Patchwork Reloaded: Circuit Optimization and Board Layout.

[MJL21193]

Originally posted by ferencz

But, as on every side of life there is always room for improvement!

The only mod I could recommend now is to change the drivers to MJE15032/33. The reason is that in practice I found that related the MJE15030/31 pair there is always a huge Hfe difference between PNP and NPN device.

Thank you again,
The drivers I have were supposed to be matched when I bought them 3 years ago. I have ten sets of these.
I'm actually thinking of using the 2SA1011/2SC2344 as the drivers and changing out the VAS to the 2SA1381. These are higher gain, higher frequency devices.

Any thoughts on the 2STC5200 / 2STA1943 as outputs? These are the ST version of the 2SA5200/1943 but with higher gain. These are faster than the MJL21193/94 and about half the price.

Originally posted by megajocke



I'd probably use zeners anyway as this noise will have a pretty hard time getting into the circuit and be amplified as the differential stage has very good rejection when voltage changes the same on both collectors. Another option is just using a resistive divider with a bypass cap. I'd probably like to have a little more voltage than possible with LEDs, at least 12-15V or so.

I'd probably reduce the output stage base resistors to 2 ohms or something similar. It might even be possible to remove them completely like you say but I believe the only thing gained is a lower component count.

As much as 15V? I'll try it in the sim. Thanks.

What could be the cause of this overshoot on the squarewave? Also quite a slope on the downward cycle of the wave. Here's a picture of the amp with a 20kHz squarewave.

[MJL21193]

15X magnification.

[MJL21193]

Originally posted by megajocke

Another option is just using a resistive divider with a bypass cap. I'd probably like to have a little more voltage than possible with LEDs, at least 12-15V or so.

I did that with good results. Better THD performance also (yeah, I know that means nothing).
Updated the schematic to show that change. This is what I like - a change that simplifies, improves and lowers the component count.

[ferencz]

MJL21193/94 > 2SC5200/2SA1943
Then you'll need more pairs to stay on the safe side when amp is loaded with 4R speaker (SOA!). Three, or even four pairs will needed!

I'm actually thinking of using the 2SA1011/2SC2344 as the drivers and changing out the VAS to the 2SA1381.

Very good idea, strongly recommended!

What could be the cause of this overshoot on the squarewave? Also quite a slope on the downward cycle of the wave. Here's a picture of the amp with a 20kHz squarewave.

I'd "play" with the compensation circuit (here it's a lag-Miller-capacitor, namely C5). Try 47pF and then 150pF. Maybe some lead compensation (10pF from the VAS's output to Q2's base) helps. Inserting the lead capacitance will increase the hi-freq THD a bit!
Or maybe C9 has wrong value? Try 100nF!

[Nico Ras]

Hi John,

Experiment with a 100 - 470 uF cap between the emitters of Q1 and Q2 and have a look see what it does with your already minuscule distortion. The results may be interesting in future designs.

Kindest regards

Nico

[megajocke]

Originally posted by MJL21193



Thank you again,
The drivers I have were supposed to be matched when I bought them 3 years ago. I have ten sets of these.
I'm actually thinking of using the 2SA1011/2SC2344 as the drivers and changing out the VAS to the 2SA1381. These are higher gain, higher frequency devices.

Any thoughts on the 2STC5200 / 2STA1943 as outputs? These are the ST version of the 2SA5200/1943 but with higher gain. These are faster than the MJL21193/94 and about half the price.




As much as 15V? I'll try it in the sim. Thanks.

What could be the cause of this overshoot on the squarewave? Also quite a slope on the downward cycle of the wave. Here's a picture of the amp with a 20kHz squarewave.

Assymetric slew rate suggests something is wrong with the current mirror. Try lowering R2, R3 - they drop quite a bit of voltage as it is now and probably doesn't leave enough headroom for the transistors in the mirror to work correctly.

What is the rail voltage?

Originally posted by ferencz
MJL21193/94 > 2SC5200/2SA1943
Then you'll need more pairs to stay on the safe side when amp is loaded with 4R speaker (SOA!). Three, or even four pairs will needed!

Depends on rail voltage...

You can go up to 60V rails or so with two pairs and 4 ohm loads. Three pairs if you want to keep down Ice where they are more linear.

Fairchild has a version to both under the name 2sc5200/2sa1943 and FJL4215/FJL4315. These are cheap from digikey. Actually the originals are cheaper than the ST copies even...

4 pairs is enough for 80V (sagging) rails and 2 ohms if you build an amplifier without insulation between transistors and heatsink.

[megajocke]

Originally posted by Nico Ras
Hi John,

Experiment with a 100 - 470 uF cap between the emitters of Q1 and Q2 and have a look see what it does with your already minuscule distortion. The results may be interesting in future designs.

Kindest regards

Nico

That would destabilize the amp if it's compensation is optimal.

I believe it is better to have emitter resistors there linearizing the input stage and use a high standing current to get the same transconductance but much more linear.

[Nico Ras]

Originally posted by megajocke


That would destabilize the amp if it's compensation is optimal.

I believe it is better to have emitter resistors there linearizing the input stage and use a high standing current to get the same transconductance but much more linear.

It actually nulls the emitters at ac and reduces distortion. A trick used by Dan Augustino in the mid seventies and I believe still use it today. I have used it in many of my designs both private and commercial with quite good results.

[MJL21193]

Originally posted by ferencz
MJL21193/94 > 2SC5200/2SA1943
Then you'll need more pairs to stay on the safe side when amp is loaded with 4R speaker (SOA!). Three, or even four pairs will needed!

I'd "play" with the compensation circuit (here it's a lag-Miller-capacitor, namely C5).

Hi,
Tried the miller cap - no go.

Yes, lower power devices, but the load for this amp will be 8 ohms.

Originally posted by Nico Ras
Hi John,

Experiment with a 100 - 470 uF cap between the emitters of Q1 and Q2 and have a look see what it does with your already minuscule distortion. The results may be interesting in future designs.

Thanks Nico,
Glad you are here. I will try this as I have seen it elsewhere too (KSA-50?). Your advice, as usual, is always valuable.
Do you have any other ideas That could improve this project?

Originally posted by megajocke


Assymetric slew rate suggests something is wrong with the current mirror. Try lowering R2, R3 - they drop quite a bit of voltage as it is now and probably doesn't leave enough headroom for the transistors in the mirror to work correctly.

BINGO! That's the beauty of this forum - there is someone who knows EXACTLY what your problem is. Thank you very much Mega, I lowered the R's to 110 ohms and all is well. Perfectly symmetrical squarewave at 20kHz. I'll post some pics soon.

Originally posted by megajocke

What is the rail voltage?



Depends on rail voltage...

You can go up to 60V rails or so with two pairs and 4 ohm loads. Three pairs if you want to keep down Ice where they are more linear.

Fairchild has a version to both under the name 2sc5200/2sa1943 and FJL4215/FJL4315. These are cheap from digikey. Actually the originals are cheaper than the ST copies even...

4 pairs is enough for 80V (sagging) rails and 2 ohms if you build an amplifier without insulation between transistors and heatsink.

The rail voltage for this is 56 - drooping to 54VDC under full load with a fairly stiff power supply (see pic below). 23500uF/rail but this PS will be shared by two other amps. There will be two PS in total, one 750VA transformer and 6 amp modules.

I saw the ST devices are higher gain, that's why I choose them. If this is no big deal, I'll use the Fairchild ones (actually just checked and they have the higher gain also)

[megajocke]

Originally posted by Nico Ras


It actually nulls the emitters at ac and reduces distortion. A trick used by Dan Augustino in the mid seventies and I believe still use it today. I have used it in many of my designs both private and commercial with quite good results.

If you design around it I guess you could make it work in an excellent way. In a symmetric circuit with two LTP:s (one per rail) it would be useful as it lets you use large emitter resistors for good DC balance in the input stage.

Though, if you take an optimized design which was designed around an LTP with emitter resistors it will increase the open loop gain bandwith product possibly making the amplifier unstable.

Originally posted by MJL21193

BINGO! That's the beauty of this forum - there is someone who knows EXACTLY what your problem is. Thank you very much Mega, I lowered the R's to 110 ohms and all is well. Perfectly symmetrical squarewave at 20kHz. I'll post some pics soon.

Glad that I could help

The rail voltage for this is 56 - drooping to 54VDC under full load with a fairly stiff power supply (see pic below). 23500uF/rail but this PS will be shared by two other amps. There will be two PS in total, one 750VA transformer and 6 amp modules.

I saw the ST devices are higher gain, that's why I choose them. If this is no big deal, I'll use the Fairchild ones (actually just checked and they have the higher gain also)

Then you will have good margin with 2 pairs. That is a pretty stiff supply, looks nice too