16 Channel Fairchildish Summing Amp (Added Documents)

[ForthMonkey]

You can download gerber files from my website.

https://analogobsession.wordpress.com/hardware/fairchild-mnnl/

Decided to design 16 channel unbalanced summing amp with Fairchild summing amp circuit.Changed transistors.Used BC550 and BC560.600 ohm input.0dB to 25dB gain.Easy to upgrade with trafos.

Spiced,breadboarded and working good.

Here is first schematic.Everything will be same board.Just internal wiring for XLRs or jacks.

Orginal

http://www.technicalaudio.com/pdf/Fairchild/Fairchild_692MNNL.pdf

My schematic

https://dl.dropboxusercontent.com/u/91809016/FAIR.pdf

 

[ruffrecords]

There seems to be no dc stabilisation of the output stage. Is there an error in the schematic? Also, is this meant to be a passive or VE summer?

Cheers

Ian

 

[ForthMonkey]

There seems to be no c stabilisation of the output stage. Is there an error in the schematic? Also, is this meant to be a passive or VE summer?

Cheers

Ian

No error in schematic.I converted Ltspice circuit directly to Eagle.

Sorry but i'm not good with English terms about electronic.Can you tell me what is c stabilisation and VE?

 

[leadbreath]

Virtual Earth?

 

[ForthMonkey]

Virtual Earth?

I believe it's virtual earth and c stabilisation is current stabilisation.But i'm not sure.

If we talking about virtual earth,yes,it's virtual earth summer.

 

[ruffrecords]

There seems to be no c stabilisation of the output stage. Is there an error in the schematic? Also, is this meant to be a passive or VE summer?

Cheers

Ian

No error in schematic.I converted Ltspice circuit directly to Eagle.

Sorry but i'm not good with English terms about electronic.Can you tell me what is c stabilisation and VE?

No problem. I meant to say dc stabilisation (I have corrected the original post)). You probably want to ensure the dc voltage at the junction of the two output  transistors is half the supply voltage. I am not sure how this circuit achieves this. Do you have a link to the original Fairchild circuit?

VE stands for virtual earth.

Cheers

Ian

 

[ForthMonkey]

There seems to be no c stabilisation of the output stage. Is there an error in the schematic? Also, is this meant to be a passive or VE summer?

Cheers

Ian

No error in schematic.I converted Ltspice circuit directly to Eagle.

Sorry but i'm not good with English terms about electronic.Can you tell me what is c stabilisation and VE?

No problem. I meant to say dc stabilisation (I have corrected the original post)). You probably want to ensure the dc voltage at the junction of the two output  transistors is half the supply voltage. I am not sure how this circuit achieves this. Do you have a link to the original Fairchild circuit?

VE stands for virtual earth.

Cheers

Ian

Thanks Ian.

I added orginal circuit to first post.Can you check it for dc stabilisation?If it's possible,how can i do it?

There is 7.5V at the junction.

 

[ruffrecords]

Now it makes sense. Looking at the original you can see that R35 needs to be selected . What this means is that the dc operating point depends on  the parameters of the actual transistors and you need to adjust R35 when you build it to get the  junction close to half the supply volts. I notice there are no instructions on how to select R35 in the original document.

In the introduction it says it is a passive mixer.

Cheers

Ian

 

[ForthMonkey]

Now it makes sense. Looking at the original you can see that R35 needs to be selected . What this means is that the dc operating point depends on  the parameters of the actual transistors and you need to adjust R35 when you build it to get the  junction close to half the supply volts. I notice there are no instructions on how to select R35 in the original document.

In the introduction it says it is a passive mixer.

Cheers

Ian

I tried this circuit on breadboard.Working good.Used 3.3M and i didn't use trimmer to get unity gain and it looks like ok.

So junction should be 12V?

Tried with spice and used 1.5meg for R35 and getting around 12-13V.

 

[Jano1954]

@ForthMonkey

Is there any particular reason why you used a 2,2 uf capacitor in all the input connections?
Would it be a little more economical to tie all the 150 kohms resistors together and connect that junction via a  single 2,2 uf capacitor with the basis of the first transistor? Like in the Original Fairchild schematics?  It saves place and gives the possibility to  design the PBC a lot smaller.
Greetz

 

[ForthMonkey]

@ForthMonkey

Is there any particular reason why you used a 2,2 uf capacitor in all the input connections?
Would it be a little more economical to tie all the 150 kohms resistors together and connect that junction via a  single 2,2 uf capacitor with the basis of the first transistor? Like in the Original Fairchild schematics?  It saves place and gives the possibility to  design the PBC a lot smaller.
Greetz

Ohh,i just saw that!Why did i put 2,2u for per channel i have no idea.It was really late when i draw it.Because of this i believe

Circuit changed.

 

[ruffrecords]

So junction should be 12V?

Difficult to say. With a 24V supply then the dc level of the output need to sit at 12V so you can get maximum output swing. Whether they did this is anyone's guess. Unfortunately the figures do not stack up too well. If the output us at 12V then the 3K9 R37 must have about 12V across it so it must pass about 3mA. This 3mA flows through R36 (560) ohms so the voltage across it must be about 1.7V  Now, even if R33 was connected to ground it would only pass 133uA so the collector current of Q1 is probably 100uA or less. This flows through R34 (150R) so the voltage drop across R34 is negligible. Assuming 0.7V for the be junction of Q1, this means there must be 1V across the 3M3 resistor so about 0.3uA flows through it into the base of Q1. As long as the beta of Q1 is a few hundred we get the correct collector current. As you can see this is quite sensitive to the beta of Q1 and the value of the 3M3 resistor. Back in the day it would probably have been even more marginal since transistor beta were generally lower.

Like all engineering it is a compromise and in this case you need to select a resistor to get the right output stage current and voltage.

Cheers

Ian

 

[ForthMonkey]

So junction should be 12V?

Difficult to say. With a 24V supply then the dc level of the output need to sit at 12V so you can get maximum output swing. Whether they did this is anyone's guess. Unfortunately the figures do not stack up too well. If the output us at 12V then the 3K9 R37 must have about 12V across it so it must pass about 3mA. This 3mA flows through R36 (560) ohms so the voltage across it must be about 1.7V  Now, even if R33 was connected to ground it would only pass 133uA so the collector current of Q1 is probably 100uA or less. This flows through R34 (150R) so the voltage drop across R34 is negligible. Assuming 0.7V for the be junction of Q1, this means there must be 1V across the 3M3 resistor so about 0.3uA flows through it into the base of Q1. As long as the beta of Q1 is a few hundred we get the correct collector current. As you can see this is quite sensitive to the beta of Q1 and the value of the 3M3 resistor. Back in the day it would probably have been even more marginal since transistor beta were generally lower.

Like all engineering it is a compromise and in this case you need to select a resistor to get the right output stage current and voltage.

Cheers

Ian

Thanks for explanation and analysis,Ian!

I will check what can i do.It looks hard to select resistor but i will figure it out.

As i said,with breadboard everything was ok.Used 3.3M resistor.But when i make PCB i will try other values.I hope it work as it should be.

 

[Monte McGuire]

I haven't done a noise analysis of the amplifier, but one issue that might be a problem is the somewhat large (150KΩ) input summing resistor value. This resistor will generate Johnson noise (from the resistor itself) as well as develop noise voltage from the input noise current of the amplifier, acting through the 16 input resistors essentially in parallel. If this value is halved, the Johnson noise will go down by 3dB, and the induced current noise will go down by 6dB. The amplifier seems to be run with a very small first stage current, so maybe the current noise issue is irrelevant, but still... 150KΩ seems large.

Since the inputs are loaded with 680Ω, is there a reason to use such a large summing resistor value? Again, I have no idea how noisy the summing amplifier is, and whether the noise added by the 150KΩ series input resistor will be significant, but since you're asking the input devices to drive 680Ω, they surely could drive a far smaller summing resistor value than 150KΩ.

Just thinking about what could go wrong or be improved...

 

[ruffrecords]

The noise from a 150K resistor is about -100dBu which does not sound too good. However, this is a passive mixer so the noise resistance is the paralle sum of the 16 x 150K resistors which is 935 ohms and the Johnson noise from this is  below -110dBu. You will be lucky if any of the sources feeding the summing amp will have a noise level this low and as a rule, Johnson noise in passive mix buses is not an issue. If this still worries you, change the resistor values to 15K and the Johnson noise will drop below -120dBu.

Input current noise is a whole different ball game but in general the lower the resistance the better.

Cheers

Ian