts, all will be well.

All this is for your considerations. The position will be much clearer when the amplifier designs are complete and I have done the thermal calculations.

One final point: you will need a thermometer to measure up to around 150 deg C. High accuracy is not needed but the sensing probe will need to be small enough to measure the temperature of the transistor base and various points on the heatsink.

I didn't know your skill level, but as you now say that you have made PCBs and built electronic units before, there is no reason why you shouldn't have a successful class A amplifier project, but think big amplifier and power supply!

 

[spec]

But 2n3055 not good at high frequencies

The 2N3055 will be fine, especially in a class A design- not the best, but fine. The high frequency performance of an audio power transistor is important for a number of reasons. With class A/B it is important to have a high gain at high frequencies so that the amps feedback mechanisms, both local and overall, can correct the distortions caused by the transistor cross-over, which can be greatly reduced by the optimum quiescent current, but not lowered to levels acceptable for hifi. In class A the transistors simply do not cross-over so this distortion is totally eliminated.

The other thing is that, in the class A/B design I previously posted, the output transistors are operated in an emitter follower configuration, which optimises their frequency response. On the other hand, the amplifier that you suggested (not getting at you) has complementary emitter followers. This arrangement is always problematic* for various reasons, one of which is poor frequency response, especially as the bases of the output transistors are not terminated with resistors. This means the output transistors will turn on fast, but will have a problem turning off. This is made much worse because the output transistors are operating in common emitter.

* I'm not trashing this arrangement per se, just in this particular application where you are trying to get the best performance from a given pair of output transistors.

 

[audioguru]

Frequently we can buy the same speaker at 4 ohms or at 8 ohms. Of course when fed the same voltage the 4 ohm one is louder at all frequencies than the 8 ohm one because it uses more current. Twice the power, but only a little louder.

An old class-A heater/amplifier is the opposite of a modern class-D cool amplifier. Class-A also has distortion that needs a lot of negative feedback for it to be low.

 

[spec]

Frequently we can buy the same speaker at 4 ohms or at 8 ohms. Of course when fed the same voltage the 4 ohm one is louder at all frequencies than the 8 ohm one because it uses more current. Twice the power, but only a little louder.

Agree, but there is a proviso: a 4 Ohm speaker will only be louder if the power amplifier can supply twice the current- that is often not the case especially at high volumes.

Also, with a lower impedance load, an audio power amp will inevitably produce more distortion for a given power output because of the increased current demand- sometimes the distortion is twice as much.

Why the bold 'at all frequencies' I didn't mention the frequency response of the amplifier, only the impact that the frequency response of the power transistors has on the distortion and other factors It would be a rare amplifier that didn't have a flat frequency response over the audio band. To a first approximation, the distortion characteristics of an audio power amplifier are not related the loudness.

An old class-A heater/amplifier is the opposite of a modern class-D cool amplifier.

I'm sure your comments are well-meant, but why all the innuendo and constant sniping from the side-lines on this thread? Class A, AB, D ... have been around for years. There is very little new under the sun in terms of design approaches- just the enabling tecnology that allows them to be realised in practice. I just can't see where you are coming from. If there is something that is bothering you please come out with it and we can discuss it man to man.

The other thing is: there is a whole sub-culture of class A amp bulders even today- just look on the net.

Class-A also has distortion that needs a lot of negative feedback for it to be low.

Totally disagree. The class A amp has lower distortion than a class AB. The constraints are less for class A. For one thing, the output drivers are feeding a much less troublesome load than with class AB . The rest of the circuitry is exactly the same. In fact you can look at a class A amplifier as a class AB amplifier with the quiescent current cranked up to (Vsupply tot/2)/Rl. In practice, forgetting power dissipation, you can turn any class A/B amp into a class A amp by simply doing just that- I have done it many times, but only for a short period though.

The other advantage with class A amps is that you don't have the huge gulps of current sucked from the supply rails, especially the nasty non-linear currents needed to sort out the cross over. With class A, the current is constant and, therefore, modulation products injected from the supply lines are also eliminated.

I would go as far as to say that class A amps have the best performance bar non- the one big drawback is obviously the power dissipation!

I know nothing about class D, but if the awful noise that comes from class D amps in TVs and laptops, for example, is an indication they don't sound too good.

But, if you are a devote of class D, please post a circuit using components that Nikolai has, so we can discuss that. Maybe that would show the best way forward. I'm just trying to help Nikolai by posting a few approaches for him to consider, which after all, is in the spirit of ETO, as far as I can tell.

 

[MrAl]

Hi,

I dont want to get into this too deep, but just to mention another possibility is that some amplifiers have a different distortion rating for 4 and 8 ohm speakers. That's probably because the transistors introduce some distortion that has more or less constant amplitude, so a lower output voltage with the same volume (4 ohm) would show up as higher distortion in the voltage wave.
I suppose it could be the other way around too...more distortion with an 8 ohm speaker.
But in any case there is sometimes a difference.

Also, just a little point, power is V^2/R so a 4 ohm speaker would theoretically put out twice the power as an 8 ohm speaker provided the amp could handle the extra current demand, and twice the power would be noticeable.

 

[spec]

I dont want to get into this too deep

Hi MrAl,

... just to mention another possibility is that some amplifiers have a different distortion rating for 4 and 8 ohm speakers. That's probably because the transistors introduce some distortion that has more or less constant amplitude, so a lower output voltage with the same volume (4 ohm) would show up as higher distortion in the voltage wave...

I hadn't remembered what you note about distortion being a constant voltage at the cross-over point. In view of that, there are two factors increasing distortion with lower impedence loads- the constant voltage aspect that you mention, and the distortion indroduced by the incresed current demand on the amplifier.

In general, I have found the the lighter the load on an amplifier, especially class A/B, the lower distortion.

I suppose it could be the other way around too...more distortion with an 8 ohm speaker.
But in any case there is sometimes a difference.

True. It all depends on the nature of the amp- not common though.

Hi,

Also, just a little point, power is V^2/R so a 4 ohm speaker would theoretically put out twice the power as an 8 ohm speaker provided the amp could handle the extra current demand, and twice the power would be noticeable.

(in your post did you mean V/2 rather than V^2)

Twice the power is +3dB, which due to the logarithmic response of the ear, is defined as the biggest increse in power that, for all intents and purposes, is not audble.

In a caparison between distortion you would adjust the volume so that the power is the same for both 8 Ohm and 4 Ohm speakers. The formula just allows you, to a first approximation, to derive what current the amplifier would be asked to supply flat out. The formula is misleading in that respect. In practice, it's more complicated because you have to take into account the amplifier overhead voltage and the output current capability.

Take an example:

8.00V @ 8 Ω load = 1.00A = 8W,
5.66V @ 4 Ω load = 1.42A = 8W.

I hope I have done the sums right- let me know if not.

 

[spec]

Mr Al,

Without getting too deep, here is a graph showing current gain (hFE) against collector current for a 2N3055. The first red arrows show the hFE at an output current of 3A with an 8 Ohm load. The second arrow shows the same power output but with a 4 Ohm load (4.24A). Notice that the Hfe has dropped from 50 to around 36.

Using the 150 deg C line is reasonable because that is about the temperature that the junction would be after driving a load for a while. This drop in Hfe will add distortion in a number of ways, depending on the amplifier design, of course. What is not shown here, is the change in Vbe, and Vsat. The first introduces more distortion and the second reduces the maximum voltage that the amp can output into a 4 Ohm load.

Interestingly, you wouldn't want to use this transistor in an audiophile amp for any ICs over about 700mA peak, because the hFE drops off so fast. And the situation is even worse if you look at the worst-case hFE on the datasheet, which is only 20@ 4A

Compare that graph to the IC/hFE graph of an audiophile transistor, where the hFE is reasonably flat from 50mA to around 5A. The worst-case hFE for the 2N3055 is 20@ 4A but the worst-case hFE for the MJL3281A is 75@ 5A and still 45@ 8A. And if that is not enough the ft of the 2N3055 is 3MHz compared to the ft of 30MHz for the MJL3281A.

 

[Tony Stewart]

An essential part of a sub-woofer design is dampening factor (DF) between 100 and 1000. Higher the better. This requires lots of open loop gain. negative feedback and low closed loop gain to reduce the ESR of the final stage, since DF = load/source impedance.

Strictly speaking ( no pun intended) it is the minimum speaker impedance as the load which can generate back EMF from motion and closed loop source impedance. Thus the goal of any "Voltage Source" is getting the source impedance close to zero or 0.1% of the load for a DF of 1000. ( 100 for consumer quality)

This same DF design rule applies to all power supply designs and is called % Load Regulation Error (%LRE) as DF is the inverse of %LRE. Generally %LRE is tested at 50% ~100% of full load or 10~90% and the %LRE is separated into DC error and overshoot.

Imagine a bass kick drum impulse. If a subwoofer AMP has a DF of only 100 , then it ok for an uneducated listener, then it can generate more resonant distortion from the overshoot generated by the back EMF and speaker resonant frequency rather than the resonant frequency of the kick drum. If distorted it would make the bass sound "muddy" or boomy. The inertia of the cone becomes the driving force of the EMF and the quality factor of the speaker is the lowest ESR or coil resistance possible relative to the acoustic impedance.

 

[audioguru]

Hi Spec,
1) I mentioned that a 4 ohm speaker could be louder than an 8 ohm speaker at all frequencies because I think many people would wrongly think that since the impedance of the speaker impedance graph you showed is all over the place at different frequencies then the frequency response would also be all over the place if the impedance was 4 ohms instead of 8 ohms.

Of course the current is twice as high for the 4 ohm speaker but we assume that the amplifier can drive it and is not overloaded with it.
Since the distortion from modern amplifiers is so low then double the distortion caused by doubled current is still not heard.

You showed the frequency response of a ported (two low frequency resonances) speaker that has a crossover to a tweeter because it also has an impedance peak then dip at mid frequencies. It has a minimum impedance of 5 ohms and you said that some speakers rated at 8 ohms have a minimum impedance of 2 ohms so maybe you imply that this speaker is rated at 8 ohms.

I disagree that the impedance of an 8 ohm speaker could be 5 ohms as you show because its impedance cannot be less than the 7 ohms resistance you stated. I think it is a speaker rated at 4 ohms.

2) I think that Nigel, the moderator will agree that his modern class-D amplifiers have excellent performance.

3) I believe that the emitter-follower output transistors used in a class-AB amplifier produce less almost unmeasureable harmonics distortion than the common-emitter transistors in a class-A amplifier because the common-emitter transconductance is exponential, not linear. The common emitter transistor compresses the tops of the sinewave when it tries to reduce its current and expands its current at the bottom of the sinewave when it increases its current too much.
My simulation is with a resistor as the collector load. The distortion is less when the common-emitter transistor has a current source for its load.

I think that all amplifiers should be Texas Instruments tiny and cool class-D amplifiers today.

 

[MrAl]

Hi MrAl,



I hadn't remembered what you note about distortion being a constant voltage at the cross-over point. In view of that, there are two factors increasing distortion with lower impedence loads- the constant voltage aspect that you mention, and the distortion indroduced by the incresed current demand on the amplifier.

In general, I have found the the lighter the load on an amplifier, especially class A/B, the lower distortion.



True. It all depends on the nature of the amp- not common though.



(in your post did you mean V/2 rather than V^2)

Twice the power is +3dB, which due to the logarithmic response of the ear, is defined as the biggest increse in power that, for all intents and purposes, is not audble.

In a caparison between distortion you would adjust the volume so that the power is the same for both 8 Ohm and 4 Ohm speakers. The formula just allows you, to a first approximation, to derive what current the amplifier would be asked to supply flat out. The formula is misleading in that respect. In practice, it's more complicated because you have to take into account the amplifier overhead voltage and the output current capability.

Take an example:

8.00V @ 8 Ω load = 1.00A = 8W,
5.66V @ 4 Ω load = 1.42A = 8W.

I hope I have done the sums right- let me know if not.

Hi,

First, power is calculated knowing the voltage and the resistance as:
P=V^2/R
and not V/2/R.

Second, you are the first person i ever heard describe a +3db change as NOT audible to a normal human ear. Normally people state that a +3db change is just detectable. But read on...

If you were an environmental scientist that knew a relationship between the wellness of a certain field in a given town was related to the number of spiders found in that field, and you needed a unit of measure so that you could relate this information to other scientists without too much trouble, which name would you prefer to give this unit of measurement:
1. The Spid
2. The ThreeSpid

#1 implies that each spider is counted as one unit, so one spider=1 Spid.
#2 implies that each group of 3 spiders count as one unit, so now 3 spiders=1 ThreeSpid
Also, if both units were already in vogue, then ThreeSpid=3 Spid's.

Now you switch to measuring loss in long transmission lines.
You find that when you have a long loop of wire of such a length that the sound heard through that line to a human listener is just noticeable as being less than it was without that line and you are told to pick a unit for that measurement, which do you choose:
1. The decibel
2. The ThreeDecibel

and here the ThreeDecibel=3 decibels.

In this case if you choose the ThreeDecible then there is no physical basis and no use for the decibel, because if you raise or lower the sound by only 1 decibel that is only 1/3 of the change that a human can just detect. So what would be the purpose of having one decibel in this physical system.

This is probably why all references to human hearing state that the smallest change that a human can hear as just detectable is 1 decibel.

This all really originates i think from a guy picking up weights, and noting how much weight is detectable to a human, but it was later applied to transmission lines and then all kinds of stuff.