BC550 vs KSC1845, BC560 vs KSA992 and others.

[Ranchu32]

OS, I've been using the KSA1015 as LTP on some of my project. Do you think these compare favourably with the BC560 for that purpose?

 

[Ranchu32]

Alan, you will be fine with those outputs at those voltages. Have a look at the SOA curves in the data sheets and you will see what Andrew means.

Another low cost option is the onsemi NJW3281 and compliment. Personally I think 5pr for 40V rails is a massive overkill unless you are driving a 2R load.

 

[AndrewT]

+-42Vdc to +-44Vdc @ 0.2A for each of 5 pairs is a THERMAL concern !
And after you have dissipated this ~86W, plus the drivers, you then have to de-rate the outputs to take account of their operating Tc.
You could be down ~3Apk @ 43Vce for a one shot 100ms transient.
5pairs gives a peak 100ms transient of 15Apk !

 

[Ranchu32]

No kidding.

Another consideration is current hogging. 4-5pr is the point where I want to increase the Re from 0R22 to 0R33 for better current sharing.

You could get a comparable "Class A area" from 3pr outputs optimally biased at 0R22 Re than 5pr at 0R33 each.

 

[AndrewT]

............. Personally I think 5pr for 40V rails is a massive overkill unless you are driving a 2R load.

150W into 4ohms is probably a target for a 5pair stage running from a 30-0-30Vac transformer.
a 4r0 dummy load will draw 8.66Apk when trying to reach that 150W target.

5pairs running hot will not be able to double that target into a reactive 4ohms speaker.
I reckon one should aim for three times the dummy load peak current, i.e. 26Apk for a 150W into 4ohms capable amplifier.
That would require each output device to pass a transient of ~5.2Apk
That amounts to a de-rating factor of only 0.86 @ 43Vce. i.e. the devices must not exceed a Tc of 42°C. The heatsink must be even cooler.
The saving grace is that a 3times peak current transient will have a duration of much less than 100ms.
But is that 50ms, or 15ms, or 1ms?

It is not overkill. It requires DESIGN input to ensure reliability.

 

[bimo]

2sa1943/c5200 have very poor SOA compared to similar 150W devices.
They do NOT suit amplifiers with higher supply rail voltage.
I would limit them to <<+-60Vdc preferably <+-50Vdc

Not here on my lovely country. Many people use 2sa1943/c5200 with +-90VDC 😎 Of course they use many of them in parallel. 2sa1943/c5200 is cheap (fake) and we love cheap components 😀
I design an amplifier use them with +-77VDC and tested by other (I have few builder for my amp just like OS 😀 )

NJW0281 have slightly better SOA but difficult to find here.

 

[Ranchu32]

Andrew, I think your "three times the dummy load peak current" is overly pessimistic test, particularly for a 4ohm speaker. You are saying that the speaker has an impedance dip equivalent to 1.3ohm! I think that rule of thumb is a pessimistic test for an 8ohm rated speaker too, although 2.6ohm equivalent dip is more likely, I suspect. I will consult a couple of speaker builders I know and solicit their thoughts.

As far as I am aware, the industry standard is that a 8 ohm rated amplifier needs to be able to drive an 8 ohm dummy load at full rated power for an hour & a 4 ohm rated amplifier needs to be able to drive a 4 ohm dummy load to full power for an hour.

 

[Ranchu32]

Not here on my lovely country. Many people use 2sa1943/c5200 with +-90VDC 😎 Of course they use many of them in parallel. 2sa1943/c5200 is cheap (fake) and we love cheap components 😀
I design an amplifier use them with +-77VDC and tested by other (I have few builder for my amp just like OS 😀 )

NJW0281 have slightly better SOA but difficult to find here.

You could always cascade them if there is no readily available alternative. That's what I would do.

 

[bimo]

You could always cascade them if there is no readily available alternative. That's what I would do.

Class G? Hmmmm... may be next time.

 

[Ranchu32]

Not class G... something simpler.

Have a look at the Super Leach schematic: http://users.ece.gatech.edu/mleach/superamp/circuit.pdf

 

[AndrewT]

Not class G... something simpler.

Have a look at the Super Leach schematic: http://users.ece.gatech.edu/mleach/superamp/circuit.pdf

This was also called the "double barreled". The sch may not be identical, but Leach does give a design description.

 

[franzm]

I don't follow what is baxandall.

Feedback amplifier circuit sometimes called Baxandall super pair

 

[Alan0354]

Alan, you will be fine with those outputs at those voltages. Have a look at the SOA curves in the data sheets and you will see what Andrew means.

Another low cost option is the onsemi NJW3281 and compliment. Personally I think 5pr for 40V rails is a massive overkill unless you are driving a 2R load.

I have the data sheet of 2SC5200 and NJW3281 and I compare the SOA graph. I don't see what you guys said.

2SC5200 SOA is 3.5A at 40V. The graph of the NJW3281 is messed up, the grid of the current scale is off by just looking at it. But if you look at the 40V point and look at the current between 1A and 10A, it is about 1/2 way, very similar to 2SC5200.

Bare in mind, 2SC is 150W vs NJW is 200W. Of cause you expect NJW can take a little higher power. But I don't see you guys said look at the graph and see what you mean.

Please explain.

Also, I am also looking at MJW3281/1302 which is pretty much the same as NJW in different package.

Thanks

 

[Alan0354]

+-42Vdc to +-44Vdc @ 0.2A for each of 5 pairs is a THERMAL concern !
And after you have dissipated this ~86W, plus the drivers, you then have to de-rate the outputs to take account of their operating Tc.
You could be down ~3Apk @ 43Vce for a one shot 100ms transient.
5pairs gives a peak 100ms transient of 15Apk !

I did not mention I am buying a chassis with 11" X 5" heat sink for each side. It's a big chassis that is going to cost $270 with shipping. I want the first few watts in class A. I need 200mA per pair and use 0.15ohm emitter resistor to come close to Oliver's optimization.

That's part of the reason why I ask a lot and still do very little. This amp is going to cost me $1000 when it's all said and done. The Chassis, PT and pcb is $500. I already tally up $500 in parts from digikey!!!

 

[AndrewT]

Re=0r15 will probably require around 22mVre for optimal ClassAB bias. That equates to ~150mA per device.

Using your lowish supply rails you should be able to use R.Cordell's formula to confirm that Re=0r1 is usable.
That is likely to end up with bias ~ 180mA to 200mA for optimal ClassAB.

5pr @ 200mA gives ClassA 2Apk. That equates to ~8W into 4r0 for your 150W into 4ohms capable amplifier. i.e. it stays in ClassA for all powers upto 13dB below maximum.

Have you done a heatsink calculation to predict the Tc of your 5pr stage?

 

[Alan0354]

Re=0r15 will probably require around 22mVre for optimal ClassAB bias. That equates to ~150mA per device.

Using your lowish supply rails you should be able to use R.Cordell's formula to confirm that Re=0r1 is usable.
That is likely to end up with bias ~ 180mA to 200mA for optimal ClassAB.

5pr @ 200mA gives ClassA 2Apk. That equates to ~8W into 4r0 for your 150W into 4ohms capable amplifier. i.e. it stays in ClassA for all powers upto 13dB below maximum.

Have you done a heatsink calculation to predict the Tc of your 5pr stage?

Mr. Cordell said the R_E should not go below 0.13ohm no matter what. So I violate a little and make it 0.15ohm.

No I have not done the thermal calculation. I up the bias current of my Acurus with 4 pairs to 100mA each pair. It makes a big difference in sound. It has +/-80V rail. It's not that hot. I can put my hand on the heat sink for as long as I want. It's been running for a few months already of 6 hours a day 7 days a week. It only has 11" X 3.xx" heat sink. The fins are shorter than the one I am going to buy. Seems like the new one is going to do a whole lot better thermally.

Thanks

 

[Alan0354]

Hi guys, thanks for all the inputs. This has been very helpful. I still yet to read more about the Baxandall VAS. Just too much to learn, I just learn how to plot FFT in LTSpice to plot distortion. Still have to simulate my designs. I still want to read more on the CFA to see what's the big fuzz about.

I have no plan to build another SS amp after this. The next one will be a tube amp. So I am being very careful in learning and asking questions, doing simulations before committing big money into it. I am going to order the chassis soon, as this govern the dimensions of the pcb and where to drill holes on the heat sink.........More important, I want to buy one piece at a time so my CFO and big boss ( say wife) doesn't get shocked all at one time!!!😀😱 Also, Thanks and no thanks to Andrew, I just ordered the book by D Self.😀😛 It's going to take me time to read.

I have another important question:

I am planning to use 3EF OPS. I already finished the layout of the OPS on a 10.5" X 2" pcb with 3EF but only in 3 NPN on top and 3PNP at the bottom. Before I sent it out, I want to run by you guys whether I should look into using one PNP on the top and one NPN at the bottom. With 3NPN/3PNP, The Vbe multiplier has to do 6Vbe where as using an opposite transistor on each side only need 2Vbe.

Please advice.

Thanks

Alan

 

[vzaichenko]

Hi Alan, this is completely unclear (at least to me 🙂):

I am planning to use 3EF OPS. I already finished the layout of the OPS on a 10.5" X 2" pcb with 3EF but only in 3 NPN on top and 3PNP at the bottom. Before I sent it out, I want to run by you guys whether I should look into using one PNP on the top and one NPN at the bottom. With 3NPN/3PNP, The Vbe multiplier has to do 6Vbe where as using an opposite transistor on each side only need 2Vbe.

Can you please drop a sketch (hand-drawing or something)?

Cheers,
Valery

 

[Alan0354]

I can do better!!! Attached is the latest schematic that I already have the board layout and gerber generated.

If I am going to redo the board, I likely will remove all the error correction stuffs as I have no plan to try MOSFET. Also, I want to see whether I can use the space to put the SS relay for speaker protection.

Please comment, this is a lot more critical than the IPS/VAS board. the components are a lot more expensive. I want to do it one time through. IPS/VAS board is cheap and components are cheap. I am planning to fab 3 different versions. One I want to try is Apexaudio SR200 design with enhancement. I still want to read the CFA design also.

Thanks

 

[vzaichenko]

OK, I probably see now 🙂
Your question is - whether you use EF3 (as it is now) with 6Vbe bias, or you use a CFP (Sziklai) arrangement, having just 2Vbe bias in this case, right?