Hi,
I have built a G SSL stereo compressor to use in my own studio.
Out of curiosity I have looked at the schematic and to me the output stage looks odd to me.
Im sure there is a real simple explanation which i just cant see.
Schematic:
http://www.gyraf.dk/gy_pd/ssl/ssl_sch.gif
In the upper right area the signal is feed from the DBX202 in to a 5532 dual op amp configuration acting as a balanced output stage.
Looks like the first inverting op-amp output is the (+) out and its feeding via a 10k resistor a second op-amp inverting output which creates the inverted (-) out signal.
Why is the negative feedback resistor value 15k on the first op-amp and 10k on the second.
What is the purpose of the 10k resistor between them?
Why is the op-amps +input to ground 470r and then 10k on the other op-amp.
Wouldn't it be better to match the feedback resistors to match the gain?
I have built a G SSL stereo compressor to use in my own studio.
Out of curiosity I have looked at the schematic and to me the output stage looks odd to me.
Im sure there is a real simple explanation which i just cant see.
Schematic:
http://www.gyraf.dk/gy_pd/ssl/ssl_sch.gif
In the upper right area the signal is feed from the DBX202 in to a 5532 dual op amp configuration acting as a balanced output stage.
Looks like the first inverting op-amp output is the (+) out and its feeding via a 10k resistor a second op-amp inverting output which creates the inverted (-) out signal.
Why is the negative feedback resistor value 15k on the first op-amp and 10k on the second.
What is the purpose of the 10k resistor between them?
Why is the op-amps +input to ground 470r and then 10k on the other op-amp.
Wouldn't it be better to match the feedback resistors to match the gain?
The first opamp is providing gain form whatever comes before it, so it has whatever feedback resistor will provide the required gain. The second opamp is a unity gain inverter, so it has to have the same value feedback resistor as input resistor. There is no reason why the two stages should have the same resistor values.
That is the input resistor for the inverter. Google 'opamp inverting configuration' to learn more.
It is usually best to have the same resistance at both input pins of an opamp, in order to minimise offset voltages caused by input bias current. I assume the first opamp sees a low impedance on the - input pin. The second opamp sees 10k on its - pin. However, there is also the 10k feedback resistor, so maybe the ground resistor here should be 4.7k?
No reason to do that. Gain is not set by resistors, but by ratios of resistors.
The first half of the 5532 gives an overall gain of the product of the output impedance (DBX202) and 15k that produces the + output swing and the rather crude inverter uses two 10k resistors to give a nominal gain of two for the - output. The 470R versus 10k makes little difference as the standard is 10k in that position but the lower value is probably because they had a lot of them to use up.
If you look at quality output stages, the 5532 is balanced with an overall gain of one. See attached;
If you look at quality output stages, the 5532 is balanced with an overall gain of one. See attached;
the two opamps in the top right give a differential output by inverting the phase of the signal and feeding the two pahse and inverted as the differential output.
The bits that makes it balanced are the 100r and the macthed capacitors between the output terminals and the opamp outputs.
Balanced impedance connection is quite different from differential voltage. They use quite different topologies to achieve their own ends.
The bits that makes it balanced are the 100r and the macthed capacitors between the output terminals and the opamp outputs.
Balanced impedance connection is quite different from differential voltage. They use quite different topologies to achieve their own ends.
John,
you are making the same mistake as Silver, concentrating on the "differential voltage section" instead of the balanced impedanceconnection.
The 47r at the two outputs is what makes your .doc into a balanced impedance connection at the output.
The clever circuitry around the two opamps is developing a differential voltage signal, nothing to do with "balanced impedance connection".
It is a single ended to differential convertor stage that happens to be balanced because the two 47r are very accurately matched. Or they should be, but I don't see any mention of that.
Use better than 0.1% for the two resistor and use better than 1% for the two sets of capacitors if the output needs to be AC coupled.
Last edited:
Wow great! Thank you all.
How could i not se that the second op-amp is a 1:1 gain to match the first one whatever gain it has, it makes perfect sense.
I guess i was thinking both op-amp pairs was fed by the same signal, which is not the case.
Still, the 470R makes no sense to me because its way lower than the 10k.
Why is there no 10k resistor in to the first op-amp after the dbx202?
Many questions, i just started to read The Art Of Electronics to expand mo knowledge.
How could i not se that the second op-amp is a 1:1 gain to match the first one whatever gain it has, it makes perfect sense.
I guess i was thinking both op-amp pairs was fed by the same signal, which is not the case.
Still, the 470R makes no sense to me because its way lower than the 10k.
Why is there no 10k resistor in to the first op-amp after the dbx202?
Many questions, i just started to read The Art Of Electronics to expand mo knowledge.
Last edited:
the second from right has 470r on the +IN pin and 10k||15k on the -IN pin.
The bias voltages across these will be different and lead to an excess of output offset.
But increasing the 470r will increase the noise of that stage so it may be that the Designer adjusted this to as low as the circuit would allow and keep reasonable noise.
Note that the output is AC coupled so an excess of output offset only affects the next stage and maybe fed back to the circuit feeding into the -IN (dbx202).
Look at the right most.
It has 10k compared to 10k||10k. That does seem more odd. High noise and higher offset. 5K might be better.
OOPs,
just noticed that he has used Polar electros on the AC coupling.
He may be deliberatly using output offset to ensure the caps never experience a reverse voltage.
I would have used a Bi-Polar electro there. (see Rane for an AC coupling using a 220uF Panasonic there).
The bias voltages across these will be different and lead to an excess of output offset.
But increasing the 470r will increase the noise of that stage so it may be that the Designer adjusted this to as low as the circuit would allow and keep reasonable noise.
Note that the output is AC coupled so an excess of output offset only affects the next stage and maybe fed back to the circuit feeding into the -IN (dbx202).
Look at the right most.
It has 10k compared to 10k||10k. That does seem more odd. High noise and higher offset. 5K might be better.
OOPs,
just noticed that he has used Polar electros on the AC coupling.
He may be deliberatly using output offset to ensure the caps never experience a reverse voltage.
I would have used a Bi-Polar electro there. (see Rane for an AC coupling using a 220uF Panasonic there).
Last edited:
I am going to guess it is because the chip produces a current output, not a voltage. This current is then fed into the virtual earth at the opamp input.silversweden said:
> first half of the 5532 gives an overall gain of the product of the output impedance (DBX202) and 15k
The '202 is a current-difference "variable gain" unit.
The real gain at zero unbalance is the 27K to the 15K. This is "about 1/2" because the output stage runs at gain 1+1 or 2, and unity-gain was desired at zero unbalance.
Yeah, 15K:27K is 0.55, about 10% high. "Unity" a bit high of 1.000 is always "better". It also compensates interface loss at the many-K input and the 200r output.
The "input current compensation" resistors are not real "wrong". The 470 is "zero" and perhaps just a jumper in-case something was needed there. '5532 input bias in the 15K produces an offset. The inverter would produce the same offset but the other way. The "wrong" 10K actually reduces this offset. Exact correction is not needed because the output is cap-coupled.
FWIW, this "crude" plan has been a favored limiter in mastering studios replacing the revered SSL mix-console limiter.
The '202 is a current-difference "variable gain" unit.
The real gain at zero unbalance is the 27K to the 15K. This is "about 1/2" because the output stage runs at gain 1+1 or 2, and unity-gain was desired at zero unbalance.
Yeah, 15K:27K is 0.55, about 10% high. "Unity" a bit high of 1.000 is always "better". It also compensates interface loss at the many-K input and the 200r output.
The "input current compensation" resistors are not real "wrong". The 470 is "zero" and perhaps just a jumper in-case something was needed there. '5532 input bias in the 15K produces an offset. The inverter would produce the same offset but the other way. The "wrong" 10K actually reduces this offset. Exact correction is not needed because the output is cap-coupled.
FWIW, this "crude" plan has been a favored limiter in mastering studios replacing the revered SSL mix-console limiter.
- Status
- Not open for further replies.