Transistor equivalent

[Nikolai Petrenko]

but we can make linear opamp from discrete high-quality transistor, why not??? My "nerd" friend even tried to make opamp from a lot of triodes and tetrodes that he had received from his brother (what a rich man!). This unique opamp may approx 3kg and need 100w transformer to work. lol

 

[Nikolai Petrenko]

Ah I have a question.
Many people (in my neighborhood) like 2SD718-2SB688 amplifier, I heared it for many time in friend's house but I feel sound worse than LA4440 or HA1392 in btl mode.
I know many ICs have good sound such as LM3886, STK....., LA..... but their power not really high, almost 45-200w. I want make power amp above 300w so only can use power transistors (tubes too expensive and high power consumption). Any suggest about what transistor should be use???
I have heared 2SC5200-2SA1943 but don't know about sound quality of it.

 

[spec]

P
...........
Vacuum tubes?? You mean those old fashioned hot things that produced tremendous amounts of even-harmonics distortion? Some people think the distortion makes a gEEtar sound better.
Transistors?? You mean those old fashioned little things that produce all kinds of distortion?

Opamps have been used for audio for many years and new ones are much better than ones designed 47 years ago. Some opamps are designed to be used in phono, tape or anything preamps and are called "audio opamps" because they have low noise and extremely low distortion.

A transistor produces a lot of distortion because it is not linear. It needs to have a circuit with very high voltage gain so that a lot of negative feedback can be used to cancel the distortion. An overloaded transistor produces more distortion than a transistor that is not overloaded. Opamps have the very high voltage gain that is needed.

but we can make linear opamp from discrete high-quality transistor, why not??? My "nerd" friend even tried to make opamp from a lot of triodes and tetrodes that he had received from his brother (what a rich man!). This unique opamp may approx 3kg and need 100w transformer to work. lol

You are both right. Of course opamps are only made from transistors, BJT and FET, but the point is that a lot of the basic development work has already been done for you with an op amp approach, and because of fabrication leads are extreemly short and the layout can be optimised. Opamps make extensive use of constant current techniques to linerise the transistors and also to get high gain, as do most hifi descrete amps. Also, the designers have spent a lot of time sorting the frequency response so the amp behaves well with feedback. But on the other hand, the fabrication process does bring some disadvantages.

 

[spec]

Ah I have a question.
Many people (in my neighborhood) like 2SD718-2SB688 amplifier, I heared it for many time in friend's house but I feel sound worse than LA4440 or HA1392 in btl mode.
I know many ICs have good sound such as LM3886, STK....., LA..... but their power not really high, almost 45-200w. I want make power amp above 300w so only can use power transistors (tubes too expensive and high power consumption). Any suggest about what transistor should be use???
I have heared 2SC5200-2SA1943 but don't know about sound quality of it.

Wow. You and your friends are into hifi in a big way; 300W... nice.

You know my warnings about developing an audiophile phono stage, well multiply that by 10x for power amps! And perhaps another 10x for 300W- good fun though. Just the power supply will be a development programme in itself. You must have some power hungry speakers.

I will have a look at your questions. But in the meantime one more bit of advice. When you buy any expensive electronic components, especially power transistors, alway get them from a reliable source because there are a lot of rip-offs on the market. Not only will they fail but they will not sound too good either.

 

[spec]

Just had a look at what 300W implies:

Assume 8 Ohm speaker

Supply rails: +- 75V (allowing 5V overhead)

Current: 18A (allowing 100% over current capacity)

That would be some PSU. Of course, you could half the current as far as the power supply is concerned because it is unlikely, with music, that the high current demands will last long. All the same, the output transistors will need to handle that current

 

[Nikolai Petrenko]

yes, I will collect high power-quality transformer from old microwave oven (the core make from steel-silicon, good). I will take speaker from electric dumps (if I lucky enough, might found an old Sony speaker!) or buy from professional amplifier repair mans. Transistor I will get them from old Japan electric machine or receiver from my friends in foreign countries. Buy them on the market 80% is fake or chinese clone.
Main power in my neighborhood not stabile so I need use zenner diode (voltage regulator), connect regulated volt to "Base" of power transistor 40411 to make high power volt regulator. I will draw circuit and post then yous will check error. Thank a lot

 

[MrAl]

Hello,

Another problem with making a differential amplifier from discrete transistors is offset drift with temperature.
When two devices are physically separated by a distance they tend to have different temperatures especially if there are any air currents. The closer they are together the better they stay similar. Not only should they be close together, they should be a matched pair. You can get matched pairs but they are quite expensive. You can also get transistors on a single chip (very close together then) but they may or may not be matched. Matching here means they both track over temperature. A mismatch causes a change in input offset, which in turn causes an output offset and so limits the useful gain of the stage.

 

[spec]

yes, I will collect high power-quality transformer from old microwave oven (the core make from steel-silicon, good). I will take speaker from electric dumps (if I lucky enough, might found an old Sony speaker!) or buy from professional amplifier repair mans. Transistor I will get them from old Japan electric machine or receiver from my friends in foreign countries. Buy them on the market 80% is fake or chinese clone.
Main power in my neighborhood not stabile so I need use zenner diode (voltage regulator), connect regulated volt to "Base" of power transistor 40411 to make high power volt regulator. I will draw circuit and post then yous will check error...

I wish I could be there and join in the fun

Thank a lot

No probs; a pleasure my friend

 

[spec]

Hello,

Another problem with making a differential amplifier from discrete transistors is offset drift with temperature.
When two devices are physically separated by a distance they tend to have different temperatures especially if there are any air currents. The closer they are together the better they stay similar. Not only should they be close together, they should be a matched pair. You can get matched pairs but they are quite expensive. You can also get transistors on a single chip (very close together then) but they may or may not be matched. Matching here means they both track over temperature. A mismatch causes a change in input offset, which in turn causes an output offset and so limits the useful gain of the stage.

You are spot on. A good way to keep the output offset to a minimum is to use the servo approach where an op amp monitors the output voltage and adjusts the input stage to make the output offset virtually zero. This is at DC so it does not affect the audio frequency range which is typically taken to be 0.1Hz to 40 KHz for audiophile amps.

Another consequence of not matching the transistors and their operating conditions is that the dstortion goes up. In hi-end amps they go to extrodanary lengths to keep the two input transistors running the same and under optimum conditions:

(1) matched transistor fabrication
(2) physically close
(3) optimise Ic (higher Ic better for minimising distortion and frequency response. low Ic better for low noise)
(4) minimise current modulation index (change in Ic for for input sigal= dic/IC)
(5) optimise VCE (high good for minimising distortion and maximising frequency response but bad for noise)
(5) minimise voltage modulation index (change in VCE for changes in input voltages = dvc/VC.)

Just the input stage of a high end amp is a subject in it's self! The first time I saw one of these circuits I couldn't make head nor tail of it. This one is fairly simple as they go and the practical aspects, decoupling etc, have been left out to make the fundamental operation clearer:

 

[Nikolai Petrenko]

2N3055-2N2955 or C5200-A1943 or TIP41-TIP42 better to use in audio amplifier?

 

[MrAl]

You are spot on. A good way to keep the output offset to a minimum is to use the servo approach where an op amp monitors the output voltage and adjusts the input stage to make the output offset virtually zero. This is at DC so it does not affect the audio frequency range which is typically taken to be 0.1Hz to 40 KHz for audiophile amps.

Another consequence of not matching the transistors and their operating conditions is that the dstortion goes up. In hi-end amps they go to extrodanary lengths to keep the two input transistors running the same and under optimum conditions:

(1) matched transistor fabrication
(2) physically close
(3) optimise Ic (higher Ic better for minimising distortion and frequency response. low Ic better for low noise)
(4) minimise current modulation index (change in Ic for for input sigal= dic/IC)
(5) optimise VCE (high good for minimising distortion and maximising frequency response but bad for noise)
(5) minimise voltage modulation index (change in VCE for changes in input voltages = dvc/VC.)

Just the input stage of a high end amp is a subject in it's self! The first time I saw one of these circuits I couldn't make head nor tail of it. This one is fairly simple as they go and the practical aspects, decoupling etc, have been left out to make the fundamental operation clearer:

View attachment 95432

Hi there,

I didnt realize we were talking about AC only amplifiers, i assumed that if we tried to design an op amp it would be general purpose which includes DC also.
But does input offset really matter that much for an AC amplifier? Typically these are AC coupled anyway, so if we had 1mv input offset it would only amount to 0.1v output if the gain of that single stage was 100. A coupling capacitor of the right size would do away with that pretty nicely.
I never had to design a really high end amp though, maybe DC coupling is better for the best audio amplifiers, that's not my area really

But for DC stabilization, a chopper stabilized amp is the way to go
I tested one of these for current sensing and low Ohms measurements, and i can say wow, they are really nice. I think the input offset was something like 10uv (10 microvolts). The amplified readings i was getting on the output of the amp set for a gain of 10 was amazing, almost perfect, and almost unbelievable how well it tracked the input.
A little more expensive but well worth it
The more typical unit i used is the LM358 but rarely for audio. If the output stage is biased right it works ok for audio, but there are a lot of other chips that can do better. For simple control circuits though it's hard to beat the low input offset drift and PRICE PRICE PRICE
I think they are some of the lowest priced op amps on the market.
I looked into a higher speed version of this one and it was like $6 USD though just for one 8 pin DIP package (with 2 amplifiers on the one chip). I'd only use that for special circuits. The intended app for this one was a specially designed super fine adjustment linear power supply, able to adjust over a range of only 10mv with the "fine adjustment" pot at any voltage up to about 20vdc. The response of the LM358 isnt really suited for power supplies except at very low gains.

 

[spec]

Ah I have a question.
.... I know many ICs have good sound such as LM3886, STK....., LA..... but their power not really high, almost 45-200w. I want make power amp above 300w so only can use power transistors (tubes too expensive and high power consumption)...

Yeah, the LM3886 is a complete power amp in one pak. It is popular and sounds very good, but perhaps not in the audiophile class I would suggest. And, as you say, it doesn't have enough power to crack the foundations of your house either.

Another approach, which is recommended, is to try one of the relatively new front end audio power amp driver chips from Texas Instruments: LME49810, LME49811, and LME49830. The last is optimised for driving MOSFET output stages.

These chips, with the appropriate output stage, will easily meet your power requirements and some. They also use audiophile design techniques in the chip. By using this approach you will only have to refine the output power stage and of course PSU; still quite a task though.

The Baker clamp mentioned in the data sheets provides soft limiting, one feature of tube amps, and can make an amp sound better when the sudden spikes of power in some types of music drive the power amp into voltage saturation. If I remember correctly, you can disable the Baker clamp if you don't want it.

I haven't heard an amp made from any of the three chips, or had any practical experience of them, but I understand that they sound pretty good.

News release
**broken link removed**
Datasheets:

LME49810
http://www.ti.com/lit/ds/symlink/lme49810.pdf

LME49811
http://www.ti.com/lit/ds/symlink/lme49811.pdf

LME49830
http://www.ti.com/lit/ds/symlink/lme49830.pdf

Application reports:

LME49830
**broken link removed**

PSU
**broken link removed**

 

[audioguru]

An LM358 was designed to be the first LOW POWER dual opamp (that is why it is slow, noisy and has crossover distortion) so it works poorly for audio. We have hearing up to and beyond 20kHz but the LM358 has slew rate trouble with high level frequencies above only 2kHz.

National Semi (now TI) has some excellent class-AB audio power ICs but their output power is limited to about 60W unless they are paralleled and bridged. National Semi (TI) has an LM4702 audio driver IC that produces more than 300W when it drives darlington power transistors. TI has some excellent class-D high power audio power amplifier ICs.

 

[MrAl]

An LM358 was designed to be the first LOW POWER dual opamp (that is why it is slow, noisy and has crossover distortion) so it works poorly for audio. We have hearing up to and beyond 20kHz but the LM358 has slew rate trouble with high level frequencies above only 2kHz.

National Semi (now TI) has some excellent class-AB audio power ICs but their output power is limited to about 60W unless they are paralleled and bridged. National Semi (TI) has an LM4702 audio driver IC that produces more than 300W when it drives darlington power transistors. TI has some excellent class-D high power audio power amplifier ICs.

Hello again ag,

The crossover distortion can be eliminated by biasing the amplifier in the right manner. The crossover distortion is not there anymore because there is no crossover anymore. Thus, it sounds OK in an audio application if done right and the gain not set too high per stage.
This price is always low for these guys that is what makes them attractive for various applications, and they have low input offset drift.

 

[spec]

2N3055-2N2955 or C5200-A1943 or TIP41-TIP42 better to use in audio amplifier?

2N3055/2N2955: Reliable work-horses, cheap and abundant. Very little spec. Runs out of hFE quite soon as IC increases. You never know what is in the can, the spec is so loose (not rip-off).

TIP41/TIP42: Nice delicate little 'power' transistor. Current gain drops off quite quickly. Don't mess with the boys in the TIP range just go for the TIP35C/TIP36C. They are about the same price, but even the TIP35C/TIP36C are delicate and not well speced. TIP3055/TIP2955 more rugged.

C5200/A1943: Now you are talking. Good for 5A maximum in your application when you take into account the SOAR and current gain graphs, so you would need least 4 in parallel. These are probably based on the Toshiba originals. It was a sad day in the audio field when Tosh stopped making audio power transistors.

Here is a recommended procedure for choosing audio power transistors:.

(1) Check if there is a compliment to the transistor. In fact, there is no such thing as a complement; it's an allusion. By definition the two types are made differently using different materials. Normally the NPN version has the better characteristics but not always. With MOSFETs the situation is not so definite. It's not unusual for so called complementary pairs to have radically different specifications.

(2) Google to see if it both NPN and PNP (N type and P) types, are available at a price you are prepared to pay.

(3) Google for any problems with either NPN or PNP versions Ask around.

(4) Check that the manufacturer is continuing to make the device, or is it on 'one-time-buy' notice or 'not recommended for new design' notice or even obsolete.

(5) Does the data sheet say that the transistor is intended for high quality audio. Alternatively is it used in any top end gear or mentioned on audio sites as being good, diyaudio for example. Do any of the text books say it is good?

(6) Check maxim current required. In your case 18A

(7) Check that VCE is it at least 2x the voltage of the power lines of your propose amp plus 10V minimum. In your case, assuming +- 75V rails, you get (75 *2)+10 = 160V

(8) Check the hFE linearity against IC, the flatter the better for two reasons: low distortion and ease of driving especially at 40KHz. Most power transistors suffer from hFE drop-off as IC increase, often severely, same for low currents which is right in the croos-over region on class AB amps. The 2N3055 suffers badly from hFE droop for example, and who knows what happens at low currents.

(9) Check the SOAR graph. To an approximation VC= one supply line/2. IC = max output current/2. In your case VC= 75/2=37.5V and IC =18/2=9A.

Those are the basics more or less in the correct order.

Once you have focused on a complimentary pair of power transistors:

(10) Check the FT: 30MHz or over should be OK.

(11) Check capacitances, the lower the better. MOSFETs have massive capacitances, 1.5 nF for example. And the capacitance changes acording to VC and IC so thy cause distortion and make life difficult in many ways.

(12) Check thermal resistance junction to case and case to heat sink

(13) Do the thermal budget to find out what size heat sink will be required

(14) Have a beer

(15) Start looking at the spec details

This may sound like an awful rigmarole, but once you have done it a few times, it isn't that bad and doesn't take too long either. In fact you should carry out the same procedure for every component you design in. Of course, the parameters to check would be different.

Notes:

(1) Because of the undesirable characteristics of power transistors, it is best to stay with an emitter-follower (source- followwer) configuration.

(2) The hFE and VEB of the complementary pair you actually use need to be matched to obtain low distortion.


After all that a bit of light relief:

**broken link removed**

Data Sheets

2N3055_MJ2955 (2N2955) (NPN_PNP)
https://www.onsemi.com/pub_link/Collateral/2N3055-D.PDF

TIP41 (NPN)
https://www.fairchildsemi.com/datasheets/TI/TIP41C.pdf

TIP42 (PNP)
https://www.fairchildsemi.com/datasheets/TI/TIP42.pdf

2SC5200_FJL4315 (NPN)
https://www.fairchildsemi.com/datasheets/2S/2SC5200.pdf

2SA1943_FJL4215 (PNP)
https://www.fairchildsemi.com/datasheets/2S/2SA1943.pdf

 

[spec]

...National Semi (TI) has an LM4702 audio driver IC that produces more than 300W when it drives darlington power transistors.....

LM4702 I knew of it but just could not remember the number or find it on the TI site. It caused quite a stir when it came out and many power amps are built with it. Thanks to you, I have now down-loaded the data sheet. I used to have it, but it probably got lost in various laptop upheavels.

The LME49810, LME49811, and LME 49830 are later developments of the same general approach, but just one channel instead of two in the chip.

 

[spec]

Nikolai, have you thought about class A.

The best approach, to limit power dissipation, would be to forget about the excess current to handle the complex impedence of the speakers so you would end up with a standing current of 4.5A. Any excess current demanded by the speakers could be provided by class B operation. A standing current of 4.5A would give a dissapation of 75V x 4.5A =337.5W in each half of the output stage, giving a total amplifier dissipation of 675W. Class A has many advantages but the big problem is obviously power dissipation. Be great in the winter but not so great in the summer. Air conditioning would be the thing.

I have done a design like that, but only 70W, and started working on the cooling. Liquid colling, as is used in high power PCs, would probably be the way to go. I ended up designing an oil cooled approach where the power transistors were immerssed in cooling oil that was circulated to a rad with a fan. You won't be surprised to hear that I never built it.

I think this approach would impress your audiophile friends

 

[spec]

Hi there,

I didnt realize we were talking about AC only amplifiers, i assumed that if we tried to design an op amp it would be general purpose which includes DC also.

Hello MrAl,

This is turning into quite a thread- my fault

The subject in this thread is audio amps but that doesn't matter most of what you say is relevant.

As I see it there are two aspects involving the LF response of an audio amplifier. It's generally assumed that, for design purposes, the human ear can go from 20Hz to 20KHz, so you would imagine that an amp that just covered that frequency range would be fine and it probably would be apart from two practical points at the low end (there are separate issues at the high end).

(1) If the amp is 3dB down at 20Hz it involves a phase change which it can be argued is audible.

(2) It is important that the amp keeps control of the speaker at all frequencies especially around the 20Hz area because that is where the speaker is liable to be resonant. Without putting a transducer on the voice coil and using feed-back from the cone, the best that can be done is to present the speaker with a zero impedance and all the current that is required to do that. Hence the 0.1Hz.

In the original transistor amps the speaker was coupled to the amp via a capacitor, typically 1mF (1000uF). This meant that at the low end the speaker was presented with and ever increasing impedance at the very time it needed most damping, not to mention the distortion and phase changes introduced by the capacitor itself. Later the feedback was derived from the speaker side of the capacitor. This sounds like a neat solution but it introduces other problems.

Split supply amps eliminate all this, and they have become the standard now for high end amps and many ordinary audio power amps too. But this meant that the standard current feedback input stage, with all its advantages, could not be used easily so over to long tail pairs. Not only do these give a small offset they also reduce distortion, but at the expense of more noise.

But does input offset really matter that much for an AC amplifier? Typically these are AC coupled anyway, so if we had 1mv input offset it would only amount to 0.1v output if the gain of that single stage was 100. A coupling capacitor of the right size would do away with that pretty nicely.

I never had to design a really high end amp though, maybe DC coupling is better for the best audio amplifiers, that's not my area really

That's right even an offset of +- 100mv only amounts to 80mW which isn't going to bother the average speaker with a power rating of 20W upwards. You do get a small click when you connect the speaker to a live amp though, which can be a bit off-putting to some people.

But for DC stabilization, a chopper stabilized amp is the way to go

I tested one of these for current sensing and low Ohms measurements, and i can say wow, they are really nice. I think the input offset was something like 10uv (10 microvolts). The amplified readings i was getting on the output of the amp set for a gain of 10 was amazing, almost perfect, and almost unbelievable how well it tracked the input.

A little more expensive but well worth it

Chopper stabilised amps are quite something , and as you say the ultimate in low offset. But opamps have improved so much that they can even better 10uV. The OPA182, for example has an input offset of +-5uV and they are not that expensive.

Back in the germanium transistor days I did a chopper amp using a reed relay. By the standard of the day it was quite good but nothing like your 10uV.

The more typical unit i used is the LM358 but rarely for audio. If the output stage is biased right it works ok for audio, but there are a lot of other chips that can do better. For simple control circuits though it's hard to beat the low input offset drift and PRICE PRICE PRICE

I think they are some of the lowest priced op amps on the market.

I looked into a higher speed version of this one and it was like $6 USD though just for one 8 pin DIP package (with 2 amplifiers on the one chip). I'd only use that for special circuits. The intended app for this one was a specially designed super fine adjustment linear power supply, able to adjust over a range of only 10mv with the "fine adjustment" pot at any voltage up to about 20vdc. The response of the LM358 isn't really suited for power supplies except at very low gains.

Yes, the LM358 and its quad counterpart, the LM324. When they were introduced they opened up all sorts of possibilities mainly because of their low cost but also because of their characteristics:

(1) IOS 3mV,

(2) IPI 20nA,

(3) Supply current 700uA.

(4) IP range down to negative rail,

(5) Supply range from 2V to 32V.

(6) Output voltage to within 5 mV of the negative supply rail.

Pretty good especially at the time.

Where they are bad is frequency response; they have pretty much had it by10kHz and I suspect that is one reason they have such high cross-over distortion. Also the output drive current isn't that good 20mA up and 10mA down. Inspite of that the the LM358 is still useful today.

Your power supply is interesting. When you say the LM358 is not good for power supplies was the frequency stability the issue or the low output current capability, probably both I expect.

When you mention cost is that because it was a hobby or were you working on a very cost conscious application. I worked on military projects were cost wasn't relevant, within reason that is. Home projects were quite the reverse!

 

[spec]

An LM358 was designed to be the first LOW POWER dual opamp (that is why it is slow, noisy and has crossover distortion) so it works poorly for audio. We have hearing up to and beyond 20kHz but the LM358 has slew rate trouble with high level frequencies above only 2kHz.

National Semi (now TI) has some excellent class-AB audio power ICs but their output power is limited to about 60W unless they are paralleled and bridged. National Semi (TI) has an LM4702 audio driver IC that produces more than 300W when it drives darlington power transistors. TI has some excellent class-D high power audio power amplifier ICs.

Hi AG

Ditto about the LM358.

I see you mention class D amps. They seem to be the current wave and are also dicused in HiFi terms. What is your view? Do they sound good. I only know them from way back when they were awful. I suppose, in therory, there is no reason why they should sound any worse than a normal A or AB amp.

 

[spec]

Nicolai,

The more I think about your power amp the more it worries me. May I suggest that you lower your sights and go for around 70W rather than 300W. You will be surprised how loud a good 70W amp is and that would be more than enough to drive most speakers to high sound levels. Our gang had a shootout with speakers amps and turntables/cartriges. There were massive transmission line speakers, class a amps, amps with massive power output and so on, all home made by some very good engineers. In the middle of this was a little 40W commercial amp with speakers about 2 feet tall on stands. Guess which one sounded the best- the 40W commercial amp. Guess which one sounded the loudest - same again. We were amazed. Power is not everything. This lower power amp had a +-20A current capability and that is the secret it seems. Later I investigated and the output transistors were specially selected versions of stock transistors and there was two in parallel- just for 40W! Think about this too; because of the way that the ear works 300W will only sound twice as loud as 30W