Preamp Design

[Jbliss]

Thanks!!

The Values I roughly Came up with for an audio taper where
R1 1
R2 4.3
R3 9
R4 10.7
R5 14.4
R6 23.5
R7 36.8
R8 57.9
R9 93.5
R10 147
R11 232
R12 370
Total = 1000.1
(approx. depending on available Resistors)
Would these values work ?

Well, theoretically they would produce gains up to 80 dB.  Are you sure you want that much gain in one stage?  Wiring the chip out to a 1 Ohm resistor for 80 dB gain will make an opportunity for the injection of noise into the signal via the wires.  I checked only the first 3 values (starting with 1) and found that they were not consistently spaced.  I think they were at 80dB, 75.5 dB, 57 dB ... or so.  Gotta get to bed now.  What Gains are you shooting for?  I'll take another look at it tomorrow.

ahhh ok, I'm looking for approx 60db of gain
Thanks you so much !
after further research I think my best option would be a 10k stepped attenuator with a series 10 Ohm resistor to set the maximum gain at 60db and allow the minimum gain set at 6db

attached is the circuit

[The Soulman]

Correct.

You want to look also at the compatible SSM2017 datasheet, it also has some info on pcb lay-out.
Or a commercial design around a SSM2017 can be downloaded here: (all Crest century series consoles use ssm2017 pre-amps):

 http://peaveycommercialaudio.com/schematics.cfm

[Jbliss]

So after some calculating if the lowest value of the attenuator (gain switch) is 10ohm I have 60db of gain. Going down in 3db steps gets me down to 27db for 12 positions. What I really need is a stepped attenuator with the lowest value being 10 Ohm and highest values being around 640 ohm (approx.) for a minimum gain of 27db (approx.)

So max gain 60db
Min Gain approx. 27db

just need to work out resistor values for each 3db Drop

[basinstreetdesign]

So after some calculating if the lowest value of the attenuator (gain switch) is 10ohm I have 60db of gain. Going down in 3db steps gets me down to 27db for 12 positions. What I really need is a stepped attenuator with the lowest value being 10 Ohm and highest values being around 640 ohm (approx.) for a minimum gain of 27db (approx.)

So max gain 60db
Min Gain approx. 27db

just need to work out resistor values for each 3db Drop

OK, so according to your latest numbers, here are my values.  They do not take into account your 4R7 resistor on your schematic.

BTW: Be wary of those switches if you go for them.  I have used them.  The body is made of thermo-plastic.  I left the iron on a terminal for 3 seconds too long once and the plastic melted around it and the terminal flopped over.  It was ruined.

[dmills]

I am guessing there is a ground pour on the back of that board?

If so, pin 1 on both XLRs (and also the earthy end of C1,C2 (Why so physically large by the way, smaller is better here) should go directly to chassis NOT the the circuit reference plane, see AES48 for the reasoning.

The reference plane should be connected to the chassis at ONE point, so that shield current does not end up flowing across this plane (You need this one connection so P48 works!).

Some bulk decoupling on the power rails would be a good thing, two of those 47uF caps you like would do nicely, also a pair of zenners to clip the excursion if you have a phantom accident....

Input caps don't really need to be non polar (Use polar ones with the positive towards the XLR)!

4148s are a bit wimpy if you ever have a real phantom power accident, 4001s are more like it (The trouble comes when a signal line with P48 on it (so cap charged) gets shorted to ground), on that subject 10R or so in series with each input cap adds little noise but does serve to limit the surge current.

How are you going to mount the board? a 3.2mm diameter hole about 10mm in from each corner is something you will thank me for later.

Lots of really skimpy traces on that thing, nothing wrong with putting some meat on them, especially the power lines.

Have a look at your placement, especially around those input coupling caps, there are simple things that would make for much cleaner routing.

Some build out resistance (maybe 47 ohms or so) right before pin 2 and 3 at the output is always a good idea, helps with capacitive loads and also improves ESD immunity (May or may not be necessary with that driver but never hurts).

Regards, Dan.

[Jbliss]

I am guessing there is a ground pour on the back of that board?

If so, pin 1 on both XLRs (and also the earthy end of C1,C2 (Why so physically large by the way, smaller is better here) should go directly to chassis NOT the the circuit reference plane, see AES48 for the reasoning.

The reference plane should be connected to the chassis at ONE point, so that shield current does not end up flowing across this plane (You need this one connection so P48 works!).

Some bulk decoupling on the power rails would be a good thing, two of those 47uF caps you like would do nicely, also a pair of zenners to clip the excursion if you have a phantom accident....

Input caps don't really need to be non polar (Use polar ones with the positive towards the XLR)!

4148s are a bit wimpy if you ever have a real phantom power accident, 4001s are more like it (The trouble comes when a signal line with P48 on it (so cap charged) gets shorted to ground), on that subject 10R or so in series with each input cap adds little noise but does serve to limit the surge current.

How are you going to mount the board? a 3.2mm diameter hole about 10mm in from each corner is something you will thank me for later.

Lots of really skimpy traces on that thing, nothing wrong with putting some meat on them, especially the power lines.

Have a look at your placement, especially around those input coupling caps, there are simple things that would make for much cleaner routing.

Some build out resistance (maybe 47 ohms or so) right before pin 2 and 3 at the output is always a good idea, helps with capacitive loads and also improves ESD immunity (May or may not be necessary with that driver but never hurts).

Regards, Dan.

Thank you heaps, this is great feedback !!
Have Implemented changes!
Just to confirm, should the ground point for the 48v switching circuit be, circuit ground or Chassis?
Regards to connecting both Pin 1 together, is the best way to do this by routing a trace across the board ?

Thanks
JBliss

[Jbliss]

So after some calculating if the lowest value of the attenuator (gain switch) is 10ohm I have 60db of gain. Going down in 3db steps gets me down to 27db for 12 positions. What I really need is a stepped attenuator with the lowest value being 10 Ohm and highest values being around 640 ohm (approx.) for a minimum gain of 27db (approx.)

So max gain 60db
Min Gain approx. 27db

just need to work out resistor values for each 3db Drop

Thanks this is great !

[dmills]

Just to confirm, should the ground point for the 48v switching circuit be, circuit ground or Chassis?
Regards to connecting both Pin 1 together, is the best way to do this by routing a trace across the board ?

P48 switching circuit, it does not really matter.
 
Connecting the pin 1's, gods no don't do that!
Pin 1 should be bonded to chassis right at the connector (Ideally it would not enter the box at all), the objective is for the screen current to flow in the box walls (In theory over the OUTSIDE of the box walls) and not set up fields inside the box to couple to your sensitive signals.

See AES48 for lots of detail on this.

Regards, Dan.

[Jbliss]

Hi All,

Just wondering about adding a insert Circuit to the preamp. what would be the easiest way to do this?

Thanks

[Richard Crowley]

Just wondering about adding a insert Circuit to the preamp. what would be the easiest way to do this?

Break the connection between the output of the INA217 and the input of the DRV134. Use a switching TRS jack and connect the INA217 output to the Ring terminal (2), and the DRV134 input to the Tip terminal (3).  That is the basic method used on virtually all audio gear.

You need a "switching" connector so that when nothing is plugged in, the signal flows through from the INA217 to the DRV134. Connect the Ring switch terminal (4) to the Tip Switch terminal (5).



The complications from adding this circuit comes from exposing to the (possibly hostile) outside world that "pristine" node between the INA217 and the DRV134.  It would be prudent to protect that node from DC injection using DC blocking capacitors.  But that rather defeats the nice DC servo circuit (the OPA2137).  So, like many things in the Real World, it is a trade-off decision which is more important to you. 

You could argue that since you are only making a standalone preamp (vs the input circuit to a mixer) you don't really need an insert jack because you can simply route the output of the preamp to wherever you want (including whatever processing you want to use, etc.).

[marcuswilson007]

C4 and C5 should be 1000uF in order to preserve balance at the low frequencies and to preserve the low noise at low frequencies.
At 20Hz 100uF is about 80 Ohms, electrolytics are +50% -20%, there could be an imbalance of up to 50 Ohms on the input legs.  What's the base spreader resistance of a 2N4403, 5-10 Ohms?  The caps would completely bugger the CMRR.
Also at 20Hz the 160 Ohms of the capacitors almost doubles the microphone's 200 Ohm source impedance, increasing the noise right in the popcorn noise region.

[Jbliss]

Just wondering about adding a insert Circuit to the preamp. what would be the easiest way to do this?

Break the connection between the output of the INA217 and the input of the DRV134. Use a switching TRS jack and connect the INA217 output to the Ring terminal (2), and the DRV134 input to the Tip terminal (3).  That is the basic method used on virtually all audio gear.

You need a "switching" connector so that when nothing is plugged in, the signal flows through from the INA217 to the DRV134. Connect the Ring switch terminal (4) to the Tip Switch terminal (5).



The complications from adding this circuit comes from exposing to the (possibly hostile) outside world that "pristine" node between the INA217 and the DRV134.  It would be prudent to protect that node from DC injection using DC blocking capacitors.  But that rather defeats the nice DC servo circuit (the OPA2137).  So, like many things in the Real World, it is a trade-off decision which is more important to you. 

You could argue that since you are only making a standalone preamp (vs the input circuit to a mixer) you don't really need an insert jack because you can simply route the output of the preamp to wherever you want (including whatever processing you want to use, etc.).

Hi Richard,
Excellent that is what I had in mind. Thank you very much for your help.

[Jbliss]

C4 and C5 should be 1000uF in order to preserve balance at the low frequencies and to preserve the low noise at low frequencies.
At 20Hz 100uF is about 80 Ohms, electrolytics are +50% -20%, there could be an imbalance of up to 50 Ohms on the input legs.  What's the base spreader resistance of a 2N4403, 5-10 Ohms?  The caps would completely bugger the CMRR.
Also at 20Hz the 160 Ohms of the capacitors almost doubles the microphone's 200 Ohm source impedance, increasing the noise right in the popcorn noise region.

Hi marcuswilson007 I believe your are referring to the original design I was planning on using I have since changed to a more simple and modern design based around the INA217.

[marcuswilson007]

True, I hadn't caught up with that.
The logic still applies to the INA217 preamp, the unbalance due to the capacitor tolerance and the impedance at low frequencies.
It's even worse as they have 47uF capacitors.
1000uF non polar caps are not common but are available.
Nichicon do a 470uF 50V bipolar (UEP1H471MHD), one or two in parallel will improve the low frequency performance.

[Jbliss]

True, I hadn't caught up with that.
The logic still applies to the INA217 preamp, the unbalance due to the capacitor tolerance and the impedance at low frequencies.
It's even worse as they have 47uF capacitors.
1000uF non polar caps are not common but are available.
Nichicon do a 470uF 50V bipolar (UEP1H471MHD), one or two in parallel will improve the low frequency performance.

As dmills mentioned in a previous post the input caps do not need to be bipolar. Would a good option be to use 1000uF polarised capacitors?

[dmills]

One nasty issue with very large input caps is that they store a lot of energy when P48 is on, and can massively increase the damage potential is you ever plug in a shorted line after powering up the P48 (1000uF with 48V across it is a whole joule give or take), if that ends up being dumped into the input stage you would definitely want low value series resistors and very butch clamp diodes to make sure you don't damage the input transistors.

Personally I usually go for a corner about a decade below the audio band (so 2Hz ish) on the basis that the increased LF noise will be taken out by a subsequent highpass filter further down the chain (The single most useful form of tone control!), and that by 50Hz (which is the place we start properly caring about CMRR the caps will have become a much smaller issue), large caps are good for the noise but raise worrying issues with stored P48, trade this one carefully.

Regards, Dan.

[marcuswilson007]

The problem is that uni-polar electrolytic capacitors need some voltage across them to keep the chemistry going and if the phantom was not turned on for a long time their characteristics can be well below spec and the worst thing is that the sound deteriorates.  Bipolar ones build up a charge from the audio voltage.

[marcuswilson007]

Any energy from these capacitors goes through the protection diodes.

By the way, 1N4001 diodes are not a good choice for op amp protection.  One, they have high capacitance. Two, they are slow. Three, they turn on at the same voltage the ESD diodes in the IC, so they are not protecting the ESD diodes. (The 1N4148's have this problem also.)

The diodes to use are Shottky diodes. 1N6263 is a good choice. They turn on at 200mV as opposed to 700mV for the 1N4001, they are very fast and have a tenth of the capacitance of the 1N4001.

[Richard Crowley]

Protection diode capacitance isn't really a factor in a low-impedance circuit like a modern microphone input.  However the voltage-drop and the speed ARE certainly factors. Schottky diodes would be an excellent choice just for those factors.

NOT recommended to put massive DC blocking capacitors on the inputs.  There are heaps of disadvantages and ZERO advantages.   

One clever design even raised the entire circuit up to a +48VDC offset from ground when phantom power was required, just to avoid those input capacitors altogether.  They are certainly the problem-child in that circuit design (along with the weird gain control requirements).  Might even be worth investigating a switching circuit that bypasses those capacitors when phantom power was switched off.